Instrumental Methods in Food and Beverage Analysis.pdf

Document technical information

Format pdf
Size 29.6 MB
First found Jun 9, 2017

Document content analysis

Language
English
Type
not defined
Concepts
no text concepts found

Persons

Organizations

Places

Transcript

DEVELOPMENTS IN FOOD SCIENCE 39
INSTRUMENTAL
METHODS
IN FOOD AND BEVERAGE
ANALYSIS
This Page Intentionally Left Blank
DEVELOPMENTS IN FOOD SCIENCE 39
INSTRUMENTAL
METHODS
IN FOOD AND BEVERAGE
ANALYSIS
Edited by
DAVID LOUIS BENTE WETZEL
GEORGE CHARALAMBOUS t
1998
ELSEVIER
Amsterdam
- Lausanne
- New York-
Oxford - Shannon
- Singapore
- Tokyo
ELSEVIER SCIENCE B.V.
Sara Burgerhartstraat 25
P.O. Box 211, 1000 AE Amsterdam, The Netherlands
Library of Congress Cataloging in Publication Data
A catalog record from the Library of Congress has been applied for.
ISBN: 0-444-82018-3
91998 Elsevier Science B.V. All rights reserved.
No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form
or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior
written permission of the publisher, Elsevier Science B.V., Copyright & Permissions Department,
P.O. Box 521, 1000 AM Amsterdam, The Netherlands.
Special regulations for readers in the USA - This publication has been registered with the Copyright
Clearance Center Inc. (CCC), 222 Rosewood Drive, Danvers, MA, 01923. Information can be obtained from the CCC about conditions under which photocopies of parts of this publication may be made
in the USA. All other copyright questions, including photocopying outside of the USA, should be referred to the copyright owner, Elsevier Science B.V., unless otherwise specified.
No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a
matter of products liability, negligence or otherwise, or from any use or operation of any methods,
products, instructions or ideas contained in the material herein.
O The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992
(Permanence of Paper)
Printed in The Netherlands
DEVELOPMENTS IN FOOD SCIENCE
Volume 1
Volume 2
Volume 3
Volume 4
Volume 5
Volume 6
Volume 7
Volume 8
Volume 9
Volume 10
Volume 11
Volume 12
Volume 13
Volume 14
Volume 15
Volume 16
Volume 17
Volume 18
Volume 19
Volume 20
J.G. Heathcote and J.R. Hibbert
Aflatoxins: Chemical and Biological Aspects
H. Chiba, M. Fujimaki, K. Iwai, H. Mitsuda and Y. Morita (Editors)
Proceedings of the Fifth International Congress of Food Science and Technology
I.D. Morton and A.J. MacLeod (Editors)
Food Flavours
Part A. Introduction
Part B. The Flavour of Beverages
Part C. The Flavour of Fruits
Y. Ueno (Editor)
Trichothecenes: Chemical, Biological and Toxicological Aspects
J. Holas and J. Kratochvil (Editors)
Progress in Cereal Chemistry and Technology. Proceedings of the VIIth World Cereal
and Bread Congress, Prague, 28 june-2 July 1982
I. Kiss
Testing Methods in Food Microbiology
H. Kurata and Y. Ueno (Editors)
Toxigenic Fungi: Their Toxins and Health Hazard. Proceedings of the Mycotoxin
Symposium, Tokyo, 30 August-3 September 1983
V. Betina (Editor)
Mycotoxins: Production, Isolation, Separation and Purification
J. Hollo (Editor)
Food Industries and the Environment. Proceedings of the International Symposium,
Budapest, Hungary, 9-11 September 1982
J. Adda (Editor)
Progress in Flavour Research 1984. Proceedings of the 4th Weurman Flavour
Research Symposium, Dourdan, France, 9-11 May 1984
J. Hollo (Editor)
Fat Science 1983. Proceedings of the 16th International Society for Fat Research
Congress, Budapest, Hungary, 4-7 October 1983
G. Charalambous (Editor)
The Shelf Life of Foods and Beverages. Proceedings of the 4th International Flavor
Conference, Rhodes, Greece, 23-26 July 1985
M. Fujimaki, M. Namiki and H. Kato (Editors)
Amino-Carbonyl Reactions in Food and Biological Systems. Proceedings of the 3rd
International Symposium on the Maillard Reaction, Susuno, Shizuoka, Japan,l-5
July 1985
J. Skoda and H. Skodova
Molecular Genetics. An Outline for Food Chemists and Biotechnologists.
D.E. Kramer and J. Liston (Editors)
Seafood Quality Determination. Proceedings of the International Symposium,
Anchorage, Alaska, U.S.A., 10-14 November 1986
R.C. Baker. P. Wong Hahn and K.R. Robbins
Fundamentals of New Food Product Development
G. Charalambous (Editor)
Frontiers of Flavor. Proceedings of the 5th International Flavor Conference, Porto Karras,
Chalkidiki, Greece, 1-3 July 1987
B.M. Lawrence, B.D. Mookherjee and B.J. Willis (Editors)
Flavors and Fragrances: A World Perspective. Proceedings of the 10th International
Congress of Essential Oils, Fragrances and Flavors, Washington, DC, U.S.A., 16-20
November 1986
G. Charalambous and G. Doxastakis (Editors)
Food Emulsifiers: Chemistry, Technology, Functional Properties and Applictations
B.W. Berry and K.F. Leddy
Meat Freezing. A Source Book
J. Davidek, J. Veli~ek and J. Pokorny (Editors)
Chemical Changes during Food Processing
V. Kyzlink
Volume 22
Principles of Food Preservation
H. Niewiadomski
Volume 23
Rapeseed. Chemistry and Technology
G. Charalambous (Editor)
Volume 24
Flavors and Off-flavors '89. Proceedings of the 6th International Flavor Conference,
Rehymnon, Crete, Greece, 5-7 July 1989
R. Rouseff (Editor)
Volume 25
Bitterness in Foods and Beverages
J. Chelkowski (Editor)
Volume 26
Cereal Grain. Mycotoxins, Fungi and Quality in Drying and Storage
M. Verzele and D. De Keukeleire
Volume 27
Chemistry and Analysis of Hop and Beer Bitter Acids
G. Charalambous (Editor)
Volume 28
Off-Flavors in Foods and Beverages
G. Charalambous (Editor)
Volume 29
Food Science and Human Nutrition
H.H. Huss, M. Jakobsen and J. Liston (Editors)
Volume 30
Quality Assurance in the Fish Industry. Proceedings of an International Conference,
Copenhagen, Denmark, 26-30 August 1991
R.A. Samson, A.D. Hocking, J.I.Pitt and A.D. King (Editors)
Volume 31
Modern Methods in Food Mycology
G. Charalambous (Editor)
Volume 32
Food Flavors, Ingredients and Composition. Proceedings of the 7th International Flavo
Conference, Pythagorion, Samos, Greece, 24-26 June 1992
G. Charalambous (Editor)
Volume 33
Shelf Life Studies of Foods and Beverages. Chemical, Biological, Physical and
Nutritional Aspects
G. Charalambous (Editor)
Volume 34
Spices, Herbs and Edible Fungi
H. Maarse and D.G. van der Heij (Editors)
Volume 35
Trends in Flavour Research. Proceedings of the 7th Weurman Flavour Research
Symposium, Noordwijkerhout, The Netherlands, 15-18 June 1993
J.J. Bimbenet, E. Dumoulin and G. Trystram (Editors)
Volume 36
Automatic Control of Food and Biological Processes. Proceedings of the ACoFoP III
Symposium, Paris, France, 25-26 October 1994
Volume 37A+B G. Charalambous (Editor)
Food Flavors: Generation, Analysis and Process Influence
Proceedings of the 8th International Flavor Conference, Cos, Greece, 6-8 July 1994
Volume 38
J.B. Luten, T. Borresen and J. Oehlenschl&ger (Editors)
Seafood from Producer to Consumer, Integrated Approach to Quality
Proceedings of the International Seafood Conference on the occasion of the 25th
anniversary of the WEFTA, held in Noordwijkerhout, The Netherlands,
13-16 November 1995
Volume 39
D. Wetzel and G. Charalambous t (Editors)
Instrumental Methods in Food and Beverage Analysis
Volume 21
vii
Dedication
This volume devoted to Instrumental Analysis for Food and Beverage Analysis is
dedicated to the late Dr. George Charalambous whose contributions to agriculture and food
chemistry were not limited to his own work in the flavor area. His vision, always forward, was
like the high beam of automobile headlights exploring the distant future down the road and
fanning out slightly to the center lane and the shoulder of the road for a glimpse of new
unexpected things that may lie ahead and become important in the future.
His curiosity made him an attentive listener and formulator of pertinent questions as he
mentally explored unfamiliar territory that those who shared his joumey in the direction of
progress may benefit from. He was interested in new and potentially useful approaches and in
new instrumental techniques. George recognized the value of one-on-one or small group
discussion as a provocative form of scientific communication. He organized scientific
conferences where travelers on the road to success in food and flavor chemistry could pause for
refreshment or charge their intellectual batteries before proceeding on their journey. He
understood the human and cultural side of those in the scientific community and appropriately
staged the conferences he organized with that in mind. His charm persuasion and persistence
brought both the published proceedings (a living text) and the conferences to fruition.
At the time that George was putting together his last two volume collection of papers
from the Intemational Conference in Cos, Greece, I received a call from him at my home at some
odd hour and he said in reference to one chapter, "It took me two hours to read and study the
thing and I'm still excited." Our dear colleague was blessed to the end of his day with a keen
mind and the curiosity that made his approach to new "tools of science" with the enthusiasm of a
boy opening a package containing a new toy. It was he who initially planned this volume and
identified the experts in each area to participate.
This Page Intentionally Left Blank
ix
Preface
Advances in instrumentation and applied instrumental analysis methods have allowed
scientists concemed with food and beverage quality, labeling, compliance, and safety to meet
ever increasing analytical demands. Texts dealing with instrumental analysis alone are usually
organized by the techniques without regard to applicaitons. The biannual review issue of
Analytical Chemistry under the topic of Food Analysis is organized by the analyte such as N and
protein, carbohydrate, inorganics, enzymes, flavor and odor, color, lipids, and vitamins. That
review introduces at least seven instrumental techniques under the first topic and in successive
topics some of the techniques are revisited and new ones are introduced as needed. Finally under
flavor and odor the subdivisions are not along the lines of the analyte but the matrix (e.g. wine,
meat, dairy, fruit) in which the analyte is being determined.
Approximately 200 references were cited that appeared in a two year period. Molecular
spectroscopy, chromatographic or other sophisticated separations as well as hyphenated
techniques such as GC-Mass spectrometry usually predominate. The reader is referred to a list of
72 entries on page III entitled "Instrumentation and Instrumental Techniques" that appear in this
book. In the text of Instrumental Methods in Food and Beverage Analysis, a few of these appear
under a chapter named for the technique. Most of the analytical techniques used for
determination, separations and sample work up prior to determination are treated in the context
of an analytical method for a specific analyte in a particular food or beverage matrix. The
instruments and techniques are applied to an analyte and sample matrix with which the author
has a professional familiarity, dedication, and authority.
In food analysis in particular it is usually the food matrix that in fact presents the research
analytical chemist involved with method development the greatest challenge. Incoming
analytical graduate students are made aware of one infamous case in particular. A heat stable
form of a dough conditioner, shelf-life extender, and nutrient had been synthesized and its value
was dependent upon proving its survival of the baking condition. It required only a day to fred
excellent HPLC conditions to separate and determine the particular surface active agent from a
concentrate, but it took the rest of the year to establish an actual method for the analyte in the
food matrix. Recovery was the key and therein was the analytical dilemma.
David L. B. Wetzel, Professor
Kansas State University
Instrumentation and Instrumental Techniques
ultraviolet spectrometry (UV)
colorimetry, visible spectrometry
Hunter reflectometer
chemiluminescence
laser induced fluorescence
pyro-chemiluminescence
infrared spectrometry
near-infrared (NIR) spectrometry
Fourier transform spectrometry
FT-IR
FT-NIR
microspectrometry
Raman spectroscopy
acousto-optic tunable filter spectrometer (AOTF)
grating monochromator
diode array
nuclear magnetic resonance spectrometry (NMR)
~3Ccross polarization magic angle spinning NMR
diffuse reflectance
circular dichroism
low angle laser light scattering photometer (LALLS)
enzymatic-spectrometric method
mass spectrometry (MS)
quadrupole mass spectrometry, 3-D
electron impact ionization
chemical ionization
negative chemical ionization
ion trapping
high performance liquid chromatography (HPLC)
reversed phase chromatography
high performance size exclusion
chromatography (HPSEC)
supercritical fluid chromatography (SFC)
supercritical fluid extraction (SFE)
field flow fractionation (FFF)
gas chromatography (GC)
fused silica capillary-GC
solid phase extraction
purge-and-trap technique
flame ionization detection
selected ion monitoring (SIM)
elemental nitrogen analyzer
chemiluminescent nitrogen detection
electrochemical detection
differential refractive index
InGaAs detector
light emitting diode (LED)
micellar electrokinetic capillary
chromatography
centrifugal partition chromatography
cyrofocusing
coelectroosmotic electrophoresis
counter electroosmotic electrophoresis
isatachophoresis
digital image analysis (DIA)
enzyme-linked immunosorbent assays
(ELISA)
Brabender amylograph
Brabender Viscograms
viscometers
viscoelasticity, linear and non-linear
differential scanning calorimetry
Coulter counter
scanning electom microscopy (SEM)
X-ray diffraction
wide angle X-ray diffraction
mass fragrnentographic SIM
small deformation oscillatory measurements
chemometrics
multivariate statistics
pattern recognition
discriminant component analysis
Euclidian distance
Mahalanobis distance
xi
LIST OF CONTRIBUTORS
Number in parenthesis indicate wherecontributionsbegin
O.R. Abou-Samaha (49) Food Science and Technology Department, Faculty of Agriculture,
Alexandria University, Alexandria, Egypt
M.H. Bekheet (49) Food Science and Technology Department, Faculty of Agriculture,
Alexandria University, Alexandria, Egypt
P.W. Bosland (347) Department of Agronomy and Horticulture, New Mexico state University,
Box 30003, Las Cruces, New Mexico, U.S.A.
J. Brady (467) Hercules Inc., Research Center, 500 Hercules Road, Wilmington, DE 19808,
U.S.A.
I. Chronakis (99) Physical Chemistry 1, Center for Chemistry and Chemical Engineering, Lund
University, P.O. Box 124, S-22 100 Lund, Sweden
F. Colon (489) Chimie des Ar6mes-Oenologie, Associ6 au CNRS, URA 1411, Facult6 des
Sciences et Techniques de St-J6r6me, Case 561, Avenue Escadrille Normandie-Ni6men
F13397 Marseille C6dex 20, France
J. Crnko (379) Antek Instruments Inc., 300 Bammel Westfield Road, Houston, TX 77090,
U.S.A.
T. Drumm Boylston (225) Southern Regional Research Cemer, U.S. Department of Agriculture,
Agriculture, Agricultural Research Service, P.O. Box 19687, 1100 Robert E. Lee
Boulevard, New Orleans, LA 70179, U.S.A. - Present address: Department of Food
Science and Human Nutrition, Washington State University, Pullman, WA 991640-6376
C. Femandes (571) Department of Food Science and Technology, Virginia Tech, Blacksburg,
VA 24061-0418, U.S.A.
G. Flick, Jr. (571) Department of Food Science and Technology, Virginia Tech, Blacksburg, VA
24061-0418, U.S.A.
E. Fujinari (377-475) 25411 Avery Hill Lane, Spring, TX 77373-6089, U.S.A.
F. Goycoolea (99) Research Center for Food and Development (C.I.A.D., A.C.), P.O. Box 1735,
83000 Hermosillo, Sonora, Mexico
A. Hoffmann (303) Gerstel GmbH, AktienstraBe 232-234, 45473 MiJlheim an der Ruhr,
Germany
xii
S. Kasapis (1) Department of Food Research and Technology, Cranfield University, Silsoe
College, Silsoe, Bedford MK45 4DT, United Kingdom
B. Kibler (379) Antek Instruments Inc., 300 Bammel Westfield Road, Houston, TX 77090,
U.S.A.
J.M. King (195) Department of Food Science and Technology, The Ohio State University, 122
Vivian Hall, 2121 Fyffe Road, Columbus, OH 43210, U.S.A.
F. Kolpak (467) Hercules Inc., Research Center, 500 Hercules Road, Wilmington, DE 19808,
U.S.A.
C. Lageot (245) Chimie des Ar6mes-Oenologie, Associ6 au CNRS, URA 1411, Facult6 des
Sciences et Techniques de St-J6r6me, Case 561, F13397 Marseille C6dex 20, France
K. MacNamara (303) Irish Distiller Limited, Bow Street Distiller, Smithfield, Dublin 7, Ireland
D.B. Min (195) Department of Food Science and Technology, The Ohio State University, 122
Vivian Hall, 2121 Fyffe Road, Columbus, OH 43210, U.S.A.
Y.G. Moharram (49) Food Science and Technology Department, Faculty of Agriculture,
Alexandria University, Alexandria, Egypt
C. Pfirkfinyi (245) Department of Chemistry, Florida Atlantic University, PO Box 3091, Boca
Raton FL 33431-0991, U.S.A.
R. Richardson (1) Department of Food Research and Technology, Cranfield University, Silsoe
College, Silsoe, Bedford MK45 4DT, United Kingdom
H. Shi (475) Department of Chemistry, Virginia Polytechnic Institute and State University,
Blacksburg, VA 24060, U.S.A.
B. Strode HI (475) Department of Chemistry, Virginia Polytechnic Institute and State University,
Blacksburg, VA 24060, U.S.A.
L. Taylor (475) Department of Chemistry, Virginia Polytechnic Institute and State University,
Blacksburg, VA 24060, U.S.A.
J. Thompson (475) Department of Chemistry, Virginia Polytechnic Institute and State University.
Blacksburg, VA 24060, U.S.A.
G. Vemin (245 & 489) Chimie des Ar6mes-Oenologie, Associ6 au CNRS, URA 1411, Facult6
des Sciences et Techniques de St-J6r6me, Case 561, F13397 Marseille C6dex 20, France
xiii
B. Vinyard (225) Southern Regional Research Center, U.S. Department of Agriculture,
Agriculture, Agricultural Research Service, P.O. Box 19687, 1100 Robert E. Lee
Boulevard, New Orleans, LA 70179, U.S.A.
M.M. Wall, (347) Department of Agronomy and Horticulture, New Mexico State University,
Box 30003, Las Cruces, New Mexico, U.S.A.
D. Wetzel (141) Department of Grain Science and Industry, Kansas State University, 201
Shellenberger Hall, Manhattan, KS 66506, U.S.A.
R. Young (425) Shell Canada Products, Ltd., Scotford Ref'mery, Fort Saskatchewan, Alberta,
Canada
xiv
ERRATUM
Capillary Electrophoresis for Food Analysis (Custy F. Femandes and George J. Flick, Jr.)
pp. 575-612
On pages 596 and 601, references in the text to figures 8 and 9 should be disregarded, as
these figures were not included by the authors.
XV
CONTENTS
Dedication ..........................................................................................................................................
vii
Preface ...............................................................................................................................................
ix
Instrumentation and Instrumental Techniques ...................................................................................
x
List of Contributors ...........................................................................................................................
xi
Rheological methods in the Characterisation of Food Biopolymers
ROBERT K. RICHARDSON and STEFAN KASAPIS .......................................................
1
Destructive and Non-Destructive Anal)r
Methods in Starch Analysis
Y.G. MOHARRAM, O.R. ABOU-SAMAHA, and M.H. BEKHEET ................................. 49
Specific Methods for the Analysis of Identity and Purity of Functional Food Polysaccharides
FRANCISCO M. GOYCOOLEA and IOANNIS S. CHRONAKIS .................................... 99
Analytical Near-Infrared Spectroscopy
DAVID L. B. WETZEL .......................................................................................................
141
Analysis of Fatty Acids
J.M. KING and D.B. MIN ...................................................................................................
195
Isolation of Volatile Flavor Compounds From Peanut Butter Using Purge-and-Trap Techniques
TERRI DRUMM BOYLSTON and BRYAN T. VINYARD .............................................. 225
GC-MS(EI, PCI, NCI, SIM, ITMS) Data Bank Analysis of Flavors and Fragrances. Kovats indices
G. VERNIN, C. LAGEOT, and C. P/kRKANYI ................................................................. 245
Gas Chromatographic Technology in Analysis of Distilled Spirits
KEVIN MacNAMARA and ANDREAS HOFFMANN .................................................... 303
xvi
Analytical Methods for Color and Pungency of Chiles (capsicums)
M.M. WALL and P.W. BOSLAND ....................................................................................
347
Chemiluminescent Nitrogen Detectors (CLND) for GC, SimDis, SFC, HPLC and SEC Applications
Dedication/Preface ............................................................................................................... 375
Part 1 Elemental Total Nitrogen Analyses by Pyro-Chemiluminescent Nitrogen Detection
JOHN CRNKO, BOB C. KIBLER, and EUGENE M. FUJINARI ..................................... 379
Part 2 Gas Chromatography- Chemiluminescent Nitrogen Detection: GC-CLND
EUGENE M. FUJINARI .....................................................................................................
385
Part 3 Simulated Distillation-Chemiluminescent Nitrogen Detection: SimDis-CLND
RICHARD J. YOUNG and EUGENE M. FUJINARI ......................................................... 425
Part 4 High Performance Liquid Chromatography- Chemiluminescent Nitrogen Detection: HPLCCLND
EUGENE M. FUJINARI .....................................................................................................
431
Part 5 The Determination of Compositional and Molecular Weight Distributions of Cationic
Polymers Using Chemiluminescent Nitrogen Detection (CLND) in Aqueous Size Exclusion
Chromatography
FRANK J. KOLPAK, JAMES E. BRADY, and EUGENE M. FUJINARI ........................ 467.
Part 6 Chemiluminescent Nitrogen Detection in Capillary SFC
HENG SHI, J. THOMPSON, B. STRODE l]], LARRY T. TAYLOR, and
EUGENE M. FUJINARI .....................................................................................................
475
The SPECMA 2000 Data Bank Applied to Flavor and Frangrance Materials
F. COLON and G. VERNIN ................................................................................................
489
Capillary Electrophoresis for Food Analysis
CUSTY F. FERNANDES and GEORGE J. FLICK, Jr ...................................................... 575
Index ................................................................................................................................................
613
D. Wetzel and G. Charalambous (Editors)
Instrumental Methods in Food and Beverage Analysis
9 1998 Elsevier Science B.V. All rights reserved
RHEOLOGICAL METHODS
FOOD BIOPOLYMERS
IN
THE
CHARACTERISATION
OF
Robert K. Richardson and Stefan Kasapis
Department of Food Research and Technology, Cranfield University,
Silsoe College, Silsoe, Bedford MK45 4DT, United Kingdom
INTRODUCTION
All natural food systems contain biopolymers, often in the form of complex
multicomponent mixtures, which play a fundamental role in their structure and
function. Similarly, in processed foodstuffs, biopolymers are used widely to create a
structured body, and in the case of high solid products (confectioneries) to maintain
the rubbery (jelly babies, gummy bears, etc.) or glassy (boiled down sweets) texture
required by the consumer. Furthermore, increasing consumer awareness of the health
implications of a high calorie, low fibre diet and in particular of excess consumption
of saturated fat has generated a large market for low fat or even zero fat substitutes for
traditional yellow fat spreads (butter and margarine) and cheeses. Developments in the
area of fat replacers centre on the use of carbohydrates, particularly maltodextrins
(hydrolysed starch) and proteins (gelatin, milk proteins, etc.) to bind water within the
product and to generate acceptable fat-like rheology (spreading behaviour) and
' mouthfeel'
Particularly important examples of polysaccharide and protein functionality, therefore,
are the capacity to alter the flow characteristics of fluids and to interact in the hydrated
state with other dispersed or dissolved molecular species which they may bind,
chelate, complex, emulsify, encapsulate, flocculate, stabilise or suspend. As a result
they perform, either by themselves or in association with hormones, lipids or other
molecules, vital biological functions by providing structural support and energy
reserve, and by mediating various other processes such as adhesion, cellular
recognition and growth. This broad range of functional properties, frequently unique
in their nature, originate from the primary structure of the individual residues and the
configuration of the linkages between them which lead to the development of
electrostatic, hydrophilic and dipole-dipole interactions, hydrogen bonding, and
covalent associations (e.g. disulphide linkages in proteins) [1,2].
In the generation of a diversity of manufactured products, industrial processing often
exploits the dramatic alteration of the physical properties of biopolymers when they
undergo transformation from the fluctuating chain geometry, typical of the solution
state, to the fixed, ordered conformations typically seen in gelling systems. At present,
the only general route to detailed characterisation of polysaccharide ordered structures
at atomic resolution is X-ray diffraction from ordered fibres. The single crystal X-ray
diffraction patterns of globular proteins contain sufficient information for direct,
unambiguous determination of macromolecular organisation and packing [3]. Things
are less clear in the case of polysaccharides, but in general it is relatively easy to
extract basic information such as helix pitch and overall dimensions of the repeating
'unit cell'. These, together with the known, and essentially invariant, ring geometry of
the constituent sugars, and the C-O-C bond angle of the inter-residue linkage,
normally restrict the stereochemically-feasible arrangements to a manageable number
of candidate structures, whose anticipated diffraction patterns can be calculated and
compared with observed intensifies [4].
Although fibre diffraction provides valuable (and indeed essential) information for
detailed interpretation of the physical properties of polysaccharides, the existence of
an ordered structure in the solid state does not necessarily imply that the same
structure persists under hydrated conditions. The advent of microcomputing in recent
years has allowed the rapid development of rheological techniques, with computer
driven rheometers becoming commonplace in the laboratory. These are now
established as the most productive line of attack for the development of functionstructure-texture relationships in food products. Rheological measurements are
performed therefore:
(i) to detect consequences of possible conformational order in solution
(ii) to characterise changes in chain geometry and packing under
hydrated conditions
(iii) to monitor any order-disorder transition behaviour [5].
Findings from such studies in vitro also provide significant insights into the probable
in vivo behaviour of polysaccharides and proteins in complex natural systems
(e.g. living plant cells). The aim of this Chapter is to describe the rheological
techniques used in food biopolymer research for investigation of dilute and
concentrated solutions, aqueous gels and high-solids systems, biphasic systems, and to
indicate the principles on which they are based.
RHEOLOGY
In this section, it is intended to give a basic introduction to rheological techniques
sufficient for the practical examination of biopolymer based sample systems,
interpretation of results and to provide enough background information for the more
detailed accounts of particular system types described later.
Definitions and Terms
The word rheology is derived from the Greek rheo meaning flow. Several defimtions
have been advanced over the years but perhaps the most simple description is that
rheology is the study of the relationship between stress and strain within a sample
material as a function of time, temperature, etc. Stress is the force per unit area acting
on a sample and therefore has the units of pressure, usually Nm ~ or Pa., strain being
the resulting fractional deformation and therefore a dimensionless ratio.
Two early laws of physics divided materials into two distinct types according to their
observed mechanical behaviour; solids and liquids. In the former case, it was observed
that stress (r is directly proportional to strain 0') [Hooke's law] and for liquids, stress
is proportional to rate of strain (dy/dt) [Newton's law]. For a perfect Hookean solid,
all the energy necessary for the deformation process is stored as recoverable potential
energy whereas, in the case of a Newtonian liquid, all the energy contributing to flow
is dissipated as heat. The proportionality constants for the two cases are termed
modulus and viscosity which for simple shear measurements are given the symbols
G = or/), and 11 = t~/(d),/dt). These classes of response represent 'extremes' of
behaviour, most materials combining both elastic and viscous properties, giving rise to
the term 'viscoelasticity'.
The simple, classical experiments used to determine viscoelastic properties involve
measuring the time-dependence of either the stress due to a given strain, or the strain
generated by a known stress. The former experiment, known as stress-relaxation,
yields a time-dependent modulus, e.g. G(t) = or(t)/7, and the latter, creep experiment,
gives the time-dependent compliance J(t) = ~,(t)/~.
Viscoelastic Behaviour
Viscoelasticity is often modelled using combinations of springs and dashpots as
mechanical analogues of the independent elastic and viscous contributions. The two
simplest forms are the Maxwell (series) element and the Voigt (parallel) element
shown in Figures l a and l b respectively. The former represents a viscoelastic liquid
and the latter, a viscoelastic solid and the response of these to a creep experiment is
illustrated in Figures 2a and 2b.
In Figure l a, where the stress in both components is equal, the spring is
instantaneously stretched and further increase in strain proceeds at a rate inversely
proportional to the viscosity component. On removal of the imposed stress, the spring
instantaneously relaxes and a residual strain, equal to the product of strain-rate and
experimental time, remains. In the case of the Voigt element, the spring and dashpot
experience equal strain at any given time and hence the initial stretching of the elastic
component is 'damped' by the dashpot, the strain at long times approaching a constant
value proportional to the compliance of the spring. Here, on removal of the stress, the
relaxation of the spring is again damped but eventually, the displacement returns to the
original zero-strain condition.
In general, the molecular processes in real viscoelastic materials are too complex to be
described directly by such simple models. As the material structure becomes more
complex, so the number of relaxation processes at different time-scales contributing to
the overall mechanical properties, increases. The relaxation and creep measurements
described above are, however, useful in determining the underlying long-range
structure, i.e. if liquid-like or solid-like behaviour is displayed at long times.
Combinations of the model elements can also assist in making realistic predictions of
overall response provided they truly represent discrete mechanical components such as
'sandwich' structures or machine/sample combinations. They are usually used,
however, to generate approximate flow curves to mimic complex systems rather than
for the detailed investigation of microstructure.
D
~
t
s
(a)
Figure 1.
I
1 -D
(b)
Simple models of viscoelastic behaviour: (a) Maxwell and (b) Voigt
elements.
Linear and Non-Linear Viscoelasticity
Classically, the relationship between stress (or) and strain (,/) is determined in the
linear strain region, that is, at strains sufficiently small that structure is not disrupted
by the local deformation. A system is said to be linear if it can be described by a linear
differential equation with constant coefficients. This means an equation relating stress
and strain with respect to time of the form, (after Arridge [6]),
Aocr + Al(dcr/dt ) + A2(d2cr/dt2) + . . . . . + Am(dmcr/dtTM)
= Bo~/ + Bl(dy/dt ) + B2(d~/dt2 ) + . . . . . + Bn(dr~/dt n)
(1)
For the simple systems already mentioned, this is greatly reduced to give:
Aocr = BoY
Aocr = Bl(d),/dt )
Aocr + Al(dcr/dt ) = Bl(dy/dt )
Aocr = Bo~, + Bl(dy/dt )
(Hooke's law; G = Bo/Ao)
(Newton's law; 11 = B1/Ao)
(Maxwell element)
(Voigt element)
(2)
(3)
(4)
(5)
euU
CP~
a
w
Time
I
.m
k
,4~
X
Y
Time
Figure 2. Resultant strain due to stress imposed at X and removed at Y, for
(a) Maxwell model (b) Voigt model.
If, for instance, strains or rates of strain become excessive for a given material,
equation (1) is no longer obeyed and the behaviour is said to be non-linear. This is
exemplified by, for instance, flow in metals above the elastic limit or non-Newtonian
flow (shear thinning) of biopolymer solutions at high shear-rates. In the modem food
industry, the non-linear (breakdown and recovery) properties of a component are often
its most important feature. Numerous applications exist where a food system must
remain thick, or even suspend particles, under low or zero shear conditions but be
capable of flow when pumped during processing or subjected to high shear stresses
and strains by the consumer. In particular, the modem trend to replace conventional
dairy products with low-fat biopolymer-based analogues requires careful matching of
both linear and non-linear properties together with other considerations such as
temperature dependency. It is not intended to delve more deeply into this subject here,
but it is essential that linear measurements, where meaningful molecular interpretation
can be employed, should be distinguishable from the non-linear regime, where, with
the exception of the simplest structural types, 'phenomenological' modelling of
particular samples is still the more accepted, common practice.
Types of Deformation
The two forms of deformation usually applied to hydrated food biopolymer systems
are uniaxial compression (or tension) and shear. If the sample is sufficiently rigid
(self-supporting), simple compression testing can be used. Here, a sample of welldefined size and shape is confined between parallel surfaces that can be made to
approach one another under controlled conditions, the relative displacement and
resulting stress in the sample being monitored. For infinitesimal strains in compression
or tension (Figure 3a), during, for instance, a stress-relaxation experiment, the timedependent stress t~(t) = F(t)/A and the strain (~,) [fractional deformation] is A1/L giving
a time-dependent elongational (Young' s) modulus E(t) = [F(t)L]/[AAI].
Figure 3.
Types of deformation commonly used to determine moduli of food
materials. Uniaxial tension/compression (a), and shear (b).
Several problems are inherent in such measurements. Under normal conditions,
samples need to be relatively 'stiff' and at higher deformations, true strain and crosssectional area are difficult to determine. The level of lubrication between instrtunent
and sample surfaces also becomes critically important under such conditions.
For these reasons, measurements are usually confined to the small-deformation
regime or, commonly in food-development applications, to produce characteristic
force/deformation (or approximate stress/strain) curves for comparison with target
products or optimisation of process conditions. Since large volumes of sample are
usually available, by averaging several replicates representative 'breakdown and flow'
properties can be determined under the compressive strain conditions which may be
applicable to end-use.
Of more general use in the investigation of food biopolymer systems is shear
deformation, illustrated in elemental form in Figure 3b. Here, the shear stress is again
or(t) = F(t)/A but the shear strain ~, is equal to tan(0) (~, 0 for small 0). Because the
sample is confined between two surfaces a fixed distance apart with strain usually
being imposed by their relative angular displacement about a common axis, systems
from simple solutions to strong gels can be analysed, and, up to moderate strain and
rate-of-strain, sample dimensions remain well defined. An added advantage of such a
geometry is that the exposed sample surface is small, allowing effective sealing
against water-loss during long-time measurement or experiments involving prolonged
heating using a suitable medium such as silicone fluid or liquid paraffin.
For the special case of a perfectly elastic body, or a viscoelastic solid at long
experimental times whereby all time-dependent processes have relaxed away and an
equilibrium modulus is measured (i.e. for a real system, the modulus of the underlying
gel network is measured, or for the Voigt model, equilibrium extension of the spring is
reached), a simple relationship between these moduli exists. The equilibrium Young's
modulus, E e and equilibrium shear modulus G e are related by the equation
E e = 2Ge(1+~t), where la is Poisson's ratio which is the ratio of lateral contraction to
axial strain in the extensional experiment. The maximum value possible for ~t is 0.5
and for some soft-solid systems, this is almost achieved giving E e ~ 3G e
Small-Deformation Oscillatory Measurements
In the transient experiments so far described, the measured parameter is a result of
both the elastic and viscous sample properties. Although all necessary information
about these is contained within, for example, a creep curve, a method whereby the
elastic and viscous components can be determined directly at a given experimental
time (or corresponding frequency) is obviously desirable. Also, it is often necessary to
obtain information about the ~iscoelastic properties of a sample system as it undergoes
structural change, e.g. gelation. Here, a method is required such that measurements
can be made in periods shorter than the process to be monitored without affecting the
natural mechanisms involved.
If a sample is subjected m a time-dependent strain-wave of the form
)' = "/o sin o)t
(6)
where ,/is the time-dependent strain, "/o is the maximum strain amplitude and o) is the
angular frequency (Figure 4a), then the resulting shear-stress wave will also be a sinewave but differing in amplitude and phase, i.e.
o = o 0 sin(cot + ~)
(7)
where cr is the time-dependent stress, cro is the maximum stress amplitude and 6 is the
phase angle difference between the two waves.
Considering a purely elastic sample, the instantaneous stress will be proportional m
the corresponding strain and the strain and stress waves will therefore be in phase with
each other, as shown in Figure 4b. For a viscous system, however, the stress will be
proportional to the strain-rate at any given time and thus the maximum stress will
occur when the slope of the strain wave is a maximum, i.e. at the zero cross-over
points. This results in a phase-shift of n/2 radians (Figure 4c). A viscoelastic sample
will therefore produce a stress-wave whose amplitude is proportional to the strain
amplitude, but having contributions from both the in-phase and out-of-phase
components (Figure 4d). Assigning separate symbols m the elastic and viscous moduli,
the stress wave function may be written in the form:
(8)
cr = "/o(G'sino3t + G"coso3t)
where G' is the in-phase, storage modulus and G" is the out-of-phase, loss modulus.
Re-writing equation (7) as
cr = Cro(COS~Ssintot+ sin5coso)t)
(9)
by comparison with (8), it is clear that
and
G' = ((rdvo)COS5
G" = ((~o/Yo)sin~5
(1 o)
(11 )
From these two basic parameters, others may usefully be derived.
G"/G' = tan5
(12)
Tan& the loss tangent, is useful in detecting structural changes particularly during
broad gelation or melting processes where large variation in the absolute values of
both G' and G" may conceal subtle effects indicative, for instance, of more than one
underlying process. Also, from (10) and (11)
(G') 2 + (G") 2 = (t~o/Yo)2(sin25 + cos2~i)
therefore
cr0/y0 = [(G') 2 + (G")2] v' =
[G* I
(13)
(14)
I
l
l
l
i
l
l
l
l
t
a
Drive strain
l
l
i
l
i
b
I
[
I
I
Elastic
response
I
I
v
\
', /
I
I
',\',
i
:
I
I
I
O
~:/2
\~
~
: \
::
: I
i
II
!
7:
I
i
I
1
I
I
3~:/2
2~:
~:
/
\
\
Viscous
response
I
\
d
Viscoelastic
response
/
,
pr-
O)t
Figure 4. Resultant stresses due to a smusoidal drive strain (a) for
elastic (b), viscous (c), and viscoelastic (d) materials.
10
]G'I, the complex modulus, is therefore the simple ratio of stress amplitude to strain
amplitude without regard for the type of mechanism involved, storage or loss.
As for the dynamic moduli, dynamic viscosity components can be defined. The real
dynamic viscosity rl' is the ratio of stress in-phase with the strain-rate to that strainrate and the imaginary dynamic viscosity, rl", is the ratio of stress n/2 out-of-phase
with the strain-rate divided by the strain-rate. The rate-of-strain is the first derivative
of the strain function with respect to time, thus
d~,/dt = ~,0o~coso~t
(15)
i.e., a sine-wave of amplitude ,/0co phase-shifted by n/2 with respect to the strain wave.
By the same procedure used to obtain equations (1 O) and (11 )
(16)
(17)
rl '= (Cro/~'oO~)sin8= G"&o
11"= (Cro/~,0~o)cos5 = G'/o~
Although rl' is sometimes compared with the steady-shear viscosity rl, it is probably at
least equally valid to use a viscosity obtained from the total resultant stress cr0.
i.e.
Cro/~,0o~= I G* l/co = II1" I
(18)
where [r I* [ is termed the complex dynamic viscosity.
Using oscillatory methods, it is therefore possible to extract the elastic and viscous
contributions to the mechanical behaviour of a sample system as the structure changes
with experimental time, temperature, strain, etc. By 'sweeping' the imposed
frequency, mechanical spectra of, for instance, G'(o~) and G"(co) are obtained from
which the time (frequency) dependent nature of the sample can be analysed to identify
the most important structural features.
Rheologicai
Techniques.
Characterisation
of
Biopolymer
Systems
Using
Oscillatory
The mechanical spectrum of a material (e.g. the storage modulus (G'), loss modulus
(G"), and dynamic viscosity (11") plotted as a function of angular frequency (co)) gives
an accurate indication of the structural type. Typical examples of the four general
classes of behaviour for low to medium concentrations of polymer over an unspecified
frequency range are shown in Figure 5.
In the case of a dilute solution, (Figure 5a), even high molecular weight chains can be
considered to be isolated so that their principal effect is to perturb the solvent flow.
For the random coil chains usually encountered in food polymers, all possible
rearrangements can take place within the period of the imposed oscillation [7].
As frequency increases, the viscosity remains essentially constant and the dominating
loss component G" (= rl'cO) increases in proportion to frequency although an elastic
contribution (G' oc o32) also begins to come into effect. Mechanical spectra of these
11
relatively simple dilute systems were the first to be successfully modelled on a
molecular basis using bead-spring arrangements for flexible chains [8,9].
As polymer concentration and/or molecular weight increase, the volurnes of influence
of individual chains begin to encroach on each other, (domain overlap), followed by
more direct chain-chain interaction (coil overlap) [7]. Above this 'coil overlap
concentration' c*, the system is said to be semi-dilute. With further increase in
concentration, polymer-polymer interaction becomes progressively more dominant
until essentially uniform polymer density is achieved throughout the system. Through
this concentration regime, polymer chains begin to interpenetrate and a new
mechanism known as entanglement coupling dominates where the rheological
properties are dictated by molecular weight rather than the hydrodynamic volume of
individual coils. In the dilute region, and using frequencies applicable to standard
laboratory rheometers, only Newtonian behaviour can be observed. Moving through
the semi-dilute region, however, the frequency and shear-rate to which rl*(co) and
rl(7) remain constant become progressively less as 'shear-thinning' behaviour is
encountered at the higher frequencies and shear-rates. In this region, both G' and G"
become much less dependent on frequency and, when true entanglement coupling
exists, G' exceeds G" (Figure 5b). At higher frequency still, the moduli are almost
independent of frequency and because this behaviour, combined with G ' > G", is
similar to that exhibited by rubber-like materials, this is called the 'rubbery' plateau
region; the low-frequency Newtonian region being the terminal zone for this type of
material.
The transition from Newtonian to shear-thinning behaviour is dictated by a terminal
relaxation process, characterised by the longest relaxation time associated with the
natural motion of the polymer chain or a polymer-polymer disentanglement time.
In the latter case, the disentanglement-re-entanglement process can occur easily within
the timescale of imposed motion provided the frequency is sufficiently low and thus
the overall entanglement density remains constant. At higher frequencies and shearrates, this is no longer possible and re-entanglement diminishes giving rise to a
progressive reduction of entanglement density and consequent lower dependency of
transmitted stress on frequency or shear-rate. For dilute solutions, measurements are
strain-independent and the same is substantially true of this region where mechanical
properties are dictated by a single, or narrow band, of relaxation times, depending on
degree of polydispersity. This gives rise to the well known empirical law that the value
of rl*(o3) is equivalent to that of rl(dl,/dt) when o3 and d~t/dt are numerically equal (i.e.
the Cox-Merz Rule [10]. Because materials possessing shear-thinning properties are of
commercial importance, numerous models have been developed to emulate their
Newtonian to power-law-dependent theology such as those of Cross [11 ] and Morris
[12]. Advanced theories based on molecular considerations (i.e. chains topologically
constrained by nearest neighbours) are those of Graessley, described by entanglement
loci with weighted frictional coefficients [13], and de Gennes and Doi and Edwards,
reptation and confining tubes [14,15].
In the systems described so far, only mechanical interaction between the constituent
polymers has been considered. Although some weak, non-covalent bonding
(hyperentanglements) is detected in aqueous food biopolymer systems such as
galactomannans [16,17], the effect on the 'entanglement spectrum' of Figure 5b is
12
minimal. If more permanent bonding is introduced, the system can maintain elastic
properties to the longest times (lowest frequencies) normally observed and the classes
of mechanical response known as gels and 'weak-gels' are obtained.
The gelling of biopolymers has been a traditional method of structuring foods for
many centuries. Heat denaturation of animal protein to form a solid (gelation of
ovalbumin when boiling an egg) and cold setting of plant polysaccharide in the
presence of high levels of sugar (pectin gelation in jam making) are obvious examples.
Extracted proteins and polysaccharides form the basic structuring agents m
commercial fabricated foods and a vast body of literature concerning gelation
mechanisms in general and mechanical properties of specific aqueous biopolymer gels
has therefore been established. This important topic is beyond the scope of the present
brief overview but those interested should refer to the excellent review of the subject
by Clark and Ross-Murphy [ 18].
Figure 5c shows the mechanical specmma of a 'true' gel with G'(o~) substantially
above G"(o~), both being almost independent of frequency so that rl*(to) decreases
with a slope of-1 on the double logarithmic plot. If the bonding is truly permanent, a
finite elastic modulus (equilibrium modulus G'e) could theoretically be measured at
infimtely long time. For many biopolymer gels, the bonding is likely to be somewhat
more transient in nature and the plateau modulus observed at the moderate times of
normal experimental procedures is referred to as the 'pseudo-equilibrium modulus'.
Many gels, although of relatively high modulus, can withstand large strains (often in
excess of 100%) before rupture.
The class of materials known as 'weak gels' are of particular importance to the
modem food industry. As shown in Figure 5d, the spectrum is similar to that of a
normal gel with, perhaps, some frequency dependence and a higher average value for
tan& The major difference between the two types is, however, that these systems
exhibit much greater strain (or corresponding stress) dependence. Food types which
are solid-like when undisturbed but can be easily poured or spread may be included in
this category. A consequence of the rapid (but usually smooth and coherent) straininduced breakdown of these materials is that superposition of rl*(09) and rl(~ ) no
longer applies, strong power-law dependency being observed but with the steady-shear
viscosity considerably below the dynamic viscosity. The archetypal 'weak-gel' food
biopolymer is the bacterial polysaccharide xanthan composed of weakly associated
extended chains. Similar properties can be generated by specialized manufacturing
techniques to reproduce a heterogeneous structure composed of strong domains linked
again by narrow regions of weak bonding in which strain is concentrated to produce
the required breakdown characteristics. The time necessary for complete recovery of
the undisturbed structure in such systems has been observed to extend to several hours
[19].
It is clear from the spectra shown in Figure 5 that a relationship exists between timedependent properties and the level of internal structure. Increasing concentration and
molecular weight or decreasing temperature have often been used to effectively shift
measurable spectra to higher frequencies. The entanglement system in the range of
frequencies normally accessible shows behaviour typical of the terminal region at low
frequencies through a transition zone to a rubber-like plateau at moderately high
13
a
b
G'
G"
41.
-It-
_
_
_
_
.
.
.
.
.
.
.
o
o
_
1
~
0
_
_
_
rl*
__-
9
,
9
I_
log m (rad s "1)
.
.
.
.
.
.
_
_
_
d
C
4t-
_
Olog m (rad s -l)
.
.
.
.
.
.
.
.
.
!
e~0
O
"l(-
G
!
G
t!
o
e~0
_
.._
G
l!
rl*
rl*
I
0
Figure 5.
-_
.
.
.
.
.
.
.
.
.
.
.
.
.
.
|
. . . . . .
_
L,
~
-
....
o
log Co (rad s -1)
log 03 (rad s"l)
The four principal categories of mechanical spectra: (a) Dilute solution,
(b) entangled solution, (c) strong gel and (d) 'weak gel'.
14
frequency. If this range could be extended to access even shorter times, a 'glassy'
region would be reached where only local chain motion is possible within the
experimental timescale. For permanently cross-linked gels, rubber-like terminal
behaviour will be observed but a similar glassy region is still accessible. Later, glassy
behaviour and 'frequency shifting' (time-temperature superposition) techniques as
applied to food biopolymers will be addressed in more detail.
From the above, it is clear that rheology is a powerful technique to assist in the study
of food biopolymer materials as well as being indispensable in the areas of product
development and quality control. Long-time creep measurements, for instance, are
particularly informative with regard to permanency of structural bonding [20], and
oscillatory frequency and strain sweeps, perhaps in combination with microscopy, are
routinely used to indicate underlying microstructure. Investigation into the gelling
mechanisms of biopolymers is greatly facilitated by monitoring small-deformation
viscoelastic properties during temperature induced gelation or melting, particularly
when results are compared with those from other physical techniques such as
differential scanning calorimetry (DSC) and optical rotation (OR). Temperature
sweeps are also extremely useful in determining, for instance, which component forms
the continuous phase in two component, or multi-component, systems.
MODERN RHEOMETER TYPES
Modem computer controlled rheometers intended for shear measurements are now
relatively commonplace in food research and development. According to their mode of
operation, two distinct types are produced; controlled strain and controlled stress.
In the former type, a powerful motor is used to impose a pre-determined strain, or
strain-rate on the sample via one side of the measuring geometry and the transmitted
stress is detected at the opposing sample fixture by a low-compliance (or active
zero-compliance) transducer. Controlled stress instruments consist of a 'drag-cup'
rotor, optical strain detector and upper sample geometry on a common shaft supported
in an air-bearing, the housing of which also contains the motor windings. With the
sample confined between the upper geometry and a lower stationary fixture, the
required sample stress can be imposed with negligible losses in the air-bearing, and the
resulting strain measured by the angular displacement of the single moving part.
By virtue of their relatively simple construction and the fact that stress is easily
controlled by the current passing through the motor coils, these latter instruments tend
to be less expensive than the strain controlled variety in which more elaborate
engineering is necessary to guarantee accurate imposed strains over a wide dynamic
range. It is, however, desirable to be able to control the strain during, for instance,
gel-cure experiments and, provided measurement in the linear region can be
maintained, it is possible to operate stress-controlled instruments in a fashion which
closely approximates to strain control for dynamic oscillatory experiments.
15
Measurement Geometries
Three standard geometries are used to contain samples in a well-defined shape during
shear measurements: cone-and plate, parallel plate and concentric cylinder (Figure 6,
below and overleaf). The cone-and-plate geometry (Figure 6a) is theoretically ideal in
that the strain, or rate of strain, is constant over the whole working surface. Since a
full cone would touch the plate, it is trtmcated but in commercial instruments, this is
slight to minimise strain error. Since small dimensional changes in the rheometer
could still cause contact, cone-and-plate geometries are not considered suitable for
temperature sweep experiments. It is, however, the best option for isothermal
measurements on both linear and non-linear systems.
Figure 6a.
Cone-and-plate geometry.
In the parallel plate system (Figure 6b), the problem of expansion is removed but the
sample now experiences a strain varying from a maximum at the periphery down to
zero at the centre. This is, however, a good general purpose geometry for the
examination of strain-independent systems over a wide temperature range. An added
advantage is that, within reasonable limits, the gap can be varied to effectively shift
the dynamic strain/strain-rate range of the rheometer. For instance, if a wide gap can
be tolerated, the strain resolution will be improved allowing more highly straindependent or higher modulus materials to be measured.
03
Figure 6b.
Parallel plate geometry.
16
The concentric cylinder geometry (Figure 6c) is often used in simple viscometers and
for dilute solution work since the sample is contained and the torque to sample-volume
ratio can be high. The strain-rate is not constant across the sample, being a maximum
at the inner surface, but the gap is usually kept small (<= 0.1 of the outer cylinder
radius) to minimise errors. Since measurement is made in the body of the sample with
a wide gap at the upper surface, 'skin' effects are small and a thick layer of barrier
liquid can be flooded over the surface to prevent water loss making this geometry ideal
for dynamic and creep experiments over long time periods, particularly at elevated
temperatures.
Figure 6c.
Concentric cylinder geometry.
Practical Considerations
Before moving on to examine specific sample types in more detail, it is perhaps
desirable to highlight potential problems which may arise in the day-to-day
rheological evaluation of food component materials. Advances in applied
microelectronics have meant that the time-consuming manual procedures prevalent
some 20 years ago, often involving equipment developed 'in-house', have been
superseded by the introduction of computer controlled commercial insmmaents with
complete integral data processing capable of running linked, consecutive experiments.
Given the impressive, automatic nature of these insmnnents, it is sometimes easy to
forget their limitations, which, together with the structural complexity of some sample
materials, can lead to misinterpretation of results.
17
With regard to the instrument itself, the dynamic range for both stress and strain
measurement is the important requirement. Sample moduli may undergo changes of
several orders of magnitude during, for instance, gelation or melting during
temperature sweeps. Although the range can be shifted by changing geometry or
torsion bars when appropriate, this may mean that more than one set of experiments is
necessary for complete characterisation of a given sample. Taking as an example a
standard stress-controlled rheometer, measurement of steady shear viscosity over a
wide range of shear-rate and sample viscosity presents no great problems. The lower
limit of viscosity and shear-rate is defined by the minimum useful stress, which is
itself ultimately dictated by the quality of the air bearing. The upper shear-rate for
highly viscous samples obviously depends on the maximum stress available but, since
imposing a stress and measuring strain or rate-of-strain is simply a creep test, low
shear-rate measurements are possible on even the highest viscosity samples likely to
be encountered, provided enough time is available.
Small-deformation oscillatory experiments represent a more rigorous test of machine
performance. Here, two sinusoidal signals must be compared in terms of amplitude
and phase and provided at least one of these is of reasonable size, results are usually
satisfactory. Difficulties do, however arise if very small signals need to be correlated,
as may be the case for some low-modulus, but highly strain-dependent food materials.
Working below the designed range of the instrument in this manner sometimes results
m an 'offset' modulus which can be falsely assigned to the sample. Although
automatic corrections for moment of inertia and, where appropriate, compliance of the
insmunent, are incorporated in modem machines, these still limit the high-frequency
performance and upper limit of sample modulus respectively. In a driven mechanical
system, the forces due to elasticity and inertia are in anti-phase, the inertial force being
proportional to co2. At high frequency, therefore, the elastic component of a lowmodulus sample may be dominated by inertial forces (measured as negative G') such
that it is reduced to the error-band of in-phase resolution. Indeed, when making high
frequency measurements of very weak system, where the normal 'narrow-gap'
conditions (stress waves traverse the sample) begin to be superseded by surface
loading or 'infinite sea' conditions (stress waves penetrate only the upper layer of
sample), the first, approximate correction to be made is for loss of sample inertia, the
dominant effect. With regard to very high modulus systems such as glasses, two
factors come into play. Unless the rheometer is designed for such samples, overall
machine compliance may become significant compared to that of the sample and, if
only limited maximum stress is available, the strain generated may be too small for
accurate measurement. The obvious ploy of using a geometry of small working surface
area and large gap may produce measurable strains but correction for insmmaent
compliance is also a likely requirement.
Rheology is unusual among the physical methods used in the modem laboratory in
that sample handling represents a major experimental event. Loading the sample
probably imposes far greater stresses and strains than will have to be endured during
the subsequent experiments. Whilst this presents no problem for simple fluids or gels
that can be set from the liquid state, more structured materials with long memories
such as weak gels require small-deformation time-sweeps to ensure complete recovery
to the undisturbed state before further testing. Trapped air can also be a problem in
such systems during temperature sweeps, as its expulsion at elevated temperature can
18
lead to water loss and 'skin' formation at the edges of parallel plate geometries.
Sample slippage is also a potential problem particularly for some biopolymer gels such
as agarose, carrageenan and gellan which set very quickly and are susceptible to
syneresis and perhaps shrinkage. Slippage is not necessarily easy to detect and,
indeed, analysis using linear models indicates that provided very thin, viscous
interfacial layers are involved, the measured elastic modulus will not be greatly
affected. However, these conditions, particularly linearity, cannot be guaranteed and
an observed collapse of G' accompanied by a high value of tan 5 usually indicates a
problem. Using small strains and slow temperature scan rates can help but it may be
necessary to resort to special non-slip geometries in such cases [21,22].
To summarise this section, it should be remembered that both the rheometer and the
sample are being measured in a rheological experiment. It is desirable to ascertain the
range, sensitivity and phase resolution of the instrument using standard materials, the
results being a useful check on future performance. As far as the samples are
concerned, measurement of complex systems should be supplemented by as much
independent structural information as possible, the minimum requirement being visual
assessment to highlight any simple problems which may not be obvious from the
generated data. Although all the considerations mentioned above are fairly obvious
and will be well known to those routinely involved in the rheological investigation of
food biopolymers, there may be some information which will prove useful to those
about to embark on such work or who have limited experience of practical
measurements.
RHEOLOGY OF SOLUTIONS UNDER STEADY SHEAR CONDITIONS
Dilute Solutions
A particularly convenient and useful experimental parameter in studies of dilute
solutions is the intrinsic viscosity [q], a measure of the volume occupied by the
individual polymer molecules in isolation, which is directly related to molecular
weight (M) and is therefore widely used in routine characterisation of polymer batches
(Mark-Houwink equation):
[11] = K M ix
(19)
where K and ot are constants whose values depend on the shape of the polymer, the
solvent used and the temperature of measurement. Typical values of ot for random
coils are 0.5 - 0.8, and for rigid rods 1.5 - 1.8 [23].
Intrinsic viscosity is defined by the standard equations:
rlrel = rl/rls
rlsp = (11- rls)/rls = rlrel- 1
[11] = lim (risp/C)
c----~0
(20)
(21)
(22)
19
where ~ and Vls denote the viscosities of the solution and solvent, respectively, and qrel
and TisD are, respectively the dimensionless parameters of relative viscosity and
specifi6viscosity. Experimental values of Vispfor extrapolation to intrinsic viscosity at
infinite dilution (equation (22)) should be in the range 0.2 to 1.0.
Treating now molecules as particles widely separated, the Einstein relationship for
laminar flow can be derived [24]:
rl = rls (1 + kl~ )
(23)
where r is the phase volume of the disperse phase and k 1 takes the value of 2.5 for
spheres. From equations (21) and (23) rlsp can be expressed as a function of phase
volume:
rlsp = klr
(24)
Combination of equations (22) and (24) for infinite dilution gives the identity:
klr = rlsp = [rl]c
(25)
At higher concentrations, where the increase in viscosity is not directly proportional to
the mass of the disperse phase, the Einstein relation is extended by including higher
order terms in equation (24):
rlsp = k1r + k2t~ 2 + k3t~ 3 + k4~ 4 + ...
(26)
Experimentally, a linear intrinsic viscosity-concentration relationship is observed for
specific viscosities up to about 1 and its numerical form is obtained by keeping terms
up to quadratic in equation (26) and substituting 0 from equation (25):
rlsp/C = [rl] + k' [ri]2c
(27)
where k' = k2/kl 2
This is the Huggins equation and the extrapolation to give intrinsic viscosity is
obtained from a plot of rlsp/C v s c [25]. An alternative extrapolation is given by the
equation of Kraemer [26]"
ln(rlrel)/C = [r I] + k" [rl ]2c
(28)
It can be readily shown that k" = k' - 0.5, and thus equations (27) and (28) may be
combined to give an expression that allows intrinsic viscosity to be estimated from
measurements at a single concentration (the single point method) [27]:
[ri] = [2(rlsp- lnrlrel)] 89
(29)
The values obtained by this method may, however, also be plotted as a function of
concentration, along with the corresponding values from the Huggins and Kraemer
treatments. In practice these extrapolations may not be strictly linear, but using all of
them, intrinsic viscosities can be well bracketed (Figure 7).
20
Huggins -
J
Single point-_
[hi
---------liD-
J
[2(TIsp ,
In T]rel)]l/2/C
,,
ln(TIrel)/C
Concentration
Figure 7.
The three extrapolations to zero concentration in the determination of
intrinsic viscosity.
Effect of Entanglement
Moving now from the case of very dilute solutions, where the intention was to acquire
information about the volume occupied by individual molecules, up to the real range
of practical viscosity behaviour (rlsp > 1), the Huggins and the other associated
extrapolations become irrelevant because higher order powers such as those in
equation (8) start to be significant. With further increase in concentration, viscosity
begins to show appreciable dependence on shear rate (~). At low shear rates viscosity
remains constant at a fixed, maximum value (the 'zero shear' viscosity, 11o), but at
higher values 'shear thinning' is observed (Figure 8). Taking the maximum 'zero
shear' value, it has been observed empirically that for a wide range of 'random coil'
polysaccharides the log of rlsp varies approximately linearly with the log of
concentration over the viscosity range 1 < rlsp < 10, with a slope of about 1.4 [28].
This is illustrated in Figure 9 for the disordered form of the capsular polysaccharide
from Rhizobium trifolii (CPS), with the zero shear specific viscosity at the critical
concentration (c = c*) being close to 10 [29]. At higher values of rlsp, however, the
concentration dependence changes to a slope of about 3.3 and solutmns are termed
semidilute. At the inflection point, the 'coil overlap parameter' (c*[rl] from equation
25) has been found to have a value between 3 and 4 regardless of polymer primary
structure and molecular weight [28]. At even higher concentrations, individual coils
will form an entangled network whose relaxation time will be heavily governed by the
polymer molecular weight.
21
T1 --1"1o
11 - 0.5
@
m
I
I
I
log "~(l/s)
Figure 8.
Shear-rate dependence of viscosity for a typical concentrated
biopolymer solution.
3.0
slope = 9.3
s l o p e - 3.3
2.0
log rls p
s l o p e - 2.2
1.0
slope = 1.4
0.0
I
I
I
I
I
Ic[rl] = 3 6
J
I
I
t
-1.0
-1.0
i
-0.5
i
I
0.0
I
1I
i
0.5
i
!
1.0
i
1.5
log clrll
Figure 9.
The variation of 'zero-shear' specific viscosity with degree of
space-occupancy for CPS in the disordered form at 55~ (D), and for
levan at 20~ (m), from [29] with permission.
22
The change in concentration-dependence of solution viscosity during the transition
from a dilute solution of independently moving coils to an entangled network can be
rationalised as follows. At concentrations below the onset of coil overlap and
entanglement (c < c*), the main effect of the polysaccharide coils is to perturb the
flow of the solvent by tumbling around and setting up 'countercurrents', with mutual
interference of countercurrents from adjacent chains giving a somewhat more than
proportional increase in viscosity with increasing concentration. At concentrations
above c*, however, where flow requires chains to move through the entangled network
of neighbouring coils, the restriction of mobility increases steeply with increasing
network density, giving rise to the higher concentration-dependence of viscosity.
Shear thinning behaviour in polysaccharide solutions can be rationalised as follows:
At concentrations below c*, shear thinning is minimal (typically less than 30% over
several decades of shear rate), and can be attributed to elongation of individual coils in
the direction of flow at high enough ~;. The 'Newtonian plateau' observed (Figure 8)
for entangled coils (c > c*) at low shear rates corresponds to a dynamic equilibrium
between forced disentanglement (to allow the solution to flow), and re-entanglement
with new partners. At higher values of ~, where the rate of disentanglement exceeds
the rate at which new entanglements can form, the overall crosslink-density of the
network is reduced, with consequent reduction in viscosity (often by several orders of
magnitude). The form of the shear-thinning for entangled polysaccharide coils can be
matched, with reasonable precision [28], by the equation:
rl = 1]~ . ( r l / ~ l/z)0.76 1,1~ 0.76
(30)
where ~ ~/~is the shear rate required to reduce rl to rio/2. Thus the two parameters rio
and 3;,/2 which, in conjunction with equation (30), completely characterise the flow
behaviour of a random coil solution can be determined from a linear plot of
vs. rl~ 0.76. Figure 10 shows linearised shear-thinning plots derived in this way for
some typical random coil polysaccharide solutions [12].
2o r
15
10
0
0
20
40
60
80
1O0
n ~ 0.76
Figure 10.
Shear-thinning plots for locust-bean gum (i), alginate (o), pectin (A),
and lambda carrageenan (o), from [12] with permission.
23
Recently the concentration dependence of zero shear viscosity was monitored for
bacterial levan [29], an extensively branched polysaccharide with a compact,
spheroidal shape ([rl] = 0.17 dl/g). This time the discontinuity in the rlsp v s . c[rl]
profile was first indicated at much lower values of the coil overlap parameter (about
0.75), a threshold which indicates the end of the dilute regime (Figure 9). Similarly, a
value of C[rl] about 1 has been proposed to signify the concentration (c = c*) where
the swept-out volume occupied by spherical coils becomes equal to the total volume
(~ = 1). The sharp rise in viscosity of levan at high concentrations (slope of 9.3 at
c >_ 19%) has been rationalised qualitatively by the reptation theory of de Gennes [14].
In this model a highly branched macromolecule is confined within a hypothetical tube
whose domain is determined by the branching points of the main backbone. Long
range movement along the tube is only allowed when a branch disentangles from the
sites of neighbouring chains thus making obvious the additional constraints to flow for
the heavily branched levan molecules as opposed to linear polysaccharides.
BINARY B I O P O L Y M E R MIXTURES
Two gelling biopolymers in the same system can create three general types of network
structure, namely:
a)
b)
c)
Interpenetrating networks
Coupled networks and
Phase-separated networks
as illustrated in Figure 11 below (from [30] with permission).
.J
a
b
C
Figure 11.
The three possible network topologies for binary gelling systems.
24
a) Interpenetrating Networks
These represent the simplest situation, rarely encountered in mixed biopolymer gels,
where the two components gel separately forming two independent network structures.
Both networks span the entire system, interpenetrating one another, but interaction is
solely topological (Figure 11 a). If each polymer forms a homogeneous network across
the whole of the single phase, and if the like segments possess identical properties in
all directions, then the modulus of the composite has been observed to be related to
the moduli of the two component networks by a logarithmic mixing rule [31 ].
Log M = ~x log M x + t~y log My
(31)
where ~x and #v are the phase volumes of polymers X and Y, and M (composite), M x
and My can be hither the Young's or the shear modulus.
Although the formation of interpenetrating networks is simple and reasonably well
understood, two dissimilar polymers present in the same system may not necessarily
form two separate interpenetrating networks for reasons of thermodynamic
incompatibility discussed in the Section on phase separated networks. One way of
ensuring bicontinuity for experimental studies is, as suggested by Morris [30], to
introduce a second polymer into the pores of a pre-existing network, and then to alter
conditions in such a way that this second species forms its own network without
disruption of the original gel. The problem of slow diffusion of a polymer solution
into a gel can be tackled by using smaller globular proteins as the diffusing species.
Dispersion may be enhanced with suitable alterations of the network charge or even
by the use of external electric fields. An alternative would be to prepare a xerogel and
then swell this gel in the protein solution. A final requirement is the thermal
irreversibility of the pre-existing polymer network in order to allow for the heat-set
process of the protein.
b) Coupled Networks
This kind of interaction involves a direct association between two different polymers
to form a single network (Figure 11b). Three different types of intermolecular binding
may then arise:
1))
iiQ
Covalent linkages
Ionic interactions
Co-operative junctions
Chemical cross-linking between different chains offers a direct way of forming a gel
network The main characteristic of a system held by covalent bonds is its
thermostability and this can be achieved even at relatively low crosslink densities.
It can withstand heating, but loses its gel-like character through disruption of chemical
bonds in a way reminiscent of the oxidative degradation of the disulphide bonds of
rubber. Rheologically, the most notable feature of covalently cross-linked gels is the
permanency of the network formed, which is typically greater than that observed in
physically-crosslinked gels [32,33]. Effectively, covalent crosslinks have infinite
relaxation time.
25
One reasonably well-understood interaction of this type involves the formation of
amide bonds between the propylene glycol esters of alginate (PGA) and uncharged
amino groups of gelatin [34]. The crosslinking reaction proceeds smoothly when
aqueous solutions of the ester and protein are mixed under mildly alkaline conditions
(z pH 9.6) giving gels that are stable to above 100~ Evidence for the chemical nature
of the crosslink is not direct, but dye binding experiments, the binding of
trinitrosulphonic acid, and studies of the formal titration show that lysine groups are
involved but that only about one in six is utilised. Gel strength increases with
increasing mannuronate content in the alginate [35], indicating that ester groups
attached to the extended polymannuronate ribbons are more accessible than those in
the highly buckled polyguluronate sequences.
Direct interactions can also occur between biopolymers of opposite net charge, by
formation of an insoluble coacervate. Complex coacervation between, for example,
gum arabic and gelatin has extensive practical use in microencapsulation [36].
Coacervation with anionic polysaccharides has also been proposed as a method for
recovery of waste protein from abattoir effluent or whey [37]. Alginate is a
particularly attractive choice for the polysaccharide component, since its calciuminduced gelation may be used to structure the recovered protein after solubilisation of
the coacervate by raising the pH to above the isoelectric point of the protein. Ionic
attraction may also be involved in the highly specific interaction of kappa carrageenan
with kappa casein, to give a weak gel network of practical importance in, for example,
preventing sedimentation of cocoa particles in chocolate milk desserts [38].
Finally, it has been suggested that in some systems the interactions of unlike
polysaccharides may involve formation of specific co-operative junction zones
analogous to those in single-component gel systems, such as carrageenans, but with
the participating chains being heterotypic rather than homotypic. The mixed gels
formed between alginate and pectin under acid conditions are believed to involve
junctions of this type [39]. Gel strength increases with the methyl ester content of the
pectin and with the polyguluronate content of the alginate. The requirement for low
pH can be explained in terms of protonation of carboxyl groups causing reduction
of electrostatic repulsion between the chains, with a high degree of esterification
in the pectin component having a similar effect. The specific involvement of
poly-L-guluronate can be rationalised in terms of its near mirror-image relationship to
the poly-D-galacturonate backbone of pectin allowing stereoregular heterotypic
junctions to be formed [40]. The most compelling lines of evidence for such mixed
junctions are the development of maximum gel strength at stoichiometric equivalence
of poly-L-guluronate and esterified poly-D-galacturonate and the circular dichroism
changes accompanying gel formation, which are very different from those observed
for either polymer in isolation.
Another example of co-operative synergism is between certain galactomannans
(notably carob and tara gum) and certain helix-forming polysaccharides (agarose,
xanthan, kappa carrageenan). In general the phenomenon is believed to involve
unsubstituted regions of the mannan backbone associating with the ordered
conformation of the second polymer to create mixed-junction zones [41 ].
26
Phase-Separated Networks
When favourable interactions (such as those in polyanion-polycation systems) are
absent, thermodynamic incompatibility between chains of dissimilar biopolymers
tends to cause each to exclude the other from its polymer domain, so that the effective
concentration of both is raised (Figure 11 c). This is true even when the energies of
interaction between the chains involved are small (disordered chain segments) in
comparison with the much stronger interactions of ordered sequences in
polysaccharide gels.
At low concentrations, thermodynamic incompatibility can promote conformational
ordering within a single phase, which, for gelling systems, can increase the rate of
network formation [42]. At higher concentrations the system may separate into two
discrete liquid phases. Generally, phase separation in protein-polysaccharide-water
systems occurs only when the total concentration of the macromolecular components
exceeds 4% [43], although there are variations from system to system. In the case of
carboxyl-containing or sulphated polysaccharides, ionic strength and pH play an
important role. For example, proteins and carboxyl-containing polysaccharides phase
separate at pH values above the isoelectric point (at any ionic strength) or when the
pH is equal to or less than the protein isoelectric point but the ionic strength is greater
than ~ 0.25.
Figure 12 presents an example of a typical thermodynamically incompatible proteinpolysaccharide system in aqueous solution [44]. The bold line is the binodal or cloud
point curve. To the left of the binodal, the system remains in a single phase whereas to
••\.•\\•
\\ ~
"'\ .....\.
"\
"-, ""',.
Initial overall
composition
_
0
i
0
I
5
II
,
I
10
1
I
15
20
!(:1(%)
Figure. 12
Phase diagram for casein (C) and sodium alginate (A) in mixed solutions
at pH 7.2 and 25~ (o) compositions of co-existing polysaccharide-rich
and protein-rich phases, (o) critical point, from [44] with permission.
27
the right the system exists in a two-phase liquid state. One of the phases is enriched in
polysaccharide and depleted in protein and vice-versa. The faint lines are tie lines
joining two points that lie on the binodal and represent the composition of each phase.
The initial composition of the mixture is a point on the tie line. All points that lie on
the same tie line will eventually separate into phases with the same concentrations
(the two terminal points of the tie line) but the relative volumes of the phases will
vary. Tie lines finally converge to a critical point and the values of its coordinates
provide a measure of the compatibility of the macromolecular components. As shown
in Figure 12 the phase diagram is asymmetric (different axis scales) with the
polysaccharide usually having a substantially lower final concentration than the
protein. This may be explained on the basis of the high volume-occupancy of an
expanded polysaccharide coil in comparison with a compact globular protein.
Application of rheological measurements to mixed systems
After gelation, phase separated elastic networks may be regarded as composites whose
mechanical properties may be derived from the moduli (G x and Gv) of the two
components considered as individual systems present at phase volume frhctions ~x and
~y (where ~x + ~y = 1). The analysis is an approximation, because it is based on binary
composites of pure, mutually insoluble, synthetic polymers whose individual
rheological properties are independent of the macroscopic amounts present [45].
According to the most simple model, by assuming extreme cases in the distribution of
strain and stress within the composite, two equations can be derived:
Gc = ~)xGx + (l)yGy
(3 2)
l/Gc = 4)x/Gx + 4)y/Gy
(33)
where G c is the shear modulus of the composite. Equation (32) applies to isostrain
conditions, where the continuous phase is more rigid than the dispersed phase so that
the strain is approximately uniform throughout the material (upper bound), whereas
equation (33) refers to isostress conditions, where the supporting phase is substantially
weaker than the discontinuous phase hence the stress may be regarded as constant in
both phases (lower bound).
By combining two simple viscoelastic models (e.g. Voigt elements in parallel and
series) in proportion to their phase volumes, it is a simple matter to predict overall
viscoelastic properties. In the original work of Takayanagi [45], such a model was
checked by performing dynamic extensional measurements on samples composed of
two layers of different synthetic polymers, the strain being imposed either parallel or
perpendicular to their common sides. For true dispersed composites, it was found that
a three element model was necessary to accurately emulate both E' and E" over the
wide temperature range employed.
In the case of water-based biopolymer composites, where structural complexity is
likely to make such a sophisticated approach inappropriate, it is common practice to
assume negligible contribution from loss processes (provided the sample is essentially
solid-like throughout) and apply the simple models shown above. For clarity, these are
28
derived here using the two-layer physical models rather than the more simple method
with mechanically descriptive viscoelastic elements.
For the isostress case, (Figure 13a), the extensional force F is applied perpendicular to
the interface between the two materials X and Y over the common area A, their
moduli being E x and Ey respectively. If the original thicknesses of the layers (L x and
Ly) are extended by amounts AL x and ALy then the modulus of the whole sample is:
Eo=
F(L• + Ly)
A(AL~ + ALy)
_
(34)
FLy
"
FLx and ALy AE~
AEy
where
A L~
hence
Ec = E•215
+ ty)
(35)
(36)
E~Ly + EyL•
By inverting both sides, we obtain
--
1
E~
=
L X
9
E~(L• + Ly)
+
L,.
E:,(L~+ Ly)
(37)
Since the area A is common to both layers, this reduces to
1 _ r
Ec
Ex
~ ,,,
Ey
(38)
For the isostrain case, the force is applied parallel to the interface as in Figure 13b.
Here, F is the sum of the forces F x and Fy acting on areas A x and Ay to produce a
common strain 7c in the two components.
i.e.
F = F x + Fy
(39)
where
F x = ExAx,/c
(40)
and
Fy = EyAxYc
(41 )
so that
F = (ExA x + EyAy)~c
(42)
Since modulus (E) = total force (F)/[(area (A) * strain (u
then
E = (ExA• + Eymy)~/~
7c(A• + Ay)
(43)
29
Figure 13.
Schematic diagrams of the (a) isostress and (b) isostrain models.
Since the sample length is common, the area of each phase is proportional to its phase
volume ~, i.e. ~x = Ax/A, ~)y = A / A
so that
E c = ,xEx + ,yEy
(44)
Although these upper and lower bound models represent a simplification of the true
situation, their usefulness lies in defining an area of modulus versus composition in
which experimental results should be confined if the rigidity of the material is
determined by simple phase separation [46].
Water distribution within mixed biopolymer systems
In the case of binary aqueous gels an extra complication is introduced by the presence
of solvent as a third component which can partition itself between the two polymer
constituents. The resulting phase volumes (r and Cy) depend on the relative powers of
the two polymers to attract solvent and must be found before equations (32) and (33)
can be used. To help surmount this difficulty, Clark and his group introduced a new
parameter, p, in a study of phase-separated agar and gelatin mixed gels [47].
This parameter is a measure of the solvent partition between the two phases and
enables the effective local polymer concentration in each phase to be estimated as p is
allocated different values. These adjusted concentrations are then used to calculate the
gel modulus of each phase based on suitable fits for the relationship between gel
modulus and concentration and hence an overall modulus, for comparison with the
experimental data, can be calculated using the simple Takayanagi treatment.
30
Considering an aqueous system of total mass W containing two polymers X and Y of
respective masses x and y, then the mass of water available for distribution within the
system is
w = W-x-y
(45)
Assuming that a fraction a of the water becomes associated with polymer X then
mass of X phase = x + txw
(46)
mass of Y phase = y + (1 - aw)
(47)
The effective concentrations of the two phases are, therefore
X
Cx(eff) = ~
(48)
(x +aw)
Cy(eff) =
Y
(y+(1-a)w)
(49)
ff the densities of the two phases are D x and Dy, respectively, then their volumes are
given by:
Vx = +(xa w )
(50)
Dx
and
Vy--
(y +(1- a)w)
D
(51)
y
Thus, the phase volumes of the components are
Wx
Dy(X +aw)
v~+Vy
and
Cy "-- ~
Vy
[ D y ( x + a w ) -I- Dx(y + (1-
-"
D~(y + ( 1 - a)w)
+ Vy [Oy(X
o (y
a)w)]
(52)
(53)
In many biopolymer gels, the polymer concentration is small so that D x ~ Dy ~ 1.
Hence the p parameter is simply defined as
p.
a/x
ay
. . .
(1-a)/y
(1-a)x
(54)
31
so that
a-
px
(55)
px+y
this can be substituted in equations for effective concentration and phase-volume to
give, for instance
Cx(efD =
(px + y)
Wp +(1-p)y
(56)
Recently, the distribution of solvent (water) between two biopolymer phases was
explored rheologically by calculating the values of storage modulus for all possible
distributions and finding which ones matched the experimental values [48].
The computerised output of this approach is shown in Figure 14 where the solvent
position is defined by Sx (aw), the fraction of water present in the polymer X phase.
.
.
.
.
.
.
.
.
.
/
I
I
i
I
I
UPPER BOUND (Gu)
~
/
/
t
/
POLYMERX
(Solid line)
//
e,.
u
o
((Dashed
D a s h eline)
~)/ /
0
Figure 14.
LONWER
LO"~WERBOUND
BO~ID (GL)
(GL)X~
Sx
1
Changes in calculated modulus as a function of Sx, the solvent fraction
in the X phase. The solid lines show the upper and lower bounds for
polymer X whereas the broken lines show the corresponding bounds for
polymer Y, from [48] with permission.
32
At very low biopolymer concentration, Sx and (1-Sx) are virtually identical to the
phase volumes 0x and r At higher concentrations the values diverge but Cx and or can
be readily calculated Irom Sx by taking account of the direct contribution ~ the
polymers to phase volume. Knowing the starting concentrations of X and Y, Sx
determines the final concentrations in the individual phases, which in turn gives the
moduli Gx and G., of the two phases from experimental calibration curves (e.g.
cascade fit [49,50]~. The overall moduli of the composite gels can then be calculated
by the isostrain and isostress blending laws. As described above, the former applies
when the continuous phase is stronger than the dispersed phase, and gives an upper
bound modulus (Gu); the latter gives a lower bound value (GL) when the dispersed
phase is stronger. At very low values of Sx, where most of the water is in the polymer
Y phase, polymer X is extremely concentrated, and thus G x >> Gy. Conversely, at
very high values of Sx, G x << Gy. At one critical value of Sx, the moduli of the two
phases cross over (with G x = G, = G U = GL). Up to this point, the 'upper bound' value
(Gu) corresponds to a p o l y m J X continuous system, with the matrix harder than the
filler, and the lower bound value (GL) to a polymer Y continuous, with the matrix
weaker than the filler (Figure 14). At higher values of Sx, G U relates to polymer Y
continuous and G L to polymer X continuous.
Kinetic influences in the gelation-phase separation of mixed biopolymer systems
The theoretical modelling of the preceding section has been applied to composites of
maltodextrin/gelatin, maltodextrin/sodium caseinate, maltodextrin/denatured milk
protein and denatured milk/soya proteins, thus providing a general outline of phase
separation in systems widely used in manufactured foodstuffs [51]. In the case of
maltodextrin/milk protein, results are summarised in Figure 15 where the milk protein
is regarded as polymer X throughout, so that the parameter Sx refers to the fraction of
solvent in the milk protein phase. Mixed gels were prepared using a fixed
concentration of milk protein (16.5% w/w) with maltodextrin concentration from 2%
to 18% w/w. Analysis of the phase separation between the two components
(Figure 15a) indicates phase inversion from a weak, protein-continuous phase (lower
bound) to a strong, maltodextrin matrix (upper bound) at about 13% maltodextrin.
Solvent fractions derived from Figure 15a were used for analysis of water partition
between the two phases (Figure 15b), yielding p = 1.7 (log p = 0.23) for the intercepts
in the milk protein continuous systems, whereas the data beyond the phase inversion
point (maltodextrin continuous systems) are better fitted with a value of p ~ 1.1 (log p
0.04).
A similar variation in the p values as a result of phase inversion was documented in
the remaining biopolymer mixtures with either component holding disproportionate
amounts of solvent in its phase when it formed the continuous matrix. Provided that
phase inversion does not affect the mechanical analysis via interfacial effects, it
follows that the difference in p values is the result of phase separated gels being
trapped away from equilibrium conditions, since the equilibrium value of 'relative
affinity' of the two polymers for water should not be affected by the geometrical
rearrangemems of their phases in a binary mixture. Therefore, the p parameter, at least
for the gels of this investigation [51 ], is not an equilibrium concept but a single point
measurement in the experimentally accessible time-temperature continuum.
5.5
4.5
2.5
1.5
0
0.2
0.4
0.6
sx
0.8
1
-1
-0.5
0
log P
0.5
1
Figure 15. Calculated bounds for 16.5%milk protein with maltodextrin concentrations as shown (Yow/w),plotted (a) as a
function of solvent fraction in the milk protein phase (SJ, and (b) solvent avidity parameter p. In the milk protein
continuous systems (2-12 % maltodextrin), only the lower (isostress) bounds are drawn, whereas above the phase
inversion point ( 13-18% maltodextrin) the upper bound (isostrain) predictions are illustrated. Experimental values
( 0 )are shown to intercept the bounds and the experimental modulus for 16.5% milk protein alone is noted by the
arrow on the right-hand axis, from [ 5 11 with permission.
W
W
34
Independent evidence of the kinetic influences on the formation of composite gels was
obtained when the gelatin-maltodextrin system was subjected to disparate cooling
regimes [52]. Figure 16 reproduces changes in the storage modulus as a function of
quench and controlled cooling regimes for gelatin-maltodextrin solutions prepared at
70~ In the case of quench cooling to 5~ there is a sharp increase in the values of
storage modulus beyond 15% maltodextrin which is coincident with the phase
inversion from gelatin to maltodextrin continuity in the co-gels. Below the phase
inversion point, mixed solutions remain clear, and upon gelation show a steady
reduction in experimental moduli with increasing maltodextrin concentration.
This was rationalised on the basis of gradual deswelling of the gelatin network due to
ordering of the polysaccharide component after gelatin gelation. For the maltodextrin
continuous combinations, however, samples become cloudy upon mixing at 70~ and
G' results are better described on the basis of immediate phase separation in solution
and separate gelation of the two biopolymers.
In the event of controlled cooling at 1~
mixed preparations with a maltodextrin
content up to 7.5% are clear during the cooling cycle (from 70 ~ to 5~ and show the
familiar reduction in moduli of deswelled gelatin networks. At concentrations of
maltodextrin between 10% and 15%, samples are still clear at 70~ but become turbid
during controlled cooling and there is an immediate reinforcing effect on the
composite strength as a result of the networks being formed at higher concentrations
than the nominal, since the whole volume is no longer available to either component.
Subsequent heating indicates gelatin continuous mixtures since a gelatin-related
heating profile is recorded (melting at ~ 30~
At maltodextrin concentrations of
17.5% and 20% turbidity is observed upon mixing (70~ but there is no question that
these gels (5~
have become maltodextrin continuous (like their quenched
counterparts), since they also collapse at an early stage during heating, at about 30~
At higher concentrations (22.5% to 30%), mixtures phase invert to give a maltodextrin
continuous situation with a prolonged melting behaviour (> 80~
In conclusion, it
was surmised that reduction in cooling rates allows for more complete phase
separation before gelation 'freezes' the system, and results in reinforcement of the
continuous phase which now can support higher volume fractions of the dispersed
phase (filler) before phase inversion finally occurs.
THE APPLICATION OF WILLIAMS, LANDEL AND FERRY KINETICS TO
THE HIGH SOLIDS BIOPOLYMER SYSTEMS
High-solids systems- those with substantial concentrations of biopolymers and/or cosolutes - are of increasing academic and industrial interest. Typically, confectioneries
have 10-20% moisture in the finished product and a high proportion of sugars.
These products are almost exclusively manufactured by the process known as 'starch
moulding' in which a hot (e.g. 80~ solution of all the ingredients at typically
20-30% moisture (liquor) is deposited into impressions made in the surface of 'dry'
starch powder filling a shallow tray. The excess moisture is extracted by 'stoving' the
sweets in the starch at a moderate temperature (about 50~ for an extended period of
typically several days. Single and mixed biopolymer systems are used to formulate
products with rubbery/glassy texture but the mechanistic understanding of viscoelastic
4
cooling
(lO/min)
Turbid on
I
cooling
(10/min)
I
- 1
I
I
Turbid
at 70°C
I
'
A
I
I
I
I
I
I
I
I
I
A
0
I
I
A
A
0
0
Gelatin continuous
4Clear on quenching
I
I
I
I
a
A
A
1 - 1 6
I
I
I
0
Maltodextrin continuous
I
Clear on
1 '
I
I
Gelatin continuous
A
A
t
0
0
I
I
b
I
I
I
I
Maltodextrin continuous
Turbid at 7OoC
I
I
I
25
30
I
5
10
15
20
Maltodextrin ( YO)
35
Figure 16. Development of modulus for gelatin-maltodextrin mixtures as maltodextrin content is varied,
when quench-cooled (0), and when steadily cooled at l"C/min (A), from 70°C to 5°C.
G values were recorded after 7h at 5°C following the cooling scan, from [ 5 2 ] with permission.
w
ul
36
and textural properties is lacking. However, the structural properties of non-crystalline
synthetic polymers are well described by Williams, Landel and Ferry kinetics (WLF)
based on free volume theory, which follows the transition from a rubbery state at high
temperature to a glassy state on cooling.
The application of the WLF approach to aqueous preparations of biopolymer gels can
fail due to development of crystallinity or intermolecular enthalpic interactions as a
function of temperature, resulting in a change in the distribution of relaxation times
not simply related to the temperature-dependence of the free volume [53]. The recent
paper by Lopes da Silva et aL on pectin dispersions only highlights the problem with
the authors stating 'but a smooth master curve could not be obtained for both moduli
simultaneously or for each one individually .... satisfactory reduction of the data to a
single curve was not obtained, irrespective of the frequency shift factor used, for each
modulus individually or with vertical shift factors higher than those calculated by the
experimental temperature-density factor' [54]. Work in this laboratory has now
documented the transition from rubber-like to glass-like consistency for high solids
gellan gum and high methoxy pectin samples. Such information might prove to be
instrumental in the development of appealing novel confectioneries. Before that,
however, we feel that a treatise on the free volume-WLF theory, tailored for the food
scientist, is necessary.
The Free Volume-WLF Kinetics Theory
Free volume is a useful concept closely related to the hole theory of liquids, and the
approach has been more successful than any other model in describing the glass
transition in synthetic polymers. The total volume per mole u is pictured as the sum of
the free volume uf and an occupied volume u o. Ferry takes u o as including not only the
van der Waals radii but also the volume associated with local vibrational motion of
atoms [55]. The free volume is therefore that extra volume required for larger scale
vibrational motions than those found between consecutive atoms of the same chain.
Flexing over several atoms, that is, transverse string-like vibrations of a chain rather
than longitudinal or rotational vibrations will obviously require extra room. The glass
transition temperature (Tz) is defined on the free volume concept as that temperature
at which uf collapses sensibly to zero, or at any rate to a fixed, low value. Large scale
mobility has therefore been totally restricted and the only movement below Tg is that
allowed by the occupied volume u o.
To move from a qualitative description to quantitative treatment, experimental work
on the viscosity of the alkanes over a wide range of temperature was carried out and
Doolittle found that the simple relation [56]:
q = A exp (B/fu)
(57)
fitted the results more precisely than the Arrhenius equation:
rl = A exp (B/T)
(58)
where the values of A and B remain unchanged during a temperature ramp.
37
Unlike equation (58), in equation (57) log viscosity is not a linear function of
temperature with fu being the fractional free volume. Qualitatively, fu is:
fu = (UT- UTo)/UTo
(59)
where u T is the specific volume at a reference temperature T o. A quantitative
definition of the fractional free volume is:
9
o
0
~
9
fu = fuo + cxf(T - To)
(60)
In equation (60), fuo denotes the fractional free volume at an arbitrary chosen
reference temperature and czf gives the coefficient of expansion of the free volume
when the sample temperature changes from T o to T [57]. The above mathematical
expression can be used to replace fu in the Doolittle equation (57) yielding the
following for viscosity:
In {rl(T)/rl(To) } = {-CI(T - To)}/(C 2 + T - To)
(61)
At any reference temperature To, therefore, C 1 and C 2 a r e functions of the fractional
free volume and of the expansion coefficient of flee volume being respectively equal
to B/fuo and f uo/etf. In synthetic polymers, the thermally-induced transition from
rubber-like to glass-like consistency maintains a characteristic spectrum of relaxation
times (absence of crystalline or enthalpic processes). In view of this, Williams, Landel
and Ferry used the above empirical equation to combine viscoelastic data from a wide
range of temperatures into a composite (or master) curve. Thus it was found that the
viscosity of synthetic polymer melts measured at shear rate ~ and temperature T, is
equivalem to viscosity measured at shear rate $ a T and the reference temperature T o.
Therefore for viscosities aT, the horizontal shift factor, is the ratio rl(T)/rl(To).
Of course, reduction of data to T o and the production of a single composite curve can
be achieved for any modulus function-G"(c0), G'(t), E(t), etc (time-temperature
superposition principle; TTS). For over forty polymers and diluted systems
(polyisobutylene, polystyrene, polybutadiene, etc) Ferry and his colleagues
demonstrated the utility of equation (61) and in conjunction with the concept of free
volume argued for the universality of relaxation processes at the glass transition
temperature regardless of chemical structure.
Although the combination of time-temperature superposition with the free volume
approach created a comprehensive model of glassy phenomena, the method of
thermorheological reducibility has been shown before to work empirically [58].
In Figure 17, a series of curves is obtained by plotting creep curves, taken at different
temperatures, against In(time). Below 25~ the experimental frequency range appears
to correspond to the glassy zone; the compliance is quite low and does not change
much with frequency. Above 55~ the behaviour appears to correspond to the plateau
zone; the compliance is characteristic of a very soft rubber-like solid, and again
changes only slowly with frequency. At intermediate temperatures, the transition zone
makes its appearance. As illustrated in Figure 18, horizontal shifting of the data along
the In(time) axis results in a continuous master curve over a long time regime with
each superposition step defining the shift factor between successive experimental
temperatures.
38
1000
~,
Z
"7
60 ~ ~ , ,
55 ~
50 ~
100
39 ~
35 ~
t",,I
45 ~
10
30 ~
E
25 ~
23 ~
0
1
~
0.1
1
10
100
Time (s)
1000
10000
The compliance of a flexibilised epoxy resin as a function of time and
temperature, from [58] with permission.
Figure 17.
1000
_--" 100
r
10
O
1
/
0.1
Reduced time
Figure 18.
Master curve for Figure 17, produced by horizontal time-temperature
shifting, from [58] with permission.
39
Figure 19 shows the temperature dependence of lnotT for the data of Figures 17 and
18. If the dependence of the relaxation time on temperature had followed the
Arrhenius relation (equation (58)) the following expression would have emerged:
(62)
lntxT = (AH/R) {(l/T) - (1/To) }
Equation (62) relates to a constant activation energy for the experimental temperature
range and is obtained from the gradiem of a linear relationship between lno~T and 1/T.
This type of relationship, of course, occurs for polymers at regimes below and above
the glass transition area. In Figure 19, however, a curve is obtained which in
accordance with the analysis of the preceding paragraph argues strongly for a
non-An'henius process.
in 13[~T
2.8
2.9
3.0
3.1
3.2
3.3
3.4
1000 K/T
Figure 19.
Plot of the shift factor w r against inverse absolute temperature for the
master curve of Figure 18, from [58] with permission.
40
The Application of the WLF Approach to High-Solids Food Biopolymer Systems
The high solids work in this laboratory originates from a programme of background
research to address the problem of gelatin-based sweets sticking together when stored
at high temperature. The immediate aim was to explore the possibility of using the
new food polysaccharide, gellan gum, to create a continuous heat-stable matrix, with
gelatin dispersed through it as a discontinuous phase. The first step was to characterise
the effect of high concentrations of dissolved solids (sucrose and corn syrup) on the
gel properties of gellan alone. In the absence of co-solutes, gellan displays a sharp
sol-gel transition on cooling, due to the formation and cation-mediated aggregation of
intermolecular double helices. Addition of co-solutes up to a concentration of about
50% w/w had the anticipated effect of raising the gelation temperature.
At higher concentrations, however, an entirely different pattern of response was
observed. Mechanical spectra recorded at high temperature were qualitatively similar
to those of gels ( G ' > G", with little frequency-dependence of either modulus).
On cooling, the spectra become more like those of biopolymer solutions (G" > G',
with both values increasing steeply with increasing frequency of oscillation) [59].
For example a mixture of 0.5% gellan, 50% sucrose and 35% corn syrup produces the
mechanical spectrum of a rubbery material at the highest experimentally accessible
temperature of 90~ (Figure 20). This contrasts strongly with the Newtonian
behaviour of single solute and gellan solutions at 90~ Subsequent cooling produces a
four decade change in the values of in-phase and out-of-phase components with the
viscous component dominating throughout the experimental frequency range
(mechanical spectrum at 5~ in Figure 20).
Although not previously seen for a biopolymer in a high-water system, such behaviour
is well known for synthetic polymer melts, as described above, and corresponds to a
transition from a 'rubbery' state at high temperature to a 'glassy' state on cooling.
Using the WLF example, therefore, frequency sweeps of the elastic and viscous
components at 90 ~ 70 ~ 50 ~ 30 ~ and 5~ were successfully superimposed according
to the TTS principle at 90~ (reference temperature), showing that the vitrification of
the above gellan-solute system commences at about 700 Hz. This is shown in
Figure 21a where it is laid alongside the classic logarithmic plots of G' and G" against
frequency near the onset of the transition zone, for poly(n-octyl methacrylate) at
100~ (Figure 21b), as shown by Ferry and co-workers [55].
Heating of the samples from 5 ~ to 90~ produces what would be an extremely unusual
'melting' profile for polysaccharide gels [60]. Figure 22 emphasises the difference in
the characteristic temperature dependence of shear modulus in gellan networks with
and without sucrose. Similarly to every single hydrophilically gelling polysaccharide,
reduction in network strength with increasing temperature produces a negative
gradient of modulus change as a function of absolute temperature (T) for aqueous
gellan gels. By contrast the positive relationship between elastic modulus and
temperature in the high solids gellan structures yields a family of constant gradients in
the G'/T v s . T graph, typical of the temperature dependence of a rubber.
41
4.3
4.3
I
m
m
mm
mmm
m
mm
3.3
m
[]
[]
;::s
= 2.3
O
[]
.....
[]
a
o
n
m
m
m
m
3.3
o
o
[]
m
o
[]
0
[]
[]
(3
m
0
90 ~
I
I
70 ~
2.3
I
I
I
I
5.3
4.7
Q
[]
mmlml
r,
9 o
m m m
3.7 -
n
13
2.7
m
o
o
13
cl
4.3
9 n
9 []
50 ~
m
3.3
0.01
0.1
1
10
a
30 ~
[]
[]
'
0.001
m I mmmmml
0.001
0.01
Frequency (Hz)
0.1
1
Frequency (Hz)
7.0
[]
0
9
c~
6.0
Q
E
0
5.0
0
0.001
[]
mm
O
4.0
|
9
[]
i
a
i
11111
0.01
I
i
i
5~
a Ilia|
A
|
0.1
i
|
ii1,!
i
1
|
i
i
lmJJ
10
Frequency (Hz)
Figure 20. Storage (m), and loss (D) moduli vs. frequency, recorded at
intervals during stepwise cooling of 0.5% gellan with
50% sucrose and 35% corn syrup, at 0.5% strata. Scan rate
between steps 1o/ram., from [59] with permission.
10
P
h)
b
a
'M 6
0
I
3 '
-4
-2
0
2
log w at 90°C (Hz)
4
6
-2
0
2
4
6
log w at 100°C (Hz)
Figure 2 1. The results of time-temperature superposition of mechanical spectra for (a) gelldcosolute
mixtures at 9OoC, and (b) poly(n -0cty1 methacrylate) at 100°C, in both cases covering most
of the plateau zone and part of the glass-transition region, from [59] and [ 5 5 ] with permission.
8
43
100
1.5% gs
10
0.5% gns
0.75% gs
--....
0.3% g ~ . . . . . , , . . , , . . , . ~ ~ - -
i
0.1
280
l
300
i
I
320
i
I
340
i
I
360
Absolute temperature (T/K)
Figure 22.
The dependence of G'/T on absolute temperature (T) for gellan networks
with and without cosolute (50% sucrose + 20% corn syrup), denoted as
gs and gns respectively alongside the individual traces, from [60] with
permission.
Application of the WLF analysis to the high-methoxy pectin samples in the presence
of high levels of sugars (from 70% to 86%) successfully combined mechanical spectra
over 110 degrees centigrade (from 90~ to -20~ thus producing the master curve
of Figure 23a [61]. Clearly the plateau zone at lower frequencies gives way to a
sharp modulus development which covers most of the glass transition region.
The transformation is also evident in the tan 8 (G"/G') values which rise continuously
up to a log frequency of = 3 thus underlining the dominance of the viscous component
(Figure 23b). The decline in tan 8 values at higher frequencies is indicative of the
eventual advent of the glassy state where the solid-like response is again dominant,
with G' and G" traces crossing over and tan 5 passing through the value of 1 for the
second time at frequencies > 10 6 Hz. Straightforward calculations produced glass
transition temperatures for the high-solids gellan and pectin systems of-26~ and
-53~ respectively. Gratifyingly, estimation of the fractional free volume fig) at Tg
gave values (0.029 + 0.0003) almost identical to those observed for the great majonty
of diluted/undiluted synthetic polymers [55], organic liquids of low molecular weight
[57], and inorganic glasses [62] (fg = 0.026 + 0.005), thus demonstrating for the first
time the generality of the free volume approach for biological glasses.
44
8.5
1.08
[]
8.0 - a
0
Om
13 m
OI
7.5
6.5
G"
6.0
j
5.5
O
J
/
cV
7.0
0
b
1.06
1.04
1.02
OI
Bill
1.00
G'
G'
5.0
0.98
4.5
0.96
4.0
1
3.5
-3 -2 -1 0
1 2 3 4
Log (Frequency/Hz)
Figure 23.
5 6
i
i
I
I
I
I
-3 -2 -1 0
1
2
3
4
5
0.94
6
Log (Frequency/Hz)
Graphs illustrating the application of the method of reduced variables to
storage modulus (G'), loss modulus (G"), and tan ~ for the 1% pectin +
86% cosolute preparation, at the reference temperature of 90~ from
[61 ] with permission.
Having made a start in understanding the behaviour of high sugars biopolymer
preparations, from a commercial point of view the intention should be to generate
fundamental understanding for a wide range of food biopolymers in order to provide
substitutes for conventional ingredients in established technology, and to design novel,
appealing low-moisture products on a more rational basis. Fundamental research in
high-viscous, low-water content materials has long been neglected by food scientists
though the area has recently seen rapid growth mainly due to work on starch and
globular proteins [63]. Such work on biogums provides an opportunity to glean and
apply information from the realm of synthetic polymer science to the specialised
problems of food biopolymers. Today, the confectionery industry is still rather
conservative and somewhat resistant to new ideas, but there is a great deal of untapped
potential for research and hopefully, as a result, commercial applications.
For example, if research can scientifically demonstrate that technologically appealing
rubbery or glassy materials can be made using otherwise brittle aqueous gels of
polysaccharides like gellan, carrageenan or agarose, it will open up the possibility of
development of suitable substitutes for traditional ingredients (gelatin and pectin) in
the jam and confectionery industries.
45
L O W FAT SPREADABLE PRODUCTS
Besides the high solids systems of the preceding paragraph, biopolymers are used
increasingly in the manufacture of low fat spreads and soft cheeses. Creation of
textures similar to those of fat-based food products such as butter and margarine
requires processing of polysaccharides and proteins to generate particles of
comparable size to fat crystals. Starch and gelling maltodextrins in mixtures with
proteins (gelatin, milk and soya proteins) have been used to generate fat-like rheology
by cooling a fluid mixture of two components, with development of discrete, microsized particles in a continuous matrix then occurring due to spontaneous phase
separation between the two biopolymers [64]. Prediction and control of the final
properties of such products, however, is in general largely empirical. In this context,
the traditional technique of compression testing has been used with some success in
identifying satisfactory substitutes of fat in spreadable samples.
Compression testing is able to differentiate between the long range properties of a
hydrocolloid gel, a spreadable dispersion and a viscous solution (Figure 24; [65,66]).
100
HYDROCOLLOID GEL
15
10
ra~
m
O'm
~p
I
I
I
-
--.
- ~ ~ I - I- -
ti
~t
I
I
I
t
l ,..
I
I.
I
I
,
I
I
~m
~m
Ep
I
~~---
J
i ....
PLASTIC DISPERSION
.
.
VISCOUS SOLUTION
. . . .
0.5
2
strain
Figure 24.
Idealised force-deformation profiles from compression testing of a
hydrocolloid gel, a plastic dispersion, and a viscous solution, from
[65] with permission.
46
In the case of gels, the yield stress (Crm) is the point where the force goes through a
maximum value before its rapid decline at higher levels of compression. A minimum
point is also observed after failure (cri), since the stress will eventually rise due to the
dosing down of the stationary and moving plates. The deformation at maximum stress
is known as yield strain (em) and is related to the elastic properties of the network.
Overall, a sharp and early profile of breakdown is characteristic of a gel as opposed to
the smooth stress-strain trace of a viscous liquid (e.g. a thick yoghurt) that does not
show apparent signs of a yield point on the curve. Between the two extremes of gels
and viscous liquids, butter and margarine show ideal spreading properties (plastic
rheology) generating a shallow shoulder followed by a plateau (crJcr m = 0.96 to 1.0),
rather than a pointed peak under a constant compression rate (typically 2"/min). In the
case of low fat products, the lack of in-depth understanding of binary biopolymer
systems has resulted in the past in commercial products with an excessive gel-like
character [65]. Intelligent manipulation of biopolymer mixtures in terms of phase
continuity, kinetics of gelation/phase separation and solvent distribution between the
two networks, as outlined in the section of phase separated networks, allowed recently
the development of spreadable very low fat products. Thus a formulation of 2.3% milk
protein, 9.5% maltodextrin, 5% fibre (inulin) yields a O'p/O"m ratio of 0.96 and a
stress-strain profile upon compression identical to that of margarine at a fat content of
only 5.2% in the formulation [67].
ACKNOWLEDGEMENTS
The authors are grateful to their colleagues, Dr M.W.N. Hember and Professor
E.R. Morris for stimulating discussions and technical assistance in the preparation of
this manuscript.
REFERENCES
.
3.
.
Cheftal JC, Cuq JL, Loivent D. In: Fenemman OR, ed. Food Chemistry.
New York: Marcel Dekker Inc., 1985; 245-371.
Rees DA. Biochem J. 1972; 126: 257-273.
Holmes KC, Blow DM. In: Glick D, ed. Methods of Biochemical Analysis.
New York: John Wiley & Sons, 1966; 113-239.
Amott S, Mitra AK. In: Amott S, Rees DA, Morris ER, eds. Molecular
Biophysics of the Extracellular Matrix. Clifton NJ: The Humana Press Inc.,
1984; 41-67.
Ross-Murphy SB. In: Chan HW-S, ed. Biophysical Methods in Food Research.
Oxford: Blackwell Scientific Publications, 1984; 138-198.
Arridge RGC. In: Mechanics of Polymers. Oxford: Clarendon Press, 1975;
51-57.
Ferry JD. In: Viscoelastic Properties of Polymers. New York: John Wiley &
Sons, 1980; 177-223.
Rouse PE Jr. The Journal of Chemical Physics 1953; 21: 1272-1280.
Zimm BH. The Journal of Chemical Physics 1956; 24: 269-278.
47
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
Cox WP, Merz EH. J Polym Sci 1958; 28:619.
Cross MM. J Colloid Sci. 1965; 20: 417-437.
Morris ER. Carbohydrate Polymers 1990; 13: 85-96.
Graessley WW. J Chem Phys 1967; 47: 1942-1953.
De Gennes PG. The Journal of Chemical Physics 1971; 55:572-579.
Doi M, Edwards SF. J Chem Soc Faraday Trans. II 1978; 74: 1789, 1802.
Robinson G, Ross-Murphy SB, Morris ER. Carbohydrate Research 1982; 107:
17-32.
Goycoolea FM, Morris ER, Gidley MJ. Carbohydrated Polymers 1995; 27:
69-71.
Clark AH, Ross-Murphy SB. Adv. Polym. Sci. 1987; 83: 57-192.
Richardson RK, Ross-Murphy SB. Int. J. Biol. Macromol. 1987; 9: 257-264.
Barnes HA, Walters K. Rheological Acta 1985; 24: 323-326.
Moldenaers P, Mewis J, Berghmans H. In: Giesekus H, Hibberd FF, eds.
Progress and Trends in Rheology II. Darmstadt, Germany: Seinkopff Verlag,
1988; 420-421.
Richardson RK, Goycoolea FM. Carbohydrate Polymers 1994; 24: 223-225.
Yalpani M. In: Polysaccharides Syntheses, Modifications and
Structure/Property Relations. Amsterdam: Elsevier, 1988; 410-413.
Barnes HA, Hutton JF, Walters K. In: An Introduction to Rheology.
Amsterdam: Elsevier, 1989; 119-131.
Huggins ML. J. Amer. Chem. Soc.a 1942; 64: 2716-2718.
Kraemer EO. Ind. Eng. Chem. 1938; 30: 1200-1203.
Solomon OF, Gut~ IZ. Journal of Applied Polymer Science 1962; 24:
683-686.
Morris ER, Cutler AN, Ross-Murphy SB, Rees DA, Price J.
Carbohydrate Polymers 1981; 1: 5-21.
Kasapis S, Morris ER. In: Nishinari K, ed. Food Hydrocolloids: Structures,
Properties, and Functions. New York: Plenum Press, 1994; 97-103.
Morris VJ. In: Phillips C~, Wedlock DJ, Williams PA, eds. Gums and
Stabilisers for the Food Industry 3. London: Elsevier, 1986; 87-99.
Nielsen LE. In: Mechanical Properties of Polymers and Composites.
New York: Marcel Dekker Inc., 1974; 496-501.
Mitchell JR, Blanshard JM . Journal of Texture Studies 1976; 7:219-234.
Mitchell JR, Blanshard JMV. Journal of Texture Studies 1976; 7:341-351.
Stainsby G. Fd. Chem. 1980; 6: 3-14.
McKay JE, Stainsby G, Wilson EL. Carbohydrate Polymers 1985; 5:
223-236.
Whistler RL. In: Whistler RL, BeMiller JN, eds. Industrial GumsPolysaccharides and their Derivatives. San Diego: Academic Press, 1993;
309-339.
Imeson AP. In: Phillips GO, Wedlock DJ, Williams PA. eds. Gums and
Stabilisers for the Food Industry 2. Oxford: Pergamon Press, 1984; 189-199.
Lin CF. In: Graham HD, ed. Food Colloids. Westport Conn.: AVI Publishing
Co, 1977; 320-346.
Toft K. In: Phillips GO, Wedlock D J, Williams PA, eds. Prog. Fd. Nutr. Sci.
Oxford: Pergamon Press, 1982; 89-96.
Thom D, Dea ICM, Morns ER, Powell DA. In: Phillips GO, Wedlock DJ,
Williams PA, eds. Prog. Fd. Nutr. Sci. Oxford: Pergamon Press, 1982; 97-108.
48
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61
62.
63.
64.
65.
66.
67.
Dea ICM, Clark AH, McCleary BV. Carbohydrate Research 1986; 147:
275-294.
Kasapis S, Morris ER, Norton IT, Brown CRT. Carbohydrate Polymers 1993;
21:261-268.
Tolstoguzov VB. In: Mitchell JR, Ledward DA, eds. Functional Properties of
Food Macromolecules. London: Elsevier, 1986; 385-415.
Suchkov VV, Grinberg VY, Tolstoguzov VB. Carbohydrate Polymers 1981; 1:
39-53.
Takayanagi M, Harima H, Iwata Y. Mem. Fac. Eng. Kyusha Univ. 1963; 23:
1-13.
Arridge RC~. In: Mechanics of Polymers. Oxford: Clarendon Press, 1975;
97-103.
Clark AH, Richardson RK, Ross-Murphy SB, Stubbs JM. Macromolecules
1983; 16: 1367-1374.
Morris ER. Carbohydrate Polymers 1992; 17: 65-70.
Gordon M, Ross-Murphy SB. Pure Applied Chemistry 1975; 43: 1-26.
Clark, AH. In: Blanshard JM , Lillford P, eds. Food Structure and Behaviour.
London: Academic Press, 1987; 13-34.
Kasapis S. In: Harding SE, Hill SE, Mitchell JR, eds. Biopolymer Mixtures.
Nottingham: Nottingham University Press, 1995; 193-224.
Kasapis S, Alevisopoulos S, Abeysekera R, Manoj P, Chronakis IS,
Papageorgiou M. In: Phillips GO, Williams PA, Wedlock DJ, eds. Gums and
Stabilisers for the Food Industry 8. Oxford: IRL Press, 1996; in press.
Kalichevsky MT, Blanshard JMV, Marsh RDL. In: Blanshard, JMV, Lillford
PJ, eds. The Glassy State in Foods. Nottingham: Nottingham University Press,
1993; 133-156.
Lopes da Silva JA, Gonqalves MP, Rao MA. Carbohydrate Polymers 1994; 23:
77-87.
Ferry JD. In: Viscoelastic Properties of Polymers. New York: John Wiley &
Sons, 1980; 264-320.
Doolittle AK. Journal of Applied Physics 1951; 22: 1031-1035.
Williams ML, Landel RF, Ferry JD. Journal of the American Chemical Society
1955; 77: 3701-3706.
Arridge RGC. In: Mechanics of Polymers. Oxford: Clarendon Press, 1975;
24-50.
Papageorgiou M, Kasapis S, Richardson RK. Carbohydrate Polymers 1994; 25:
101-109.
Papageorgiou M, Kasapis S. Food Hydrocolloids 1995; 9:211-220.
A1-Ruqaie IM, Kasapis S, Richardson RK, Mitchell G. Polymer, submitted.
Shen MC, Eisenberg A. In: Reiss H, ed. Prog. Solid State Chem. Oxford:
Pergamon Press, 1966; 3:407-481.
Kalichevsky MT, Blanshard JMV. Carbohydrate Polymers 1992; 19: 271-278.
Goutefongea R, Semur J-P. French Patent 1986; FR 2 580 471 A1.
Chronakis IS, Kasapis S. Carbohydrate Polymers 1995; 24: 000.
Chronakis IS, Kasapis S. Lebensm.-Wiss. u. Technol. 1995; 28: 488-494.
Gupta BB, Kasapis S. European Patent 1995; 0 672 350 A2.
D. Wetzel and G. Charalambous (Editors)
Instrumental Methods in Food and Beverage Analysis
9 1998 Elsevier Science B.V. All rights reserved
49
Destructive
and
Non-Destructive
in S t a r c h
Analysis
Y.G.
Moharram,
O.R.
Abou-Samaha
Analytical
and M . H .
Bekheet
Food
Science
and
Technology
Department,
Agriculture, A l e x a n d r i a University, Alexandria,
I
-- I N T R O D U C T I O N
Methods
Faculty
Egypt.
of
Starch is
one of
many h y d r o c o l l o i d s or
water soluble
resins. It is available
from a variety of field
crops such
as corn, wheat,
rice, potato
and cassava or
tapioca.
It
represents the storage carbohydrate in these plants. [i].
Starch present
in a granular
form. It is
embedded in
protein matrix.
It has
various shapes and sizes
according
to the source of starch. [2].
Starch
granules
have
a heterogenous
structure.
It
constructs
from
an
amorphous
and
crystalline
zones.
Basically,
it contains
two
main
components, amylose
and
amylopectin
and
other
minor
constituents,
particularly
lipids and minerals. [3].
Amylose
is
a linear
chain
molecule
with a
limited
branching.
It forms f r o m ~ a n h d r o g l u c o s e units connected by
-1,4
linkages.
It
is
the
component
of
starch
which
responsible
about
the
characteristics
of
gelling,
the
p r e c i p i t a t i o n or
c r y s t a l l i z a t i o n and
a set-back
or retrog
radation. [4].
A m y l o p e c t i n is
a non
linear or branched
molecule. It
consists of repeating a n h y d r o g l u c o s e units connected by
-1,4
bonds but having
~-1,6
linkages
at selected sites
generating a branch
point.
It is a non gelling portion of
the starch and generally
contributes a pituitous or stringy
consistency
to food
products
because
of its
solubility.
[1,4].
The method of starch
isolation and the levels of
nonstarch components can affect the
physicochemical properties
of the starch.
Phosphate
is believed to
be an
important
factor in determing the
ganular strength, by forming crosslinkages, in isolated starches. [5].
G e n e r a l l y native starches
lack the extended
stability
required by processed
food.
Therefore, it
uses as dusting
and
flow
agents
for preparing
fruit
fillings,
gravies,
sauces,
and
confectionary.
Processed
food
need
pH,
viscosity stability processing t o l e r a n c e , t e x t u r a l properties
shelf
stability
and
good
surface
appearance.
These
characteristics
can be obtained
by m o d i f i c a t i o n the native
starch with the proper
method, acid hydrolysis or thinning,
oxidation
process,
cross-linking,
substitution,
pregelatinization and
cold water
(CWS). hydration in
cold water,
instant starch. [6.9].
The
complexity
of
starch
structure
requires
some
50
methods to
ensure its quality. Three
met hods are generally
used
viscosity profile
chemical anal~sis and determination
a new
non
of
~unctional
~ropert~es.
[i0]
~ c e ntly.
to estimate
destructive
tec.niques are
developed and usea
stress or
shear
the quality
of starch such
as controlled
rheometers.
Such procedures
need a smal 1 quantity of raw
materials
and give
more
accurate and
rapid results
when
compared with the destructive methods. The objective of this
paper
is
to
give
knowledge on
the
utilization
of such
techniques, destructive and
non destructive, in
estimating
the quality
of starch after subjecting
for food processing
and/or modification.
2-
Starch
Isolation
:
In particular, there
are r.) strict rules
for starch
production.
Also there
are no two starch plants
exactly a
like. Local conditions, raw materials, available capital and
preference influence
the design and operation
of the plant
more than theoretical data. [ii].
Generally, the
method of
starch isolation can
affect
both the
physicochemical properties
of the starch
and the
levels
of
non
starch components.
In
turn, the
latter
components
may also
influence
the starch
characteristics
indirectly. [12, 13].
Fig. (i) summarizes the process of m a n u f a c t u r i n g starch
from both cereals, and tuDers or roots.
Tuber
Cereals
Washing
Cleanning
Coarse milling
Coarse
Oil Extraction
Separating fibre from
gluten and starch
Separating the gluten
and starch
the starch
Storage, Packaging
and Transporting
(i)
screening
Settling
Degermination
Fig.
and peeling
Milling
Steeping
Drying
or Roots
Fine
screening
Settling
Centrifuge or vacuum
filtering
Drying
the starch
Storage, Packaging
and Transporting
The
process
of
manufacturing
cereals and tubers or roots.
starch
from
51
3-
Starch
Composition:
Most of the common
starches are c o m p o s e d of two
major
polymeric components,
amylose and a m y l o p e c t i n .
The amount
of amylose v a r i e s with the b o t a n i c a l source of starch. [14].
In 1984,
C o l o n n a and M e r c i e r
[15] added a
third c o m p o n e n t
known
as
intermediate
fraction.
It
did
not
fit
the
d s c r i p t i o n of either a m y l o s e
or a m y l o p e c t i n and ranged from
5 to 10% in most cereal starches.
3.1Amylose :
It
can
be
determined
by
potentiontiometric
titration
[16], a m p e r o m e t r i c
titration
[17],
spectrophotometric
determination
of
blue
colour
intensity
of iodine
complexes.
[18] and
the s o r p t i o n
of
Congo Red. [19].
Takeda et
al. [13] reported that
amylose is g e n e r a l l y
assumed
to be a linear polymer at ~ - 1 , 4 a n h y d r o g l u c o s e link
and
limited
amount
of
the b r a n c h i n g
long-chains.
The
molecular
w e i g h t of amylose is about 250,000.
It is varied
not only b e t w e e n
species but also
w i t h i n the same
species
according
to
the
maturity
stage.
It
has
an
unique
p r o p e r t i e s such as the a b i l i t y
to form complex with iodine,
organic a l c o h o l s
and/or acids to form
c l a t h r a t e or helical
inclusion
components.
[20].
The
results
of
different
r e s e a r c h e r s indicated that the
amylose content of maize and
rice
starches was ranged from 22 to
32% with an a v e r a g e of
27% and from 15 to 23% with an a v e r a g e of 18.5% r e s p e c t i v e l y
a c c o r d i n g to the v a r i e t y
and the e n v i r o n m e n t a l
conditions.
[14,
21-24].
Schoch
[25]
classified
the n o n - w a x y
rice
varieties
into
two
groups
according
to
their
amylose
content, I~dica
with 21-23% and J a p o n i c a
with a r e l a t i v e l y
lower amylose content.
Sweet
potato starch amylose content is s l i g h t l y higher
than that of cassava but less th~n of wheat, maize or potato
[26].
3.2Amylopectin:
It
is
c o m p o s e d of
~ - D glucose
linked p r i m a r i l y
by ~ - , i - 4
and b r a n c h e d
by
~ - i - 6 bonds.
[27].
It is
c o n s i d e r e d one of the largest
m o l e c u l e s found
in
nature with a m o l e c u l a r w e i g h t of about l0 s . [28]. Robin
et al. [29]
assumed that the
a m y l o p e c t i n forms from
three
chains;
A, composes of glucose linked b y ~ - l , 4 ,
B, c o n s i s t s
of glucose
linked by ~ - 1 , 4
and ~ - 1 , 6 ;
and
C, forms
from
glucose with ~ - 1 , 4 and K - I , 6 linkages plus a reducing group.
The p o r t i o n
of the ~ - 1 , 6
linkage in
a mylopectin
is 4-5%
[30].
3.3-
Minor
components:
3.3.1Fat content:
The old
results of Taylor and
Nelson [31] r e p o r t e d that there was a r e l a t i v e l y
low a m o u n t
of fatty s u b s t a n c e s in c o m m e r c i a l starches.
They found that
maize,
rice, sago,
cassava and
potato starches
contained
0.61,
0.83, 0.ii,
0.12 and
0.04% of
lipids r e s p e c t i v e l y .
A c c o r d i n g to Gracza [32], the lipids content
of rice starch
52
o b t a i n e d from J a p o n i c a v a r i e t i e s
was higher (0.4-0.8%) than
that p r o d u c e d
from Indica ty~es
(0 i-0 3%).
In
Ir:~, AIBayatl and Lorenz [33] found
hat the lipids content
rice
starch ranged from 0.II to 0.14%.
In maize
starch, the range
of the lipids
c o n t e n t was
0.2 to 0.92%
a c c o r d i n g to
the 9 e n o t y p e and
the m e t h o d
of
determination.
[34-38].
It was v a r i e d from 0.05 to 0.6% in
sweet potato starch. [39].
3.3.2Protein:
The p r o t e i n
content
of n o n - w a x y
rice
and maize starches was
varied from 0.24
to 1.93% and
from 0.35 to 1.25%
respectively. [34, 36-38, 40, 41].
The
data of A i - B a y a t i
and Lorenz [33]
showed that the
protein
was p r e s e n t in trace amounts in Iraqi rice starches.
On the
other hand, W a n k h e d e and Umadevi [35] found that rice starch
c o n t a i n e d 2% protein.
In case of sweet potato
starch, the
level of crude protein was ranged from 0 . 0 2 - 0 . 5 4 % [39].
3.3.3Ash:
The
ash
content of
rice
starch was
varied between
0.38 and 0.78% [42]. The results of W a n k h e d e
and
Umadevi [35]
indicated that
the ash
content of
rice
starch did not increase than 0.32%.
It ranged
from 0.09 to
0.54% a c c o r d i n g
to the m e t h o d
used for
its p r o d u c t i o n
as
well
as v a r i e t y of rice
[36-38].
Sweet
potato starch had
0.008 to 1.3% ash. [39].
G e n e r a l l y , total c a r b o h y d r a t e s are the
main c o m p o n e n t s
of both native
and m o d i f i e d sta:ches.
It r e p r e s e n t s
more
than
98.5%
in both
maize and
rice
starches.
The other
components,
protein, fat,
and
ash are
minor and
usually
p r e s e n t in bound form
with starch molecules.
Their levels
depend
upon
the
source
of
starch
and
the
degree
of
refinement. [38, 43].
Sweet potato starches contain amount
of p h o s p h o r u s (8-28
mg/100 g) similar
to those in
cassava
and less than that
in Irish potato.
Amylose
c o n t a i n s less
phosphate
than the a m y l o p e c t i n
[44].
It
is b e l i e v e d that
phosphate
plays
an
important
factor
in
d e t e r m i n i n g the
granular
strength,
by forming
c r o s s - l i n k a g e s , i n starches.
[5].
4-
Starch
4.1-
Analysis
Physical
:
Methods:
4.1.1Size and
shape:
The
size
of the
starch
g r a n u l e s may be
e s t i m a t e d by the
rate of s e d i m e n t a t i o n
by
the
use
an
instrument
such
as
Coulter
counter
or
by
microscopic
analysis.
Particle
size
is
one
of
the
characteristics
that most
m a r k e d l y affects
the functional
properties
of
starch
granules.
Smaller
granules
are
reported to
have both high s o l u b i l i t y
and water a b s o r p t i o n
capacity
[45, 46].
Small g r a n u l e s of wheat starch (1-8 Um)
are more
d i g e s t i b l e , have
a lower amylose
content, iodine
affinity,
and
molecular
weight,
higher
gelatinization
53
t e m p e r a t u r e range;
water b i n d i n g
capacity
, viscosity and
more h y g r o s c o p i c than large (25-35 Um) ones. [47-49].
Starch g r a n u l e s
from d i f f e r e n t plant s o u r c e s
showed a
wide v a r i e t y in shape and size.
The size of starch g r a n u l e
is
u s u a l l y e x p r e s s e d as
a range
or as
an a v e r a g e
of the
length of the longest axis in m i c r o n s [14, 50]. The shape of
starch
g r a n u l e s of
the
plants differ
a c c o r d i n g to
their
growing
climatic
conditions.
The
damp
conditions
give
starch
granules
large
in
size,
regular
in
shape,
and
c o n s i s t i n g of c o n c e n t r i c layers around a dark spot known
as
the
hilum.
This
h i l u m does
l,ot always
distinguishable,
e s p e c i a l l y in v e r y small granules. [14, 51-53].
The results of
Waldt and Kehoe
[54] showed that
rice
starch g r a n u l e s are polygonal in shape, small in size(3-8 u)
and s o m e t i m e s
a d h e r e d together
to form the
clusters shape
[48].
While the maize starch g r a n u l e s appear in two shapes,
round
in floury portion, and p o l y g o n a l in the h o r n y part of
the e n d o s p e r m [55], with a size range of 5-25 u [20].
The
size of
starch
g r a n u l e s of
sweet potato
starch
(10.2-21.5 Um) was found
to be similar to those
of c a s s a v a
and corn but
are smaller
than those of
potato w h i c h
also
have
a
larger range
of g r a n u l a r
size
[56].
A negative
correlation
was
observed
between
particle
size
and
susceptibility
to
~(.-amylase and
acid d e g r a d a t i o n . J57].
A b o u - S a m a h a [43] using Scanning E l e c t r o n M i c r o s c o p y (SEM) to
get i n f o r m a t i o n about t h e ~ s t r u c t u r a l
detail, e s p e c i a l l y the
surface topography, of both maize and rice
starches.
Figs.
(2-8) illustrate
the influence of thin acid
and moist heat
modification
on
the
structure
of these
starches.
The
granule
of maize
starch
(Fig. 2)
exhibits the
polygonal
shape.
The flinty part
of this
g r a n u l e is an
angular in
outline.
The g r a n u l e s are n e a r l y
u n i f o r m in size 6.7-16.8
um,
and
no
concentric
rings
found
in
it
under
light
microscope.
It
has a fairly
regular p o l y h e d r a l shape
and
with well m a r k e d central hilum.
However, the surface of the
g r a n u l e s is a p p e a r e d b a s i c a l l y smooth.
Some of the g r a n u l e s
have
an i d e n t a t i o n s on their surface due to the m e t h o d used
for isolating of starch. Fuwa et al. [58] r e p o r t e d that, the
small p i n - h o l e s w h i c h noticed on the surface of maize starch
are irregular and resemble the surface of a golf-ball.
This
m i g h t be due to
enzyme attack during long s t e e p i n g
used in
isolation
of the
starch.
In
case
of rice,
the
starch
g r a n u l e s (Fig. 2) are smaller in size c o m p a r i n g with that of
maize.
It is m o s t l y ranged
from 3.3 to
8.3 um
in their
size.
The g r a n u l e s are more or less p o l y g o n a l in shape with
an angular outline. It is d i f f i c u l t to identify the
central
hilum
or stria of the granule under light m i c r o s c o p e .
Some
of the g r a n u l e s are
a d h e r e d together to form a
compound or
clusters shape.
These results are
in a g r e e m e n t with
that
reported by
Waldt and
kenoe
[54].
As in
case of
maize
starch, the rice
starch g r a n u l e s were found to
have smooth
surface with
small indentations.
Nearly
slight changes in
the
size
and
shape of
the
granules
of
maize and
rice
starches
were noticed after
treating, with either
1 or 2%
54
Fig.
(3)
Scanning elecT.ron micrograpt~s of acid modified maize starch
(Acid concentrat;on 1%, 2-5x2000x) a=0.75 hr, b=1.5 hrs, c=6 hrs,
d= 24 hrs.
55
HCl
for 0.75 to 24
hrs. (Figs. 3,
4, 6 and 7).
The acid
m o d i f i c a t i o n caused
a slight
hydrolysis of small
parts of
each of superposed concentric layers of the granules of both
maize
and rice starches.
This may be
removed por~lons os
the less crystalline or
less dense starch molecules outside
the granules.
Therefore, the surface of the granules became
slightly
rough.
These changes
were more
pronounced with
increasing the acid concentration
and extending the time of
m o d i f i c a t i o n especially in rice starch.
On
other hand, the
moist
heat treatment
did
not
affect
the shape
and
the
surface characteristics
of the
granules of both
maize and
rice starches (Figs. 5 and 6). Only slight increase
in the
size os the granules of both maize and rice starches, by 2.0
and
3.1% respectively, was observed,
such increase may be
due
to the swelling of
the starch granules.
Because this
swelling did not destruct the outer layers, the granules are
kept its smooth
surface.
Exten0ing the
moist heating time
from 5 to 21 hrs did not affect ~he ultra structure of maize
and rice starches.
Bekheet [59]
studied the effect of
food processing in
the ultra-structure of wheat starch.
The SEM micrographs of
native,
steeped and
boiled steeped
wheat starch
(Fig. 9)
indicate that
the granules
are biconvex discs
with fairly
regular
circular
outlines,
and f r e e
from
polarization
crosses.
The size of native starches are ranged from 2 to 8
um for small granules and, from
25 to 55 mu. for large ones.
When the starch granules soaked in water
and diluted alkali
at
room
temp.,
they
absorbed
solution,
swelled,
and
increased 22 and 55
times in size respectively without
any
changes in
their crystalline
nature.
The boiling
of the
granules
for
5 min.
before soaking
increased
from
the
solvent absorbing,
swelling and
size
enlargement.
These
changes
are more noticeable in dilute alkali than in water.
Also, the
lye treated starch granules had a
rough surfaces
and more flattened than
those treated with water.
This is
may
be due
to
the influence
of
diluted lye solution
on
solubilization some
of the protein
matrix surrounding
the
starch
granules.
As
shown
from
Fig.
(i0)
the
starch
granules lost their birefringence, crystallinity
and became
nearly formless sacs having complex puckered structure after
cooking.
This can
be contributed to the influence
of heat
on gelatinization
of
starch which
causes
distorting
the
crystalline
structure
and contraction
of
the dissociated
molecules to
a random
coil conformation.
Further heating
and
hydration, as show
in Fig.
(ii), caused
weakness for
this structure and produced a sol.
4.1.2Density:
The range of the density (g/cc) of
some common starches
was 1.49-1.517, 1.504-1.521,
1.479-18
and
1.5-1.528
for maize,
tapioca,
waxy
maize and
wheat
respectively. [60].
Generally,
the utilization of
an air
compression
pycnometer
for estimating
the
starch density
gave lower
value than xylene displacement
method.
Hussein
et al. [61] found that the density
of rice starch was 1.472
55
Fig.
(5) Scanning electron micrographs of moist heated modified maize
starch (2-5x2000x). a=5 hrs, b=10 hrs, c=15 hrs, d=21 hrs.
57
Fig.
(6) Scanning electron micrographs of acid modified rice starch (Acid
concentration 1%,2-5X2000x). a=0.75 hr, b=1.5 hrs, c=6 hrs, d=24 hrs.
Fig. (7)
Scanning electron micrographs of acid modified rice starch (Acid
concentration 2%, 2-5x2000x). a= 0.75 hr, b=1.5 hrs, c-~6|lrs, d=.24 hrs.
58
"O
.~.
O
E
II
U")
L_
*.-.
II
O
.~
O
~
E
O
II
O
E
~.
•
O
O
o
,
x
c=
O
~"
O
o')
O
CO
~
59
a
native
Water
Boiled
b
d
and water steeped
Fig
c
Alkali
steeped
,
Boiled
e
steeped
and alkali
steeped
(9)"
Scanning electron micrographs of native
steeped and boiled steeped wheat starch
60
Fig.
(i0) ." Scanning electron micrographs
of cooked wheat starch
6!
Fig.
(ii)
"
Scanning
electron
mechanically
dried
micrographs
wheat
starch
of
sun
and
62
and
1.439
g/cc
when
measured
by
burette
and
by
the
v o l u m e t r i c flask m e t h o d s r e s p e c t i v e l y .
EI-Refai [62] showed
that the d e n s i t y of the E g y p t i a n rice starch was 1 . 4 2 8 - 1 . 4 9 2
g/cc.
Munk
et
al.
[24]stated
that
rice
starch
had
r e l a t i v e l y higher d e n s i t y (1.518 g/cc) than maize one (1.514
g/cc).
According
to
Abou-Samaha
[43]
modification
process
either with acid
at 1 and 2% and/or by m o i s t heat t r e a t m e n t
caused a slight
increase in
the d e n s i t y of
both rice
and
maize
starches.
Generally,
the d e n s i t y
of m a i z e
starch
(1.635-1.656 g/c.c) was r e l a t i v e l y
higher than that of rice
(1.432-1.464 g/c.c) one.
4.1.3Molecular
weight
of c o n s t i t u e n t
polymers
Both the
low angle
Laser Light S c a t t e r i n g
T e c h n i q u e s [63]
and the
u l t r a c e n t r i f u g a t i o n [64]
can be used
to d e t e r m i n e
the
average
molecular
weights
of
starch
constituent
polymers.
The latter
can also be e s t i m a t e d from
the total
c a r b o h y d r a t e content and the
number of r e d u c i n g end groups.
Photometry,
periodate
oxidation,
polarimetry,
osmometry,
radiometry
and
enzyme
analysis
[65 ] besides
the
most
sensitive
methods
which
involves
the
reduction
of
f e r r i c y a n i d e ions [66] can be u t i l i z e d to m e a s u r e the number
of r e d u c i n g end groups.
Amylopectin
consists
of three
chain
types; C - c h a i n s
which have reducing
ends, B - c h a i n s w h i c h are
linked to two
or
more other chains, and A - c h a i n s linked to only one other
chain by their reducing ends [67]. The d i s t r i b u t i o n of chain
lengths can
also be d e t e r m i n e d
by gel f i l t e r a t i o n
or high
p e r f o r m a n c e liquid c h r o m a t o g r a p h y ( H P L C ) f o l l o w i n g d e b r a n c h i n g
with isoamylase [68, 69].
Takeda et al. [44] c o n c l u d e d that
sweet
potato has a higher p r o p o r t i o n
of A - c h a i n s and short
B-chains
than
has
potato.
Also,
sweet
potato
amylose
appears to have more b r a n c h e s per a m y l o s e m o l e c u l e than that
from cassava, potato, wheat or m6ize, and a higher m o l e c u l a r
weight than maize, wheat and
cassava but less than
potato.
This was the
reason for the
lower r e t r o g r a d a t i o n of
sweet
potato amylose.
The
degree of p o l y m e r i z a t i o n and b r a n c h i n g
have s u b s t a n t i a l effect on the p h y s i c o c h e m i c a l p r o p e r t i e s of
a m y l o s e and a m y l o p e c t i n [70].
4.1.4Crystallinity
: Zaslow [71]
s u g g e s t e d the
use of the X-ray d i f f r a c t o n to provide i n f o r m a t i o n about the
orientation
and
crystallinity
of
the
starch.
X-ray
d i f f r a c t i o n technique
can be used to
differentiate between
the native starches, to
detect the changes in c r y s t a l l i n i t y
during
the
physical
and/or
chemical
modification
[72].
Three d i f f e r e n t
X-ray p a t t e r n s
known as
(A), (B)
and (C)
were obtained for plant starches. Therefore, the c r y s t a l l i n e
nature of a starch
can be defined by the
X-ray d i f f r a c t i o n
peaks. [70, 73]. The
c r y s t a l l i n e p a t t e r n of cereals, tubers
and
both roots
and seeds
starches have
A, B and
C shape
respectively.
The (A)
shape shows
three strong
peaks at
5.8,
5.2 and 3.8 A n g e s t r o m s
(A~), w h e r e a s (B)
shape has a
63
m e d i u m peaks at 1.58-16
A ~ in a d d i t i o n to three
peaks, two
at 4 and
3.7 A ~ and
strong one
at
5.16
A~
The
(C)
crystalline
p a t t e r n has
the similar
peaks of
(A) p a t t e r n
beside m e d i u m
to strong one at
16 A ~ [70].
Hizukuri [74]
demonstrated
that
mixtures
of
Aand
B-type
starches
p r o d u c e d i n t e r m e d i a t e p a t t e r n s (c-type).
c-type p a t t e r n can
be further d i v i d e d into
C., C~ and Cb d e p e n d i n g
on w h e t h e r
the p a t t e r n is closer to
A or te B. Levels of c r y s t a l l i n i t y
in
granular
starch can
be
determined
by s e p a r a t i n g
and
i n t e g r a t i n g the areas under the c r F s t a l l i n X-ray d i f f r a c t i o n
peaks [70].
Type
B
starches
tend
to
have
lower
levels
of
c r y s t a l l i n i t y (15.28%)and lower g e l a t i n i z a t i o n t e m p e r a t u r e s .
Type A starches tend to have higher levels
of c r y s t a l l i n i t y
(33-45%) and higher g e l a t i n i z a t i o n
temperature.
The latter
temperatures
increase
with i n c r e a s i n g
amylose
c o n t e n t of
type A- s t a r c h e s [70].
Sweet potato starch has a v a r i e t y x - r a y p a t t e r n b e t w e e n
C and A,
in c o n t r a s t to cereal
starches such as
w h e a t and
corn
which
have Atype and
potato
which have
a B-type
pattern. [75, 76].
The
heating
o f starch
in
excess of
water
causes a
higher loss of c r y s t a l l i n i t y
c o m p a r i n g with that happens in
less
of water
[77].
on
other
hand, soaking
of
starch
g r a n u l e s in water at
room t e m p e r a t u r e has no effect
in its
c r y s t a l l i n e structure, wh{le
using a
dilute lye,
dimethyl
s u l f o x i d e and c o n c e n t r a t e d urea as a s t e e p i n g m e d i u m lead to
a loss in b i r e f r i n g e n c e of starch at room temperature. [78].
Nearly
the same results were reported by Bekheet [59].
She
found that s t e e p i n g in water and d i l u t e d a l k a l i led to slight
changes in the c r y s t a l l i n e
s t r u c t u r e of wheat starch.
The
same results were o b t a i n e d when the wheat grain, were boiled
for 5 min. before soaking in water.
While,
a m a r k e d damage
in the
crystalline structure
of whear starch
was o b s e r v e d
after, soaking
the 5 min.
boiled grains in
d i l u t e d alkali
(Fig. 12). She a t t r i b u t e d their results to the r e a r r a n g e m e n t
of
chains of amylose fraction.
Cooking and drying of water
and Lye soaked and
u n s o a k e d grains a f f e c t e d the c r y s t a l l i n e
pattern
of
wheat starch
(Figs. 13
a,b,c
and 14).
This
influence was v a r i e d a c c o r d i n g to cooking m e t h o d and soaking
medium.
G e n e r a l l y , the changes in the c r y s t a l l i n e s t r u c t u r e
was
r e l a t i v e l y slight when cooking was c a r r i e d out in water
for u n s o a k e d and water soaked grains c o m p a r i n g with p r e s s u r e
and steam cooking m e t h o d s
e s p e c i a l l y for d i l u t e d lye soaked
grains [59].
Abou-Samaha
[43]
used
the
general
electric
X-ray
g e n e r a t o r to
study the changes
in the o r i e n t a t i o n
and the
c r y s t a l l i n i t y of maize and rice strahces before and after 24
hrs
of acid,
5 and 21
hrs
of moist
heat
modification.
Results in Figs. (15)
and (16) show that both
native maize
and rice starches
had (A) c r y s t a l l i n e pattern.
This shape
was
c h a r a c t e r i z e d with
three strong
peaks and 2-3 weak peaks.
The i n t e r p l a n a r
distances
of
these
peaks
were
differed
s l i g h t l y b e t w e e n the two types of starch.
Acid m o d i f i c a t i o n
64
AkKALi SLEEP
wATER
HAlIVE
~ "
,
I~ESSUI~E
WATER
FIG (13 "a ) I-RAY 01FFAJCTIONOF COOKEDWttEA! S[ARCH
J
65
PRESSUR~
j
51EAI4
J'~ s ~ ~
j
~"
~A,.~/
FICa | I~- b ) J -nJi'~ O|FT~ACTIONOF W&TER STEEPED CO0~EDWHEAT STARCH
I~ESSURE
STEAM
FIO |l)-C) X-P~Y OIFFRACTION OF ALKALI SlrEEPf[O COOKEO WHEAT STARCH
66
MECHANIC AL Ol:~I1'~
.
~ , nattve
I~, ~
~
24 I ~ , 1%.
c - I~c~d mo~f, e0. 24 hrs 2%
A
o
e= moist heateci 21 his
.~
\
,,j
Q
~o
Fig.
( 1 5 ) x-~o,. .... og . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
maize sla~cn
,--o~,::
Fi~.~
.
( ~ 6 )
I
~s
jo
:~::r,cli,~
2~
,It:,
~,
I~
~" ~" ,,4ro~l
.
to
x - fay o'~'actog'3m~ o' na~,ve =.oa aria too,st neat mOO,heO .,cu
s:a,ct~
67
either with 1 and 2% HCI for 24 hrs caused slight changes in
the p o s i t i o n of
the three
main strong peaks
of m a i z e
and
rice
starches.
While, th e m o i s t heat t r e a t m e n t
led to an
increase
in
the
numbers
of
the strong
peaks
to
five.
Therefore
the
crystallin e pattern
of
these
s t a r c h e s was
c o n v e r t e d from A
to AC sh ape.
On other hand,
Biliaderis
et
al [79] stated that th e cereal s t a r c h e s retained their A
crystallinity
type
aft er
acid
modification.
Acid
m o d i f i c a t i o n caused cleava ge of a
few starch chains in
the
amorphous
regions
and
t he
water
molecules
replace
the
obtained crystalline
cavities.
These changes gave
only a
sharp peaks. [79, 80].
W h i s t l e r and Paschall [55] r e p o r t e d
that
h e a t i n g of
starches in
the p r e s e n c e
of i n s u f f i c i e n t
m o i s t u r e gave
a
compact c r y s t a l l i n e
structure
and
water
swelling
r e s i s t a n c e granules.
Zobel
[70] found
that the
m o d i f i c a t i o n of
maize, rice
and wheat starches
with m o i s t
heat t r e a t m e n t gave c r y s t a l l i n e p a t t e r n having V shape.
The
same author
s u g g e s t e d that
the c o m p l e x i n g of
a m y l o s e with
fatty acids is behined this change.
4.1.5-
Gelatinization
:
4.1.5.1Gelatinization
Temperature:
The term
gelatinization
is
used
to
describe
the
swelling
and
hydration
of
starch
granules
or the
melting
of
starch
c r y s t a l l i t e s [70].
In m a n y fooa p r o c e s s i n g o p e r a t i o n s such
as baking of bread and cakes, e x t r u s i o n
of cereals p r o d u c t s
thickening
and
gelling
of
sauces and
pie
filling,
the
g e l a t i n i z a t i o n of starch is r e q u i r e d (81,82)
Despite
of the
fact
that starch
m o l e c u l e is
highly
h y d r o x y l a t e d and
very h y d r o p h i l i c ,
but still
insoluble in
cold water.
This is due to the p r e s e n c e and d i s t r i b u t i o n of
starch
g r a n u l e s in
network s t r u c t u r e
[83].
A c c o r d i n g to
Kerr
[14] heating
an a q u e o u s
s u s p e n s i o n of
starch caused
changes include the f o l l o w i n g three phases:
i
--
_
Slowly
and r e v e r s i b l e water
absorption.
Through
this
stage a limit swelling was o c c u r r e d w i t h o u t increase in
v i s c o s i t y and/or change in the g r a n u l e s shape.
With the i n c r e a s i n g the t e m p e r a t u r e of the starch
suspension
to about 65 ~
In this stage, the g r a n u l e s
are:
a) Sudenly swell or increase many times in size due to
a b s o r b i n g of a great amount of water.
b) R a p i d l y losing its b i r e f r i n g e n c e and that u s u a l l y
a s s o c i a t e d with an increase in its viscosity.
_
With
i n c r e a s i n g the
t e m p e r a t u r e , the
starch g r a n u l e s
become almost formless sacs, and s u b s t a n t i a l p o r t i o n of
its soluble c o m p o n e n t is leached out.
These
changes
give
the
swelling
starch
granules
the
68
ability
to form
a gel after
cooling.
Also, Donovan
~841
summarized
the
stimultaneous
changes
occurred
ur ng gelat nization process as follows;
a) Uptake of heat.
b) Loss of crystallinity associated with a loss of
birefringence and X-ray diffraction patterns.
c) Hydration of starch followed by granule swelling and
increasing in suspension ciscosity.
d) Lowering in relaxation time of water molecules as
measured by pulsed nuclear magnetic resonance.
Gelatinization may
happen at room temperature by using
saturated
solutions
of
certain
salts,
such
as
calcium
chloride
or by
alkalies,
such as
caustic soda
solution.
These agents break the hydrogen bonds and allow hydration of
liberated
hydroxyl groups.
[42].
Cooking
of starch for
70-90
sec
under
30
pounds of
pressure
caused
complete
gelatinization
for
starch
and
gave
grains
without
discoloration. [85].
According to Neufeld et al. [86], the
main purpose
of soaking is to
introduce rapidly sufficient
amount
of moisture
into the
wheat kernel
to help
in the
gelatinization
of
its
starch
during
cooking.
Moisture
content of
37 to 40% after soaking was found to be adequate
for
complete gelatinization
when
cooking was
done at
70
p.s.i., but
moisture con~ent of
41% or more
was necessary
with cooking at 20 p.s.i.
The data
of Smith [87] indicated
that soaking of
wheat grains to
reach around 45%
moisture
reduce the cooking time necessary for complete gelanization.
Generally, the g e l a t i n i z a t i o n temperature is controlled
not only
by the water content
but also by the
presence of
salts, sugars
and other small
molecules.
A major
factor
controlling
swelling
is
the
strength
of
the
internal
structure
of
the
granules.
The
stronger
the
internal
molecular structure, the higher the temperature required for
gelatinization.
[88].
Granule
size,
amylose
content,
molecular
weight
crystalline pattern and the internal organization all affect
gelatinization. [89].
The gelatinization temperature appears to be greater in
sweet potato starch than in that of cassava, potato or wheat
but similar to that of rice. [26].
4.1.5.2Gelatinization
Transition:
Starch
gelatinization may
be described either in
structural terms
as a loss of
macromolecular organization and order or
as a
swelling process which also has major theological effects. A
number
of
methods,
have
been
described
to
study
the
gelatinization transition of starch.
The
irreversible loss
of birefringence, which can be observed conveniently using a
Kofler hot-stage
microscope. [90], is a well known method.
These events may also
be followed by observing the
loss of
X-ray
crystallinity.
[91].
Differential
scanning
calorimetry (DSC)
can also be u~ed
since gelatinization is
69
an e n d o t h e r m i c process r e f l e c t i n g the change of order w i t h i n
the granule.
The
t e m p e r a t u r e s a s s o c i a t e d with
the onset,
peak and end point
of the e n d o t h e r m are noted and the total
enthalpy
change
~ H calculated
from
the area
under the
t h e r m o g r a m peak.
These
p a r a m e t e r s v a r y with
variety, and
also with
the e n v i r o n m e n t a l c o n d i t i o n s in w h i c h the plants
are grown. [92].
Also g e l a t i n i z a t i o n e n d o t h e r m only a p p e a r s
when water in
excess of
4 m o l e c u l e s per
g l u c o s e unit
was
present. [93].
The loss of order w i t h i n the g r a n u l e p e r m i t s
s t a i n i n g r e a c t i o n s to take place with dyes such as Congo Red
[94].
The i n t e r a c t i o n s with iodine and with enzymes such as
glucoamylase
have
also
been
suggested
to
monitor
g e l a t i n i z a t i o n . [95, 96].
The
study of
W o o t t o n and
B a m u n u a r a c h c h i [97]
on the
g e l a t i n i z a t i o n behavior of u n m o d i f i e d starch
from d i f f e r e n t
botanical
sources and
different
types of
modified
wheat
starch
by DSC
indicated
that the
values
of the
started
gelatinization temperature
(To) was
50, 70, and
57~
the
peak
t e m p e r a t u r e (Tp)
was 68.
78 and
72~
and
the final
gelatinization
t e m p e r a t u r e (Tc)
was
86, 89
and 87~
for
wheat,
maize and potato starches
respectively.
The
( ~ H)
values
was
4.7, 4.3
and 6.6
cal/g
for wheat,
maize and
potato starches respectively.
The g e l a t i n i z a t i o n e n d o t h e r m
of
some types of starches
was o b s e r v e d by
K u j i m i y a et al.
[98].
They stated that
the peak t e m p e r a t u r e
(Tp) was 65,
73,
72 and
65~
for
potato, w a x y
maize, m a i z e
and wheat
starches
respectively.
The
enthalpy
change
after
the
g e l a t i n i z a t i o n was 4.5, 3.8, 3.2 and 4.3 cal./g for
potato,
waxy
maize, maize
and
wheat starches
respectively.
The
t e m p e r a t u r e To, Tp, Tc and the e n t h a l p y of the e n d o t h e r m s of
8-rice starches ranged from 55 to 67~C, from 61.7 to 74.2~C,
from 103 to I04~C and from 2.68 to 3.23 cal./g, r e s p e c t i v e l y
[99].
The values of
peak t e m p e r a t u r e (Tp)
was 71.6~C and
63.5~
the e n t h a l p h y ( ~ H) was
2.42 and 1.94
cal./g for
maize and wheat starches
respectively. (i00].
The s w e l l i n g
of
w h e a t starch started
at 45-50~C and
c o n t i n u e d to 85~
The loss
of b i r e f r i n g e n c e and the
r e d u c t i o n in g e l a t i n i z a tion
e n t h a l p y were
mainly attributed
to the
dissociation
crystalline
clusters of
starch at
50-55~
The residual
reduction
of
enthalpy
could
be
contributed
to
the
dissociation
of the
double helices
of starch
g r a n u l e s at
55-60~
[i01].
A c c o r d i n g to
A b o u - S a m a h [43]
and as shown
from Figs.
(17 & 18) both
native and m o d i f i e d maize and
rice starches
had only a single
e n d o t h e r m t r a n s i t i o n c o r r e s p o n d i n g to the
g e l a t i n i z a t i o n process.
The
t e m p e r a t u r e s , To, Tp,
Tc and
the e n t h a l p y ( ~ H) of this e n d o t h e r m were d i f f e r e d a c c o r d i n g
to the
source of starch,
m e t h o d and time
of m o d i f i c a t i o n .
Generally,
both
native
and
modified
rice
starches
had
s i g n i f i c a n t l y higher t r a n s i t i o n t e m p e r a t u r e s , To, Tp, Tc and
lower ( ~ H) than maize one.
The m o d i f i c a t i o n process either
with
acid
and/or
moist
heat
treatment
shifted
the
temperatures
of
gelatinization
endotherm
to
relatively
higher
degrees.
This effect was more p r o n o u n c e a in case of
70
#'rod moddiud
b - 0.,b hr
c, I 5 his
,t= 3 hrs
o- 6 hrs
I= t 2 hrs
0.75 h~
2%
J= l.b hr
K= 3 rVS
L= 6 his
m. 12 hrs
n. 24 hrs
p , 5hr$
r,. t0 hrS
t- 21 hrs
|e
10
80
tO
v)o
jtR
rio
Temperature oc
Fig.
( 1 7 ) osc
thermograms el nalive, aod and me,st heat modified ma,ze
starch. (march: water rat0o. 1:4.
P,tting rate t0~
I'
O- 0 75 hr
p.md meddled
c,, 1 5 his
d= 3 hrS
e= 6 his
1.1, 12 hrS
n,, 24 his
,F
%
n,., 0 75 hr
,h, 1.5 hr
K,, 3 his
r
L- 6 hr5
m= 12 hrs
g= 24 hrs
p - 5hrS
r= 10 his
S,, 1,5 hrs
t,, 21 hrs
~
Fig.
fO
(18)
10
el)
gO
',,OO
t~O
Temperature Oc
D$C thermograms of natwe, acid and mOiSt heat moddJeO r,ce
s t a / ~ . (starch. water rat,o. 1 4. heating rate 10Oc/mm)
2%
7l
moist
heat than acid m o d i f i c a t i o n ,
the e n t h a l p y ( ~ H) was
markedly
decreased
with
increasing
the
time
and
c o n c e n t r a t i o n of
acid m o d i f i c a t i o n
process.
While, m o i s t
heat m o d i f i c a t i o n
led to an increase
in~H.
A polynomial
relationship
with
a
second
order e q u a t i o n
was
observed
b e t w e e n each of To, Tp, Tc, ~ H; and time of acid and
moist
heat
m o d i f i c a t i o n process.
Acid
t r e a t m e n t of
the starch
selectively
c l e a v e s the
amorphous
regions
of the
starch
granules.
As a result of p r o g r e s s i v e h y d r o l y s i s , i n c r e a s i n g
either time
and/or acid c o n c e n t r a t i o n , the
non c r y s t a l l i n e
parts of the g r a n u l e s were increased, and a d e c r e a s e in the
H of the o b t a i n e d e n d o t h e r m was noticed.
In case of m o i s t
heat
treatment
a
reduction
in
the
intramolecular
and
spherulitic
intermolecuiar
ordering
may
be
occurred.
According
to
Whistler
and
Easchall
[55]
moist
heat
modification
gained
starches
the
compact
crystalline
structure.
The results of D o n o v a n et al. [102] showed that
the range of the
g e l a t i n i z a t i o n t e m p e r a t u r e of m o i s t heated
s t a r c h e s was
broadened.
Also, two
peaks of g e l a t i n i z a t i o n
endotherm
m a y be
noticed.
Zobel [70] s u g g e s t e d
that the
c o m p l e x i n g of a m y l o s e with fatty acids behind this change.
B e k h e e t [59]
studied the effect of
food p r o c e s s i n g on
g e l a t i n i z a t i o n of w h e a t starch.
She found that
s t e e p i n g of
wheat
grains either in water and/or in d i l u t e d lye s o l u t i o n
before
and/or after 5 min.
b o i l i n g r e d u c e d from the energy
required for starch g e l a t ~ n i z a t i o n . She o b s e r v e d only single
e n d o t h e r m peak for g e l a t i n i z a t i o n of u n s o a k e d and water, lye
and 5 min. boiled soaked wheat starches.
The To,
Tp, Tc of
these peaks
were ranged from
30 to 35,
70 to 75
and 115120~
The cooking
and drying of the p r e v i o u s
samples led
to a s i g n i f i c a n c e
high r e d u c t i o n
in the peak
area of
the
native, u n c o o k e d
undried, e n d o t h e r m
peak starch.
This is
due
to
the i r r e v e r s i b l e
damage
occurred
during heating,
c o o k i n g and drying of starch.
4.1.6Colour:
Several
m e t h o d s are
in use of
the
determination
of colour
in
starch products,
ranging from
visual i n s p e c t i o n
to the
use of
a
photometer.
Lovibond
t i n t o m e t e r is often used.
The T e n t a t i v e Standard M e t h o d ii18-57 is s p e c i f i c a l l y d e v o t e d
to d e t e r m i n e colour at starch
by
B e c k m a n Model
B
spectrophotometer
equipped
with
the
i n t e g r a t i n g sphere diffuse r e f l a c t a n t a t t a c h m e n t , with beamexpending
lenses
and
a
blue
sensitive
photo-tube
or
e q u i v a l e n t equipment.
The r e f l e c t a n c e (%R) at
450 mu, 550
mu, and 600 mu. is m e a s u r e d with this
instrument, and
the
c a l c u l a t i o n s involved are;
C o l o u r = log % R at 600 mu - Log % R at
B r i g h t n e s = % R at 550 mu.
Greyness
= Z - Log % R at 550 mu.
450 mu.
In USA the B r i c e - K e e n e
p h o t o m e t e r is used to d e t e r m i n e
the p h o t o e l e c t r i c r e f l e c t a n c e of starch samples, and express
it
as a % of w h i t e n e s s of standard plate c a l i b r a t e d a g a i n s t
72
pure
precipitated magnesium
oxide.
In Japan,
the Hunter
reflectometer
a t t a c h e d to
a photoelectric
colorimeter AKA
No. 50 of the Kotaki
M a n u f a c t u r i n g Co., and the i n t e g r a t i n g
sphere of the same c o m p a n y has beed used. [48].
4.2-
Chemical
Methods:
4.2.1Water
soluble
materials:
D e t e r m i n a t i o n of
water s o l u b l e s
m a t t e r in
starch gives an
i n d i c a t i o n about
the added m a t e r i a l s , and the extent of c o n v e r s i o n of
starch
and dextrin.
The m e t h o d s used for d e t e r m i n i n g these m a t t e r
depend
on s u s p e n d i n g
the
starch in
water
at a
definite
temperature
and
concentration
then
filtering
and
the
filtrate
is e v a p o r a t e d to d r y n e s s on a s t e a m bath, dried to
c o n s t a n t w e i g h t in an oven at I05~C and w e i g h t e d .
4.2.2Acidity
and
pH:
Acidity
o~
starch
is
c h i e f l y related to the
p r e s e n c e of a m y l o - p h o s p h o r i c acid as
h y d r o l y s a b l e salts
and p a r t l y to the r e s i d u e s
of S02 w h i c h
used
as a
preservative,
or p r o p p i o n i c
and other
organic
acids formed by the c o n t r o l l e d f e r m e n t a t i o n
of c a r b p h y d r a t e
during steeping. [48].
A c o m p l e t e p i c t u r e could be o b t a i n e d
by d e t e r m i n i n g the e l e c t r o m e t r i c t i t r a t i o n curve.
According
to B e k h e e t
[59]
s t e e p i n g of
grains
in water
and
dilute
alkali
i n c r e a s e d the pH of
w h e a t starch from
5.69 to 5.73
and 5.81 r e s p e c t i v e l y .
On other hand, c o o k i n g and d r y i n g of
these
grains increase from
pH to
6.1-6.2 and
r e d u c e d the
t i t r a t a b l e a c i d i t y of wheat star, h.
4.2.3Alkali-Labile
value:
Native
starches,
modified
starches,
and
dextrin
contain
a
portion
of
s u b s t a n c e s termed
as alkali labile, w h i c h
is r e a d i l y acted
as
alkali.
A good grade corn starch gives an a l k a l i - l a b i l e
value of about 22, t a p i o c a starch 14, t h i n - b o i l i n g starch 60
and a y e l l o w d e x t r i n e 20. [48].
4.2.4Alkali-Number:
It is used
to e s t i m a t e
the
relative h y d r o l y t i c
d e g r a d a t i o n of starch or
the number of
reducing
end
groups
in starch.
It
is
r e l a t e d to
the
molecular
weight
and
has
no
relation
to
viscosity
or
s o l u b i l i t y of starch [19, 48].
Commercial
corn and
wheat s t a r c h e s
have c o n s i s t e n t l y
higher alkali
numbers than those ot
common tuber starches.
The alkali
number varies from 5.3-6.9
for tapioca, 5.7-6.9
for
potato,
6.7-7.5
for
wheat
and
9.8-12.1
for
corn
starches. [56].
4.2.5Damaged
starch
grains:
The m e t h o d used for
e s t i m a t i n g the
d a m a g e d starch
grain d e p e n d s on
their more
r e a d i l y swell in cold water and digest with
~-amylase.
The
p r o p o r t i o n of
d a m a g e d starch gives
some i n f o r m a t i o n
about
the t r e a t m e n t w h i c h starch
has r e c e i v e d during m a n u f a c t u r e .
It is
a f f e c t e d the
r e h e o l o g i c a l p r o p e r t i e s of
starch gel.
[48].
73
4.2.6Reducing
power:
Schoch
[19] suggested the
use
ofof t ~ a r ~ ~ i c y ~ i ~ e m ~ ~ d
power
to d e E e ~ m ~
~e
reduci
9
does not in
en
by s a m p ~
size, time of digestion and the
amouht of oxidant.
It aoes
not
give an idea about the
m o d i f i c a t i o n degree of oxidized
starches and/or
those treated in granular
state with acid.
These
types
of
modified
starches are
washed
free
from
modified reagents.
Such washing removes
the water soluble
reducing substances. [103].
El-Saadany
et al.
[104]
found
that gamma
radiation
hydrolyzed
the rice
starch molecules into
small molecular
weight units and increased its reducing power.
The reducing
power of commercial corn starch ranged from 7.7 to 11.6 Rcu.
mg/g
Wankhede
an~
Uma~evi
[3~]
s~u~i~
~h~
Gbang~
in
reducing
power os
pyrodextrins os ragi,
wheat and rice
at
different
intervals
of
time
at
200~C.
Their
results
indicated
that reducing
power increased with
extending of
heating
time.
Abou-Samaha
[43] found
that the
reducing
power
of
modified
starches
was
significantly
differed
according to the source of starch, methods and conditions of
modification.
Generally,
modified
and
unmodified
rice
starches
had lower
reducing
power than
maize one.
Acid
modified starches had significant higher reducing power than
moist heated one.
Reducing power increased with the raising
of acid concentration a n d , e x t e n d i n g the time of m o d i f l c a t i o n
A
polynomial
relationship
of second
order
equatlons was
found between this chemical property and
m o a i f i c a t i o n time.
Bekheet [59] studied the influence of food
processing,
steeping, cooking and drying, on the reducing power of wheat
starch.
She found
that steeping of wheat grains
either in
water
or in dilute alkali did not affect the reducing power
of starch.
While cooking and
drying increased this
value
from 0.0 to 3.07-4.4 Rcu. mg/g.
This was due to the thermal
hydrolysis of
the bonds of starch
molecules during cooking
and drying.
4.3-
Functional
and
rheological
methods:
4.3.1Swelling
power
and solubility:
When
starch
is heated in
the presence of
water, the individual
granules
swell and a portion of the starch dissolves in the
surrounding aqueous medium.
The
degree of swelling and the
amount of
s o l u b i l i z a t i o n depend
on the extent
of chemical
c r o s s - b o n d i n g within
the
granules,
the
presence
of
non
carbohydrate
substances
in starch
such
as
lipids and/or
phosphate, a high amylose content and the greater numbers of
intermolecular bonds. [105].
Badenhuizen
[106]
reported
that
during
heating
of
starch in w a t e r , p a r t s of its molecules in amorphous
regions
are
g r a d u a l l y liberated
until the
shortest linear
chains
become able to diffuse
out the walls of sac
shaped swollen
starch granules.
The
latter granules consisted
m a i n l y of
?4
branched molecules, amylopectin.
A c c o r d i n g to M i l l e r et al.
[107]
and
French,
[73]
swelling
starts
in
the
least
organized
amorphous intercrystalline regions
of the s t a r c h
granules.
This
leads
to
extent
a
tension
on
the
n e i g h b o u r i n g c r y s t a l l i t e s and d i s t o r t i n g them.
With further
heating,
the
double
helical
region
of
the
amylopectin
crystallites
structure
become
an
uncoiling
and/or
dissociate.
H o w e v e r , the
liberated
side
c h a i n s of
the
a m y l o p e c t i n h y d r a t e and swell l a t e r a l l y to exert m o r e s t r e s s
on
the
remaining
crystallites.
Because
of
the
starch
m o l e c u l e s are u n a b l e to
stretch longitudinally,
it m a y have
a
t e n d e n c y to c o n t r a c t to a r a n d o m coil c o n f o r m a t i o n .
This
may prevent
s t a r c h s w e l l i n g in the
d i r e c t i o n of m o l e c u l a r
chains.
W h i l e the i n c r e a s e of
the m o l e c u l a r m o b i l i t y w i t h
further
hydration
permits
a
redistribution
of m o l e c u l e s
w h i c h a l l o w the s m a l l e r
linear a m y l o s e m o l e c u l e s to d i f f u s e
out
the s w o l l e n
granules.
However,
further heating
and
hydration
are
w e a k e n the
structure
and
produces a
sol.
Bowler
et
al. [108]
suggested
that
the s w e l l i n g
occurs
essentially
in the plane of
two m a j o r axes
of the g r a n u l e
and v e r y
little in its
thickness direction.
The s w e l l i n g
includes
a
radial
expansion
to form
a
flattened
disc,
followed
by
tangential
expansion
to
produce
a
complex
puckered granule.
L e a c h et
al.
[109]
found that
the
maize
and
milo
s t a r c h e s gave two s t a g e s ' o f s w e l l i n g .
This is an i n d i c a t i o n
that there are two sets of b o n d i n g forces w h i c h relax at two
different
temperatures.
M a d a m b a et
al. [ii0]
found that
sweet p o t a t o
s t a r c h e x h i b i t e d s i n g l e stage
swelling, which
s u g g e s t e d the p r e s e n c e of
u n i f o r m i n t e r m o i e c u l a r bonds.
In
contrast,
Delpeach
and
Favier
[i05] found
a
two
stage
swelling
p a t t e r n for s w e e t p o t a t o e s starch, w h i c h s u g g e s t e d
that this type of s t a r c h has a high d e g r e e of i n t e r m o l e c u l a r
bonds
in
its
granules.
Therefore,
the
swelling
and
solubility
of this s t a r c h are less than those of p o t a t o and
c a s s a v a but g e n e r a l l y m o r e than those of corn s t a r c h e s .
A c c o r d i n g to
A b o u - S a m a h a [43] the
native maize starch
had h i g h e r
s w e l l i n g power
than rice starch.
Modification
p r o c e s s , e s p e c i a l l y w i t h 2% HCI, r e d u c e d the s w e l l i n g power.
This
reduction
increased with
extending
the m o d i f i c a t i o n
time.
A p o l y n o m i a l r e l a t i o n s h i ~ with an e q u a t i o n
of s e c o n d
order was
observed between
this p r o p e r t y and
the time
of
modification. Also
n a t i v e and m o d i f i e d m a i z e
starches were
more
s o l u b l e than
that of
rice one.
Acid
modification,
e s p e c i a l l y w i t h 2% HCl for longer p e r i o d s
gave m o r e s o l u b l e
starch
than m o i s t
heat
treatment.
The r e l a t i o n
between
s t a r c h s o l u b i l i t y , and
m o d i f i c a t i o n time was e i t h e r
linear
and/or
polynomial.
He
attributed
his
results
to
the
redistribution
and/or
partial
hydrolysis
in
the
starch
molecules
due to
both
acid and
m o i s t heat
modification.
These changes
are b e h i n d
the r e d u c t i o n of
the a b i l i t y
of
s t a r c h g r a n u l e s to
swell and the i n c r e a s e of the s o l u b i l i t y
of m o d i f i e d
starches.
The
same r e s u l t s were
r e p o r t e d for
moist heated
s t a r c h e s of corn, potato,
Parley, red m i l l e t ,
75
w h e a t and
c a s s a v a as well as potato
acid m o d i f i e d s t a r c h e s
[80, I 0 ~
iii, 112].
The changes in s w e l l i n g and s o l u b i l i t y
of
the
starches
after
modlfication may
be
due to
the
changes
occurred
in
the
physical state
of
the
amylose
c o m p o n e n t of the native starch.
Bekheet (59] found that
soaking of wheat grains either
in water or d i l u t e d alkali
before and after
5 min b o i l i n g
gave
starches
having
lower
swelling
capacity
and
less
soluble c o m p a r i n g with native wheat starch.
The d e c r e a s e in
both p r o p e r t i e s
were more in starches of
boiled lye soaked
w h e a t grains than those
of water soaked one.
Also c o o k i n g
and d r y i n g lowered from the swelling power and s o l u b i l i t y of
w h e a t starch.
H o s e n e y et ai. [7~] reported that when starch
was
placed
in water,
it a b s o r b e d
part
of the
water and
swelled
slightly.
At
room t e m p e r a t u r e ,
the p r o c e s s
was
reversible.
At
50~
the internal
s t r u c t u r e of the
wheat
starch
granules
was
altered.
According
to
Tester
and
Morrison
[i01],
s w e l l i n g of
wheat
begun
at 45-50~C
and
continued
to 85~C.
At 50-55~C, loss of b i r e f r i n g e n c e and a
large d e c r e a s e in g e l a t l n i z a t l o n
e n t h a l p y occurred.
These
changes
were a t t r i b u t e d to
d i s s o c i a t i o n of the c r y s t a l l i n e
clusters and the double helixes. L e a c h i n g of p o l y s a c c h a r i d e s
amylose
and a m y l o p e c t i n from starch was
found to be h i g h l y
c o r r e l a t e d with
swelling factor.
The
formation of a m y l o s e
lipid c o m p l e x e s inhibited the swelling.
4.3.2Viscosity
and consistency
: The p r o p e r t i e s
of starch d i s p e r s i o n s in water and its p a s t i n g behavior
are
u s u a l l y studies by o b s e r v i n g
changes in v i s c o s i t y of starch
systems [77, 113]
Kerr [114] s u g g e s t e d the d e t e r m i n a t i o n of the v i s c o s i t y
of
hot paste, cold paste and
a l k a l i n e s o l u t i o n or f l u i d i t y
test to evaluate the starch d i s p e r s i n g and p a s t i n g behavior.
Generally,
different
instruments including
the capillary,
Stromer,
Mac
Micael,
brookfield,
the
falling
sphere.
B r a b e n d e r a m y l o g r a p h and corn
industries v i s c o m e t e r s can be
used
to estimate
the v i s c o s i t y
of starch
d i s p e r s i o n s and
pastes [48].
ity
: The i n t r i n s i c
v i s c o s i t y is
4.3.2.1Viscos
increase the
related to the a b i l i t y of polymer m o l e c u l e s to
in
the
absence
of
any
viscosity
of
the
s olvent
It is
directly
related
to
intermolecular
interac tion.
the degree
of p o l y m e r i s a t i o n
molecular
size and
he nce to
v i s c o s i t y of u n m o d i f i e d
[81].
A c c o r d i n g to Rad ley [48] the
starches
even after co n s i d e r a b l e cooking are not c o n s i d e r e d
phenomenon, but
may be due
to the
primarily
as colloidal
including
undisintegrated
presence
of
larger
aggregates
starch g r a n u l e s
form a
structure
granules.
The
swolle n
including p o r t i o n of liquid phase.
This phase incrased the
v i s c o s i t y of native starch than of an acid and or moist heat
treated one.
Abd-Allah
et
al.
[ 75 ] stated
that
the
relative
viscosity
of
both
maize
and s o r g h u m
was
reduced
after
76
modification
either with
acid
a n d / o r by
oxidation.
EIS a a d a n y et
al. [104] found
that the r e l a t i v e
v i s c o s i t y of
rice s t a r c h was 2.328.
T a k e d a et al. [44] s h o w e d that s w e e t
p o t a t o a m y l o s e has a l i m i t i n g v i s c o s i t y
h i g h e r than that of
wheat
but
lower
than
that
of
cassava
or Irish
potato
amylose.
Also,
sweet
potato
amylopectin
has
a
lower
limiting
viscosity number
than
Irish
potato
amylopectin
s u g g e s t i n g s m a l l e r or m o r e s p h e r i c a l m o l e c u l e s .
Abou-Samaha
[43]
found
that n a t i v e
and
modified
maize starches
had
higher
r e l a t i v e and
inherent viscosities,
lower f l u i d i t y ,
than rice one.
M o d i f i c a t i o n of starch, e s p e c i a l l y w i t h
2%
acid r e d u c e d
the r e l a t i v e
and i n h e r e n t v i s c o s i t i e s .
This
r e d u c t i o n i n c r e a s e d w i t h e x t e n d i n g the m o d i f i c a t i o n time.
A
pol y n o m i a l r e l a t i o n s h i p was n o t i c e d b e t w e e n these p r o p e r t i e s
and
time of m o d i f i c a t i o n .
He a t t r i b u t e d his r e s u l t s to the
red u c t i o n
in the
a b i l i t y of
starch granules
to a g g r e g a t e
aft er m o d i f i c a t i o n .
According
to B e k h e e t [59] the i n h e r e n t
and
relative
v i s c o s i t i e s of
wheat
starch were
decreased
after s o a k i n g e i t h e r in w a t e r or d i l u t e a l k a l i , c o o k i n g
and
d r y i n g the grains.
4.3.2.2Paste
characteristics:
The v i s c o s i t y of
the paste of s t a r c h d e p e n d s on the s w e l l i n g of the g r a n u l e s ,
the
percentage
of
t he
dissolved
molecules
and
the
orientation
of
these
molecules.
Also,
the
type,
c o n c e n t r a t i o n , h e a t i n g t e m p e r a t u r e , pH, and d r y i n g of s t a r c h
a f f e c t its v i s c o s i t y [ii 5].
W h e n an
a q u e o u s su s p e n s i o n of s t a r c h is
h e a t e d , it is
p r e s u m e d that
the a m o r p hous
parts of the
granules undergo
p r o g r e s s i v e h y d r a t i o n an d s w e l l i n g .
This gives an e x p a n d e d
network structure
w h i c h is held t o g e t h e r
by the p e r s i s t e n t
and
intact
micelles.
This
structure
gives the
swollen
g r a n u l e s their e l a s t i c p r o p e r t i e s and m a y be r e s p o n s i b l e for
their
p a s t i n g . [109].
Maximum swelling
of g r a n u l e s
and
formation
of a paste of
w h e a t s t a r c h is
o b t a i n e d at w a t e r
c o n t e n t m o r e than 70% [116].
The s t a r c h paste c o n s i s t s of a
mixture
of
starch
and
water
as
two
phases.
The
d i s c o n t i n u o u s solid
or s e m i s o l i d
phase u s u a l l y
forms from
the s w o l l e n
starch granules
and nearly,
all the
w a t e r in
this
system
is
absorbed
by
the s t a r c h
granules.
The
remaining
is
scarely
s u f f i c i e n t for
acting
as l u b r i c a n t
between
the m o v i n g
particles.
When
the s t a r c h
paste is
s u b j e c t e d for
a
high
p r e s s u r e or
a
high
stirring,
the
s w o l l e n p a r t i c l e s are d i s t o r t e d or d i s r u p t e d [43].
The
Brabender amylograph
provides a
good m e t h o d
for
d e f i n i n g these c h a r a c t e r i s t i c s .
It m e a s u r e s the c h a n g e s in
v i s c o s i t y as
a function
of t e m p e r a t u r e
and
time.
After
gelatinization
the v i s c o s i t y i n c r e a s e d
b e c a u s e of g r a n u l a r
swelling
and
the
effects
of
soluble
matter
which
are
r e l e a s e d from
swollen granules
through further
h e a t i n g or
mechanical disruption.
The t e m p e r a t u r e is u s u a l l y r a i s e d at
a rate
of 1 . 5 ~
per min. until 95~
where
it is held for a
given length
of time b e f o r e
the t e m p e r a t u r e is
l o w e r e d to
50~
at the rate of 1.5oC per mi~.. and then held at 50~
for
??
another
given
length
of time.
The
first
part of
the
Brabender Viscograms
of starches d e s c r i b e s
the s w e l l i n g of
the
starch g r a n u l e s
which
increased with
the raising
in
temperature.
While
second
part
indicates
the
maximum
c o n s i s t e n c y , when the
g r a n u l e s become more h i g h l y
swollen.
The third part
of these
curves shows the
d e c r e a s e in
the
viscosity
due to
the d e f o r m a t i o n
and the
rupture
of the
swollen
starch granules.
The last
part i l l u s t r a t e s
the
setback that occurrs as
a result of the o r i e n t a t i o n
of the
amylose molecules
in a parallel fashion
to form a g g r e g a t e s
of
low solubility.
The f o l l o w i n g important c h a r a c t e r i s t i c s
can be o b s e r v e d from these curve:
i-
The h i g h e s t
or peak
v i s c o s i t y (p)
which is showed
a
n o t i c e a b l e i r r e s p e c t i v e trend with t e m p e r a t u r e applied.
This p r o p e r t y
is important during
p r e p a r a t i o n of
the
usable starch paste.
2-
The v i s c o s i t y
of the
t e m p e r a t u r e of 95~
paste
(M)
when
it reaches
3-
The s t a b i l i t y
or b r e a k d o w g n of the hot
(H) after c o o k i n g for i0 min. at 95~C.
4-
The v i s c o s i t y
of the cooked paste after c o o l i n g down
50~C or the setback v i s c o s i t y (C).
It is a m e a s u r e
the t h i c k e n i n g p r o d u c e d by cooling.
5-
The b r e a k d o w n
or t h i c k e n i n g
or s t a b i l i t y
ratio (C/P).
ratio
(H/P)
paste
and
the
viscosity
to
of
the setback
The
pasting viscosities
are
d e p e n d e d on
preparation
method, impurities,
concentration
and v a r i e t y
of
starch.
[117].
In 1991, A b o u - S a m a h a
[43] studied the influence of
acid
and
moist
heat
modification
on
the
reheological
behaviour, paste
c h a r a c t e r i s t i c s , of both native
maize and
rice
starches using
the
Brabender visco/amylograph
at
a
c o n s t a n t c o n c e n t r a t i o n , 10%.
Figs. (19, 20 & 21) illustrate
the
complete
cooking
and
cooling
curves
obtained
from
B r e n b e n d e r instrument for both native and m o d i f i e d maize and
rice starches.
The data in these figs indicate that:
1-
Generall, the
rheological
properties
of native
and
modified
rice starches,
namely
P,M,H,C, H/P
and C/P
values, were
lesser
than those
of maize
one.
This
means
that the
s w e l l i n g of rice
starch is much less
than maize.
Hence its g r a n u l e s stay more rigid and are
embeded in more fluid.
Therefore, it is expected that
a lesser q u a n t i t y of
rice starch g r a n u l e s b r e a k a g e are
o c c u r r e d during cooking
c o m p a r i n g with m a i z e granules.
A c c o r d i n g l y , the
rice starch
paste are built
up from
entire g r a n u l e s
or well
o r g a n i z e d parts
of granules,
low starch is present in
the i n t e r s t i t i a l liquid.
So,
the p a r t i c l e s
are easier to separate.
Therefore, the
78
"i..,
, , ' ; 7 ~.'
+
ll
B
9
v,+'
j
vISCOS,~y ol ma,ze slarc~ (conc( r ,,~,o~ ~o%,
,,-
o
9
,,,,or,,,,+
i
" ~
0
~
I+~
~'~
B
1,.f~
r
;
qJt
i,,.,, p...,,
Fig.
( 2 0 )
Eflec~ ol i c ~ mo~,hcal,on (A ~ o and B. 2/.) On tr~e arny,og,ap,
viscosity of rice $1arcn (concer~lrahOrl 10e~)
siarches (conconiralon 10%)
f
79
rice s t a r c h p a s t e is d e s c r i b e d as s h o r t e r
and on c o o l i n g m o r e f r a g m e n t s c o n t r o l the
retrogradation process.
_
_
than of m a i z e
n a t u r e of the
Modification
either
with
acid
and/or
moist
heat
treatment
r e d u c e d the paste
c h a r a c t e r i s t i c s of n a t i v e
starches.
This
effect
was
more
pronounced
with
extending
the m o d i f i c a t i o n
time.
Also, the
rate of
r e d u c t i o n in
these p a r a m e t e r s was
more noticeable
in
case of acid m o d i f i e d , e s p e c i a l l y w h e n 2% HCl was
used
for
up to
3 hrs,
than that
of m o i s t
h e a t e d starch.
This
e f f e c t m a y be due
to the n a k e d
r e d u c t i o n in the
s w e l l i n g power and the
i n c r e a s i n g in the s o l u b i l i t y of
starch granules.
It is i n t e r e s t e d to n o t i c e that m o i s t heat t r e a t m e n t for
21 hrs at II0oC, gave s t a r c h p a s t e s with h i g h e r m a x i m u m
cooking
and
setback
viscosities
than
that o b t a i n e d
after acid m o d i f i c a t i o n
w i t h 1 or 2%
HCl for up
to 1
hr.
This is
an i n d i c a t i o n that a p a r t i a l
destruction
m a y be o c c u r r e d in s t a r c h g r a n u l e s ,
especially amylose
f r a c t i o n , after
m o i s t heat
treatment.
These samples
showed
a high s t a b i l i t y d u r i n g c o o k i n g
of starch.
He
attributed
the
paste
stability
behaviour
to
a
conversion
of a m o r p h o u s a m y l o s e to h e l i c a l form, w h i c h
can
acts
as
a
weak
centers
of
crystallinity
for
s t a b i l i z i n g the
starch granules.
On the
other hand,
the acid m o d i f i c a t i o n up 3 hrs, e s p e c i a l l y w i t h 2% acid
destroyed
the
structure
of
the
granules.
This
destruction
affected
the
paste
characteristics,
e s p e c i a l l y at h i g h e r t e m p e r a t u r e s .
The c h a r a c t e r i s t i c s
of
maize
and
rice
starch
pastes
were
completely
d i s a p p e a r e d after
12 and
24 hrs of
acid m o d i f i c a t i o n
with 1
and
2% HC i.
This m e a n s
that
such type
of
modified
starches
gave
solutions
almost
free
of
structure either through heating
and or after c o o l i n g .
In c o n c l u s i o n , the m o i s t heat m o d i f i e d s t a r c h e s
showed
a
good s t a b i l i t y ,
setback
paste c h a r a c t e r i s t i c ,
and
higher
c o n s i s t e n c y than acid m o d i f i e d one.
Therefore,
such
products
a re
appreciated
in
food
industry
especially
w h e n t he
swollen starch
granules
m u s t be
left as
fully
in tact as
possible.
While
the
acid
m o d i f i e d s t a r c h es p e c i a l l y that t r e a t e d for m o r e than 3
hrs can be s u g g e s t ed
to use in i n d u s t r i a l a p p l i c a t i o n s
to p e n e t r a t e fiber , e.g. paper and t e x t i l e i n d u s t r i e s .
Seog et
al. [118] found no peak v i s c o s i t y was
w i t h 4-6%- (w/v)
sweet p o t a t o s t a r c h s u s p e n s i o n s .
m o d e r a t e peak v i s c o s i t y d u r i n g
c o o k i n g and a high
on c o o l i n g w i t h a s t a r c h s u s p e n s i o n of 7% (w/v).
obtained
While a
set back
4.3.2.3Alkaline
fluidity:
It
estimates
the
v o l u m e of s t a r c h paste w h i c h flows in 7 seconds.
It uses to
c o m p a r e b e t w e e n the acid m o d i f i e d s ~ a r c n e s [1i8,
1i9].
The
80
p r o d u c t s have high a l k a l i n i e
f l u i d i t y having low t h i c k e n i n g
power and~ v i s c o s i t y . . [ 1 2 0 ] " Seib and M a n i n g a t [121] o b s e r v e d
a strong
c o r r e l a t l o n between the a l k a l i n e
fluidit
and the
Vi~O~!
y
Of
~a~h
A b o u - ~ a m a h a [433
s
hat
the
alkaline
fluidity
os the
acid
modis
rice starch
was
higher ~han
that os m a i z e one.
In both sources of starch,
the value of this
c h a r a c t e r i s t i c increased with raising the
HCl c o n c e n t r a t i o n and e x t e n d i n g
the time of m o d i f i c a t i o n . A
polynomial
relationship
with
second
and
third
order
equations
was found
between the
alkaline fluidity
and an
acid m o d i f i c a t i o n time.
A c c o r d i n g to Bekheet [59] s t e e p i n g
of wheat grains
in water or dilute
alkali s l i g h t l y r e d u c e d
the
alkaline
f l u i d i t y of
its starch.
While
c o o k i n g and
drying
caused a m a r k e d
d e c r e a s e in
this
property.
She
attributed
these
results
s
the
irreverslble
damage
o c c u r r e d in starch d u r i n g heating.
4.3.2.4Visco
elastic
properties
of s t a r c h
gels
The c h a r a c t e r i z a t i o n
of starch paste by
v i s c o m e t r y such as
amylograph
and corn
industries
viscometers
may
cause
a
destruction
of the starch system
and do not
cover all the
required
i n f o r m a t i o n s to c h a r a c t e r i z e the starch s t r u c t u r e
[122].
Therefore,
Evans
and H a i s m a n
[123] and
Wong and
Lelievre [124] s u g g e s t e d the d e t e r m i n a t i o n of the m e c h a n i c a l
b e h a v i o u r or the v i s c o u s
and e l a s t i c p a r a m e t e r s of starches
either in native state or ~ during and after m o d i f i c a t i o n and
gelatinization.
The starch
paste at a
c o n c e n t r a t i o n > 5%
showed
a certain amount of r i g i d i t y due to swelling and the
a b i l i t y of
g r a n u l e fragments to act
as c r o s s - l i n k i n g sites
for
the p o l y m e r i c
exudate
of
the
paste.
This
is
an
indication
that
some
of
the
applied
stress
is
not
d i s s i p a t e d , but is stored. [115, 122, 125].
Therefore, the
pastes have v i s c o e l a s t i c c h a r a c t e r i s t i c s .
In this case, it
is
possible to resolve the
stress in terms
of an in-phase
component
and o u t - p h a s e
term
related by
the phase
angle
b e t w e e n the sinusoidal functions.
This d y n a m i c r h e o l o g i c a l
m e a s u r e m e n t can
be followed by two
p a r a m e t e r s , the storage
m o d u l e s (G ~) and the loss m o d u l e s (G ~ ) respectively. [126].
These p r o p e r t i e s can be e s t i m a t e d
by s u b j e c t i n g the gels to
an o s c i l l a t i n g strain, and the
v i s c o - e l a s t i c p a r a m e t e r s (G'
and
G") were
extracted
by c o m p a r i n g
the strain
with the
resultant
o s c i l l a t i n g stress. (127].
Bell [126] s u g g e s t e d
the
use of
the
dynamic mechanical
testing t e c h n i q u e
for
determining
these properties.
In this technique the strain
a m p l i t u d e and f r e q u e n c y can be i n d e p e n d e n t l y controlled.
By
this way, the changes in the s t r u c t u r e by b u i l d i n g up and/or
b r e a k i n g down, r e c o v e r y c h a r a c t e r i s t i c s , can
be m e a s u r e d in
real time and continously.
A b o u - S a m a h a [43]
studied
the influence
of
acid
and
moist heat
m o d i f i c a t i o n on
the v i s c o - e l a s t i c
behaviour of
both
native and m o d i f i e d
maize and rice
starch gels using
the R h e - t e c h I n t e r n a t i o n a l Rheometer. The range of f r e q u e n c y
was 0.01-0.5 HZ and a m p l i t u d e was 0.01-0.i mNm.
The storage
modules
(G') , the loss
m o d u l u s (G")
and tan
delta (G"/G)
8!
were calculated.
The o b t a i n e d d a t a w e r e i l l u s t r a t e d
in Fig.
(22).
The c h a n g e s
in
G' ( s t o r a g e
modulus)
and G"
(loss
modulus) with
the c o n c e n t r a t i o n
as a
f u n c t i o n of f r e q u e n c y
and
a m p l i t u d e are s h o w n in Fig. (22).
The r e s u l t s i n d i c a t e
that
at a c o n c e n t r a t i o n
of
5% of m a i z e
and rice s t a r c h e s ,
b o t h G ' a n d G" are h i g h l y f r e q u e n c y d e p e n d e n t .
The v a l u e s of
G" are
h i g h e r t h a n of
G over all
the m e a s u r i n g r a n g e s
of
frequency
and s t r a i n .
This
is an i n d i c a t i o n
that at this
concentration,
rice s t a r c h had
a l m o s t , the c h a r a c t e r i s t i c s
of a v i s c o u s s y s t e m .
I n c r e a s i n g the c o n c e n t r a t i o n
to 7.5 or
10%
w/v
was
associated with
an
increase
in
G, e l a s t i c
properties.
The s t a r c h p a s t e at a c o n c e n t r a t i o n
> 5% s h o w e d
a c e r t a i n a m o u n t of r i g i d i t y due to s w e l l i n g and the a b i l i t y
of
g r a n u l e f r a g m e n t s to act as
cross-linking
s i t e s for the
p o l y m e r i c e x u d a t e of the
paste.
[115, 125].
In
this case,
it is p o s s i b l e to r e s o l v e the s t r e s s in
t e r m s of an i n p h a s e
component
and o u t - p h a s e
term
r e l a t e d by
the p h a s e
angle
b e t w e e n the s i n u s o i d a l f u n c t i o n s .
I n c r e a s i n g the
concentration
to
7.5 or 10%
(w/v) was
associated
w i t h an i n c r e a s e in G
( e l a s t i c c h a r a c t e r ) and a
d e c r e a s e in Tan delta.
T h i s is
an i n d i c a t i o n t h a t at t h e s e
concentrations
a gel
like s t r u c t u r e
can De
obtained.
In
this
case
the
G
was
almost
frequency
independent.
Generally
t h e s e c h a n g e s in
the t h e o l o g i c a l c h a r a c t e r i s t i c s
w e r e b a s e d on the c o n c e n t r a t i o n
at w h i c h the s t a r c h g r a n u l e s
fill the total v o l u m e of the s y s t e m . E v a n s and H a i s m a n [123]
and E l i a s s o n
[115] r e p o r t e d
the same c o n c l u s i o n .
According
to t h e i r
r e s u l t s the i n c r e a s i n g of
the c o n c e n t r a t i o n s
than
2.8, 2.5 and 3 . 8 %
for m a i z e , p o t a t o and w h e a t
respectively
was a s s o c i a t e d
w i t h an
i n c r e a s e in
the e l a s t i c i t y .
This
i n c r e a s e was
due to
the g r a n u l e - g r a n u l e
interaction.
The
diagrams
in
Fig.
(23)
illustrates
that
acid
modified
starches gave
weaker gels
c o m p a r i n g w i t h the
n a t i v e ones.
T h e r e f o r e the g e l s
of a c i d m o d i f i e d s t a r c h e s had low v a l u e s
of
G , G"
and a h i g h
over all
v a l u e of tan
delta.
This
effect
was
markedly
observed
with
increasing
acid
concentration
from 1
to 2% and
e x t e n d i n g the c o n t a c t
time
b e t w e e n a c i d and s t a r c h .
The r e d u c t i o n in
G' and G" v a l u e s
may
be a t t r i b u t e d
to
the d i s i n t e g r a t i o n
o c c u r r e d in
the
a m o r p h o u s r e g i o n s of s t a r c h g r a n u l e s
during acid treatment.
This
led to
weak
the s t r u c t u r e
and
a l s o to
reduce
the
swelling
p o w e r of the s t a r c h g r a n u •
T h e s e c h a n g e s had a
n e g a t i v e e f f e c t on the p a s t e c h a r a c t e r i s t i c s .
G e n e r a l l y the
results
a l s o s h o w e d that m o d l f i c a t i o n
w i t h a c i d up to 6 hrs
g a v e a v e r y w e a k gel e v e n at a c o n c e n t r a t i o n
of 10%.
A l s o , the d a t a
in Figs.
(24 & 25)
r e v e a l that
moist
heat
t r e a t m e n t of rice s t a r c h had n e a r l y the same e f f e c t of
acid
modification.
It r e d u c e d
the v a l u e
of G
(elastic
c h a r a c t e r ) , G" ( v i s c o u s c h a r a c t e r )
and g a v e w e a k gel.
The
same
o b s e r v a t i o n was n o t i c e d in case
of m a i z e s t a r c h a f t e r
t r e a t e d w i t h a m o i s t heat, e x c e p t that G" s l i g h t l y i n c r e a s e .
In b o t h s o u r c e s of s t a r c h the
tan d e l t a ( G " / G )
was
higher
after
the
treatment.
This
effect
was
increased
with
e x t e n d i n g the time of m o i s t h e a t t r e a t m e n t .
82
;oomI
B
mo~+
+.o'I
,,.4
. ~ - - - ' - " . . . . . jr._..,%"
~+~..%
,.-:. . . . . . . .
,~
'-......~
,
,
i
,-----~.
""....
l
i
~
D
, , ~,-,
~
i ............
:------~
I+
,.
,
.
.
.
Jn
. . . .
~,
~
!"'""
-~-~ .... ~
~,
' :g,,;,
.-,
i
&.- ~.
.
.-"
I
llm~Iilu0e (~,~) sweep o| ma,z@ anO rice $1atr eel (cOncunlrahon
~0"/.)
B
,31
'q
.
,~
D
'k"-.
aml~,UOe (8.0) sweep ot ma,ze anti nco slarcn gels
,+,
x~uu~
""... ""
F i g ( 2 2 ) +"'~ o, .......,..... c, a~ c,"o. . . . ~ .........t^ u, ,,,',o
+.._.
++r
+
h
J
o
",.
j
;'i
J
~ [,/'/~'5/
....
d +
i : o . , ';.
Z'
'
:
long
C
_
l
-"
r
"...)..
b
: .....
A
A,
" """
1" ,,
.27
"'-~..,._.--.,-~e .........
101~,
r
"'I
o ,,,~o~1+
"l
~/,'"
o ,,,-~,,ir
,'
I~ . , a
1~.~/
;+
,~. ""
Q+,
41
.]
.~ .I~
,0G
~
03
s
~o
~
~
m"-';4
)
~i
~'
I
~ . o + , o ~ .,.., t ,,+ :
F i g
9( 2 5 ) E , , ~
of aoo IA) ...o ~o,~, . . . .
<oi ~o0,,,r
.....
(G'IO') Ol ma, ze ancl tic6 $1atCh i~ll$ {concenlralion ~g-l~)
F ~g
9( 2
4
) ~"~
o'
,,,o,m ,+ . . . . .
,. . . . . . . . .
G .+o o" ~ .... ~
anO ampl~luOe (b) sweep OV mmze ancI r~c~ sla'chu~
.......
. .9. .
,.o o .....
83
4.3.3Gel conslstency
ano retrograoatlon:
i~ is
Known ~na~ wnen mo~e c o n c e ~ r a ~ e ~
past~s or cer~aln s~arc~es
s~ano at room ~emp ranure
or a
ew no rs, ~ney set to rlglo
gels
The s~rengtn of gels
oepenas on ~ne ~ime o~ se~tlng,
the t e m p e r a t u r e
a~ w~Ich ~ney
are storeo
an~ ~es~eo,
~ne
Olmenslon
o~ ~ne m o u l d ano
solnetlmes on ~ne
nature o~ ~ne
surface.
The o~taineo
gels
develop
~neir s t r u c t u r e
by
retrogradatlon.
Thls
galns
the ge•
s~ructure a
certaln
rigidity, e i a s t l c p r o p e r t i e s , ano r e s l s t a n c e a g a i n s t stress.
The main force in the gei s ~ r u c t u r e is 0ue ~0
ire volume of
the
remnants of granule fragments in the warm paste.
These
serve as a framework
for c r y s t a l l i z a t i o n during coo~ing an0
around them the
starch solution
is able
to r e t r o g r a d e
to
give
a gel.
Also, a c e r t a i n
m i n i m u m degree of s w e l l i n g of
the starch g r a n u l e s is n e c e s s a r i l y for a c o h e r e n t gel.
When
swelling
is too low, a
s u s p e n s i o n of more
or less s w o l l e n
p a r t i c l e s will resui~
ano the a v a i l a b l e water
is not fully
immobilized.
This
leads
to
lose
water
leaving
a more
concentrated
gel during
cooling.
This process
is called
syneresis, the exude some
of the water a b s o r b e d on
pasting
during
cooling
period.
This s i t u a t i o n
may
be
further
complicated
where
the
starch
granules
are
ruptured
by
s h e a r i n g or
other m e t h o d s of thermal
or m e c h a n i c a l damage.
[128].
Further
changes
occur
on
storage,
involving
r e c r y s t a l l i s a t i o n or r e t r o g r a d a t i o n
of the polymer
chains.
R e t r o g r a d a t i o n is a f f e c t e d
by the
a m y l o s e and
amylopectin
concentrations,
the
presence
of other
molecules
such as
sugars,
salts and e m u l s i f i e r s , m o l e c u l a r size, t e m p e r a t u r e ,
pH and other non starch components. [129].
A c c o r d i n g to Del R o s a r i o and P o n t i v e r o s [129] the sweet
potato starch
r e t r o g r a d e d more slowly than
wheat, corn and
cassava starches.
This is
the reason for
the o b s e r v a t i o n
that bread
c o n t a i n i n g sweet
potato flour as
a substituent
staled at a
slower rate than other breads.
The results of
the
study of Rasper [130] on the changes in gel c o n s i s t e n c y
of d i f f e r e n t starch gels over 7 days storage using the FIRA,
Jelly Tester were:
Starch
source
Sweet potato
Maize
Cassava
Xanthosoma
Colocasia
Concentration
(g/450 ml)
26
34
25
32
25
C o n s i s t e n c y in
milliliters
1 day
4 days
7 days
8.5
10.5
12.5
10.5
17.0
14.8
Too low for m e a s u r m e n t
16.3
18.7
20.5
2.7
3.6
2.9
Radley
[ 48 ] stated
that
starch
gels
increased
in
s t r e n g t h r a p i d l y during the first i0 hrs.
Slight changes in
strength of gels are occurred after 18 to 24 hrs.
Creda and
Wosiaki
[131] found that the lost water of maize starch gel
84
was
41.6%
after
7 days
of
storage at
4=C.
Hoover
and
found
that
st~rcht~l , t~r~atat
~5~
4h~d 9
$osulki
~132~f
syneresls
lower va ue
, los
wa
ke
This
may
be due
to the
low
klnetl~ energy
of seg~entai
aotion 0s starch chains at -15~
than at -4~
Abou-Samaha [43] studied the syneresis property os gels
prepared from
native and
modified maize and
rice starches
during storage at 2 and -18~
for different aged periods, 216
days.
He found
that the syneresis
exudated water, was
higher in the gel of native and modified maize starches than
of
rice.
Modification
process e~ither
with acid
and/or
either moist
heat treatment increased the
syneresis.
This
e~s
was
increased
with the
extending
of m o d i f i c a t i o n
time.
A polynomial
relationship was obtained
between the
synthesis
ahd ~odis
time.
The level of the exudated
water was higher
when the
starch gels wer~
kept at
-18~C
than
at 2~c
and also
with extending
os the
gel age.
A
polynomial relationship
was
observed between
the
exudate
water and
modiflcation time.
He supposed
that the changes
in exudate
water during setting
the gels at
-18~
lowered
its ability to
bind water leading to
increase of syneresis
values
than that
at- 2~
The
storing at
-18~C may
be
affected the
retrogradation of starch
by rearrangement
of
starch molecules to be less branched or less differed.
4.3.4- Water bindiag
cap:Jcity (WBC):
It estimates
the
ability of
starch to
bind water.
Because
there are
different methods to determine the WBC of starch, the values
of
this property
are
varied
between researchers.
[133].
According to
Mac Arthur
and D'appolonia
[134] the
WBC of
wheat, oat and legume
starches differed from 83 to
107; 85
to 87 and 78.2
to 92.4% respectively.
Dreher et
al. [135]
found
that the pindak be~n and pinto bean starches had 88.7
and 98.5% WBC respectively.
According to Abd Allah
et al.
[75]
the WBC of the
starch of yellow
maize, sorghum (Olza
3), sorghum
(Giza
114), sordan
(79)
and millet
was
98,
68.15, 79.16, 71.14 and 89.16% respectively.
The value
for
sweet
potato
ranged
from 66.3
to
211.6%.
In general,
tuberous
starches have
higher
WBCs than
those of
cereal
origin.
Also, this
property
is higher
in
sweet potato
starch
than
potato
(93%) and
cassava
(72-92%) starches.
[26].
The moist
heat treatment of wheat
and potato starches
increased the
watert binding
capacity from 89.1
to 182.6%
and
from
102
to
108.7%
respectively
[136].
The
same
observations were
reported by Donovan et al.
[102] for the
WBC of potato, barley, red millet and cassava starches after
the
moist heat treatment.
Mok and Dick
[137] stated that
the hydroxyl group
are the
water binding
sites of
starch
molecules.
Boiling
may
increase
these
groups
and
consequently the WBC of starch.
Abou-Samah
[43] found
that native
and modified
rice
starches
had higher
levels of
WBC than
maize. Also,
the
moist heated starch had higher levels of
WBC comparing with
85
acid m o d i f i e d one.
A strong p o s i t i v e c o r r e l a t i o n was
found
between the
WBC and
time of
modification
A
~olynomial
r e l a t i o n s h i p with third order e q u a t i o n was found b _ t w e e n the
WBC
and time of m o d i f i c a t i o n .
He a t t r i b u t e d his results to
the
rearrangement
of the
starch m o l e c u l e s
e s p e c i a l l y the
amylose
fraction in
addition
to
the
partial
hydrolysis
during the moist
heat and acid m o d i f i c a t i o n treatments.
In
1992, Bekheet
showed that cooking
and drying of
water and
d i l u t e d lye steeped wheat grains increased the
WBC of their
starches.
Nutritional
Methods:
It
is
known that
starch
can be
easily
d i g e s t e d by
monogastric
animals
first
by
action
of
salivary
and
p a n c r e a t i c alpha amylase to produce m a i n l y m a l t o s e and alpha
limit dextrins, and second by the action of intestinal brush
border g l u c o s i d a s e to produce glucose [138].
The enzymes c a t a l y z i n g starch d i g e s t i o n included;
4.4-
i
-amylaser, which attack only the internalgZol,4 links in
starch chains r a n d o m l y to form
dextins.
The latter is
h y d r o l y z e d first
to d e x t r i n s
and finally
to maltose.
[139].
_
B-amylase, which attack ~ - 1 , 4 links of starch m o l e c u l e s
starting
from the
hon reducing
ends.
Also, it
can
split
off m a l t o s e
units
until 1,6
or 1,3
branching
points
are
reached.
This type
of
endwise
action
permits complete
d i g e s t i o n of pure linear
amylose and
permits
the
removal
of
the
external
branches
of
a m y l o p e c t i n . [138].
_
Debranching
enzymes: R-enzyme
breaks
the 1,6 links and
that
continue its action [140].
_
_
_
or 1 , 6
allows
g l u c o s i d a s e can
the B - a m y l a s e to
Amyloglucosidases:
It
produces
by
certain types
of
molds, n a m e l y
Aspergillus ni_~r
A ysami
and R h i z 0 p u s
telemark.
It is able
to h y d r o l y s e starch
to g l u c o s e
units
as an end product
of the digesta.
Also, these
enzymes remove
glucose s t e p w i s l y from the
ends of the
starch chains [141].
P h o s p h o r y l a s e ~ In
the presence of inorganic phosphate,
it transfers
glucose from the non
reducing chain ends
of starch
to form g l u c o s e - l - p h o s p h a t e .
The action of
this enzyme
is inhibited by the presence of 1,6 links.
[142].
The degree
of a m y l o l y s i s
is d e p e n d e n t on
the source,
method of p r e p a r a t i o n , physical and chemical p r o p e r t r i e s
of
starch. [143].
Both Leach and Schoch [144] and S a n d s t e d t et
al.
[141] a t t r i b u t e d the v a r i a t i o n s in starch d i g e s t i b i l i t y
to the following:
86
i
_
_
The d i f f e r e n c e s
in the s t r u c t u r e
of starch
variations
in the
nature of
the b o n d i n g
molecules.
The d i f f e r e n c e s in
the h y d r o g e n
layers of the starch granules.
bonds
due to
between
among
the
its
the outer
U n c o o k e d root and tuber
starches were less s u s c e p t i b l e
to
a m y l o l y s i s than
the
u n c o o k e d cereals
starches. [145].
A c c o r d i n g to Sugimoto et
al. [146] and Dreher et
al. [138]
the
uncooked
starches can
be
divided
into three
groups
a c c o r d i n g to their s u s c e p t i b i l i t y to
the a m y l o l y s i s enzymes
as follows:
(a)
(b)
(c)
The least
d i g e s t i b l e starches, w h i c h
include potato,
canna,
a r r o w root,
sago
palm, a m y l o m a i z e
and banana
s~arches.
The
i n t e r m e d i a t e l y d i g e s t i b l e starches,
w h i c h include
sweet potato and various legumes starches.
The most
digestible
starches
which include
wheat,
normal maize, waxy maize, rice and cassava starches.
The m o d i f i c a t i o n p r o c e s s
a f f e c t e d the d i g e s t i b i l i t y of
starch.
The
in-vitro
digestibility
of
gelatinized
hydroxypropyl
starch by
p a n c r e a t i c was d e c r e a s e d
with the
increase in the degree of' propyl substitution.
The forming
of the
hydroxypropyl
glucopyranose
molecules
during
the
m o d i f i c a t i o n process prevents the h y d r o l y s i s of a d j a c e n t
0([-1,4 g l u c o s i d i c bonds
by these
enzymes. [147,
148]. The
same o b s e r v a t i o n
was reported for
the a c e t y l a t e d starches.
[149].
The results
of W o o t t o n
and
C h a u d h r y [150]
indicated
that
in-vitro
digestibility
of
substituted
starches
by
h y d r o x y p r o p y l a t i o n was more lower than the c r o s s - l i n k e d one.
The c o m b i n a t i o n of both techniques, s u b s t i t u t i o n
and crosslinking,
for m o d i f i c a t i o n
had
an
a c c u m u l a t i v e effect
in
reducing starch d i g e s t i b i l i t y .
The study
of Fanco et
al.
[151]
showed
that
the
large starch
granules
were
more
susceptible
t o , C - a m y l a s e s and a n y l o g l u c o s i d a s e enzymes than
small one.
A b o u - S a m a h a [43] found
that the in-vitro d i g e s t i b i l i t y
of native and
m o d i f i e d maize starches was ~igher
than that
of
rice.
Modification
of
starch,
e s p e c i a l l y with
acid
improved the
starch d i g e s t i b i l i t y .
Also, this
effect was
increased with e x t e n d i n g the m o d i f i c a t i o n time.
The d o m e s t i c p r o c e s s i n g and cooking t r e a t m e n t s improved
the starch d i g e s t i b i l i t y . A u t o c l a v i n g was the most e f f e c t i v e
method
of
increasing
starch d i g e s t i b i l i t y
of
pulses and
wheat,
followed
by
sprouting,
cooking
of soaked
seeds,
cooking of unsoaked seeds,
cooking of sprounts and soaking.
These
t r e a t m e n t s reduced
the
level of
amylase i n h i b i t o r s
which
may
be
responsible
for
the
increase
in
starch
d i g e s t i b i l i t y of
p r o c e s s e d and cooked legume
grains.
Also
the g e l a t i n i z a t i o n
process allows for more
rapid attack of
87
s t a r c h g r a n u l e s by
d i g e s t i v e enzymes.
[59,
152].
The
~ o t a t o and
high amy lose
s t a r c h e s were less
digestible
ovided
lower g r o w t h
r e s p o n s e than
the other
cereal
legume s t a r c h e s . [149].
raw
and
and
4.5Microbiological
Methods:
Tanner
[153]
examined
the flat sour
spores, t h e r m o p h i l i c
a n a e r o b i c and
s u l f i d e s p o i l a g e o r g a n i s m s in c o m m e r c i a l s t a r c h samples.
It
w a s found that
the n u m b e r
of the flat
sour s p o r e s
ranged
from 131 to
193 per grain.
A b o u t 50% and 6 to
30% of the
samples
had
thermophilic
b a c t e r i a and
sulphide
spoilage
organisms respectively.
The
tos
bacterial
count
varied
from 65 to
3 . 1 7 x i 0 6 per g r a m
in c o m m e r c i a l starch.
[154].
In Egypt, the s a m p l e s of c o m m e r c i a l rice s t a r c h e s
were free
from
the gas
producing,
coliform
b a c t e r i a and
contained
relatively
high count
of flat
sour spores.
[62].
AbouSamaha
[43]
found
that
the
flat
sour
bacteria
spores
(F.S.S.)
were
higher
and
total
viable
count
bacteria
(T.V.C.) was lower
in n a t i v e and
m o d i f i e d rice than
maize
starch.
M o d i f i c a t i o n of
starches either
with
acid a n d / o r
m o i s t heat
treatment, r e d u c e d
the T.V.C.
and F.S.S.
This
e f f e c t was m o r e
n o t i c e a b l e in case
of m o i s t h e a t e d
starch
than acid
m o d i f i e d one.
Also, this e f f e c t
increased with
e x t e n d i n g the m o d i f i c a t i o n time.
5-
Starch
Modification
~ Methods:
Native
starches
are
insoluble
in
water
at
room
temperature,
highly
r e s i s t a n c e to
enzymic
h y d r o l y s i s and
lack
specific functional
p r o p e r t i e s [3].
To
a c h i e v e the
desired
characteristics
from
native
starch,
Vogel [155]
suggested
the
using
of
chemicals,
enzymes
and/or
combinations
of them
to
m o d i f y it.
The
modified starch
products
can
be
used
according
to
their
functional
c h a r a c t e r i s t i c s as t h i c k e n i n g , g e l l i n g and b i n d i n g a g e n t s in
a
v a r i e t y of food such as c a n n e d and f r o z e n p u d d i n g s , fruit
pie f i l l i n g s ,
gravies, whipped
t o p p i n g and
c a n d i e s [156].
The
I n t e r n a t i o n a l O r g a n i z a t i o n for S t a n d a r d i z a t i o n has been
d e f i n e d the m o d i f i e d s t a r c h
as a n a t i v e one after
treating
w i t h p h y s i c a l , c h e m i c a l or b i o c h e m i c a l m e a n s to alter one or
more of its
original physical
and/or chemical
properties.
This
definition
includes,
pregelatinization, moist
heat,
o x i d i z e and s u b s t i t u t e s t a r c h e s . J48].
M o d i f i e d s t a r c h e s are
considered
t o x i c o l o g i c a l l y safe
and m a y
be used
in foods
w i t h o u t l i m i t a t i o n s or r e s t r i c t i o n s .
[149] .
5.1-
Hydrolysed
starches:
5.1.1Acid
thinning
starch:
Acid modified starch
is
defined
as
a
starch
materials
in
the
form
of
superficially unchanged granules.
[157i.
It is p r e p a r e d Dy
the a c t i o n of acid
on the a q u e o u s s t a r c h s u s p e n s i o n
at sub
gelatinization temperature
and
c h a r a c t e r i z e d by
less
hot
paste v i s c o s i t y ,
high r e d u c i n g value, low
iodine a f f i n i t y ,
less g r a n u l e
s w e l l i n g , high g e l a t i n i z a t i o n
t e m p e r a t u r e and
low m o l e c u l a r weight. [69].
88
The
first p r e p a r a t i o n
of
an acid
m o d i f i e d or
thinboiling
s t a r c h was
in 1886
by L i n t i n e r
and was
known as
Lintiner
starch.
The L i n t i n e r m e t h o d has been a d o p t e d as a
s t a n d a r d p r o c e d u r e in w h i c h the n a t i v e
s t a r c h is t r e a t e d by
hydrochloric
acid (7.5%) at room t e m p e r a t u r e
for 7 days or
at 40~C for
3 days.
S h o p m e y e r and F e l t o n
[158] used
the
waxy
m a i z e to
p r e p a r e an
acid m o d i f i e d
starch
h a v i n g 62
fluidity.
He
was t r e a t e d a s l u r y of 22 ~ Baume m a i z e s t a r c h
with
s u l f u r i c acid at a t e m p e r a t u r e of 4 8 - 5 5 ~ C for 5 hours.
L a n s k y et al.
[159] i n c u b a t e d a
40% m a i z e s t a r c h
slurries
e i t h e r in 0.07 or in 0.3
N of h y d r o c h l o r i c acid at 50~
for
a p e r i o d s t a r t e d from
2-16 h o u r s and e x t e n d e d to
5-40 d a y s
to p r e p a r e d i f f e r e n t
types of acid
modified starches.
In
1950 Kerr [14] p r e p a r e d acid m o d i f i e d s t a r c h by t r e a t i n g the
starch
slurries
with
0.1-0.2
N
sulfuric
acid
at
a
temperature
of
50-55~
until
the
desired
fluidity
was
obtained.
D u r i n g this p r o c e s s the a m y l o p e c t i n c h a i n s in the
amorphous intermicellar regions
of the s t a r c h
g r a n u l e s are
c l o v e and the
large b r a n c h e d c h a i n s
are e x t e n d e d from
one
crystalline micellar
r e g i o n to a n o t h e r one.
[109].
During
acid m o d i f i c a t i o n
s o l u b l e s u g a r s are g e n e r a t e d
but m o s t of
the
starch
remains
in
the
g r a n u l a r form.
The
linear
molecules produced
d u r i n g the h y d r o l y s i s
are a r r a n g e d into
b u n d l e form and are
r e s p o n s i b l e for the set back
of s t a r c h
p a s t e on c o o l i n g [157]. B u t t r o s e [160] used the h y d r o c h l o r i c
acid
at 8%
concentration
and a
temperature
of 39~
for
preparing
acid
modified
starch
from
tabioca
and
other
s o u r c e s c o n t a i n i n g high a m y l o s e content.
The o b t a i n e d thinb o i l i n g s t a r c h e s had the f o l l o w i n g c h a r a c t e r i s t i c s , r e d u c t i o n
in
hot v i s c o s i t y , r e t e n t i o n of gel s t r u c t u r e and high in an
adhesiveness.
5.1.2Dextrins:
Caesar
[161]
suggested
the
spraying
of the acid over the dry r o a s t e d s t a r c h to p r e p a r e
m o d i f i e d p r o d u c t s i m i l a r in
its p r o p e r t i e s to that p r o d u c e d
by
wet
acid process.
This
method
also c a u s e s
a random
c l e a v a g e of
l i n k a g e s a l o n g the
starch molecule,
producing
d e x t r i n s d i f f e r than that
of an acid t h i n n e d s t a r c h e s ,
It
contains more
w a t e r s o l u b l e solids, f o r m e d
t h r o u g h the dry
roasting
process,
light-tan
to
y e l l o w colour,
very
low
viscosity,
high s o l u b i l i t y
and
m a y or
m a y not
form gel,
according
to the d e g r e e
of d e x t r i n i z a t i o n ,
comparing with
a c i d t h i n n e d one. [9].
5.2Cross-linked
starch:
The
t r e a t i n g of
native
starches
with
di-or
polyfunctional
reagents,
the
cross
l i n k i n g was o c c u r r e d .
T h e s e r e a g e n t s will react w i t h s t a r c h
molecules
at s e l e c t e d
hydroxyl groups
and c r e a t e
a cross
bond
b e t w e e n two
starch molecular
chains.
C o m m o n l y used
reagents
are
adipic
acid,
epichlorohydrin,
phosphorus
o x y c h l o r i d e and
trimetaphosphate.
These
types of m o d i f i e d
s t a r c h e s are high in
m o l e c u l a r w e i g h t ~han n a t i v e one.
It
can
be used w h e n a
s t a b l e nigh v i s c o s i n y
s t a r c h p a s t e are
needed, e s p e c i a l l y at p r o l o n g e d s e v e r e heat t r e a t m e n t , shear
89
force
and/or low pH. [162,
163].
The
m e t h o d of p r e p a r i n g
these
p r o d u c t s is
based on
treating the
a~ueous alkaline
starch s u s p e n s l o n at 20-50~
with
a cross-li k reagent at a
level
of
0.005-0.1%
for the
proper
time.
The
treated
s u s p e n s i o n are
filtered
washd and dried.
G e n e r a l l y , this
process causes
a d r a s t i c changes in starch c h a r a c t e r i s t i c s .
It increase
the
t e m p e r a t u r e of
hydration,
the
stability
under acidic
conditions, both
heat tolerance
and s h e a r i n g
r e s i s t a n c e of starch [139,
162].
C r o s s - l i n k e d starches are
used
by foods that heated for an e x t e n d e d time or s u b j e c t e d
to high shear.
It does not show s u b s t a n t i a l
refrigeration
or
freezing stability.
Therefore,
it can De
used in high
acid
food
systems
such
as sauces
os
pizza,
spaghetti,
cheese, barbeque and hot filled systems such as pie fillings
and
bakery
glazes in
addition
to
a specially
processed
products such as puddings. [9].
5.3- Substituted
starch:
The s u b s t i t u t i o n r e a c t i o n s
introduce m o n o f u n c t i o n a l group at
the hydroxyl group of the
starch
molecule. C o m m o n l y
used reagents
are
acetic acid,
acetic
anhydride, vinyl
acetate, acetyl
guanidine, acetyl
phosphate and
p r o p y l e n e oxide. [7].
R u t e n b e r g and Solarek
[162] s u g g e s t e d the using
of starch a c e t a t e c o n t a i n i n g 0.51.5%
acetyl
groups
in
food
industry
to
provide
sol
stability,
h y d r o p h o b i c c h a r a c t e r i s t i c s and good c l a r i t y for
the food p r o d u c t s , e s p e c i a l ~ y that storage at low t e m p e r a t u r e
S u b s t i t u t e d starches had the following
c h a r a c t e r i s t i c s ; low
temperature
of hydration, good
clarity, low
s y n e r e s i s and
less f r e e z e - t h a w stability.
Therefore, it is only suitable
for p r e p a r i n g thick textured
foods.
It is not
suitable to
utilize in
acid foods s u b j e c t e d to
either high t e m p e r a t u r e
and/or shear forces. [9].
Paschall
[164] p r e p a r e d sta::ch
p h o s p h a t e m o n o e s t e r by
heating
intimate
blends
of
10%
moisture
starch
and
orthophosphates
at pH 5-6.5 for 0.5 to 6 hours at 1 2 0 - 1 6 0 ~ C
The obtained
product
haa a
gooa
paste c l a r i t y
and
high
s t a b i l i t y to r e t r o g r a d a t i o n .
The v i s c o s i t y of this type
of
starch
can
be
controlled
by
adjusting
the
amount
of
p h o s p h a t e salts, reaction t e m p e r a t u r e , t i m e and pH. R u t e n b e r g
and
Solarek
[162]
suggested
the
using
of
sodium
t r i p o l y p h o s p h a t e and urea with o r t h o p h o s p h a t e s for p r e p a r i n g
starch p h o s p h a t e products.
5.4- Oxidized
starch: Oxidized
starch is m a n u f a c t u r e d
by
treating
the
starch
slurry
with
alkaline
sodium
hypochlorite.
This
reaction causes
a random
cleavage of
linkages
along
the
starch
m o l e c u l e and
bleaches
starch
colour. [162].
According
to H u l l i n g e r [165]
the oxidized
starches
have limited food
f u n c t i o n a l i t y compared to other
types
of m o d i f i e d starches.
It has low viscosity, low heat
for hydration,
less g e l l i n g p r o p e r t i e s and
good dry powder
flow.
Thereforew, it can be used as d u s t i n g agents for such
foods
as
m a r s h m a l l o w , chewing
gum
and
in p r e p a r i n g
the
tablets of the p h a r m a c e u t i c a l industry. [9].
90
5.5- Moist-heat
treated
starches:
All the
above
types
of s t a r c h e s
r e q u i r e heat
to c a u s e d the
g r a n u l e s to
hydrate
and swell.
However,
there are other
two types of
modified
starches which
can d i s p e r s e in w a t e r
s y s t e m s and
known
as m o i s t
heated
starches.
These
p r o d u c t s can
be
classified
a c c o r d i n g to the c h a n g e s in the g r a n u l e shape of
native starches during
the m o d i f i c a t i o n .
The first type is
substantially
degraded during
the m a n u f a c t u r i n g
and known
c o m m e r c i a l l y as p r e g e l a t i n i z e d starch.
The other type keeps
its g r a n u l a r shape and r e f e r r e d to as cold w a t e r swell (CWS)
or
cooked
up
starch
[9].
Sair
[iii]
suggested
the
a d j u s t m e n t of
the s t a r c h
m o i s t u r e to
a definite
level to
c o n t r o l the w a t e r a b s o r p t i o n
of the g r a n u l e s d u r i n g h e a t i n g
in c l o s e d
container.
The
m o i s t heat
treatment
of
27%
moisture content maize
and p o t a t o s t a r c h e s for 16
hours at
I 0 5 ~ C led to:
i234-
Slight
changes
in
the
microscopic
appearance
and
r e d u c t i o n in the s w e l l i n g power of p o t a t o starch, w h i l e
no c h a n g e s were o b s e r v e d rot tne m a i z e starch.
The g e l a t i n i z a t ~ o n
~empera~ure
of
p o t a t o and
maize
s t a r c h e s . w a s i n c r e a s e d from
61 to 80~C and 67
to 75~C
respectively.
The v i s c o s i t y was
r e d u c e d as c o m p a r e d
w i t h the n a t i v e
one.
The
X-ray
crystalli~e pattern
of
potato starch
was
changed
from (B) to (A)
and (C) types.
H o w e v e r , the
m o d i f i e d m a i z e s t a r c h had the same
crystalline pattern
(A) type of the n a t i v e one.
The
a d j u s t m e n t of
moisture content
of n a t i v e
maize,
wheat
and rice s t a r c h e s to 25% b e f o r e h e a t i n g for m o r e than
3 hrs at 205~
are r e q u i r e d to p r e p a r e m o i s t
heat m o d i f i e d
starches.
[27].
The m o i s t h e a t e d s t a r c h can be p r o d u c e d by
h e a t i n g of w h e a t and p o t a t o
s t a r c h e s of 18, 21, 24
and 27%
m o i s t u r e c o n t e n t for 16 hours at 100~
in a s e a l e d c o n t a i n e r
[136].
The
work of D o n o v a n et al. [102] on h e a t i n g both w h e a t
and
p o t a t o s t a r c h e s of 27% m o i s t u r e c o n t e n t for 16 hours at
100~
s h o w e d that:
i-
2-
No c h a n g e s in
the c r y s ~ a l i i n i t y
pattern.
A
type, of
wheat starch
was noticed.
While the
(B) c r y s t a l l i n e
pattern
of
potato
starch
was
changed
to
a
i:i
c o m b i n a t i o n of (A) and (B) types.
This i n d i c a t e d that
such heat t r e a t m e n t c a u s e d
a d r a s t i c a l t e r a t i o n in the
p h y s i c a l n a t u r e of p o t a t o g r a n u l e s .
The DSC r e s u l t s p r o v e d that ~he g e l a t i n i z a t i o n range of
these types
of
moist heated
s t a r c h e s was
broadened.
Their g e l a t i n i z a t i o n
t r a n s i t i o n could be
d e s c r i b e d as
biphasic
or two peaks in the e n d o t h e r m y region.
These
r e s u l t s m a y be a t t r i b u t e d e i t h e r to the p r e s e n c e of two
types of s t a r c h
g r a n u l e s s t r u c t u r e or one
kind but in
two d i f f e r e n t e n v i r o n m e n t s .
91
_
Generally,
the
final
geiatinlzation temperature
was
for both
hincreasedwandeat
antis poSWtato
~eliin~t~rches~
in. w a t e r was r e d u c e d
Hoseny
[30] s t a t e d
that,
the (B)
p a t t e r n of
potato
s t a r c h can be c o n v e r t e d to A p a t t e r n by a p p l i c a t i o n of m o i s t
heat t r e a t m e n t .
The c o m m e r c i a l p r e g e l a t i n i z e d s t a r c h can be
produced
by h e a t i n g the 40%
s t a r c h s l u r r y on
drums.
This
p r o c e s s gave c o o k e d s t a r c h d i s p e r s i b l e in cold
water with a
less
g e l l i n g power
comparing
with n a t i v e
s t a r c h used
in
b a k e r y and e x t r u d e d p r o d u c t s .
[15].
6-
Food
starch
applications
:
The
food
induustry
comprises
one
of
the
largest
c o n s u m e r of
s t a r c h and
s t a r c h p r o d u c t s [166].
In
sugary
c o n f e c t i o n e r y , s t a r c h is used as a b a s i c g e l l i n g i n g r e d i e n t ,
as a
filler and as a m o u l d i n g
base.
D i f f e r e n t t e x t u r e s of
gums, j e l l i e s and p a s t i l l e s can be o b t a i n e d by using v a r i o u s
types of w a t e r
binding gelling
agents, principally
arabic
gum,
modified
starch,
gelatine
and
pectin.
Using
of
unmodified
starch gives
an u n s a t i s f a c t o r y
products having
The c r o s s - l i n k e d m o d i f i e d
short t e x t u r e a n d weak b o d y [9].
food t h i c k e n e r in
c a n n e d pie filling.
starch
can u se as a
O x i d i z e d star ch can be used to p r o d u c e a v e r y tender gum and
d o u g h as
a
carotene
s t a b i l i z e r to
prevent
in
spaghetti
L o r e n z and Kulp [•
used the m o i s t heat
oxidation.
[I 67].
starch
~o p r e p a r e
b r e a d and
CaKe.
They
m o d i f i e d pota to
products containing
modified starch
had a
found
that t he
than
the control.
According
to
higher
water absorption
of
modified
s t a r c h e s had
the
Galliard
[166],
each type
f o l l o w i n g a d v a n t a g e and uses:
i
_
_
_
4 -
5-
A c i d - t h i n n e d m o d i f i e d s t a r c h has low hot p a s t e v i s c o s i t y
and,
high gel v i s c o s i t y .
T h e r e f o r e , it can be used in
gums and jellies.
Cross-linked starch
i n c r e a s e s the s t a b i l i t y
of f r o z e n
p r o d u c t , the pH and the shear r e s i s t a n c e of the
canned
and frozen foods.
E s t e r i f i e d s t a r c h r e d u c e s the s e t - b a c k and i n c r e a s e s the
c l a r i t y of frozen foods.
O x i i z e d s t a r c h i n c r e a s e s the s t a b i l i t y of f r o z e n foods.
Pregelatinized starch dissolves
in cold water.
So it
can be used in p r e p a r i n g pie s
coat.
According
to
the
results
of
Abou-Samaha
(43j
the
p a n e l i s t s did
not find
any v a r i a t i o n s b e t w e e n
the s e n s o r y
properties
of y e l l o w
cake made
~rom 100% w h e a t
flour and
those p r e p a r e d from 100% 6 and
iu hrs m o i s t h e a t e d of m a i z e
and
5
hr m o i s t
h e a t e d of
rice s t a r c h e s .
The l e a v e n i n g ,
general
appearance,
crust
color,
crumb
colour,
crumb
texture,
porous
distribution
and
taste
of
cake
were
negatively
affected when
n a t i v e and
other's moist
heated
m a i z e and rice s t a r c h e s were used.
92
7-
Conclusions:
This
review
has
shown
that
the
physico-chemical
c h a r a c t e r i s t i c s of s t a r c h are
a f f e c t e d by v a r i e t y , c l i m a t e ,
sources, method
of i s o l a t i o n
and m o d i f i c a t i o n .
More than
one
m e t h o d are
needed
to c n a r a c t e r i s e
and to
s e l e c t the
s t a r c h of a g i v e n g u a l i ~ y for a s p e c i f i c u t i l i z a t i o n .
8i23-
REFERENCES:
L i n e b a c k D.R. B a k e r s Digest. 1984; 58; 16.
A n o n y m o u s . W a s h i n g t o n : Corn R e f r i n e r s Assoc. Inc.1983.
H e c k m a n E.
in: Graham. D.G.
ed. W e s t p o r t :
Avi. Pub.
Co. Inc. 1977; 464-499.
4Kalichevisky
M.T;
Orford.
P.D.;
Ring.
S.G.
C a r b o h y d r a t e Res. 1990; 198; 49.
5M o o r t h y S.N; R a m a n u j a m T. Strarke. 1986; 36; 58.
6O'Dell.
J.
In: B l a n s h a r u ;
Mitchell.;
eds. London:
B u t t e r w o r t h . 1979, 171.
7Biliaderis C.G.J.
Agric. Food Chem. 1982; 30; 925.
8Smith P.S.
In:
Lineback
D.R.;
I n g l e t t G.E.
eds.
W e s t p o r t : AVI Pub. Co. Inc. 1982; 237-269.
9L u a l l e n T.E. Food Tech. 1985; 39; 59.
i0A n d e r s o n R.A. C e r e a l Chem. 1982; 59; 265.
iiBesso R. London: A p p l i e d Sci. Pub. LTD. 1976; 1-25.
12Lii
C.Y; C h a n g S.M.
Proc. Natl Sci
C o u n c Roc. 1978;
2; 146.
13Takeda
Y;
S h i t a o z o n o T;
Hizukuri
S. Starke.
1988;
40; 51.
14Kerr R.W. New York: A c a d e m i c Press. 1950; 3-25.
15C o l o n n a P;
M e r c i e r C.
Carbohydrate
Res. 1984;
131;
233.
16Banks
S.W; G r e e n w o o d
C.T., Muir.
D.D. Starke.
1974;
26; 73.
17Larson
BI; G i l i e s
K.A; J e n n e s s R.
Annal Chem.
1953;
25; 802.
18McCready
R.M; H a s s i d
W.H. J. Amer.
Chem. Soc.
1943;
65; 1154.
19Schoch T.J. In"
W h i s t l e r R.L. ed.
New York: A c a d e m i c
Press. 1964, 64-66.
20Lineback D.R.U.K.:
A A C C - F M B R A Joint C o u r s e on Starch.
1990.
21Kamal
M.A.M.
M.Sc.
Thesis. Ain
Shams U n i v . ; E g y p t .
1963.
22Kamal M.A.M,
R a s h e d A.G.E.
Hadidy. Z.A.F. Ain
Shams
Univ., Egypt. Res. Bull. 1968; 155.
23J u l i a n o B.O. Food R e v i e w s I n t e r n a t i n a l . 1986; 1;423.
24Munck L; Rexen F.; H a a s t r u p L. Starke. 1988; 40;
81.
25Schoch
T.J.
In: W h i s t l e r
R.L.
P a s c h a l l L.E.,
eds.
New York: A c a d e m i c Press. 1967. 79-235.
26Rickard
J.E.; Asaoka. M;
B l a n s h a r d J.M.V. Trop.
Sci.
1991; 31; 189.
27Fukui T; Nikuni
Z. Agric. &
Biolog. Chem. 1969;
33;
460.
28F r e n c h D. J. Jap. Soc. S t a r c h Sci. 1972; 19; 8.
29Robin
J.P; M e r c i e r .
C;
c h a r b o n n i e v ~;
GuiiDo~.
A.
93
303132333435363738394041424344454647484950-
5152535455-
5657585960616263-
Cereal Chem. 1974; 51; 289.
l~A~'_vv USA. A m e r i c a n
Chem1~~H~
Association
of
Cereal
Taylor T.C; Nelson
J.M.J.
Am. Chem. Soc.
1920; 42;
1738.
Gracza
R. In:
W h i s t l e r R.L;
Pascall A.F.
New York:
A c a d e m i c Press. 1965; 105-131.
Ai-Bayati
Lorenz
K. L e b e n s m i t t e l
Wissenschaft
und Tech. l~Ib; ~; 163.
E i - S a i e d H.M. Starke. 1979; 31; 84.
W a n k h e d e D.B; Umadevi S. S~arke. 1982, 34, ~62.
Soni O.; A g r a w a l A. Starke. 19~3; 35; 4.
Swinkels. J.J.M. StarKe. •
37;1.
Leelavathi K; ~ndrani D. $~arKe. i ~ 7 7
39; 378.
Tian
S.J; RicKard
J.E; B i a n s n a r d
J.M. J.
~ci. Food
Agric. 1991, 57; 459.
Reyes
A.E; Briones
V.P; Juiiano B.O.
J. Agric.
Food
Chem. 1965; 13; 438.
Briones
V.P; M a g b a n v a L.G;
Juiiano B.O. Cereal
Chem.
1968; 45; 351.
W a l l i s T.E. London: Churchill. J.A. Ltd. 1967; 7-20.
Abou-Samaha
O.R.M.
Ph.D.
Thesis,
Alex.
Univ.,
A l e x a n d r i a , Egypt. 1991.
Takeda Y;
T o k u n a g a N.T; Takeda C; Hizukuri
S. Starke.
1968; 38, 345.
Rasper V. J. Sci. Food Agric. 1971; 22; 572.
Goering K.J; Dehaas B. Cereal Chem. 1972; 49; 712.
Kulp K. Cereal Chem. 1973; 50; 666.
R a d l e y J.A.
London: A p p l i e d Sci. Publ. Ltd.
1976: 51107.
M o r r i s o n W.R; Gadan H. J. Cereal Sci. 1987; 5; 263.
Lopez
O.P;
Calvino E.C;
(:astaneda J.G;
Rivera R.M.
Starke. 1988; 40; ii.
Schoch T.J; M a y w a l d E.C. Anal. Chem. 1956; 28; 382.
M a c - M a s t e r s M.M; Wold. M.J; Seckinger H . L . J .
Agric.
Food Chem. 1957; 5; 455.
Bechtel
D.B;
Zayas
I,
Kaieikan
L.
Pomerans.
Y.
Cereal Chem. 1990; 6'I; 59.
Waldt L; Kehoe D. Food Tech. 1959; 13; i.
W h i s t l e r R.L;
Paschall E.F. New York.
A c a a e m l c Press.
1967; 65-78; 293-334.
Dreher M.L; Berry J.W. StarKe. 1983; 35, 76.
B o u w k a m p J.C. (ed.). Boca Raton. FL.: CRC Press. 1985.
Fuwa H; Sugimoto. Y; Tanaka M. Starke. 1978; 30; 186.
Bekheet M.H.
Ph.D Thesis, A l x a n d r i a U n i v e r s i t y , Alex.,
Egypt. 1992.
Smith P.S. In:
W h i s t l e r R.L., ed.
New York: A c a d e m i c
Press. 1964; 41-42; 103-106.
Hussein
M;
Ei-Waraki
A.G;
Saied.
A;
Toma.
R.B;
Hassan
Y.M. Ain
Shams Univ.,
Egypt. Ann.
Res. Bull.
1970, 157.
Ei-Refai A.A.
M.Sc. Thesis, Alex.
Univ., A l e x a n d r i a ,
Egypt. 1970.
Hizukuri S; Takagi T. C a r b o h y d r a t e Res. 1984; 134; i.
94
646566676869707172737475767778-
79808182838485868788899091929394959697-
Kainuma
K. In:
Stumpf
P.K;
Conn.
E.E. New
York:
A c a d e m i c Press 1988; 141-180.
Hizukuri
S;
Takeda
Y.;
Yasuda
M;
Suzuki.
A.
C a r b o h y d r a t e Res. 1981; 94; 205.
Park J; J o h n s o n M.J. Biol. Chem. 1949: 181; 149.
H i z u k u r i S. C a r b o h y d r a t e Res. 1985; 141; 295.
H i z u k u r i S. C a r b o h y d r a t e Res. 1986; 147; 342.
Inouchi N; Glover. D.V; Fuwa H. Starke. 1987; 39; 259
Zobel H.F. Starke. 1988; 40; 44.
Zaslow
B. In:
W h i s t l e r R.L;
P a s c h a l l E.F.
eds. New
York: A c a d e m i c Press. 1965; 179-286.
Zobel H.F; M e t h o d s in c a r b o h y d r a t e Chem. 1964; 4; 109.
French
D. In: W h i s t l e r
R.L; B e - M i l l e r J.N;
Paschall.
E.F. eds. New York. A c a d e m i c Press 1984; 184.
H i z u k u r i S. J. Jap. Soc. S t a r c h Sci. 1969; 17; 73.
A b d - A l l a h M.A; Foda Y.H; M a h m o u d R.M; A b o u - A r a b
A.A.
Starke. 1987; 39; 9.
K o m i y a T; Y a m a d a T; Nara S. S~arKe. •
29; 308/
Ghiasi
K; H o s e n e y R.C;
Varriano- Mors~on.
E. C e r e a l
Chem. 1982; 60; 5~.
H o s e n e y R.C;
L ~ n e p a c k D.R;
Seib P.A.
B a K e r s Digest.
1983, J u l y / A u g u s t , 65.
Biliaderis
C.G;
G r a n t D.R;
Vose. J.R.
C e r e a l Chem.
1981; 58; 502.
K o m i y a T; Nara S. Starke. 1986, 38, 9.
Olku J; Rha. C. F o o d ~ C h e m . 1978; 3; 293.
Biliaderis
C.G; M a u r i c e
T.J; Vose J.R.
J. Food
Sci.
1980; 45; 1669.
Leach
H.W. In:
W h i s t l e r R.L; P a s c j o l l
E.F; eds.
New
York: A c a d e m i c Press. 1965; 289-397.
D o n o v a n J.W. B i o p o L y m e r s . 1979; 18; 263.
Brown
A.H. In:
R e s e a r c h R e p o r t of
the C o n f e r e n c e
on
Role of W h e a t
in W o r l d ' s Food.
WRRL.
Albany; Calif.;
USA. 1962.
N e u f e l d C.H.H; W e i n s t e i n N.E; M e c h a m D.K. C e r e a l
Chem.
1957; 34; 360.
Smith P.S;
Barta E.J;
Lczar
M.E. Food
Tech. 1964;
18; 89.
Hari P.K; Garye S; G a r y S.K. StarKe. 1989; 41, 88.
Banks
W;
G r e e n w o o d C.T.
Edinburgh: Eainburgh
Univ.
Press. 1975.
W a t s o n S.A. In: W h i s t l e r . R.L., ea. New YorK: A c a d e m i c
Press. 1964, 240-242.
Song
E;
Shi.
M.S.;
Hang
Y.H.
Han'guk
Nanghwa
H a k h o e c h i . 1987; 3U: 242.
Kitada
Y; Sasaki
M; Y a m a z o e
Y; N a K a z a w a .
H. N i p p o n
S h i k o - k u h u Kogyo G a k k a i s h i . 1 9 ~ ;
35; 135.
D u c k w o r t h R.B. J. Food T e c h n o l o g y . 1971; 6; 317.
McMasters
M.M.
~n"
Whistler
R.L;
ed.
New
York"
A c a d e m i c Press. 1964, 4, 233-240.
Toyama
T; H i z u k u r i
S;
Nikuni
Z.J.
Tech.
Soc.
Starch. 1966; 13, 69.
T a k e d a Y; H i z u k u r i S. C e r e a l Chem. 1983; 60; 212.
W o o t o n N; B a m u n u a r a c h c h i .
Starke. 1979; 31; 262.
95
98-
Kujimiya
M; Donovan
W; Wong
k.f. StarKe.
1980; 32;
265.
99Russel P.L; Juliano B.O. StarKe. i983; 35; 382.
100- Buck J.S; Walker C.E. StarKe. 1988; Yo, 35~.
i01- Tester R . F ; M o r r i s o n W.R. Cereal Chem. 1990; 67; 551.
102D o n o v a n J.W;
Lorenz
K.L; Kulp K.
Cereal Chem. 1983;
60; 381.
103Lyne F.A.
In: R a d l e y
J.A; ed.
London: A p p l i e d
Sci.
Pub. LTD. 1976; 133-167.
104E l - S a a d a n y R.M.A;
Foda Y.H;
E l - S a a d a n y F.M.
Starke.
1979; 31; 230.
105D e l p e u c h F;
Favier
J.C. Ann.
Technol. Agric.
1980;
29; 53.
106- B a d e n h u i z e n
N.P. In:
R u h l a n d W. ed.
Berlin, Germany:
Sprmger- V e r l a g 1958; 6; 137.
107Miller B.S; Derby. P.I;
Trimbo H.B. Starke. 1973; 32;
186.
108Bowler P; W i l i a m s
M.R; A n g o l d R.E.
Starke. 1980; 32;
186.
109Leach H.W;
Mc
Cowen L.D;
Schoch
T.J. Cereal
Chem.
1959; 36; 534.
ii0- M a d a m b a
L.S.P; ~ u s t r i l l o s A.R; San
Pedro E.L. P n i l i p p
Agric. 1975; 5~; 338.
iii- Sair L. Cereal Chem. 1967; 44; ~.
112- Lorenz K; Kulp K. Cereal Chem. 19~2; 9~; 46.
113- C h r i s t i a n s o n
D.D; Bagiey
E.B. Cereal Chem.
19~;
60;
116.
114- Kerr R.W. New York: A c a d e m i c Press. 1944.
115- Eliasson A . C . J .
Texture Studies. 198%; 17; 25J.
116- Takahashi R; Ojima T. StarKe. i969; 21; 318.
117- Cereda
M.P; Wosiacki G; Gonceicao.
F.A. Cienc Technol
Aliment. 1985; 5; 61.
118Seog
H.M;
Park Y.K;
Nam
Y.J;
Shin
D.H; Kim
J.P.
Han'guk Naghwa Hokhoech, 1987; 20; 179.
119Rohwer R.G;
Klem
R.E. In'
W h i s t l e r R.L;
Be-Miller.
I.N;
Paschall
E.F;
eds. New
York"
Academic
Press.
1984; 529-541.
120- Sing V; Zakiudin A.S. Starko . 1987; 39; 277.
121- Seib
P.A; M a n i n g a t
C.C.U.K:
A A C C - F M B R A Joint
course
on starch. 1990.
122- Myers
R.R; Knauss C.J;
H o f f m a n R.D. J.
Appl. Polymer
Sci. 1962; 6; 659.
123Evans I.D; Haisman D.R.
J. Texture Studies. 1979; i0;
347.
124- Wong
R.B.K;
Lelierre J.
J. App.
Polymer Sci.
1982;
27; 1433.
125- Ferry J.D. Chichester.
John W i •
and Sons Inc.
1980;
1-55.
126Bell
A.E.
In:
Hadman
T.M.
ed.
London:
Elsevier
A p p l i e d Sci. 1989; 251-175.
127Ross-Murphy
S.B.
In:
Cnan
H.W.S;
ed.
London:
Blackwell S c i e n t i f i c Publ. 198~; 13~-i99.
128M e s t r e s C; C o l o n n a P,
Bulean A. J.
Cereal Sci. 1988;
7; 123.
96
129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164-
Del R o s a r i o
R.R;
P o n t i v e r o s C.R.
Starke. 1983;
35;
86.
R a s p e r V. J. Sci. Food Agric. 1969; 20; 165.
C e r e d a M.P; W o s i a c K i G. Starke. 1985; 37; 404.
H o o v e r R; S o s u i s k k i F. StarKe. 1985; 37, 397.
R a s p e r V; De Man J.M. C e r e a l Chem. 19~0; 57; 331.
Mac-Athur
L.A;
D ' a p p o l o n i a B.L.
Cereal
Chem.
1979;
56; 455.
D r e h e r M.L; P a d m a n a b a n G; F r a z i e r A.R; J. Agric.
Food
Chem. 1983; 31; 844.
Kulp K; L o r e n z K. C e r e a l Chem. 19~i; 58; 46.
Mok C; Dick J.W. C e r e a l Chem. 1991; 68' 405.
D r e h e r M.L; D r e h e r C.J; B e r r y J.W. CRC. C r i t i c a l
Rev.
Food Sci.& Nutr. 1984; 20; 47.
R a d l e y J.A. London: C h a m p e n and Hall Ltd. 1968; 247.
S a n d s t e d t R.M. C e r e a l Sci. ~oday. 1965; i0; 305.
Sandstedt
R.M. S t r a h a m D;
Veda. S; A b b o t
R.L. C e r e a l
Chem. 1962; 39; 123.
F l e m i n g S.E; V o s e J.R; J. Nutr. 1979; 109; 2067.
Abbas
I.R; S c h e e r e s
J.C;
Berry
J.G. Starke.
1987;
39; 280.
Leach H.W; S c h o c h T.J. C e r e a l Chem. 1961; 38; 34.
S u g i m o t a Y;
T a k a y a T; Fuwa H. J.
Jap. Soc. Starh Sci.
1976; 23; 117.
S u g i m o t a Y; O h m i s h
K; T a k a y a T; Fuwa
H. J. Jap.
Soc.
S t a r c h Sci. 1979; 26~ 182.
L e e g w a t e r D.C; Luten J.M. Die StarKe. 1971; 23; 43U.
W o o t t o n N; C h a u d h r y M.A. StarKe. ~979; 31; 224.
WHO. Tech. Rep. Ser. No. 439. WHO; Geneva. i974.
W o o t t o n N; C h a u d h r y M.A. StarKe. •
33; i35.
F r a n c o G.M.L;
P r e t o S.J.~;
C i a c c o C.F;
T a v a r e s D.Q.
Starke. 1988; 4; 29.
Jood S; C h a n h a n
B.M; Kapoor. A.C. J. Food
Chem. 19~8;
30; 113.
T a n n e r F.W. USA: G a r r a d Press. 1994; 673-674.
S p i c h e r G. Bortu. Geback. 1960; i~; 130.
Vogel W.F. C o n s e r v a . 1969; 18; i.
Hood L.F; S e i r f r i e d A.S;
Meyer R. J.
Food Sci. 1974;
39; 117.
S h i l d n e c h P; Smith C.E. In:
W h i s t l e r R.L; P a s c a l l E.F;
eds. New York: A c a d e m i c Press. 1967; 217-234.
S h o p m e y e r H.H; F e l t o n G.E. In: R a d l e y J.A. ed. L o n d o n ;
C h a m p a n and Hall. 1943, 451.
L a n s k y S;
Kooi. M.
S c h o c h T.J.
J. Amer.
Chem. Soc.
1949; 71; 4066.
B u t t r o s e M.S. Die Starke 1963; 14; 85.
C h a m p e n and
C a e s a r G.V.
In: R a d l e y
J.A. ed. London:
Hall Ltd. 1968; 282.
R.L;
BeRutenberg
M.W;
Solarek
D. In:
Whistler.
York : A c a d e m i c
Miller
J.N;
Paschail
E.F; eds.
New
Press. 1984; 312-366.
und T e c h n o l o g i e .
C o l a s B.
Lebensmi~el.
Wissenscna~t
1986; 19; 308.
Paschall
E.F.
In:
Wnis~ler
~.L. , ed.
New
YorK:
97
A c a d e m i c Press. 1964; 294.
Hullinger
C H.
In"
Whistler
R.L.
A c a d e m l c Fress. 1964, 313.
166- G a l l i a r d T. C h i c h e s t e r : John W i l e y and
167- Z u s m a n S. Nahrung. 1985; 29; 905.
165-
ed.
New
Sons
Inc.
York1987.
This Page Intentionally Left Blank
D. Wetzel and G. Charalambous (Editors)
Instrumental Methods in Food and Beverage Analysis
9 1998 Elsevier Science B.V. All rights reserved
99
Specific methods for the analysis of identity and purity of functional
food polysaccharides
Francisco M. Goycoolea a and Ioannis S. Chronakis b
Research Center for Food and Development (C.I.A.D., A.C.),
P.O. Box 1735, 83000 Hermosillo, Sonora, Mexico
a
b Physical Chemistry 1, Center for Chemistry and Chemical Engineering,
Lund University P.O. Box 124, S-22 100 Lund, Sweeden
INTRODUCTION
All foods with few exceptions contain biopolymers, namely polysaccharides and
proteins, occurring either as natural constituents of edible living tissues (e.g.
connective tissue and muscle fibers in meat and fish, cell walls of fruits and
vegetables, etc.) or as aid-agents intentionally added to manufactured food in order to
increase sensory acceptability and physical stability. The term functional has been
coined to the range of physico-chemical properties largely affected by polysaccharide
and protein macromolecules in the architecture of foods, namely those related to one
or more of the following:
9 Generation and control of product rheology (e.g. viscoelastic and plastic properties,
etc.)
9 Melting and setting temperature of food gels
* Water and fat binding capacity
9 Glass transition temperature
9 Crystallization inhibition
. Surface activity (e.g. emulsions and foams formation and stabilization)
. Thermodynamic properties, (e.g. heat capacity, heat stability, phase behavior, etc.)
9 Flavor encapsulation (e.g. spray drying, microencapsulation)
9 Sensory properties (e.g. texture, mouthfeel and flavor release)
Due to their ability to form colloidal aqueous solutions, functional polysaccharide
gums are also referred to as hydrocolloids. The effectiveness of hydrocolloids to
modify and/or control functional properties of foods is modulated by their molecular
structure and conformation. The replacement of fat by modified starches in many new
'low fat' versions of foodstuffs in today's market place, is just an example of the huge
commercial importance of hydrocolloids gmns in the engineering of new food
products [ 1]. There is hardly any manufactured food or drink that does not contain a
stabilizing hydrocolloid as a part of its formulation.
100
There is a well established structure-function relationship underlying the individual
behavior of different hydrocolloids. However, the great complexity of multicomponent real foods, allow at present only limited understanding of the precise
behavior of each of the individual pieces in the food system machinery. The choice of
a functional additive is still largely done on an empirical basis. The main reason for
this is because of the limited information available between structural parameters of
the different functional hydrocolloids and
key characteristics of commercial
importance such as heat stability, thixotropy, shear viscosity, etc.. Once a given
hydrocolloid(s) is set to perform successfully in the product, the manufacturer aims at
a constant quality from the supplier of that ingredient. Quite often though, the food
manufacturer does not even know about the nature of the purchased gums, since these
are sold as commercial blends of two or more ingredients under trademark names (e.g.
'ice cream mix'). However, problems can arise for instance, when a change in
supplier is made or a new different batch is received from the same supplier.
Certainly, in many instances, it is the hydrocolloid producer the one who assists food
companies with the expertise and understanding about the use of their products, since
they must have a soundly based and detailed knowledge of the chemical and steric
structure of the materials they deal with.
With the notable exception of gelatin, industrial hydrocolloids are all
polysaccharides. These include starches and a
fair number of
different
polysaccharide materials from plant, seaweeds and microbial (e.g. by use of controlled
liquid fermentation) sources. It is also worthy of mention that the majority of
functional food polysaccharides, digestible starches apart, belong to the non-digestible
carbohydrate fraction of food (i.e. the dietary fiber) from the nutritional viewpoint, so
many of the analytical tests provided here might be relevant to the analysis of the
individual components of dietary fiber. Nevertheless, coverage of the current methods
for the determination of dietary fiber as such, lays beyond the scope of this chapter.
The focus of this review is thus, on selected instrumental analytical methods, which
can provide 'fmgerprints'
of the free structure characteristics leading to the
unambiguous identification of some industrial polysaccharides. First, an overview is
presented on the different techniques and strategies available for analysis of
polysaccharide primary structure, which is followed by a section on specific examples
of different analytical problems found in food polysaccharides. The systems covered
are: i) starch, iii) gum arabic, iv) alginate, and iv) carrageenan. A summary of
analytical tests to identify the presence of hydrocolloid gums in foods, is also included
as the final section of this review. Although the major structure features of each of the
polysaccharide systems covered in the above examples and current analytical
strategies to determine such features are outlined in this review, coverage by no means
pretends to be exhaustive. The reader is also referred to excellent treatises on
polysaccharide structure analysis [2, 3].
I01
ANALYTICAL STRATEGIES
Carbohydrate Chemistry
For those readers unfamiliar with carbohydrate chemistry, a few brief concepts on
primary structure and chain geometry are introduced here. The basic building block in
all industrial polysaccharides is a 6-membered (pyranose) sugar ring, composed of five
carbon atoms and one oxygen. In the projection shown in Fig. 1 carbon atoms are
customary numbered clockwise from the ring oxygen, with C(6) lying outside the
ring. The stable conformations of the pyranose ring are chair forms (4C1 and IC4) in
which all bonds are fully staggered [4]. In the chair conformations, substituents at each
carbon atom may be present in either equatorial locations, or in crowded axial
positions above or below the ring. Hexose monosaccharides (sugars with 6 carbon
atoms), are classified in two groups, according to the steric configuration at C(5), the
position of ring closure. In the D series C(6) is equatorial in the 4C1 ring form, while in
the mirror image L series (i.e. their enatiomers), the corresponding stable chair form is
IC4. Out of the possible isomers given by the configurations at C(2), C(3) and C(4),
only some of them occur in nature. In the stable ring form of glucose (4C~ for D and
1C4 for L) O(2), 0(3) and 0(4) are all equatorial; in mannose 0(2) is axial; in
galactose 0(4) is axial; in gulose 0(3) and 0(4) are axial and in idose all three are
axial. Configuration at C(1) is denoted as ~ when O(1) is axial and as 13when O(1) is
equatorial. O(1) is chemically different from the other pendant oxygens of the ring,
since it forms part of a hemiacetal group with the ring oxygen, 0(5). Since all
polysaccharides are composed of monosaccharide residues, the first consideration in
describing the structure of the polysaccharide is the component units. Polysaccharides
comprised by identical sugar units are called homopolysaccahrides (e.g. amylose and
amylopectin in starch, cellulose); otherwise if sugar units of different kind occur
together in the same structure is a heteropolysaccharide. Other than amylose and
amylopectin
most
industrial
polysaccharides
are
heteropolysaccharides.
Heteropolysaccharides may have greater structural complexity, including the
following type of arrangements:
a) linear homopolymeric structures interrupted by insertion of residues of a different
type (e.g. pectin, carboxymethyl cellulose);
b) regular alternating linear structures where two sugar units repeat periodically over
long stretches in the same skeleton (e.g. carrageenans and agarose);
c) more complex linear repeats formed by more than two different alternating
sequences (e.g. gellan);
d) linear structures in which two or more units are present in non-regular patterns (e.g.
konjac glucomannan);
e) those in which one sugar unit is present in the main backbone and the other in the
sidechains (e.g. galactomannans);
f) more complex periodic branched sequences (e.g. xanthan, which has a cellulosic
backbone with three-sugar (trisaccharide) sidechains on alternating residues).
102
g) irregular, non-periodic, complex structures comprising several types of sugar units
and sometimes non-polysaccharide appendages (proteins), often having dense
branching (e.g. gum arabic).
H
HO
~" CH20H
0
H
H
OH
H
H
Figure l. The 4C 1 conformation of 13-D-glucopyranose.
Structure and shape of polysaccharide chains
Linkage of adjacent sugars in carbohydrate chains involves condensation between
the hemiacetal OH group at C(1) on one residue and one of the alcohol OH groups of
the next residue, with formal elimination of water (Fig. 2). Therefore, glycosidic
linkage geometry can either be axial (a) or equatorial (13). Based on the assumption
that the ring conformation is fixed, knowledge of the carbon to which the other
glycosidic O is attached to def'mes the structure of a linear homopolymer with
identical linkages. Thus, in contrast with proteins, different polysaccharides can share
a similar primary structure. For instance, polymers of D-glucose, include cellulose [3(1--->4) glycan, amylose a or-(1--->4), dextran or-D-(1--->6) and curdlan 13-(1--->3), and
yet their chains arrange themselves into different higher order structures and indeed
behave in markedly different ways. Polysaccharides thus, show low diversity in
primary structure due to the few monosaccharide residues amenable for their
biosynthesis. Also, polysaccharides tend to have regularly repeating sequences.
Therefore, chemical diversity often arise from the degree and branching pattern and
the geometry of the of glycosidic linkages. This fact restricts the number of unique
analytical methods that can resolve effectively the type of polysaccharides present in
food, which in fact means that often several methods must be used.
103
o
o
.o-J
~
o
"
""' '
I
I
CH20H
Figure 2. Conformation of two adjacent 13-D-glucose residues linked together by a
glycosidic linkage in a cellulose chain. Rotation angles about the glycosidic bond are
indicated.
Identity and purity
Identity defines the structural features which characterizes in chemical terms a
population of molecular species which distinguish them among others of the similar or
related type. In polysaccharides, chemical identity is dictated by the type and
configuration of the sugar rings, the geometry and position of the glycosidic linkage,
the sequence of monosaccharide residues also regarded as the fine structure, the
chain length and the immunological response to specific antibodies
Purity in turn, in the context of the present review, is referred to as the extent to
which polysaccharide species with a well established chemical identity are free of
mixture with related and~or different compounds. In polysaccharide chemistry,
impurities often arise from residual protein, lipids and mineral salts (chelated by
uronic acid residues). Purity is closely associated to the extraction/isolation process
by virtue of which a polysaccharide is produced, while identity is related to the
botanical source where it was obtained from and to the possible chemical/enzymatic
modifications carried out on the native materials. Both attributes are therefore assessed
from a number of chemotaxonomic characteristics which allow to describe the sample
features in very precise terms and to distinguish it unequivocally among other related
compounds or sub-fractions. In the context of food additives legislation and
certification, identity and purity are the prime concepts used to lay down
specifications for food additives.
General criteria for polysaccharides purity include [5]:
i) constancy in monosaccharide composition,
ii) constancy in quantitative values of unique structural constituents,
iii) constancy in the molar ratio of monosaccharide constituents,
104
iv) uniform sedimentation rate on ultracentrifugation (through calibration
membranes), and
v) uniform behavior on gel permeation chromatography or ion-exchange
chromatography
(i.e. a symmetrical elution peak is indicative of homogeneity).
Molecular Weight Distribution
While many proteins, which are synthesized under direct genetic control, are
monodisperse, (i.e., all molecules, isotopic variations apart, are identical in structure
and molecular weight), few polysaccharides (chiefly of microbial origin), if any, are
synthesized in this manner and even for those which are chemically homogeneous,
variation occur from molecule to molecule. If this variations are continuous in respect
of all parameters, such as molecular size, proportions of sugar constituents and
particular linkage types, separation into discrete molecular species is impossible and
the material is said to be polydisperse. Hence, the molecular weight distribution
descriptors are expressed in statistical terms: number average (&r weight average
(-~4w), z average (Mz), and intrinsic viscosity (My) [6]. Polydisperse species have ratios
M w / - ~ greater than 1, and in general,
~r,, < ~ < Mz+I<M v
Well-known experimental methods for determining molecular weight and molecular
weight distribution include absolute methods: osmotic pressure, light scattering,
ultracentrifuge sedimentation, diffusion; and relative methods: intrinsic viscosity,
fractional precipitation, size exclusion permeation chromatography) high performance
liquid chromatography (SE-HPLC) with refractive index (RI) detection,
electrophoresis, thermal field flow fractionation (FFF). Use of relative methos to
access molecular weight of polysaccharides, requires precise calibration with
polysaccharides of known molecular weight (i.e. determined by an absolute
technique), typically microbial pullulans or dextrans.
General separation and identification strategies
Before addressing specific tests for analysis of polysaccharide identity and purity,
two broad strategies for identification of polysaccharide macromolecules are worth of
mention. The f'trst one is to isolate first the unknown molecules from the substrate and
cleave them subsequently, to further identify the resulting residues and fractions. The
second strategy is to isolate and to characterize the molecules in their intact form in a
single operation.
105
Both gas chromatography (GC) and HPLC techniques are now commonly coupled
to far UV, differential refractometry (RI), and mass spectrometry (MS) detection, and
sensitivity has increased by orders of magnitude in recent years. Unfortunately, this
approach demands laborious chemical work done on the material such as
derivatisation reactions (e.g. periodate oxidation, methylation, hydrolysis, etc.) and
also gives little information about the anomericity of the glycosidic linkages. Also,
conventional (PAGE) and capillary electrophoresis (CE), using fluorophore labeling
agents (e.g.8-aminonaphtalene-l,3,6-trisulphonate ANTS or 2-aminoacridone
AMAC), i.e. flurophore-assisted carbohydrate electrphoresis (FACE), has started to be
explored [7].
In the second approach, techniques often used as 'fingerprint' determinations
predominate. These include spectroscopic methods such as Fourier transformed
infrared (FTIR) and IH and 13C nuclear magnetic resonance (NMR) spectroscopy.
The advent of high resolution proton IH-NMR (600 MHz) and cross polarisation
magic angle spinning 13C-CP/MAS NMR in modem equipments have revolutionized
the study of polysaccharide structure and conformation in the solid and in solution
state [8]. In general, solid-state NMR studies are most informative in systems with
relatively simple major molecular structure features such as polysaccharides [9].
Current NMR equipments provide a structural information with great detail, including
anomeric configuration, and chain conformation. Also, available information directly
available from NMR concerns the purity of the polysaccharide. For example, the
detection of upfield 1H and 13C signals attributable to aromatic moieties may be
indicative of residual protein, or it may show the presence of a contaminant introduced
during a chromatographic step during isolation. If uronic acid containing
polysaccharides chelate paramagnetic impurities during processing, severe line
broadening of their 1H and 13C spectra is observed [10]. NMR is undoubtedly the
most powerful instrumental method to achieve detailed knowledge about
polysaccharides structure, however it has the disadvantage that the equipment is
expensive and requires specialist technical support.
A third type of tests which are complementary to those described above, are based
on specific proteins as biomolecular probes. One strategy uses lectins and monoclonal
antibodies which are reactive with specific polysaccharide sites or terminus in the
chain, and thus effectively, the corresponding Enzyme-linkedImmunsorbent Assays
(ELISA) can be developed specifically for the identification of a family of
polysaccharides [11]. Yet another powerful biochemical approach, has been through
the use of exo- and endoglycosidases. These enzymes are highly specific not only for
a particular polysaccharide but also for the anomericity of the glycosidic bonds. They
also show preferential cleavage for particular linkage positions.
106
Degree of Polymerisation (DP) Determination
Both gel filtration methods [12,13] and methylation analysis [14], have been
important methods for the structural analysis and determination of molecular weight
and degradation products of polysaccharides. In the filtration methods the use of gels
of various pore sizes have become available. The preparation is passed through a
column of the gel and the eluant from the column are analyzed for carbohydrate.
Calibration curves are generated from standard polysaccharides of known molecular
weight size. The behavior of the polysaccharides on filtration on gels also yields
information about the purity of a polysaccharide preparation as well as for assessing
homogeneity.
Preferably using a strong base (e.g. methylsulfinyl methyl sodium), methylation
analysis of polysaccharide involves complete methylation and hydrolysis to the
constituent monosaccharides which are converted to partially methylated alditol
acetates. It is essential that a complete methylation of all of the hydroxyl groups of a
polysaccharide be achieved. Identification of the partially methylated alditol acetates
and the types of the glycosidic linkages in the polysaccharides are based on the gasliquid chromatography and mass spectrometry (GLC-MS)[ 15].
GLC has been used by Morrison [16], to determine the DP. The polymer was
reduced, hydrolyzed and the released reducing sugars converted to oximes prior to
acetylation. The degree of polymerisation is given by the ratio of aldonitrile to alditol.
For linear polysaccharides the degree of polymerisation is calculated from the yield of
the methylated derivative from the terminal residue. For branched molecules the
average chain length of terminal chains can be determined.
If the molecular weight is adequately high, ultracentrifugation methods through
membranes can be used [17,18]. The technique of sedimentation equilibrium in the
analytical ultracentrifuge can provide absolute size and size distribution information in
terms of molecular weight averages and molecular weight distributions. Problems
commonly associated with light scattering techniques such as dust or aggregates do
not interfer with centrifugation techniques. Density gradient analysis is important for
assaying the purity of a polysaccharide preparation (i.e. freedom from unconjugated
protein, lipid or nucleic acids) [ 19], and the sample can easily be recovered by further
centifugation steps and by collecting the fractions.
The molecular weight of polysaccharides of low degree of polymerisation can be
determined by methods based on the quantitative determination of a functional group
of the polysaccharide and appropriate standard curves. Colorimetric procedures [20]
can be measure the reducing groups of polysaccharides or can be reacted with
radioactive reagents [21 ].
107
Sample homogeneity/purity
Sedimentation velocity
Shape information
Interaction information
Sedimentation equilibrium
f
Size (Molecular weight)
Size distribution
Density Gradient eqm.
~
Samplepurity
Small mols. thro' polysaccharide matrix
Diffusion analysis
Interface transport
Figm'e 3. Ultracentrifuge methods and the potential information available (from
Harding [ 19], with permission)
Isatachophoresis
Isatachophoresis technique has been used for the separation of small ionic
molecules, proteins and peptides [22]. The separation of ionic polymers proceeds
according to their electrophoretic mobilities and the extent to which counterion
binding reduces their net effective change. The method has been used for the
determination of the chemical heterogeneity of carboxymethyl cellulose (CMS). CMS
samples with various degree of substitution (DS), show different isatachopherograms.
Higher DS values give sharper zone boundaries, indicative of a high degree of
homogeneity, in comparison to lower DS values with broader substitution range [23].
However, samples with the same DS value indicated differences in carboxylmethyl
distribution in all cases in addition to the major component.
108
Acetolysis
Acetolysis of polysaccharides results in the complete acetylation of the flee
hydroxyl groups of the polysaccharide and the selective cleavage of glycosidic bonds
[2, 24]. The ( 1 ~ 6) glycosidic linkage is highly susceptible to acetolysis whereas the
(1-->2) and the (1---> 3) linkages are comparatively resistant.
It is a useful method for investigating the structure of polysaccharides from microorganisms. Yeast mannans structure has been studied from the nature of hydrolysis
products, identified by gel permeation chromatography methods [25].
Purity Determination by Phase Solubility Analysis
Conclusions can be drawn on the purity and identity of a substance by means of
phase solubility analysis (PSA) without a priori knowledge of the chemical structure
of the sample. The technique is derived from Gibbs phase rule and involves the
analysis of the composition in solution as a function of the total amount of solid added
and yields a phase diagram [26]. When phase equilibration and solubility analysis are
used to prepare a pure solid, separated from its impurities, the process is often called
"swish purification". This technique can also be used to enrich impurities in solution
phase, for their further identification [27].
Analysis of mixtures of mono- and oligosaccharides
For many analytical purposes a value for total free sugars, expressed possibly in
terms of a monosaccharide, may be sufficient, in other cases a detailed analysis of the
various carbohydrate species may be required. The analysis of sugars found in foods
or as components of polysaccharides, depended on the use of both physical and
chemical properties. The measurement of the free sugar in most foodstuffs therefore
involve the analysis of mixtures. The analysis depends primarily on whether the sugars
are in solution or whether they need to be extracted from the foodstuff. Extraction
methods of polysaccharides usually involve the combined use of various types of
alcohols [28].
The detection of total sugar content relies on the use of specific assay methods (e.g.
anthrone and phenol methods). Reagents such as concentrated acid to hydrolyze the
oligosaccharides glycosidic linkages to monosaccharides and to produce a suitable
chromogen (e.g. hexoses produce 5-hydrohymethyl fiirfiaraldehyde) [29] are often
used. A number of other assays have been reported and the development of suitable
automated assay systems monitoring of chromatographic columns is a major
advantage.
D-Glucose, fructose, sucrose, lactose and maltose, which occur in the diet, can
often be analyzed quite well by reducing sugar methods. However such analysis is
109
difficult and a combination of enzymatic and acid hydrolysis needs to be used. The
most useful approach is either chromatographic separation and analysis, or the specific
enzymatic methods. Quantitatively, gas chromatography is the most suitable method
[28], using a wide range of column conditions. Two classes of derivatives seem to
provide the most satisfactory separation of sugar mixtures: the trimethylsilyl (TMS)
derivatives of the sugars themselves and the alditol acetates [30] prepared after
reducing the sugars to alditols with sodium borohydride.
ELISA Techniques
Immunological methods are becoming increasingly more specific and powerful for
the structural characterization of macromolecules. Polysaccharides elicit an immune
response by virtue of which antibodies which recognise specific carbohydrate residues
can be raised (e.g. in rabitt's blood serum) and subsequently purified. More recently,
methods using enzymes to detect and amplify the antigen-antibody response have been
developed. ELISA techniques have developed for the structural characterization and
analysis of carrageenans [31] and other polysaccharides [32], as discussed below.
'In situ' identification/characterization and quantification
Combination of techniques including fluorescence, bright-field and near infrared
(NIR) microscopy, microscope photometry (MP) and digital image analysis (DIA),
allow in situ identification and/or characterization and quantification of carbohydrates
in cells, tissues and food products [33, 34].
With the development of epi-illuminating systems for fluorescence microscopes, and
the increased availability of improved bright-field and fluorescent probes for specific
groups and reactions [35], it is now possible to carry out relatively specific chemical
determinations on microscopic structures and analysis of carbohydrates. Bright-field
microscopy is very useful technique for visualizing carbohydrate structure and
composition, while NIR and microspectrophotometry offers the ability for mapping
distributions of carbohydrates in raw and processed food materials [33]. Many
microscopic methods can be used in combination with digital image analysis (DIA) to
determine the sizes and shapes of polymeric systems in foods (properties of starches,
cell walls etc.) [36, 37].
110
EXAMPLES
STARCH
Starch, the major reserve storage polysaccharide of most plants, occurs as waterinsoluble granules, the size and shape of which vary with species and maturity of the
plant. Under molecular scales, the starch granule is a giant highly organized structure.
A typical 15kmi corn starch granule is composed of over a billion molecules, which is
composed essentially of polymers of u-D-glucose with trace amounts of protein and
lipid componems. The majority of starch granules contain two polysaccharide
fractions: amylose and amylopectin. Amylose is a linear chain consisting of up to 4000
glucosyl residues connected by cz-(1-~4) glycosidic linkages. Amylopectin is a
branched polymer of repeating glucose units connected by tx-(1-~4) linkages and
branched with ct-(1-->6) linkages. The ratio of amylose to amylopectin varies in plants
from regular (1:3), high amylose (1:1) to waxy (up to 100% amylopectin) [38]. The
characteristic blue color with iodine in potassium iodide is caused by complexing of
amylose with iodine. The amylopectin molecule does not appear to form stable
complexes with iodine but gives a very pale red color in its presence. Amylose and
amylopectin in the starch granule are packed together in a way that is not yet fully
understood.
Starch is a particularly attractive hydrocolloid for textural modification because it is
both natural and safe. There are many types of starch, derived from corn, waxy maize,
wheat, potato, rice, tapioca, pea, among other. Different starches have different
properties and are applied in the food industry as thickening and gelling agents. Native
starch, although widely used in the food industry, has limited resistance to the physical
conditions applied in modem food processing. In order to improve this resistance,
native starches are chemically modified. These modifications can either be chemical or
physical. Chemical modifications include acid hydrolysis, oxidation, esterification or
etherification and cross-linking [39]. Partial hydrolysis of starch is employed to
prepare maltodextrins, which are polymers of amylose and amylopectin of shorter
chain length than that in the native starch.
Physical modification of native starches is mainly achieved either by drum drying
or extrusion of a native or chemically-modified starch slma3,, whereby starch is
obtained in a so-called pregelatimsed state [40]. Starch molecules differ from those of
other polysaccharide hydrocolloids in that they are made functionally useful only by
disrupting the granule structure, i.e. by gelatinisation [41]. During heating in the
presence of water at a characteristic temperature, known as the gelatinasation
temperature, the granule irreversibly swells to many times its original size, crystalline
order is lost and amylose is preferentially solubilised. On cooling to room temperature,
the solubilised or partially solubilised amylose aggregates and some crystalline order
is recovered, and as a consequence a gel is formed. These molecular processes are
collectively known as retrogradation and have important dietary and textural
111
implications. The overall benefit of subjecting starch to a gelatinasation/retrogradation
cycle is to obtain a starch with the ability to form a paste in cold water. The term
resistant starch (RS), has been coined to gelatinizect/retrograded and physically
unavailable starch, which is indigestible by amylase m vitro and m vivo [42]. Novel
concentrated sources of RS with a claimed 30% fraction which analyses as dietary
fiber have recently been launched commercially in the US and Europe under trade
names such as Novelose | and CrystaLean~.
This large number of different commercially available starches, illustrates the
importance of the correct choice of analytical strategies in order to access the key
molecular features underpinning their identity and performance.
Ewers method for starch purity analysis
The Ewers method [43] used to determine the starch content, based on the partial
acid hydrolysis of starch followed by measurement of the optical rotation of the
resulting solution. This based on the controlled acid degradation of the starch, in
which firstly the granules are fully gelatinized and subsequently the solubilised starch
is hydrolyzed. Hydrolysis is stopped by fast cooling. A different specific optical
rotation can be applied for different starch sources. True starch content was calculated
by determining the non-starch components, and the residue was assumed to be starch.
Clarification and removal of protein is commonly done by addition of Carrez reagent.
Although this method dates back from 1908, with few modifications, it is still used
and in fact is the official European Commission method for determination of starch
purity (regulation 2169/86).
Alternatively, an approach to disperse the starch in hot calcium chloride solution to
clarify and to calculate starch content from the measurement of optical rotation [44] it
is used.
Dextrose Equivalent (DE) Determination
Maltodextrins (and indeed, glucose syrups) are produced by acid/enzymatic partial
hydrolysis treatments of native starch, resulting in products with improved functional
properties. Their molecular weights (and physico-chemical properties) are related to
the degree of hydrolysis which is characterized by one parameter, the 'dextrose
equivalent' (i.e. D-glucose) or DE value, and is a measure of the total reducing power
of all sugars present towards Fehlings solution. Thus a degradation product with a
high dextrose equivalent has been subjected to a greater degree of hydrolysis than one
of a lower dextrose equivalent. Degrees of esterification (DE) in typical commercial
maltodextrins vary from 2-19. Glucose syrups usually contain glucose, maltose and
higher maltose oligosaccharides, maltotriose and maltotetrose mixtures.
112
Any method for reducing sugar determination can be used, but traditionally the
Lane and Eynon method [45] is used to determine the content of reducing sugars in a
sample and still is the method of choice for some industrial applications. Alkaline
cupric salts giving cuprous oxide after the reaction of reducing sugars. Their
concentration is monitored titrimetrically, compared with a reference included in
standard tables and calculated as a percentage of the dry substance [28]. Careful
control of the heating is required and for most accurate analysis two titration readings
are necessary; the first to establish the approximate volume of the test solution to
effect reduction, and the second to measure the precise volume required. Corrections
to the standard table values have been also reported improving the original method
[46].
A number of specific assay methods have been developed for the quantification of
individual oligosaccharides, among colorimetric methods, which are used for the gross
determination of total carbohydrate content or total reducing sugar content [29]. Such
assays use alkaline 3,5-dinitrosalicylic acid [47], alkaline ferricyanide [48] or alkaline
picric acid [49]. However the use of oligosaccharide fractionation by gel permeation
chromatography is now recommended as the best method for characterization of
starch hydrolysates and the determination of dextrose equivalent is based on the actual
composition of oligosaccharides [50]. Fractionation of starch and its hydrolysis
products using Bio-Gel P-2 [51], microspherical cellulose [52] and porous glass beads
(CPG-10) [53] have also been reported.
Infrared and NMR spectroscopy of starch
The infrared spectroscopy of potato starch is different from that of cornstarch,
particularly in absorption regions for oxygen-containing groups. Thus in cornstarch,
absorption is stronger at 1681, 1053-952 and 855 cm~, whereas in potato starch the
band at 926 cm1 is stronger [54]. Modified starches show bands which are
characteristic for the different derivatives, and thus effectively the degree of
modification can be determined quantitatively [55].
Water absorption properties of wheat flour depend on the degree of mechanical
damage of the starch after the milling of hard wheat. These properties are critical in
flour quality, as far as performance on automatic dough handling equipment during the
bread baking process is concerned. Near infrared (NIR) reflectance measurements
have been used successfully to determine the degree of starch damage of commercially
milled flours [56, 57]. The basis for this determination is consistent with the
hypothesis that the mechanical damage is associated with cleavage of hydrogen bonds
between starch molecules and water molecules. Furthermore, observations that NIR
reflectance measurement correlated with both digestibility and extractability
characteristics of damaged starch which, form the basis of chemical methods for its
determination, strengthen confidence in the NIR determination [58].
113
Investigative high resolution NMR techniques have also been applied to delucidate
in detail the structure of starch components. Using anomeric proton signals it is
possible to quantify ratios of both branch points and reducing tenninii to main chain
residues. This, the degree of branching of amylopectin (and potentially amylose), and
the DE value together with the degree of branching for starch degradation products
can readily be obtained using this technique. As some starches (notably potato)
contain covalently-bound phosphate groups, they are amenable to 31p NMR analysis.
First results [59] indicate that, for a number of potato varieties, C-3 phosphorilation is
nearly constant, but C-6 shows significant differences.
Bright-field microscopy technique as well is useful for visualizing starch granules in
wheat flour which have been damaged by milling, and allows routine measurements of
the degree of starch damage in diverse flours using image analysis. A relatively
unknown dye, Hessian Bordeaux is used, replacing the more traditional stain, Congo
Red [34, 60].
Starch crystallinity
In the native form, the starch granule exhibits considerable cristallinity (about 40%
by X-ray diffraction), which in turn distinguishes the following polymorphic forms: A
(in cereals), B (in tubers) [61]. An intermediate C type, is less frequently observed.
The different forms are readily and unambiguously identifiable by wide angle x-ray
diffraction (WAXS) (Fig. 4) [62], or by 13C CP MAS NMR [63]. In both polymorphs
the polysaccharide chains are thought to associate in six-fold left-handed double helix
conformations with a pitch of 2.138 nm, but to differ in their unit cell dimensions (i.e.
the lattice) and in water association [64, 65, 66].
The most notable difference between the 13C CP MAS NMR spectra of the A and B
forms is the occurrence of the C-1 peak as a triplet (-~ 102.5, 101.5, and 100.6 ppm) in
the A form but as a doublet (-~ 101.5 and 100.5 ppm) in the B form [67, 68]. Both
forms comain a distinct C-6 peak at -~ 62.7 ppm. Another form, the V form, can be
obtained by precipitation of amylose with a variety of organic solvems and
compounds. Solid state NMR spectra of V forms have been reported [60, 63, 69, 70].
Amyiose and Amylopectin Fractionation
Due to their different extent of branching and effective molecular mass, amylose
and amylopectin behave in markedly different ways, particularly in the manner in
which they gel upon cooling from a gelatinized starch solution [71]. The different
solubility of amylose and amylopectin in aqueous alcohols (thymol or n-butanol), is
exploited in order to achieve the fractionation of isolated starch [72]. Other
techniques used to fractionate both polymers include gel permeation chromatography
(GPC) and recently by thermal field fractionation (FFF) (Fig.5)[73].
114
Although amylose and amylopectin do not differ each other in terms of their
residue unit, u-D-glucose. Chemically amylopectin offers substantially greater
structural diversity than does amylopectin, due to the extensive branching in the
former. Selective cleavage of amylopectin chain, using highly purified enzymes,
namely pullulanases, has been used in order to elucidate the free structure of
amylopectin from different botanical sources of starch [74]. Based on this approach,
an important technique which is now widely used, involves the complete debranching
of amylopectin followed by fractionation of the linear chains by gel filtration. The
resulting elution pattern reveals the size distribution of the constituent chains, i.e. the
chain profile. The technique originally introduced by Lee et al. [75], it has now been
ref'med by Hizukuri [76,77]. Use of GP- HPLC with monitoring by low angle laserlight scattering photometer (LALLS) and a differential refractometer (RI), it has been
shown that amylopectin structure is related to botanical source of the starch. The
multimodal distribution of the peaks obtained has led to the proposal of the 'cluster'
model [76].
GUM ARABIC
Gum arabic is a highly branched acidic heteropolysaccharide produced as an
exudate from various species of the genus Acacia. The main source of gum arabic of
commerce is Acacia senegal L. Willdenow, which undoubtedly is the most available
gum in the commercial producer countries of Africa, namely Sudan, Nigeria, Chad,
Ethiopia, Senegal, with lesser amounts available from Ghana and Zimbabwe. Gum
arabic has long been used in a variety of foods and beverages, especially as a natural
emulsifier and encapsulating agent of citrus flavors and also in confectionery
applications where it inhibits sugar crystallization and confers a glossy appealing
appearance.
Since 1991, cold weather, foliage attack by locusts and changes in export duties on
gum arabic have severely reduced exudation and gum supplies coming to market. This
has led to the search for 'substitutes' of gum arabic, particularly in modified
maltodextrins and 13-cyclodextrins [78,79]. However, the unique combined
functionality of gum arabic as a natural emulsifier, good mouthfeel characteristics and
low solution viscosity cannot be matched by any other gum and satisfactory alternative
materials have not been developed. An obvious searching route would be to seek for
alternative botanical sources for exudate gums of similar type, and indeed gum arabic
under trade is known in practice to be a mixture of several A cacia related species. This
has resulted in increasing concern about the precise definition of gum arabic, chiefly
among regulation official bodies, gum producing companies and end-users in the food
industry. The precise chemical identity of gum arabic poses a very difficult problem
115
.
.
.
.
.
.
.
.
.
.
.
i
PO,O,os,orch I
R,O
r
,,.,,
I-.i
ATopiocoStorch
6"
10"
14"
18'
2 0 ----~
22'
26"
30"
3/."
Figure 4. Wide-angle X-ray diffraction for two starches (potato and tapioca) in
powdered granular form. The amylopectin fraction crystallize as distinguishable A and
B polymorphs (from Clark and Ross-Murphy [62], with permission).
withoutsalt
Z
9
L
withsalt
amylopectin
t*
i r a - 1
0
.
20
|
.
.
.
i
l
~
40
60
80
I1
100
]
120
l
140
TIME (rain)
Figure 5. Thermal field flow fractionation fractogram of a cationic starch in DMSO
with added salt (right peaks; 1.0 • 10"~ M LiNO3) and without salt (left peaks). From
Lou et al. [73], with permission.
116
for the food and beverages industry, mainly because the detection of adulterating non~ermitted gmns has proved extremely difficult. Indeed, even by the use of expensive
3C solid state NMR spectroscopic methods, large variations even between
authenticated gum arabic samples can still be found due to the expected natural
variability [80].
The Joint Expert Committee for Food Additives (JECFA) of FAO defmed 'gum
arabic of commerce' as the dried exudation of Acacia senegal (L. Willdenow) or
related species of Acacia Fam. Leguminosae [81], while in the US the Food and Drug
Administration defmes it as the exudate gum from various species of the genus Acacia
family Leguminosae [82]. Both defmitions did not include any specifications about the
chemical identity of gum arabic. Recently, JECFA proposed new specifications [83],
and introduced three significant additional criteria: i) gum arabic should include gum
from Acacia senegal and only "closely" related species (which according to the early
Bentham's classification includes two sub-genera, namely sub-genus Acacia (= series
Gummiferae Benth. e.g. A senegal) and sub-genus Aculeiferum (= series Vulgares.
Benth. e.g.A, seyaI)) ; ii) optical rotation limits (-26 to -34 ~ should be adopted;
iii) nitrogen content should be set between 0.27 and 0.39%.
These criteria, along with the battery of physico-chemical analytical tests typically
conducted on exudate gums, namely intrinsic viscosity, equivalent weight, individual
sugar residues and amino acid composition, have been indicated as to be inadequate to
unambiguously differentiate between closely related species of Acacia senegal and to
render inadmissible gums which fulfill the description [84], unless the sets of
analytical data are considered globally using multivariate statistical procedures as it is
described below.
Three different methodological strategies have been pursued in recent years, in
order to address the problem of gum arabic identity, namely
9 chemometric methods,
9 structural characterization studies, and
9 immunological techniques
The potential of each of these is discussed in fin~er detail next.
Chemometric methods
Chemometric methods make use of multivariate statistics, namely principal
component analysis (PCA), discfiminant component analysis (DCA) and cluster
analysis (CA) in order to evaluate similarities between patterns of a series of analytical
attributes in a set of samples. These statistical techniques are widely used for both
research and quality control purposes in order to establish a classification of items
using a projection of multivariate data sets [85-87]. The implementation of
chemometric methods to classify exudate gums from Leguminous botanical sources
based on a large number of analytical data was first done by Jurasek and co-workers
117
[88-91]. Fig.6 is a DCA loading-loading plot showing how gum samples from the
Combretum species can be identified and discriminated from Acacia gums based on
nine carbohydrate parameters.
Gum exudates from the genus Prosopis represent another group of exudate gums
which have evoked commercial and agricultural interest [92]. Prosopis gums have also
been successfully identified and distinguished from Acacia gums by this technique
[91]. In a recent paper [93 ], the classification procedure has been refined using a PCA
method called multiple nearest neighbor (MNN), based exclusively on the amino acid
composition of a set of commercial Afifcan Acacia gums. This has allowed to
compute an empirical index ('Senegal Number') of closeness of the species to Acacia
senegal. This approach appears to be a promising option, which may lead to include
an operational parameter in the official def'mition of gum arabic in the future based on
a chemometric classification.
Structural analysis
Structural studies aimed to characterize the detailed chemical and steric features of
gum arabic molecule have included analysis of the composition of the constituent
sugar residues by 13C NMR [94,95]; profiling of the aminoacids of the proteic minor
component by chromatographic techniques [78, 80, 96, 97]; composition and size of
the individual molecular fractions by gel permeation chromatography (GPC) [98-100];
high performance size exclusion chromatography (HPSEC) [101] coupled to UV, RI
and light scattering (both in static and dynamic mode). GPC procedures have also been
allied with immunological techniques (ELISA and immunoblotting) for the specific
identification of gum arabic fractions by purified antibodies [ 100, 102, 103].
It is well established that in addition to the carbohydrate components, gum arabic
contains -~ 2.2 % protein [97]. As well as having a crucial role in emulsification [104],
it is now evidem that the protein component is central to the overall primary structure.
GPC chromatograms of gum arabic show it to be an heteropolymolecular material,
formed by three distinctive components (Fig. 7).
The main fraction (Fraction AG Fig. 7B), whose contribution to the total is --88%,
has a lower protein content than unfractionated gum (<0.4%) and an average
molecular mass of-~2.8 • 105 , which does not change significantly on extensive
proteolysis. A second fraction (Fraction AGP, Fig 8C), represents -~10.4% of the total,
has a much higher protein content (~12%) and an average molecular mass about 5
times that of the major fraction (14.5 • 10s). On proteolysis, the molecular mass of
fraction 2 drops to that of fraction 1 [98,99,105] suggesting a 'wattle blossom'
structure, with on average, five branched carbohydrate assemblies similar, or identical,
to those in fraction 1, linked together by a polypeptide chain. A third, minor (-~1.24 %)
glycoproteic fraction (GP, Fig 8D) contains 47% protein which is 25% of the total
protein and a molecular mass of 2.5 • 105.
118
[]
[]
/
%
13
[]
[]
[]
DD~
D
[]
%
CXl
oa
DCl
Figure 6. DCA utilising two categories of objects (Acacia gums, 33 samples and
Cobetrmn gums, 30 samples). Features are nine mainly carbohydrate parameters.
Axes: first and second discriminant components with corresponding t-values 19.7 and
5., [], Acacia; A, Combretum; A, comercial Combretum nigricans and commercial
Combretum samples. From Jurasek and Phillips [91], with permission.
GUM ARABIC
9
A
c
IV0
DECREASING MOLECULAR MASS
V,,
Figure 7. Gel permeation chromatograms of gum arabic and its fractions" (A) whole
gum arabic; (B) fraction 1; (C) fraction 2; (D) fraction 3A. From Randall et al. [99],
with permission.
119
I
I
3
.
.
.
.
-
I
I
I
I
-
-
I
5
Vt
6
Vo
Vt
J
Figure 8. GPC elution profiles of 1% w/v gum solutions: (a) samples 1-5 as monitored
by RI; (b) samples 6-8 as monitored by RI; (c) samples 1-5 as monitored by UV at 206
nm; (d) samples 6-8 as monitored by UV at 206 nm. From Osman et al. [ 103], with
permission.
120
GP chromatograms (Fig. 8) of gum arabic, show the elution profiles of the various
samples of gum arabic. Figs. 8a and 8b show the molar mass distributions of gum
arabic as identified by refractive index (RI) and in Figs. 8c and 8d by UV detection
[103]. The reason the two elution profiles are different using the two detection
techniques is that RI is essentially sensitive to the total concentration of material
present, whereas UV is more sensitive to the chemical nature of the various molecular
mass fractions. At 206 nm, UV spectroscopy detects the carboxyl groups associated
with the polysaccharide and also the amino acids which make up the proteinaceous
component; previous NMR and methylation analysis [106], suggests that all three
fractions have similar branched structures based on a 13-(1---~3)-linked galactan core.
The presence of these three components (particularly of the AGP fraction), whose
exact proportion is characteristic of the sample, is suggested to be a more accurate
diagnostic chemical technique for identifying commercial gum arabic than the
aggregate properties optical rotation and %N, currently considered in JECFA's
def'mition [84].
ELISA techniques
The fact the gum arabic and other chemically related gums can elicit immune
response is well recognized [ 107,108]. However, the first detailed study of gum arabic
using immunoassay technology was described recently [ 109]. Following their method
an enzyme linked immunosorbent assay (ELISA) has been developed for the specific
identification of Acacia senegal gum [ 102,103]. The technique is based on the use of
antibodies that have been raised in rabbits which recognize and interact with a specific
binding site in the gum molecule, i.e. epitopes. The precise nature and size of the
epitopes to many of these antibodies is unknown, with contradictory views. Some of
these point to specific carbohydrate terminii at the epitopes, surface of low and high
molecular weight components [5, 109,110,111], and some others to the amino acids
present in the fractions AG and AGP, since very little, if any, aff'mity was found for
the AG fraction [112]. Regardless of the precise location of the immuno-active site,
the ELISA test can reliably detect concentrations as low as 10 ~tg ml ~ of gum arabic
with a good reproducibility within the range 10-100 ~tg ml ~. The ELISA approach has
been used to differentiate gmn arabic samples from A. senegal from other exudate
gums (ghatti, tragacanth, karaya) currently used as food additives. However, some
cross reaction occurs between gum arabic samples from the same genus series. The
method undoubtedly, offers great potential to develop a quick an easy test to identify
the presence of gum arabic in foodstuffs as well as the common adulterants.
121
ALGINATES
Algae, just as terrestrial plants do, rely on cell wall matrix polysaccharides in order
to build up their physical structure. Unlike terrestrial plants though, algae somehow
need greater flexibility in order to withstand the wave and tidal forces that they are
permanently subjected to. Nature achieves this, by the biosynthesis of a 'soft'
polysaccharide matrix around rigid cellulose fibers. This matrix is composed largely
by polysaccharides, collectively known as phycocolloids, whose prime property is to
be able to form gel networks. Man has learned to exploit the gelling capacity of
phycocolloids in order to modify the mechanical properties of foods. As for other food
hydrocoUoids, the extraction and purification of seaweed polysaccharides constitutes
today a large industrial activity. In this and the following section, the analysis of two
different kinds of phycocolloid families with a large bearing in food industry are
discussed, namely alginates and carrageenans.
Alginate is among the major industrial polysaccharides. The world's demand of
alginate is of about 23,000 ton per year, all of which is currently obtained from brown
seaweeds (Phaeophyceae) of the species Laminaria hiperborea, Macrocystis pyrifera
and Ascophuyllum nodosum. Alginate is available commercially principally as the
sodium salt form, although alginic acid, other metal salts (potassium, calcium,
ammonium) and derivatives (propylene glycol alginate) can be obtained. Sodium
alginate is soluble in water, producing viscous solutions. The especial properties of
alginate, however, are based on its ability to form gels in the presence of certain
cations, notably calcium ions, a property that has successfully been utilized in a
variety of food applications.
,
1
".. ~ ~
COOH
COOH
0
i
i
o
O H ~
Uo'
!
--o---
o
o
i|
==
..-
1.03 nm .-
(a)
='
'~-------- 0.87 nm ----------~'
(b)
Figure 9. The constituent carbohydrate residues in alginate are: (a) [3-D mannuronic
and (b) ct-L-guluronic acid residues
122
Alginate is a (1--+4)-linked linear co-polymer of 13-D-mannuronic (M) and its C5
epimer ct-L-guluronic acids (G; Fig. 9). The monomer residues are arranged in
homopolymeric blockwise patterns of three different types (Fig. 10): poly-Lmannuronate (M blocks), poly-L-guluronate (G blocks) and in heteropolymeric
sequences which approximate to a disaccharide repeating structure (MG blocks).
Enzymatic studies [113], however, have shown departures from this idealized,
regular, alternating sequence. The relative content of M, G and MG blocks as well as
the alginate molecular weight distribution depend on the alginate source, namely the
type of seaweed as well as the precise tissue in the algae (e.g. fronds, stipe or fruiting
bodies) where the alginate has been harvested from.
M-G-M-M-M-M-M-M-M-M-G-M-G-. ......
M-block
G-M-G-G-G-G-G-G-G-G-G-M-G-M-. ......
G-block
M-G-M-M-M-G-M-G-M-G-M-G-G-M-M-. ......
MG-block
Figure 10. Schematic illustration of the main monosaccharide block structures of
alginate
Due to the different glycosidic linkage geometry between adjacent sugar residues
on each block, the chains are arranged as flat ribbons in poly-L-mannuronate, and as
buckled ribbons in poly-L-guluronate. The technologically and biologically important
physical properties of alginates in solution, gels and algal tissue are largely determined
by the relative amounts of the three types of block present, and hence by the molar
ratio of M:G. There is strong evidence that formation of calcium alginate gels occurs
predominantly by association of poly-L-gu|uronate sequences [114-116]. This has
been rationalized as a result of specific chelation of calcium ions (or other ion of
similar size) in the cavities formed between adjacent residues in poly-L-guluronate
sequences as in the so-called 'egg box' model [ 117].
123
Analysis of chemical identity and fine structure
Since the functional properties (i.e. the rheological behavior), of alginates are
dependent upon the chemical composition, it is clearly desirable to be able to analyze
the M:G ratio and ideally the block structure of the polymer. These three parameters
too are the major pieces of information so as to establish the chemical identity of an
alginate sample. Although the monomer composition (M:G ratio) is in itself a useful
parameter, the properties of a sample of alginate can be predicted more accurately
from knowledge of the block structure.
One approach is to isolate the three types of block structure by partial acid
hydrolysis and fractional acid precipitation and quantify each by the a chemical assay
method such as phenol sulfuric test [118]. The blocks, however should not be
considered as ideal, since typically the poly-L-guluronate sequences will contain about
10-15% mannuronate, so an allowance for it must be made for accordingly. There are
two well established instrmnental methods to determine the block composition of
alginate, namely by circular dicroism (CD) and by nuclear magnetic resonance
(NMR). Although molecular probes (polyclonal and monoclonal antibodies) to locate
alginate in living tissues, have been raised in rabbit serum, the technique needs fin-ther
refinement, especially to identify the structures to which the antibodies bind [ 119].
Circular dicroism (CD), is the difference in absorption of left and right circularly
polarized light by a dissymmetric molecule, and is typically only about 10-2- 104 of
the total absorption of light by the material (for a comprehensive review on principles
and applications of CD to carbohydrate structure analysis, see for example Morris and
Frangou [ 120]). CD bands may be either positive or negative depending on whether
left or right circulary polarized light is absorbed preferentially. Electron transitions
(with a small electric dipole moment but a large magnetic dipole give strong CD
bands, n---~* transitions of carboxylic acids, salts, esters, and amides, in which a nonbonding electron of oxygen or nitrogen is prompted to an antibonding orbital, falls into
this category. The CD may be def'med quantitatively as the angle of ellipticity (0),
whose tangent is equal to the ratio of the minor to the major axes of the ellipse traced
by the net electric vector as a result of the differential retardation of the right and left
components of the polarized light beam, as it passes through an optically active
substance. The major axis being proportional to the sum of the residual intensities of
the left and right circularly polarized components (IL + IR), and the minor axis to the
difference between them (IL - IR), Thus
tan (0)= (IL- IR)/(IL + IR)
(2)
CD is commonly expressed as a molar quantity, according to
[0] = 0M/cl
(3)
124
where M is molecular weight, c is the concentration (g/ml) and 1 is the path length
(mm).
The presence of carboxylic substituents in poly-L-mannuronate and poly-Lguluronate structures of alginates makes them especially amenable to this technique.
Indeed, CD allows to determine the block structure of less than 1 mg of sodium
alginate [116]. Alginates exhibit CD behavior in the UV region 190-245 nm, the
spectra of D mannuronate and L-guluronate being very different. The CD spectra is
also modified by the presence of adjacent residues in the polymer chain, so it is
sensitive to block structure as well as to overall composition. It follows from Figure 11
that the CD spectra of alginate show a peak at 200 nm and a through at 215 nm.,
whose height and depth are dependent upon monomer composition. Equations are
given to compute the monomer M/G ratio [116], without
measurement of
concentration. Most commercial alginates show CD spectra entirely in the negative
region of molar ellipticity [0], and the estimate of the ratio of mannuronate to
guluronate residues can be obtained by simply doubling the measured peak/through
ratio (0through " 0peak / 0through)- In order to f'md the block structure it is necessary to
consider the whole shape of the spectrum and use an iterative, least square
minimization computer routine. Unlike NMR, CD van be used without any
degradation of the samples. The solution must be adjusted to a pH of 7.0 + 0.3 and
divalent cations must be absent. Solutions of < 1 mg ml ~ are used and for block
structure analysis (but not for M/G ratio) the polymer concentration must be
accurately known. This method has been used by Craigie et al. [121], to examine
tissues from a number of algae.
~H-NMR for the analysis of block structure in alginate has been developed into a
routine technique, the possibility to work at high temperatures and solvent suppression
techniques, allow at present minimally depolymerized samples (DP 20-30) using mild
hydrolysis with acid (30 min, 100 C, pH 3.0) in order to reduce the viscosity of their
solutions, Grasdalen et al. [122], were able to assign peaks in the proton NMR
spectrum which differentiated G residues with neighboring G from those with M
residues as neighbors. Data from IH NMR and from chemical analysis correlate well
although there are some discrepancies in samples with a high proportion of mixed
doublets [122]. Further work using 400 MHz ~H NMR [123], gave information on the
proportions of the G-centered triplets (GGG, MGG, GGM and MGM).
125
I
I
I
i
1001
I
l
/
l
l
t
/
'
I
2o0
0
'II
'k fnm)
. ~2~0
[01
220
2~0
[o']
~,(nml
210
220
200
0
230
2/.,0
I
I
\
/--\
-1000
/
I
i
4'/
/I
I" "\', 1"I'/
i
'V,,;//
~
//
/
,!
I/"\\ l II
,"
Ii /
I/ I
/
I
I9
!
-2000
-3000
230
/
/
i
',
', \
/
/
/
-20
i
-30
#
I
-aO00
0[
Figure 11. Left: CD of alginate blocks approximating in structure to poly-L-guluronate
( - - ) , poly-D-mannuronate (- - -), and mixed ( - - - ) chain sequences. Right:
Comparison of the CD behavior of alginate mixed-sequences ( - - ), and spectrum
(- - -) synthesized by linear combination of 50% of the spectra for each type of
homopolymeric block. Reprinted from Morris and co-workers [ 116], with permission.
Natural abundance ~3C NMR has also been used in order to increase the resolving
power of the technique. However, it is not suited for routine analysis because of the
extended accumulation time that is required to record good resolution spectra. On
balance, IH NMR is probably the most suitable technique for routine analysis of
alginate structure and a 100 MHz Fourier transform machine is perfectly adequate,
although clearly more information can be obtained from higher power machines. The
composition of alginate extracts from different algae species is shown in Table 1.
126
Table 1
Fractional Composition of Alginates as Determined by ~H-NMRa
Fractional Composition b
Botanical origin
Laminaria digitata
Laminaria hyperborea
Macrocystis pyrifera
Ascophyllumnodosum
FG
FM
FC,G
FMM
FMG
FGM
Fc,c~ FMMM
0.42
0.70
0.39
0.43
0.68
0.30
0.61
0.57
0.27
0.60
0.21
0.18
0.43
0.20
0.43
0.32
0.15
0.10
0.18
0.25
0.15
0.10
0.18
0.25
0.22 0.10
0.52 0.06
0.20 0.20
0.13 0.17
a From Grasdalen [123], with permission.
bThe following relationships betweenmonad, diads and triad frequencies hold:
Fc~ + F~ea+ FGM+ FM~= 1
F~ + FMG= FG+ Fr~ + FMG+ FM
F~ = Fc,c~ + FM~ + F~M + FMGM
FMG= FGM= FC,-GM+ FMGM
FMGG= FGGM
CARRAGEENAN
The red seaweeds (Rhodophyceae) produce a family of phycocolloid
polysaccharides, collectively known as galactans, since all of them have a galactose
backbone joined together by glycosidic linkages of alternating type. Members of this
family of polysaccharides include agar, furcellaran and carrageenans. Carrageenans
bear special importance in food systems and have been used to thicken foods for
centuries, the earliest records of use being in Ireland, where the red seaweed Chondrus
crispus ('Irish Moss') was boiled with milk to give a thickened product. In the
meantime, these seaweeds were also collected along the French coast of Brittany,
where the bleached "lichen" (a blend of Chondrus crispus and Gigartina stellata) was
used to prepare a milk-gel known as blanc-mange. Today, more than three-quarters of
the carrageenan production goes into food industry. A broad range of rheological
properties can be generated and f'mely modulated in foods by making use of the
thickening and gelling properties of the various types of carrageenans [124]. These
superior properties are a direct consequence of their ability to form gels at extremely
low concentrations and to increase the viscosity in food systems, along with their
capacity to interact synergistically with other polysaccharides (konjac flour, locust
bean gum, starch, etc.). Also, carrageenan exhibits a strong reactivity for casein, a
property which is exploited in many dairy applications (e.g. stabilization of cocoa
powder particles in chocolate drinks imparting a pleasant mouthfeel). Carrageenanyielding seaweed species (carrageenophytes) have been reported as occurring in seven
127
different families of seaweeds, namely: Solieriaceae, Gigartinaceae, Furcellariaceae,
Phyllophoraceae, Hypneaceae, Rhabdoniaceae, and most recently, Rhodophyllidaceae
[ 125]. Depending on the region of the world, some species of these families are cast
on beaches, while others are maricultured (e.g. Phillipines).
Residue composition
Carrageenans are water soluble linear sulfated polysaccharides which occur in the
cell wall of marine red seaweeds. They are composed of an alternating idealized
(A-B), structure of ct-l,3-1inked and [3-1,4-1inked D-galactose units substituted and
modified to various extents (Fig. 12). Three primary forms (K-, k-, t-) of carrageenan
which are commercially important are identified based on the modification of the
disaccharide repeating unit resulting from the more or less regular occurrence of
"masking" substituents in their structure (heterounits), chiefly sulfate hemiester
groups, but also methoxy and pyruvate groups. Also the (1--->4)-linked B residues
may, to a varying extent and depending on the algal source, be converted to the 3,6anhydro form. K-Carrageenan has an idealized alternating disaccharide repeating
structure of ~t-l,3-1inked D-galactose-4-O-sulphated and [3-1,4-1inked 3,6-anhydro-Dgalactose, it is precipitated from whole extracts of Chondrus crispus, Euchema
cottoni and Gigartma spp. and forms brittle gels in the presence of potassium ions.
os%
o~
OR~
OH
~ o--------~
b
Figure 12. Structure of the repeat units of ~:-, t- and k-carrageenan. (a) K-carrageenan
Ra = H, t-carrageenan Ra = SO3, (b) k-carrageenan Rb = 30% H, 70% SO3.
128
While t-carrageenan shares the same ideal disaccharide structure of K-carrageenan
A residues, it has an extra sulfate ester group at 0(2) of the B residues, t-Carrageenan
is sensitive to Ca 2+ ions, in the presence of which it forms elastic transparent gels. The
major source of t-carrageenan is Euchema spinosum. Both K- and t-carrageenans
backbone structure is able to adopt a helical ordered structure (i.e. to undergo a coilto-helix conformational transition), which may be induced by lowering the
temperature and/or by adding salt to the solution [126]. This transition usually
precedes the formation of a gel network (for an enjoyable review on carrageenans
gelation mechanisms see ref. 127). Agar shares the same backbone structural geometry
in all respects but the ( 1 ~ 4) linked 3,6 anhydro-D-galactose residues are in the Lrather than the D- enatiomeric form.
In gelling K- and t-carrageenans, the repeating sequence in the helix-forming
backbone is interrupted by partial replacement of the 3,6 anhydro-D-galactose by Dgalactose sulfate or disulfate residues. In t-carrageenan, for example the replacement
level is typically about 10% [ 126]. These residues cannot be accommodated within the
helical conformation adopted on ordering of ~:- and t-carrageenan chains but
introduce a backbone "kink" in the helix structure. Most of these anomalous
"kinking" residues can be converted to the helix-compatible anhydride form by
treatment with alkali [126, 128], with the residual proportion of unbridged rings
depending on the pattern of sulfation [126]. Alkali modification is known to
substantially enhance gel properties [129].
The third industrially produced non-gelling galactan is ~.-carrageenan. It is
obtained as a soluble fraction after the selective precipitation of K-carrageenan with
0.25M potassium chloride from the extracts of tetrasporophytes of the families
Gigartina, Chondrus, and lridea, (although it has also been identified
in
Acanthophora and Laurencia spp.), t-carrageenan has a different backbone structure
than the one of ~:- and t-carrageenan, it is composed predominantly by fully sulfated
(1---~3)-linked residues at O(2), and (l~4)-linked sulfated at 0(2) and 0(6) and a few
anhydride residues. Due to the lack of the helix-compatible (1---~4)-linked 3,6anhydro-D-galactose residues in k-carrageenan, it does not gel, so that it is used
primarily as a thickener.
It is evident thus, that apparently minor variations in primary structure, have
dramatic effects on the functionality of naturally biosynthesized carrageenans such as
their gelling capacity [130].
Hence, knowledge of the primary structure of
commercial carrageenans is extremely useful in anticipating their behavior in food
systems.
Analysis of chemical composition.
The variation in primary structure with algal source, has been studied by chemical,
enzymatic, immunological, chromatographic and spectroscopic methods (refer also to
129
comprehensive reviews, 131-135). It is well established that carrageenans as they are
extracted from algal tissue and purified by physical methods are never quite pure. In
particular, native samples of t-carrageenan normally contain a small fraction (5-10%
or more, depending on the algal source) of ~:-carrageenan fraction, and viceversa
[130].
General chemical analysis of carrageenans includes the quantitative determination
of their basic components, namely galactose (Gal) and 3,6-anhydro-D-galactose
(AnGal) units, as well as the sulfate contents. AnGal is determined using a
colorimetric assay using the resorcinol reagent [136] and Gal is measured by
difference after determination of total sugars by the phenol-sulphuric acid method
[ 137] and calculated as galactose after correction for the AnGal content; ester sulfate
content can be determined by a gravimetric method following precipitation with
BaSO4 [ 138], or turbidimetrically after hydrolysis of the polysaccharide with 1M HC1
at 100 ~ for 6h. 3,6-Anhydro-galactose, galactose and sulfate contets cannot be
considered as specific enough critera to establish the chemical identity of galactans.
They only provide a very broad indication of the kind of carrageenan that is being
dealt with. Typical proportions of these for polysaccharides from Euchemma cottoni
(predominantly K-carrageenan) are 66% galactose, 34% 3,6 anhydro galactose and
ester 25% ester sulphate; values for Euchema spniosum carrageenan (t- form ) are Gal
70%, AnGal 30% and 32% ester sulfate; the non-gelling ~-carrageenan is
distinguished by low proportions of AnGal of ca. 5.0% and high ester sulphate
contents (ca. 35%) [139].
Analysis of carrageenan identity
Detailed knowledge of the identity of carrageenans involves the following analysis:
9 Constituent sugars analysis: Gal, AnGal and anomalous residues
9 Location of ester sulphate and O-methyl masking units
9 Fine structure (determination of hybrid structures)
A number of techniques are available in order to addreess specific aspects of the
chemical identity of carrageenan. A summary of these techniques along with the
information derived is presented below.
Constituent sugar analysis by GC and GC/MS
Gas cromatographic methods for carbohydrate analysis involve the formation of
derivatives of sufficient volatility and adequate thermal stability. Trimethylsilyl esters
and acetate or trifluoroacetate esters are the most common derivatives. Mass
spectrometry as commonly practicised uses electron impact most frequently as the
ionization mode. Carbohydrate derivatives rarely give molecular ions in electron
130
impact spectra, although molecular weights may often be inferred from electron
impact spectra. At present, oligosaccharide derivatives of molecular weights of up to
1000 are convenient for mass spectral analysis. In terms of thermal stability and ease
of interpretation of spectra, permethylated compuonds and TMS ethers are the
derivatives of choice for mass spectrometry [3]. Such derivatives are prepared by
methylation of the polysaccharide and subsequent production of the methylated alditol
acetates.
Constituent sugars of carrageenan and galactans in general, have been analysed by
chromatographic procedures GLC and GLC-MS [132, 140]. In contrast with other
polysaccharide systems, preparation of carrageenan derivatives amenable for these
techniques, involve special considerations and meticulous chemical work as briefly
discussed here. The two main problems are that volatile derivatives of galactans,
containing up to 50% AnGal residues, cannot be prepared directly by the above
procedures (methylation ~ alditol acetates formation), due to acid-labile 3,6
anhydrogalactosyl residues are rapidly destroyed (both in the native and in the
methylated form) under the harsh conditions typically used to hydrolyse their
constituent monosaccharides. Also, during the methylation of carrageenan and
agaroids, some of these polysaccharides are highly charged by sulphate groups and
thus are insoluble in dimethylslphoxide (DMSO), and thereby difficult to methylate
fully by the Hakomori procedure [141]. These methods have been substantially
improved by incorporating in situ reduction during hydrolysis, using the acid-stable
reductant, 4-methylmorpholine-borane (MMB) [132,142]. Under relatively mild
hydrolysis conditions, virtually all the 3,6 anhydrogalactosidic bonds can be cleaved
while most of the galactosidic bonds remain intact. In the presence of the borane
reducing agent this leads to the production of "biitols", namely 3,6-anhydro-4-O-13-Dgalactosyl-galactitols [ 140]. A range of partially methylated, partially acetylated biitols
has been produced from galactans of known composition and characterised by GLC
and GLC-MS total ion anlysis. From the retention times and fragment ions obtained it
is possible to obtain information about the configuration of the AnGal moiety and the
positions of O-methylated groups, highly valuable on the identification of algal
polysaccharides.
Infrared Analysis
Infrared analysis provide a 'fmgerprint' picture of the different carrageenan types.
Such technique is particularly apt for locating the site of the hemiester sulphate
groups in the chain [143]. A broad band at 1230-1255 cm ~ due to S-O stretching
vibrations is common to all sulphated polysaccharides and increases in size with
sulphate content. The peaks at 930-94 cm ~ have been assigned to AnGal. Peaks at 820
cm ~ are characteristic of sulphate at a primary group, that is at C-6 of D-galactose; at
830 cm ~ of equatorial sulphate, that is at C-2 of D-galactose, and at 845-850 cm ~ of
axial sulphate, that is at C-4 of D-galactose. A band of low absorbance at 805 cm ~,
131
has been assigned to a sulphate at C-2 of the 3,6-anhydrogalactose unit [130], its
presence in Euchema cottoni carrageenan, is an evidence that some t-carrageenan units
co-exist with ~c- units. The bands at 1069 or 1053 cm 4 appear in K- but not in t- forms
(bands VIII and IX, respectively, in Fig. 13, ref. 144). Films of solutions of the
polysaccharide are made on AgCI plates or KBr disk prepared. Thin transparent films
can also be prepared by drying seaweed extract at 37 ~ and mounted in cardboard
with a center hole before scanning [ 145].
A
8
c
I
....
I,
J,
1200
,
a
1000
X-~ c m 4
A
t..-
I
,
__
|
1200
,
-,
14
Eli
cm 4
~
b
1000
Figure 13. Infrared spectra of carrageenan in water in the K+ form: (a) K-carrageenan
(b) t- carrageenan (from ref. 144, with permission)
132
NMR spectroscopy
The complete interpretation of 13C NMR spectra for the best known regular
structures of red seaweed galactans, was described recently by comparison with the
spectra of numerous model compounds [134]. K-Carrageenan from Euchema and
Chondrus spp. and t-carrageenan from Euchema hybrid carrageenan were found to
yield well resolved ~3C NMR spectra in the anomeric region in each case: the peak at
103.2-103.6 ppm can be attributed to C-1 of the D-galactose and D-galactose
4-sulphate residue. K-Carrageenan gave a signal at 96.2 ppm and t-carrageenan at
93.1 ppm, assigned to C-1 of 3,6-anhydrogalactose and 3,6-anhydrogalactose-2sulphate residues, respectively (Fig. 14). In )~-carrageenan, the anomeric C-1
resonances for the D-galactose and 3,6-anhydro-D-galactose residues occur,
respectively, at 103.4 and 91.6 ppm. The latter signal in particular, is different from
those of ~:- and t-carragageenan above. The ~.-carrageenan spectrum also contains a
peak at 64.2 ppm, which has been tentatively assigned to C-4 of the 3-1inked residue
[ 146]. The presence of masking units, namely pyruvate and O-methyl substituents in
carrageenans has also been identified using ~H NMR spectroscopy [147, 148].
Y
a ~
-_~.---~.;. J
b -~~,.100
90
80
70
ppm
Figure 14. 13C spectra of polysaccharides from (a) Euchema spinosum and (b)
Euchema cottom (with permission from Rochas et al. [ 130]).
133
Carrageenan hybrid structure analysis
It is not safe to assume that the properties of a given carrageenan fraction can be
understood merely in terms of its dominating structural component. The amount,
chemical nature and distribution of heterounits in carrageenan are overriding factors
which have a large influence on their behavior, even in those which closely approach
the ideal structure [ 127]. Heterounits may, in principle, be present, either in the same
chain or in separate chains. In the latter case, the sample is regarded as a mixture of
different polymeric species. In the former case, the heterounits may be distributed
randomly, regularly or in blocks. Hence, commercial 'native' carrageenans must
always be thought as hybrids of the various ideal forms [130]. The use of enzymatic
procedures coupled to ~3C NMR identification has proved successful in order to
determine the complex f'me structure in native ~:- and t-carrageenans [ 130, 149]. This
is an active field of research with important implications on the understanding of the
functionality of commercial carrageenans. Indeed, marked synergistic effects have
been reported in the rheology of t-carrageenan gels (in the K+ form) in the presence of
small fractions of ~c-carrageenan, whether added or present as impurities [ 150].
IDENTIFICATION OF HYDROCOLLOID GUMS IN FOODSTUFFS
Although it is beyond the scope of this chapter, we considered pertinent to include a
brief fmal section addressing analytical tests to identify the presence of gums and
stabilizers in foods.
Qualitative identification of hydrocolloid agents in foods
The presence of certain polysaccharides in foods and beverages can be
diagnosticated by the use of specific 'spot tests'. Thus carrageenan indicated by
positive result using the methylene blue test [ 151 ], carboxymethyl cellulose using the
2,7-dihydroxynapthalene test [152] and acidic polysaccharides (i.e. xanthan, pectin)
using the modified carbazole assay [ 153].
A method for the identification of natural thickening agents in food has been
reported [154]. Isolated thickening agent is treated by methanolysis using hydrogen
chloride in anhydrous methanol. The sugar units will occur as 1-methly glycosides
and the uronic acid units as 1-methyl glycoside 6-methyl esters. The released methyl
glycosides can be analyzed by capillary gas-liquid chromatography (GLC, Fig. 15).
More specifically, after silylation treatment with BSA (N,O-Bis trimethylsilyl
acetamide and pyridine) the volatile O-trimethylsilyl derivatives (TMS), afford
quantitative yields of the constituent monosaccharides [155]. There 11 sugars and 4
134
uronic acids which can be expected as monosaccharide units in polysaccharides
stabilizers and their peaks units are separated to a satisfactory extent (Fig. 15).
GulA
ManA
GIcA
Figure 15. Glass capillary GLC chromatogram of all monosaccharide units to be
expected in stabilizers (from Thier [155], with permission).
A wide range of stabilizers, which are ingredients of foodstuffs, can be identified
including agar, gum arabic (LBG), pectin, carrageenan and gum tragacanth. This
method yields considerable savings of time and labor when compared with former
analytical approaches, however the polysaccharides must previously be isolated from
the substrate in question using various clean-up steps [ 155].
Indirect detection of polysaccharides is also commonly achieved, using
fluorescence methods, by detection of naturally fluorescing compounds such as low
molecular weight phenolic acids (e.g. coumaric acids) which are esterified to specific
structures such as the pericarp cell walls in all cereals [156,157].
135
Quantitative methods
There is no general methods available of the polysaccharides analysis and the
chosen method must be based on the structure and properties of the individual
polysaccharides. Particularly it depends on the qualitative identification of their
presence in a foodstuff, followed in most cases by hydrolysis and measurement of the
sugars thus produced. The application of methods for total sugars to the isolated
precipitates, form the basis of a quantitative method if suitably calibrated [28].
Pectin and xanthan are acid polysaccharides and used in many foods as
thickening agent to modify the texture and the properties. An industrial method for
quantitative determination of acidic polysaccharides has been developed using the
hexamethylene polymer solution, (acidic polymers undergo cross-linking reactions),
and read UV absorption at 235 nm [158]. Rapid quantitative assays of alginates can be
achieved, using a procedure based on complexation with poly(hexamethylene
biguanidinium) chloride [ 159].
REFERENCES
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Alexander ILl. Cereal Foods World,1995; 40: 366-368.
Pazur JH. In: Chaplin MF, Kennedy JF, eds. Carbohydrate Analysis. A Practical
Approach. Oxford: IRL Press, 1994.
Aspinall GO. In: Aspinall GO, ed. The Polysaccharides, Vol. 1. New York:
Academic Press; 1982, 19-124.
Rees DA, Morris ER, Thom D, Madden JK. In: Aspinall GO, ed. The
Polysaccharides, Vol.1. New York: Academic Press, 1982; 195-290.
Pazur JH. In: White CA, ed. Advances in Carbohydrate Analysis Vol. 1, London:
JAI Press Ltd., 1991.
Sun SF. Physical Chemistry of Macromolecules. New York: John Willey and
Sons Inc., 1994.
Klockow A, Widmer HM, Amado R, Paulus A. Fresenius J Anal Chem 1994;
350:415-425.
Gidley MJ. In: Phillips GO, Williams PA, Wedlock DJ, eds. Gums and
Stabilisers for the Food Industry 4. Oxford: IRL Press, 1988; 71-80.
Gidley MJ. In: Alexander ILl, Zobel HF, eds. Developments in Carbohydrate
Chemistry. St. Paul MN, USA: American Association of Cereal Chemists, 1992.
Dietrich CO, Nader HB, Perlin AS. Carbohydr Res 1975; 41: 334-338.
Pazur JH, Kelly-Delcourt SA. Biochemistry 1985; 24: 3368-3369.
Whistler RL, Anisuzzaman AK. In: Whistler RL, BeMiller JN, eds. Methods in
Carbohydrate Chemistry Vol. 8, New York : Academic Press, 1980.
Kennedy JF, D.L. Stevenson, White CA. Starch/Starke 1988; 40: 396-404.
Haworth WH. J Chem Soc 1915; 107: 8-16.
136
15 Bjomdal H, Hellerqvist CG, Lindgren H, Lindberg B, et al. Angew Chem
Intemat Edit 1970; 9:610.
16 Morrison IM. J Chromatogr 1975; 108:361-364.
17 Fox JD, Robyt JF, Carbohydr Res 1992; 227: 163-170.
18 Harding SE, Vanun KM, Stokke BT, Smidsred O. In: White CA, ed. Advances
in Carbohydrate Analysis Vol 1. London: JAI Press Ltd., 1991.
19 Harding SE. In: Phillips GO, Wedlock DJ, Williams PA, eds. Gums and
Stabilizers for the Food Industry 7. Oxford, UK: IRL Press, 1993; 55-65.
20 Fox JD, Robyt JF. Anal Biochem 1991; 195: 93.
21 Isbell HS. In: Whistler RL, ed. Methods in Carbohydrate Chemistry Vol 5. New
York: Academic Press, 1973.
22 Boeek P, Gebauer P, Dolnik V, Foret F. J Chromatogr 1985; 334: 157-195.
23 Whitloek LR, J Chromatogr 1986; 363: 267-276.
24 Lindberg B, Lonngren J, Svensson SS. In: Tipson RS, Horton S, eds. Advances
in Carbohydrate Chemistry and Biochemistry Vol 3. New York: Academic Press,
1975.
25 Stewart TS, Mendershausen PB, Ballou CE, Biochemistry 1968; 7: 1843-1854.
26 Mader WJ. Crit Rev Anal Chem 1970; 1: 194.
27 Smith GB, Downing GV. Anal Chem 1979; 51: 2290-2293.
28 Southgate DAT. Determination of Food Carbohydrates. Barking, England:
Applied Science Publishers Ltd., 1976.
29 Kennedy JF, Pagliuca G. In: Chaplin MF, Kennedy JF, eds. Carbohydrate
Analysis. A prctieal Approach 2nd ed., Oxford: IRL Press, 1995.
30 Blakeney AB, Harris PJ, Henry RJ, Stone BA. Carbohydr Res 1983; 113: 291299.
31 Rittenburg JH, Smith CJ, Ghaffar A, Allen JC. In: Phillips GO, Wedlock DJ,
Williams PA, eds.Gums and Stabilisers for the Food Industry 2, New York:
Pergamon Press, 1984.
32 Tikami S, Sum~ S, Schuerch C, Suzuki M. Chem Pharm Bull 1986; 34: 3933.
33 Fulcher RG, Irving DW, de Francisco A. In: Munck L, ed. Fluorescence Analysis
of Foods. Essex, UK: Longman Scientific and Technical, 1989.
34 Fuleher RG, Faubion JM, Ruan R, Miller S. Carbohyd Polym 1994; 25: 285293.
35 Haugland RP. Handbook of Fluorescent Probes and Research Chemicals, 5th
edn. Eugene OR, USA: Molecular Probes Inc, 1992.
36 Symons SJ, Fulcher RG. J Cereal Sci 1988; 8: 211-218.
37 Symons SJ, Fulcher RG. J Cereal Sci 1988; 8:219-229.
38 Hoseney RC. In: Principles of Cereal Science and Technology. St Paul MN:
American Association of Cereal Chemists, 1986.
39 Wong DWS. Mechanisms and Theory in Food Chemistry. New York: Van
Nostran Reinhold, 1988.
40 Rapaille A, Vanhemelrijck J. In: Imeson A, ed. Thickening and Gelling Agents
for Food. Glasgow, UK: Blackie and Professional, 1992; 171-201.
137
41 Atwell WA, Hood LF, Lineback DR, Varriano-Marston DR, et al. Cereal Foods
World 1988; 33:306-311.
42 Englyst HN, Cummings JH. In: Cereals in an European Context. Chichester:
Ellis Horwood Ltd., 1987; 221-233.
43 Ewers E. Ztschr Offentliche Chemie 1908; 14: 150.
44 Clendenning KA, Wright DE. Can J Res 1945; 23B: 131-138.
45 Lane JH, Eynon L, J Chem Soc Ind 1923; 42: 32-37T.
46 Egan H, Kirk RS, Sawyer R. Persons Chemical Analysis of Foods, 8th ed. Essex,
England: Longman Scientific and Technical, 1981.
47 Bernfeld P. In: Colowick SP, Kaplan ND, eds. Methods in Enzymology Vol 1.
London and New York: Academic Press, 1955.
48 Kidby D, Davidson DL. Anal Chem 1973, 55:231.
49 Osborne DG, Voogt P. The Analysis of Nutrients in Food. London: Academic
Press, 1978.
50 Kiser DL, Hagy RL. In: Charalambous G, ed. Liquid Chromatographic Analysis
of Food and Beverages. New York: Academic Press, 1979.
51 Sloan HR, Kerzner B, Seckel C. Prep Biochem 1984; 14: 245.
52 Eltekov YA, Strakhova NM, Kalal J, Peska J, Stamberg J. J Polym Sci Polym
Symp 1980; 68:247-251.
53 Klingler RW. Starch/St~ke 1985; 37:111-115.
54 Samec M. J Polym Sci 1957; 23: 801.
55 Klaushofer H, Berghofer E, Diesner L. Starch/St~ke 1976; 28: 298-303.
56 Osborne DG, Douglas S. J Sci Food Agr 1981; 32: 328-332.
57 Osborne DG, Douglas S, Fearn T. J Food Techn 1982; 17: 355-363.
58 Evers AD, Stevens DJ Morton ID. In: Pomeranz Y, ed. Advances in Cereal
Science and Technology. St. Paul MN" American Association of Cereal Chemists
Inc., 1985.
59 Muhrbeck P, Tellier C. Starch/St~ke 1991; 43: 5-27.
60 Armstrong E. PhD Thesis, University of Ottawa, Ottawa, Ontario, 1992.
61 French D. In: Whistler RL, BeMiller JN, Paschall ED, eds. Chemistry and
Technolgy, New York: Academic Press, 1984.
62 Clark AH, Ross-Murphy SB. Adv Polym Sci 1987; 33: 57-192.
63 Gidley MJ, Bociek SM. J Am Chem Soc 1988; 110: 3820-3829.
64 Wu H, Sarko A. Carbohydr Res 1978; 61: 7-25.
65 Imberty A, P6rez S. Biopolymers 1988; 27: 1205-1221.
66 Imberty A, Chnazy H, P6rez S, Buldon A, et al. Mol Biol 1988; 201:365-378.
67 Gidley MJ, Bociek SM. J Am Chem Soc 1985; 107: 7040-7044.
68 Veregin RP, Fyfe CA, Marchessault RH, Taylor MG. Macromolecules 1986;
19:1030-1034.
69 Horii F, Yamamoto H, Hirai A, Kitamaru R. Carbohydr Res 1987; 160: 29-40.
70 Veregin RP, Fyfe CA, Marchessault RH. Macromolecules 1987; 30: 3007-3012.
71 Guennet JM. Thermoreversible Gelation of Polymers and Biopolymers. London:
Academic Press Ltd., 1992.
138
72
73
74
75
76
77
78
79
Schoch TJ. Adv Carb Chem 1945; 1: 247.
Lou J, Myers MN, Giddings JC. J Liq Chromatogr 1994; 17: 3239-3260.
Manners DJ. Carbohydr Polym 1989; 11: 82-112.
Lee EYC, Mercier C, Whelan WJ. Arch Biochem Biophys 1968; 125: 1028-1030.
Hizukttri S. Carbohydr Res 1985; 141: 295-306.
Hizukuri S. Carbohydr Res 1986; 147: 342-347.
Anderson DMW, Weiping W. Int Tree Crops J 1991; 7: 29-40.
Immeson AP. In: Immeson AP, ed. Thickening and Gelling Agents for Food.
Glasgow: Blackie Academic and Professional, 1992.
80 Anderson DMW, Millar JRA, Weiping W. Biochem Syst Ecol 1991; 19: 447-452.
81 JECFA-FAO. Food and Nutrition Paper, Rome 1986; 34.
82 USFDA GRAS, Chapter 184; 1987; 1330.
83 FAO. Food and Nutrition Paper 49; Rome: FAO, 1990, 23-25.
84 Phillips GO, Williams PA. In: Nishinari K, Doi E, eds. Food Hydrocolloids:
Structures, Properties and Functions. New York: Plenum Press, 1994; 45-63.
85 Pigott JR, Sharman K. In: Pigott JR, ed. Statistical Procedures in Food Research.
Essex, England: Elsevier Applied Science, 1986; 181-232.
86 Powers JJ, Ware GO. In: Pigott JR, ed. Statistical Procedures in Food Research.
Essex. England: Elsevier Applied Science, 1986; 125-180.
87 Fleiss JL, Zubin J. Multivar Behav Res, 1969; 4:235-250.
88 Jurasek P, Kosik M,, Phillips GO. Food Hydrolloids 1993; 7:73-85.
89 Jurasek P, Kosik M, Phillips GO. Food Hydrolloids 1993; 7:157-174.
90 Jurasek P, Kosik M, Phillips GO. Food Hydrolloids 1993; 7:255-280.
91 Jurasek P, Phillips GO. Food Hydrocolloids 1993; 7:337-352.
92 Saunders RM, Becker R. In: Wickens GE, Haq N, Day P, eds. New Crops for
Food and Industry. London: Chapman and Hall, 1989; 288-302.
93 Biswas S, Biswas B, Phillips GO. Food Hydrocolloids 1995; 9:151-163.
94 Defaye J, Wong E. Carbohydr Res 1986; 150: 221-231.
95 Anderson DMW. Food Addit Contam 1985; 3: 159-162.
96 Anderson DMW, Howelett JF, McNab CJA. Food Addit Contam 1985;2: 159-164.
97 Anderson DMW, Douglas DM, Morrison NAy Weiping W. Food Addit Contam
1990; 3: 303-321.
98 Connolly S, Fenyo JC, Vandevelde MC. Food Hydrocolloids 1987; 5/6: 477-480.
99 Randall RC, Phillips GO, Williams PA. Food Hydrocolloids 1989; 3: 65-75.
100 Osman ME, Menzies AR, Williams PA, Phillips GO, et al. Carbohydr Res 1993;
246: 303-308.
101 Ray AK, Bird PB, Iacobucci GA, Clark BC Jr. Food Hydrocolloids 1995; 9: 123131.
102 Menzies AR, Osman ME, Phillips GO, Williams PA. In: Phillips GO, Wedlock
DJ, Williams PA, eds. Gums and Stabilisers for the Food Industry 6. Oxford, UK:
Oxford University Press, 1992; 507-512.
103 Osman ME, Williams PA, Menzies AR, Phillips GO. J Agric Food Chem 1993;
41: 71-77.
139
104 Dickinson E, Galaszka VB, Anderson DMW. Carbohydr Polym 1991; 14: 373-383.
105 Connolly S, Fenyo JC, Vandevelde MC. Carbohydr Polym 1988; 8: 23-32.
106 Williams PA, Phillips GO, Stephen AM. Food Hydrocolloids 1990; 4:305-311.
107 Heidelberger M. J Am Chem Soc 1955; 77:4308-4311.
108 Strobel S, Ferguson A, Anderson DMW. Tox Lett 1982; 14:247-252.
109 Pazur JH, Kelly-Delcourt SA, Miskiel FJ, Burdett L, et al. J Immunol Meth
1986; 89: 19-25.
110 Miskiel FJ, Pazur JH. Carbohydr Polym 1991; 16:17-35.
111Yates EA, Knox JP. Carbohydr Polym 1994; 24: 281-286.
112 Williams PA, Menzies AR, Phillips GO, Smith CJ. In: Morgan MRA, Smith CJ,
Williams PA, eds. Food Safety and Quality Assurance Applications of
Immunoassay Systems. Barking, Essex, UK: Elsevier Science Publishers, 1992;
385-392.
113 Boyd J, Turvey JR. Carbohydr Res 1978; 66: 187-194.
114 Smidsrod O. Faraday Discuss Chem Soc 1974; 57: 262-274.
115 Morris ER, Rees DA, Thom D, Boyd J. Carbohydr Res 1978; 66: 145-154.
116 Morris ER, Rees DA, Robinson G, Young GA. 1980; J Mol Bio119~ 138:363-374.
117 Grant GT, Morris ER, Rees DA, Smith PJC et al. FEBS Lett 1973; 32: 195-198.
118 Haug A, Larsen B, Smidsrod O. Acta Chem Scand 1967; 21: 691-704.
119 Vreeland V, Slomich M, Laetsch WM. Planta 1984; 162:506.
120 Morns ER, Frangou SA. In: Northcote DH, ed. Techniques in Carbohydrate
Metabolism. Amsterdam: Elsevier; B308, 1981; 1-51.
121Craigie JS, Morris ER, Rees DA Thorn D. Carbohydr Polym 1984; 4: 237-252.
122 Grasdalen H, Larsen B, Smidsrod O. 1978; Proc Int Seaweed Symp, 9th 1977;
309-317.
123 Grasdalen H. Carbohydr Res 1983; 118: 255-260.
124 Thomas WR. In: Imeson A, ed. Thickening and Gelling Agents for Foods.
Glasgow UK: Blackie Academic and Professional, 1992; 25-39.
125 Stanley NF. In: Harris P, ed. Food Gels. Exeter, UK: Elsevier Applied Science,
1990.
126 Rees DA, Williamson FB, Frangou SA, Morris ER. Eur J Biochem 1982;
122:71-79.
127 Piculell L. In: Stephen AM, ed. Food Polysaccharides and their Applications.
New York: Marcel Dekker Inc, 1995;205-245.
128 Stanley US Patent 1963; 3094517.
129 Ainsworth PA, Blanshard JMV. Lebensmittel Wissenchaft und Technology
1978; 11: 279-282.
130 Rochas C, Rinaudo M, Landry S. Carbohydr Polym 1989; 10:115-127.
131 Bodeau-Bellion C. Physiol V6g 1983; 21: 785-793.
132 Stevensson THE, Furnaux RH. Carbohydr Res 1991; 210: 277-298.
133 Painter TJ. In: Aspinall GO, ed. The Polysaccharides Vol. 2. New York:
Academic Press, 1983; 195-285, and references cited therein.
134 Usov AI, Yarotsky V, Shaskov AS. Biopolymers 1980; 19: 977-990.
140
135 Percival E, McDowell RH In: Methods in Plant Biochemistry Vol 2. London:
Academic Press Ltd., 1990.
136 Yaphe W, Arsenault GP. Anal Biochem 1965; 13: 142-148.
137 Dubois M, Gilles KA, Hamilton JK, Rebers PA, Smith F. Anal Chem 1956, 28:
350-356.
138 Vogel AI, Bassett J, Denney RC, Jeffery GH et al. Quantitative Inorganic
Analysis including Elementary Instnunent Analysis. London: Longman Group
Ltd, 1976.
139 Fassihudin BA, Williams PA. In: Phillips GO, Wedlock DJ, Williams PA, eds.
Gums and Stabilisers for the Food Industry 6. Oxford, UK: Oxford University
Press, 1992; 145-148.
140 Falshaw R, Furneaux RH. Carbohydr Res 1995; 269:183-189.
141Hakomori S. J Biochem (Tokyo) 1964; 55: 205-208.
142 Garegg PJ, Lindberg B, Konradsson P, Kvarnstr6m I. Carbohydr Res 1988; 176:
145-148.
143 Rochas C, Lahaye M, Yaphe W. Bot Mar 1986; 29: 335-340.
144 Belton PS, Wilston RH, Chenery DH. Int J Macromol 1986; 8:247-251.
145 Knutsen SH, Grasdalen H. Bot Marina 1987; 30: 497-505.
146Falshaw R, Furneaux RH. Carbohydr Res 1994; 252: 171-182.
147Izumi K. Biochem Biophys Acta 1973; 320: 311-317.
148Knutsen SH, Grasdalen H. Carbohydr Polym 1992; 19: 199-210.
149Greer CW, Yaphe W. Botanica Marina 1984; 27: 473.
150Piculell L, Nilsson S, Muhrbeck P. Carbohydr Polym 1992; 18: 199-208.
151 Williamson FB, Ph.D. Thesis, University of Edinburgh, 1960.
152 Strange TE, J. Ass Off Agric Chem 1957; 40 482.
153 Bitter T, Muir HM, Anal Biochem 1962; 4: 330.
154 Preuss A, Thier HP. Z Lebensm-Unters Forsch 1983; 176: 5.
155 Thier HP. In: Phillips GO, DJ Wedlock, Williams PA eds. Gums and Stabilisers
for the Food Industry 2. New York: Pergamon Press, 1984.
156 Fulcher RG, OBrien TP, Lee JW. Austr J Biol Sci 1972; 25: 23.
157 Alkin DE, Ames-Gottfred N, Hartley RD, Fulcher RG et al. Crop Sci 1990; 30:
396.
158 Kennedy JF, Melo EHM, Crescenzi V, Dentini M, et al. Carbohydr Polym 1992;
17: 199-203.
159 Kenedy JF, Bradshaw IL. Carbohydr Polym 1987; 7: 35-50.
D. Wetzel and G. Charalambous (Editors)
Instrumental Methods in Food and Beverage Analysis
9 1998 Elsevier Science B.V. All rights reserved
141
ANALYTICAL NEAR INFRARED SPECTROSCOPY
David L. B. Wetzel, Kansas State University, Shellenberger Hall, Manhattan, KS 66506 (USA)
SUMMARY
Every day thousands of near infrared analyses are performed mostly by non-spectroscopists on a routine basis primarily using this technique as an inspection tool. One of the
earliest modern applications of chemometrically based near-IR was in the commodity and
food area. In the past three decades, use of near-IR has been due to it's, 1)speed, requiring
20 seconds or less; 2) user friendliness in that sample preparation or pre-separation is not
required; 3) the fact that once the chemometric relationship is established and calibration
has been done, data can be collected routinely by technicians with a minimum of training.
Besides the continuous expansion of applications as near-IR has matured much attention has
been given to software by 3rd party providers, instrument users, and manufacturers. Recent
hardware advances and solid state detector technologies have revolutionized instrument
design by making rapid electronic wavelength switching (with no moving parts) readily
available with acousto-optic tunable filter monochromomators or with grating polychromator
near-IR diode array systems. FT-NIR instruments, offered also by traditional FT-IR
manufacturers, provide alternatives to grating monochromators for scanning or random
wavelength access. Analytical near infrared spectroscopy is a useful and cost effective
method of food analysis at ingredient, processing, and product stages of production.
INTRODUCTION
In the past three decades, practical near-IR spectroscopic analysis has become well
established as a routine inspection tool in a number of industries. This includes agriculture and
food. It is not only a fast method, but also very little sample preparation is required and even
granular solids can be analyzed directly using statistically based, quantitative, chemometric
relationships. Although structural information is less apparent from near-IR data than from
mid-IR, a qualitative analysis inspection function is available in addition to quantitative determination via a multiterm pattern recognition (discriminate analysis) approach.
The range of near-IR is from 850 nm to 2500 nm (11,000-4,000 cm-l). At the lower
wavelengths or higher frequencies it is acknowledged that some electronic transitions are possible
but that this is not the major source of near-IR absorption. In general, the origin of near-IR is the
overtones and combination bands of fundamental vibrations in the mid-infrared spectrum
from
2.5 - 15 ~tm (4,000-600 cm'l). Traditionally, quantitative infrared spectroscopy applied
142
to mixtures in powered form has been assumed to be very difficult and perhaps entirely impractical
for such analytical procedures and therefore, chromatography or various other separations have
been necessary before determination of individual components could be done. In the mid-infrared
region of the spectrum, diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy
introduced by Fuller and Griffiths (1) has received considerable attention in the last 20 years.
Usually spectral subtraction is used to uncover hidden features within mixtures and it is necessary
to dilute the sample with solid granular KC1 in order to avoid complete saturation of absorption
bands by this technique. Some general applications of near-infrared may be reviewed in Table 1.
On that list, compiled in the past when near-IR was being newly developed, greater emphasis was
given to the handling of solid samples by diffuse reflectance. Of course, clear liquids and slurries
can be handled even more readily. Near-IR quantitative techniques are useful in many cases, but
in general, it is not considered for trace analysis applications. Where components are present in
the amount of 1% or more, weakly absorbing samples may be analyzed using multiple wavelengths
with multivariate statistics.
The convenience of sample handling, computer assistance, and
additivity of near-infrared responses makes this possible. Although the subject of this text involves
analytical near-infrared spectroscopy for analysis of food and beverages, originally it was applied
by industrial chemists working for
Table I. Facts Concerning Near-Infrared
Tennessee Eastman in the 1950's.
The work of Wilbur Kaye (2) led
9 Agricultural commodities in worldwide commerce are analyzed by the way to future activities. This
near-IR for their content of water, protein, lipid, etc.
work was subsequently also re9 Pharmaceuticalshave been identified by applying discriminant analysis viewed by Whetsel (3) also ofTenexpressions from multiwavelength data.
nessee Eastman. These industrial
9 Rapid and timely analysis for process control in various processing chemists, faced with either neat
industries is based upon near-IR methods,
chemicals or highly concentrated
9 Samplepreparation prior to spectroscopic measurementis minimal even solutions, encountered strong abin the case of powders.
sorption bands in the mid-infrared
Industrial applications of near-IR include determining:
9 molecular weight of propylene and ethylene glycol polymers
9 moisture content of coal and hematite
9 textile blends of cotton-polyester and of rayon-polyester
9 finishes on textile fibers
9 amount of cross linkage in chemically modified starch
9 hydrogenationof unsaturated fatty acid esters
9 hydrocarbon mixture of n-hexane, benzene, cyclobenzene, iso-octane
9 volatiles (loss on drying) and moisture in cosmetics
9 total alkaloids, nicotine, and reducing sugar in tobacco processing
9 moisture in pharmaceutical excipients and in detergent powders
and they moved to the near-infrared as a way of avoiding saturation
of the bands and production of
non-linear quantitative relationships. (This approach is similar to
that routinely employed by atomic
spectroscopists when an emission
line for one element is too strong,
they merely select a weaker line.)
In the case of vibrational spectroscopy, either overtone or combina-
143
tion bands have greatly diminished absorptivity in comparison to their fundamentals. A British
physicist named Goulden (4) applied near-infrared spectroscopy to a number of dairy products as
early as 1956. He also used what he referred to as diffuse "absorptance" and published the spectra
of various solids including nonfat dry milk. With perhaps few exceptions, most workers from the
ranks of spectroscopists and analytical chemists prior to mid-1960's dismissed the near-IR region
as having only weak (and probably uninteresting) absorption bands that overlapped. This same
group also dismissed the notion of obtaining anything meaningful, let alone quantitative analysis
from the diffuse reflectance procedure because the effect of scattering was presumed to complicate
the issue hopelessly. One persistent agricultural engineer, Karl Norris, with the United States
Department of Agriculture, and his associate, David Massey, who built the specialized instrument
used by Norris, proved to be the exceptions (5).
This persistence paid off through the introduction of a new way of treating spectroscopic
data. Chemometrics enabled using diffuse reflectance techniques on granular solid commodities
(to avoid the use of toxic solvents such as carbontetrachloride) that produced spectra that could
not be treated with a simple one-wavelength Beer's Law approach. Unlike spectroscopy of clear
liquids or dilute solutions, diffuse reflectance does not have the luxury of having 100%
transmission as a reference, and, therefore, an arbitrary standard reflector is used as the
background reference.
Absorbance = log [Intensity (of background without sample)/Intensity (with the sample)]
Multiple overlapping absorption bands that are weak in intensity require the use of perhaps multiple
wavelength optical terms each with its own coefficient in an analytical equation. Furthermore, in
diffuse reflectance the effect of scattering works to advantage by returning light back to the detector
but it will add uncertain experimental variables by producing a multipath effect because it returns
by a circuitous route (multiple bounces) once it has penetrated the specimen. Thus, the variability
of the baseline must be compensated for in the equation with a baseline correction that deals with
the scattering component. With these data treatment techniques that allow compensation for optical
effects and with some of the basic features of near-infrared absorption that will be discussed later,
modern near-infrared was rediscovered three decades ago and has found its way into a prominent
position at present in applied spectroscopy of industrial and agricultural materials primarily for
quantitative purposes but also to a lesser degree for identification.
SPECTRAL FEATURES OF THE NEAR-INFRARED
In comparison with other optical analysis methods, let us consider the relative sensitivity
of the method, the instrument signal-to-noise ratio, and the selectivity. The sensitivity of an optical
absorption spectroscopic analytical method is dependent on the probability of a transition from
the ground state to an elevated energy state within the molecule. Ultraviolet has a high sensitivity,
followed by mid-infrared. In both cases, the absorptivity coefficients are high. In contrast, the
144
sensitivity of near-infrared is much lower, and the absorptions are weak. The energetic difference
between high energy vibrations of excited molecules and the energy state at the lower vibrational
levels is less. Furthermore, the probability of vibrational excitation into the higher energy state
in the near-infrared is less than it is in the mid-infrared. As will be discussed later, this apparent
weakness is not necessarily a disadvantage and in fact, it makes possible some of the additivity
required to process the data to produce quantitative results.
The instrument signal-to-noise ratio (S/N), although it cannot compare with the visible
region of the spectrum where photomultiplier detection is used, is nevertheless much, much greater
than that of the mid-infrared, the ultraviolet, or the far-infrared. Because of the relatively low
sensitivity of near infrared, high S/N operation is a necessity. If one is making analytical decisions
based on a few milliabsorbance units of difference in the spectra of various samples then the noise
level must be held down to a few microabsorbance units. This instrumental requirement exceeds
that of most other optical instruments.
Selectivity permits considering absorptions from the analyte in a specific part of the
spectrum that are not affected by the absorptions of a molecule of the matrix or possible interfering
material that is present in the matrix. In contrast to the mid-infrared, which is famous for
qualitative information content and for the ability to relate specific absorption bands to molecular
structure, near-infrared rates in selectivity about one-third of the value as that of mid-infrared.
The selectivity of near-infrared, however, exceeds that of ultraviolet, visible, and far-infrared
spectroscopy.
Figure 1 shows absorption spectra in the most commonly used portion of the near infrared
region. These absorption spectra of 10 mixtures of organic fluids are plotted as Absorbance against
O. 7000.
0,6000.
0.5000
tu
u
z
<
o. 4000-
o
In
(D
<
0.3000-
0.2000'
O. t000
0,000t
~
~ !
1200.0
--
.1400.0
-----t-----a------I
t600.0
t800.0
~
!
2000.0
"
;
2200.0
'
: ~ -2400.0
HAVELENGTH
Fig. 1. Spectra of mixtures of organic fluids from Kansas State Universityresearch model Near-IR Acousto-optic
Tunable Filter Spectrometer.
145
wavelength from 1100 to 2500 nm. Note that there are four distinct regions of the spectra going
from the strongest bands at the right in the region above 2100 nm where the most prominent CH
combination band of hydrocarbons is found. Of the 10 spectra that are superimposed, a drastic
excursion in the absorbance between 2100 nm and 2200 nm is noted. A reverse excursion occurs
at the taller band in the region of 2250-2500 nm. A reciprocal arrangement is noted in that there
is a decline in the band at 2300 nm when the band at 2150 nm increases.
In this particular
illustration, the two bands referred to are due to combinations of aromatic CH and aliphatic CH
and when among the mixtures of hydrocarbons there is an increase in the one form there is a
corresponding decrease in the other form. In the region of 1600-1800 nm there are overtones of
the fundamental CH vibrations. Again, an excursion is noted among the 10 spectra that are
superimposed. The shape of the band that occurs in this region is changed dramatically and the
positions of the three prominent peaks have shifted depending on whether there is more aromatic
or aliphatic hydrocarbon present. Note that when plotted on the same scale as the first combination
and first overtone, the differences on the second combination at 1400 nm and the second overtone
in the 1100-1200 nm regions appear to be very small. It should be mentioned that upon scale
expansion, these regions of the spectrum (1100 - 850 nm), where the resulting absorption for the
higher combinations and overtones occur, are also useful.
In food and beverage analysis, we are concerned not only with hydrocarbons but also with
protein (amides), carbohydrates, lipids, and moisture. When these materials are found in a typical
food matrix and diffuse reflectance spectra are run, the contributions of individual constituents
are less obvious than when we look at the spectra for known mixtures of pure chemicals.
In
Figures 2 - 5 we observe the spectra of several hydrocarbons in the mid-infrared region for both
aliphatic and aromatic hydrocarbons. The unsaturation is readily apparent in the spectrum of
benzene by looking at the bands at 3037 and 1480 cm -1. The fundamental vibrations at the longer
wavelengths (higher frequencies) of the mid-infrared spectra will not appear in the near-IR
spectrum.
However, the CH vibration occurring at 3037 cm -1 from the CH attached to the
carbon-carbon double bond is at a sufficiently high frequency that there will still be measurable
absorption in the near-infrared that is observed graphically when scale expansion is used. Note
also in the region of 2927 cm -1 and 2862 cm -1 are the fundamental absorption bands of CH2 and
CH3. In the interpretation of mid-infrared spectra, the relative population of CH2 and CH3 will
tell us whether the structure is linear with many CH2's and only terminal CH3, or whether many
branch chains are presnet giving rise to more CH3's and less CH2's.
Now let us examine the near-infrared spectra of some simple hydrocarbons obtained on a
Fourier transform near-infrared (FT-NIR) instrument. In a spectrum of hexane (Figure 6) plotted
in wavenumbers (cm-1), we see the same four regions of CH bands that were observed in Figure
1. In hexane, we have CH2 and CH3 groups present. There are also carbon-carbon bonds. The
contribution of the carbon-carbon fundamental vibration is not observable as a harmonic in the
146
near-IR spectrum of hexane, and, thus, we are
100.0.
9o.oo
simply counting CH's and making a slight dis-
8o.o0
tinction between the CH2 and CH3 intensities.
The place to look for the slight distinction is in
the 6000-5500 cm -1 region. In contrast, the
7o.oo
60.00
.5O.OO
40.00
spectrum of benzene, shown in Figure 7, which
30.00
20.00
,
,
,
,
,
Wovenumber Ccrn-1)
XT
lOO.O~
95.00~
90.00~
toluene (Figure 8) where one of the aromatic
CH's of benzene have been replaced by a methyl
group with its three aliphatic CH's, we observe
a mixture of aromatic and aliphatic CH's and it
80.00.
75.0070.0065.0G
is readily apparent that the toluene spectrum is
more like the benzene spectrum than the hexane
w~n~,t,~
(~-1)
spectrum. In xylene (Figure 9), two aromatic
CH groups had been replaced by methyl groups
with the introduction of six aliphatic CH's.
100.095.0(
Thus, the spectrum of xylene has become more
like hexane than the original benzene.
By plotting the spectra of liquids with an
90.0(
8,5.0080.00
75.00-
expanded scale and plotting them linear in
70.00-
wavenumbers, we see that the early presump-
6,5.00
Wovenurnber {a'n-1)
100.1~
tions of spectroscopists is incorrect and that the
weak bands are not necessarily uninteresting nor
are they useless. In dealing with food products,
the health of heart patients (where heart disease
is the biggest killer) is concerned with the degree
of unsaturation of fats and oils. When we
compare the spectrum of oleic acid (Figure 10)
to that of hexane, distinct similarities are noted
due to the C18 hydrocarbon chain versus the
95.1X~
~.~
oo.~
75.0ff
70.0065.00.
60.00
W~numl~r Co'm- 1)
Fig. 2.
Fig. 3.
rig. 4.
Fig. 5.
1
has 5 aromatic CH's, has a tall band in the
region of 4750-4500 cm -1. Also the band at
6000 cm -1 is shifted and has a distinctly different
shape than that from the straight chain hydrocarbon hexane. Examining the spectrum of
FT-IR spectrum of hexane (transmittance).
FT-IR spectrum of oenzene.
FT-IR spectrum of toluene.
FT-IR spectrum of xylene.
shorter chain C6. There are two things to note
in contrast; first of all the CH2 to CH3 ratio
affects the shape of the band between 6000 and
5500 cm -1 and more importantly there is a band
147
2.11111-
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
""
"
'
"
in the same region where we observe aromatic
character due to the presence of the double bond ~.~
in oleic acid. Oleic acid is designated as C18:1.
For the 18 carbons in the chain there is one~ ''|
double bond present.
In Figure 11 a greater~.0,,
degree of unsaturation shows up in the spectrum
of linoleic acid because there are two double .3~
bonds instead of the single one found in oleic
~
~
1
~
--0.~0.
acid. Because of the high signal-to-noise ratio,
,
,
~*.
9/ ~ . " '
i~l."
i~.
,
9
,
9
2.
~ . _~l
in good quality near-infrared instruments, it is2"~'~ . . . . . . . . . . . . . . . . . . . . . . .
possible to scale expand the region at a p p r o x i - ' - ~ 1
mately 4600 cm -1 and to quantitate the differences and use these data to determine the iodined'2~
|q
value (amount of unsaturation) of the fat or oil.,~.0t~
:1.r
A cursory look at the spectrum of corn oil
(Figure 12) plotted on the same scale as the oleic .~
L
and !inoleic acid spectra, readily shows that the_, . . . . . .
corn oil has unsaturation nearer to that oflinoleic ~*~
2 . 1 ~
.
.
.
.
.
'
.
. . . . .
7***. " i~.
2.
~.
.
.
.
.
.
'
"
"
'"
"
c~-t
"
than of oleic acid.
In the early days of modern near-infrared "~1
spectroscopic analysis, the author of this chapter
did successful calibrations of fats and oils to~t2~
iodine value obtained by fatty acid methyl ester~.0t,~
gas chromatographic data. Solid fat index data
were also used, as well as average chain length . ~ J
determined from high performance liquid chro-_0 , ~ , - , - - - _ _
matography data. All of these oil analyses were " ~ " " i~. "
~.
~
" " ~
" " ~
' ~-~
successful, and all of them were due to absorp-21~ 1 . . . . . . . . . . . . . . . . . . . . . .
tions in the region that we are presently examining. We have contrasted these to spectra of other
,.6~
i
1.241~
unsaturated materials such as toluene, xy~ene,~
and benzene and the industrial samples found ln~.0~0."
g "
Figure 1.
The amide group that is present in protein "~
produces an NH stretch absorption band that
appears in the mid-infrared spectrum at
_, Fig.
~ . 6.
3280 cm-l(Figure 13) and strong amide I and II Fig. 7.
Fig. 8.
bands at 1650 and 1550 cm -1. Examination of Fig. 9.
~.
"absorbance
~."
ispectrum
~."
~of
. " hexane.
~ . "~-~
FT-NIR
FT-NIR spectrum of benzene.
FT-NIR spectrum ot to.mene.
FT-NIR spectrum ot xylene.
148
2. | 0 0
',
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
the near-infrared spectrum of butylamine (Figure 14) in contrast to that of hexane showed the
! .670
presence of additional bands near 5000 cm- 1 and
!.240
6500 cm 1 .
These are obviously near-infrared
manifestations of the fundamental NH stretch-
.810
ing vibrations in the combination and overtone
=*
9
[~
-,.~ ~
8=,.. . ". ~. ..
2.t00
.
.
". ~. . ' "
',
,.j
forms. These bands are in addition to the famil-
_..J
~'---~
iar CH stretch combinations and overtones.
~9 . '9 '
~. . .' ".
~. . 'c,-i Figure 15 shows the mid-infrared OH stretching
.
.
.
.
.
vibration at 3300 cm -1 in the spectrum of an
,.~,o
alcohol. The near-IR manifestation of this appears in 1-propanol and 1-octanol spectra in
i
.2,o
: , . F i g u r e s 16 and 17 at approximately 4750 cm "1
.,to
and again at 6300 cm 1 .
Note also that the
relative prominence of the OH band compared
~9
/~
-0.~ ~
f
l
ej
to the CH band in the combination region is
"-----"
greater when one OH and three carbons are
=oo. ." . ". ~. .. . ' " .~x].
' . '. .i~.. . ""
~_~i
" +-;~.
" ~ t present versus one OH and eight carbons. Ad2.t00
.
.
.
.
.
.
,
,
,
ditional differences are observed in the 6000i.67,
5500 cm -1 region in terms of the band shape.
This, furthermore, is due to the relative n u m b e r
t.240
, . o f methyl and methylene groups on the threei.m~
carbon alcohol versus the eight-carbon straight
chain alcohol.
~9
Now that we have looked at spectra of
-0.~
== ' ' imd.' " ~ d . '
x-r
.
liquids of simple chemicals, we are better pre' ~x~.' ' ~=d.' " iod.'c~t
2.tooU. . . . . . . . . . . . . . . . . . . . .
100.
90.
1.6"/O-
80.
70.
t.z,lo-
ii
.380-
,
35oo
Fig.
Fig.
Fig.
Fig.
10.
11.
12.
13.
x~o
~mo
..
,
~c~
,~'oo
|
,ooo
w. . . . . b,,C:m-;)
FT-NIR spectrum of oleic acid ~one C =C).
FT-NIR spectrum of linoleic acid (two C =C).
FT-NIR spectrum of corn oil.
FT-IR spectrum of wheat gluten (NH 3300 cm").
+
_,.~
=~J~"" imd."" ~ . ' "
i m L " " ~X~.'" ;=d. "CM-1
Fig. 14. FT-NIR spectrum of butylamine.
149
pared to examine the spectra obtained from com- x,
modities that are mixtures of many materials. ?
Figure 18 shows the diffuse reflectance spectra of
~
~
granulated solids. One spectrumisofanoil seed i. 1
tl
and the other is of a cereal grain. Although these
spectra are plotted in wavelength, as are most of +0.o~,i-~
the spectra obtained prior to the involvement of
1
FT-NIRin the near-infrared region, the bands due
~
xi~ " 2~ ..... 2~ r 1/~ ~
to the CH vibrations are readily apparent. The
,,,.~ r176
oil seed contains lipids; it nevertheless, has long 2"**t . . . . . . . . . . . . . . . . . . . . .
!
chains of C16 to C18 carbons and their CH's ,~1
1
A
produce what the near-infrared analysts refer to t
,,,.11 i
as the oil bands. A pair of bands appears at~"~*~
I/~ z
approximately 2310 nm and this pair is repeated~ mo~
I~ I i
in the region of 1700 nm. These same bands are ~
A I
i
recognized as those discussed in the previous ~
~
~+
/~
i
figures whether they were plotted by wavelength + ~,,
' ~ . " ..... imod."" +rod.' " i~."" ~1." " ;+oo+. c.-:t
or by wavenumber. The figure shows prominent
bands at 2050 nm and at 2180 nm for the oil seeds 2"~I . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
which also happen to be high in protein as evi- ,.m~
It
1
denced by their average protein value shown in ~
I~. I
9
, h t2
Tablell. Cereal gralns
are known to have a h,g
~2~
I'~ 1'~
carbohydrate content, and the contribution ot~.,,q
+
1
carbohydrate results in the large band at 2100 nm. ~
This band is so strong that the contribution of the -~
/~___f~
t
protein bands just described are nearly obscured_, ~_
_1
mod." " imd." " +oo+." ' imo+." " ~od." " kn+. t:.--:t
by the contribution of the carbohydrate. In spite
of this visual fact, equations are routinely produced from the digitized data that reflect the
difference in the slope, particularly at the region
of 2180 nm vs. 2100 nm. In fact, in an analytical
equation, both the 2100 nm and the 2180 nm
.s .
"~
.5
,t,
,3.
Table II. Compositionof Wheat and Soybean
Starch Protein Li0id Fiber
(%)
(~)
(%)
(~)
Wheat
70
14.0
2.0 3.5
Soybean
40.0 21.0 5.5
.t i;
w,,,-te,,au, ( ~ )
LL
ig. 15. FT-IRspectrumof 1-octanol.
!g. 16. FT-NIRspectrumof 1-propan,ol.
tg. 17. FT-NIRspectrum of i-OC,t~Ol.
ig. 18. NIR spectraof soybean(top~and wheat(bottom).
150
wavelengths typically are used because with a
high protein ground grain or flour the 2180 nm
0.090
goes up and the 2100 nm declines. Typically,
0.060
an equation for protein determination in a cereal
0.040
grain will have a positive coefficient times the
optical value representing the protein and a nega-
O.Ok~
a.Ol~
tive coefficient at the wavelength representing
the carbohydrate. To complete this determination, at least one more wavelength term is usually
;ii;;l
i
added to adjust for the baseline effects, particularly when dealing with granular solids or other
causes of scattering. When a scanning instrument is used that is capable of producing a full
spectrum, there are other statistical means available for looking at the entire pattern and produc-
100110 ~
9000 ~
BOO0 ~
m
7000 ~
~
50(10 ~
~
Ic~l
(____
ing quantitative results as well. It should be
pointed out that at least 80% of all qualitative
near-infrared analyses performed now are done
with filter instruments with which a group of
0.~0
O.llO
0.100
0.090
O.OaO
0.0~0
discreet wavelengths is used to produce terms
for a multi-term analytical equation.
In the previous text, we have seen that it
O.mtn
0.0'30
0.040
0.11130.
is possible by scale expansion of the near-infra//~
red spectra of hydrocarbons and other simple
chemicals to show relatively sharp peaks in the
--0.020
near-infrared spectra that have their origins in
the mid-infrared. With a grating monochromator scanning instrument near-infrared spectra
typically are plotted in wavelength rather than
wavenumber. The plotted wavenumber spectra
that we have shown for the hydrocarbons in
Figures 19 through 22 serve as a context for
examination of the spectra of wheat gluten,
wheat starch, cellulose, and wheat germ that
,
,~ ,5**
,~
have OH stretching absorptions at 5150 and
cm -1. Unlike the spectra of liquids, these
Baseline corr. FT-NIR spectrum of wheat gluten.6920
Baselinecorr. FT-NIR spectrum of wheat starch.were obtained in a diffuse reflectance mode and
Baseline corr. FT-NIR spectrum of cellulose.
Baseline corr. FT-NIR spectrum of wheat germ. had serious baseline incline from low to high
kll--tl
Fig. 19.
Fig. 20.
Fig. 21.
Fig. 22.
151
wavenumbers and a highly upward displaced baseline. The four bands observed on the spectrum
of hexane in Figure 7, can serve as a marker for the first CH combination at the extreme right
(4030 cm'l), then the first CH overtone (5500 - 6000 cm-1); the weak second combination (near
7300 cm-1), and lastly the weak second overtone at 8000 - 8500 cm "1 in the figure. With these
spectral landmarks in mind, let us examine the baseline-corrected spectrum of fractions of the
common commodity, wheat. In Figure 19, we observe the characteristic amide groups at 4611
and 4865 cm -1 (approximately 2180 nm and 2055 nm). Examination of the spectra of starch
(Figure 20) and cellulose (Figure 21) show that the bands for the amide groups are absent.
Obviously, the well-known success of determining protein in wheat, whether it is in ground whole
wheat or in wheat flour, is due to the presence of these amide groups shown in the gluten spectrum
and their absence in the background matrix material of mostly carbohydrates. A strong bands
occurs at 4762 cm -1 in the spectra of both carbohydrates (starch and cellulose).
The practical utility of near-infrared used to examine wheat milling fractions for the
presence of non-endosperm material is evidenced by the difference between spectral features of
cellulose and starch. Two bands occur for cellulose at 4240 and 4388 cm -1, whereas for starch
the main absorption band is a doublet centered at at 4388 cm -1. Most of the lipid content in wheat
is found in the wheat germ. Figure 22 shows the spectrum of wheat germ in which the carbohydrate
band is diminished in comparison to the magnitude in starch and a CH overtone band appears at
5810 cm -1. At high lipid content, a second CH combination bands appears at 4300 cm -1. The
baseline-corrected scans here were produced from an FT-NIR instrument equipped with a fiber
optic probe that was dipped directly in the granulated solid sample. The purpose of baseline
correction is to allow for scale expansion, so that minor peaks or minor differences of major peaks
can be observed. Whenever a serious baseline slope occurs, much of the vertical scale is consumed
with the shift in the baseline.
Additionally, a plot linear in wavenumber is advantageous for
discussing the theory and origin of a particular near-infrared spectrum in reference to its
mid-infrared counterpart. Even multiples or fractions of corresponding frequencies are readily
observed from these linear plots.
Because most collections of near-infrared spectra obtained on grating monochromator
instruments are expressed in wavelength, most of the rest of the spectra presented in this chapter
will be displayed in that manner, and when this is done, we will refer to particular bands by
wavelength. It is likely that this dual terminology will persist for some reasonable period of time,
because old habits are difficult to change. For this reason, we ask the indulgence from the reader.
WHY NEAR-INFRARED ANALYSIS WORKS
So far in this chapter, we have presented near-infrared as a normal extension of the
mid-infrared spectrum and we have gone out of our way to show the sharp absorption bands of
pure chemicals to prove to the reader that such absorption bands exist. At the time of this writing,
152
near-infrared sessions appear on the programs of national and international spectroscopy conferences. Near-infrared spectroscopy finally has come into its own. This was not always the case.
More than a decade ago, this author described near-infrared as a "sleeper" among spectroscopic
techniques (6). It was stated that it is a sleeper because "it is unknown, illogical, or presumed a
priori to be illegitimate by spectroscopists and analytical chemists". At that time, it was used
almost exclusively by workers in the agriculture and food areas. Seldom were there complete
spectra available to examine because only filter instruments were used. When they were, one
observed only some nonspectacular broad and overlapping bands. In a diffuse reflectance mode
of solid samples, the bands were broadened considerably, and the baseline slope was a confusing
feature. If one had diffuse reflectance spectra of high protein wheat flour and lower protein wheat
flour plotted as the log 1/reflectance vs. wavelength, the difference in the protein level between
the high and low protein samples was hidden within the width of the pen used to plot each spectrum.
Typically the near-infrared analyzers available were filter instruments, and they were
treated as a mysterious black box, because the coefficients on the analytical equations were
produced purely by statistical means. Little effort was made to communicate the background of
these absorbance differences or to relate them in a spectroscopic manner to their origins. Quite
often, the purely statistical approach would turn up a useful correlation between some region of
the spectrum (or a particular filter in a filter instrument) and some property of the material with
no apparent understandable relationship that an analytical chemist or spectroscopist could cling
to. People who sold and used early near-IR instruments took a rather cavalier attitude toward the
statistically established method for which few theoretical explanations were provided. Fortunately, more than a decade ago, a few classical spectroscopists were brought into the field and
because of their established reputations and prestige, other professional spectroscopists began to
consider that near infrared was perhaps not a mysterious black art but, in fact, did have a scientific
basis.
It is the nature of absorptions in the near-infrared to be weak since, as we have mentioned,
they consist of overtones or combinations of fundamental vibrations. The subtle differences among
samples previously referred to require a careful measurement of the signal. Near-IR is concerned
with observing differences between two or more samples in milliabsorbance units. When this is
done, it is meaningful only if the noise level is kept to a few microabsorbance units. Such
instruments require more stringent control for routine spectroscopic quantitation. When solid
sampling is used, diffuse reflectance requires special consideration in areas of the optical design,
operation sequence, sample handling, data accumulations, and statistical treatment.
Having examined the near infrared spectra of simple chemicals and compared them to their
mid-infrared spectra, we note that a few simple fundamental vibrational bands in the mid-infrared
region of the spectrum of a particular compound will produce multiple overtones and multiple
combination bands. The overtones yield many higher frequency bands in the near-infrared region.
153
These overtones from different fundamentals are broader and will tend to overlap. Therefore the
interpretation is much more difficult. The mid-infrared region is well known as a valuable tool
for obtaining structural information. In the near-infrared region such structural information is
obscured. We have demonstrated with simple hydrocarbons that some structural information will
explain subtle differences in the near-infrared spectra. Nevertheless, structural information is
relatively obscure for near-infrared when comparing it to mid-infrared. In the mid-infrared,
decades of data collection have yielded huge libraries of mid-infrared spectra and dedicated
workers have compiled catalogs of absorption bands in reference to certain functional groups.
Information about mid-infrared absorption frequencies of certain functional groups is helpful in
knowing where to look in the near-infrared for overtones. Analytical information also can be
obtained even from components that do not absorb in the near-infrared by observing a shift in the
frequency of known bands due to the influence of neighboring molecules.
Workinev with Weak Absomtion Bands
Weak absorptions in the near-infrared have been considered as shortcomings for this region
of the spectrum in the past. Instrumentally this is true, but relative absorption strength also serves
as a convenient self-limiting factor to restrict which vibrations are observed. We stated earlier
that a few fundamentals will produce many overtones and combinations. Fortunately, not all of
the overtones or combinations are observed, because the energetic self-limiting factor simplifies
the spectrum. For example, a fundamental vibration occurring at 15 lam with overtones at
approximately 7.5, 5.0, 3.75, and 3.0 is not likely to be observed in the near-infrared because the
fifth or sixth overtones lack intensity. Weak absorption in the near-infrared provides selectivity.
The commonly used region from 1000 - 2500 nm (10000 - 4000 cm -1) contains overtones of
fundamental vibrations no higher than 5 to 8 ~m depending upon the intensity of the fundamental.
The long wavelength end of the mid-infrared beyond 8 ~m (1250 cm -1) does not contribute to the
near-infrared. This means that the overlapping bands in the near-infrared produced by many
combinations of overtones although spectroscopically complicated are from only a few molecular
groups. In the near infrared we predominately see the result of vibrations of light atoms that have
strong molecular bonds.
The following equation shows the relationship of the frequency (v), expressed in wavenumbers, to the force constant (f), expressing the strength of the chemical bond and to the reduced
mass which is related to the product of the masses (m) of the vibrating atoms divided by the sum
of their masses.
f
f-21c.//"mi-m2 ,~
~ 1 + m2/
In the near-infrared we predominately see the results of vibrations of light atoms that have strong
molecular bonds. If the chemical bond is weak (low force constant) or the atoms are heavy (the
products of their masses is great) the vibrational frequency is low and its overtone will not be
154
detectable in the near-infrared. Therefore, we primarily see chemical bonds containing hydrogen
attached to atoms such as nitrogen, oxygen, or carbon thereby limiting the chemical structures
that are observable to fairly simple ones that are common in many organic compounds. There
are also big differences in the magnitudes of these absorption bands. Overtones and combinations
of the OH stretching vibration are fairly strong. OSi vibrations are very weak. This relationship
indicates that analyzing water in the presence of sand should be readily possible, but detecting
sand in a matrix of water is highly unlikely.
These weak overtone and combination bands are more subject to their environment than
the fundamental bands resulting from the same vibration would be. Examination of the equation
that shows frequency as a function of force constant and reduced mass makes this obvious. For
a given vibration of two atoms bound together with a chemical bond, the masses of the atoms
remain constant, but the force constant is subject to the environment of the chemical bond that is
acting as the restoring force. Therefore, the overtones are more subject to the environment than
is the same fundamental. A slight perturbation in the bonding scheme causes small changes in
the fundamental but drastic frequency shifts and amplitude changes in the near infrared. Salinity
is determined by the effect of salt in water on the absorption band for the OH of water. The
competition of these ionic species thus affects the force constant that restores the vibrational
position of the hydrogen and oxygen atoms with respect to each other. Since the weak bands in
the near-infrared are broad and overlapping, spectroscopic resolution is usually not a problem,
but reproduction of the same wavelength is essential for quantitative analysis. The practical result
of this is that numerous regions occur in the near-infrared where wavelength reproducibility is
high, signals are maximized, and noise is minimized. Then the optical responses are sensitive to
the environment of the absorbing molecules and the number of those molecules present. Thus,
quantitative measurement can be made and successfully correlated to chemical data obtained by
other means. From these data and application of a suitable statistical relationship with appropriate
constants, determination of analyte concentrations of unknowns can be made with surprising
Success.
An important spectral feature in the near-infrared is that the overtones occurring at higher
frequencies, which are produced by the same chemical bond, tend to have much weaker absorptions
than the fundamentals. A look at the spectrum of water shows how each successive harmonic has
a drastically reduced absorption, going from higher wavelengths to lower wavelengths. For any
given compound that has fairly intense bands in the vicinity of 2-2.5 micrometers, by the time a
harmonic wavelength at 1 micrometer is reached, the bands are so weak they can barely be
detected. This gradation of intensity from strong bands at long wavelengths to weak ones at short
wavelengths is typical for all of near-infrared spectroscopy.
In the infrared region a slight shift occurs for the vibration of a particular atom if it is in
a different environment. Because the frequency depends on the force constant, it is not surprising
155
that the force constant would change with perturbation of the bonding by the presence of other
competing groups. Thus, in the infrared and the near-infrared, interpretation is aided by knowing
that although the methyl band is supposed to appear at exactly 1470 nanometers, it actually can
appear shifted from that location. The shift observed depends on the nearest neighbors to the
group. Consider that a molecule vibrating in response to electromagnetic radiation is trying to
vibrate freely, but in actual fact, the molecule vibrates in its surroundings. Constraint in its ability
to move freely causes a shift in its absorption spectrum. Water, for example, has a strong band
at 1900 nanometers. If approximately 5% salt is added, the band suddenly occurs at 1980
nanometers. It is well known that salt has no absorption in the near-infrared, yet the salt can be
detected by the effect on absorption of water, whose bands shift from one location to another.
The use of such shifts often explains why near-infrared analysis works to detect and quantitate
compounds that do not have near-infrared absorption bands to start with. It is analytically sufficient
for the analyte to merely affect the absorption bands of another molecule in the mixture that makes
up the sample matrix.
FUNDAMENTALS OF QUANTITATIVE ANALYTICAL NEAR-IR SPECTROSCOPY
Absomtion Effect
In order to obtain quantitative results on the composition of samples, we want to perform
optical measurements directly related to the composition. Primarily, we are concerned with the
phenomenon of absorption of radiation. However, when we are doing measurements on samples,
there are many modes of interaction of radiation with the sample. If the sample is simply
illuminated with one wavelength of radiation after another, one at a time, and the variation of
transmission with wavelength is measured, the spectrum is obtained. The absorption intensity is
a function of three factors. Absorbance (A) is defined as the log of 1/transmittance (T) of the
sample. A form of Beer's law states that A =abc, where a is the intrinsic constant of the material
in question or absorptivity coefficient, b is the thickness of the sample, and ~: is the concentration
of the sample. This intrinsic constant (absorptivity coefficient) of the sample is an expression of
the sensitivity of an absorption measurement at a particular wavelength. The thickness of the
sample should be held constant. Note that in samples where considerable light scattering occurs,
maintaining constant thickness is sometimes a challenge. The concentration is the number to be
determined from this measurement. In diffuse reflectance, b and c quite often will vary together
during any given measurement. Thus, both terms will affect the perceived measurement recorded
as absorbance.
Refractive Index Effect
In pursuit of quantitative measurement of the absorption phenomenon, other optical modes
of interaction with the sample must be considered. One of these is the refractive index of the
sample material itself. The refractive index of a sample will differ at different wavelengths. As
156
the spectrum is being scanned for absorption, another response will be obtained for refractive
index. The refractive index increases very sharply and then decreases sharply in the region of the
absorption band. A change in refractive index in the sample affects the measurement by affecting
surface reflection from the sample. Diamonds, for example, which have a very high refraction,
would have very high surface reflection. On the other hand, glass, which has a much lower
refractive index, will have a much lower reflection. Particularly when a sample is being measured
in a reflectance mode, variation in reflectance will occur that is not due to absorption of light but
to variation in the refractive index of the sample itself.
Absorption has one property that is exceedingly important for near-infrared analysis. The
absorption term, usually expressed logarithmically as Absorbance, must have the property of
additivity. If five different compounds are absorbing at the same wavelength, it must be possible
to calculate the Absorbance total value by summing up the individual Absorbances of each
compound. Under ideal circumstances, when working with solids, it may be possible to achieve
Absorbance additivity, providing that reasonable linearity occurs for each of the absorbing species
within the concentration range in which they appear.
When dealing with solid samples, it is most unfortunate that their refractive index effect
does not share the additivity property that we are trying to achieve with respect to Absorbance.
In a mixture of compounds A and B, the Absorbance of the whole is the Absorbance of A plus
the Absorbance of B, assuming that the additivity is adhered to. The refractive index effect can
produce what is called anomalous dispersion. This produces a change in reflectance. Unfortunately a very complex relationship exists for combining the effects of those components. Another
mode of light interaction with the sample that produces variation is reflection. The amount of
reflectance of an incident ray normal to the surface is a function of the refractive index of the
material. For example, a material with an index of refraction of 1.5 would reflect approximately
4 % of the light, whereas a material with an index of refraction of 2 would reflect as much as 11%.
Thus, differences in indices of refraction of analyte vs. matrix, or of calibration samples vs.
analysis samples, might produce variations that are quite significant. The whole area of reflectance
is of major consideration when scattering is high compared to absorption. For the present, it is
sufficient to keep in mind that one of the results of change in refractive index with wavelength is
that it also produces a difference in reflection. This can become a serious problem for doing
multicomponent analysis and making measurements at a number of different wavelengths.
Scattering Effect
The other mode of interaction of radiation with the sample is scattering. Scattering occurs
when radiation hits the sample, enters into the body of the sample, and comes out in very many
directions. Such scattering back toward the direction in which the radiation came is referred to
as diffuse reflectance. In a slurry in which a considerable amount of particulate material is in a
liquid or in a pulverized solid (a powder sample), reflection occurs at the various surfaces. This
157
reflection occurs in random directions dependent on that fraction of the surface where the ray
struck, and the result is randomization in the direction of the returning beam
Refraction off of the individual particles again changes the direction of the beam. As the
particles get very small, diffraction also could occur where the size of the wavelength of light is
comparable to the size of the particles. In such a case, the amount of scattering that takes place
has a very strong localized maximum. It is readily observed that the scattering intensity for a
sample increases as a function of particle size. This increase is gradual and as the particle size
approaches the wavelength, a strong peak occurs in the vicinity where the wavelength dimension
starts and then it diminishes to almost nothing. In most cases, particle sizes of wavelength size
dimensions are not readily achievable. In the near-infrared, the wavelengths under consideration
are from 1 to 2.5 micrometers. Even in this case, a very fine powder is needed to use diffuse
reflectance. In general, in near-infrared reflectance, the finer the powder, the better the scattering.
There are some exceptions. In near-infrared analysis of milk, if milk containing particles is
homogenized, it becomes more transparent, not less transparent. So there are at least a few
exceptions to the rule that smaller particles are better for doing diffuse transmittance or diffuse
reflectance measurements.
Virtues of Near-IR
Before considering the virtues, let us first point out the apparent disadvantages of the
near-infrared region of the spectrum and see how some of these can be overcome. The first
disadvantage is the complexity of the spectrum, where so many combinations of bands are possible
at any given wavelength. The spectrum is overfilled and everything overlaps, so it is very difficult
to isolate just the band of the material to be measured. This overlapping also makes interpretation
of spectra in qualitative terms very difficult. This is in contrast to the classical mid-infrared,
where considerable interpretation is done with the use of large, library reference collections. At
the present time, the library of spectra in near-infrared is just beginning. The fact that scattering
is higher in the near-infrared than in the mid-infrared may appear to be a disadvantage, but it could
also be used as an advantage. The near-infrared absorption is produced because of a second
overtone, third overtone, etc. of the fundamental vibrations that occur in the mid-infrared. It has
been pointed out that each successive harmonic is much, much weaker in intensity than the previous
harmonic or the fundamental. The same statement is true with regard to combinations of
fundamentals or combinations of overtones with fundamentals or with each other. The whole area
of the near-infrared is characterized by weak absorptions. This has been thought of by many as
being a good reason not to bother making measurements in this region of the spectrum.
We see, however, that this weak absorption characteristic actually can help to simplify, in
a practical way, the spectra observed in this region. If a fundamental vibration in the mid-infrared
occurs at 15 micrometers, its first overtone is at 7.5, its second is at 5, its third is at 3.75, its
fourth is at 3, and only its 5th gets into the near-infrared region. By the time the 5th overtone is
158
reached, there is practically no intensity left. This effect is beneficial. In the chemical sense, it
can be stated that in the near-infrared the observable bands contain only light atoms and have
strong molecular bonds. Typically in near-infrared, one observes strong bonds involving
hydrogen. If the chemical bond is weak, then vibrational frequency is low, and overtones of that
vibration never appear in the near-infrared. If the atom is heavy, the frequency is low, and
overtones will never appear in the near-infrared. If one had to choose groupings to observe that
are pretty much represented among organic compounds, one could not make a better choice than
CH, NH, and OH.
One of the practical virtues of the near-infrared region is that the intensity of bands is not
very temperature dependent. An exception to this rule of thumb is a situation where a chemical
equilibrium that is temperature dependent can produce a change in the absorbing species. Water
of hydration, for example, could have an equilibrium shift with temperature, so the spectrum also
could change from one temperature to another. For this reason, to get good quantitation in certain
cases, some degree of temperature control is required.
Now let us look at other virtues. In working in the near-infrared region of the spectrum,
there is good news instrumentally. For a source to illuminate the sample, a quartz tungsten halogen
lamp could be used.
This is a very simple device and an excellent source having a typical
emissitivity that approaches black body radiation. For a detector, lead sulfide may be used. Lead
sulfide is well characterized, fairly rugged, and involves no particular problems (other detectors
with a rapid response are discussed in the instrumentation section). The use of glass for cells or
lenses is very convenient in comparison to using alkali halide salts. Excellent instrument
performance can be achieved. In fact, the success of modem near-infrared analysis is very much
dependent on making decisions based on tens of milliabsorbance units, thus, a noise requirement
of a few microabsorbance units is standard.
The fact that the spectrum is rich with bands in the near-infrared means that conceivably
a band can be found for almost anything. Furthermore, not only is that band found once, but the
vibration of that band is repeated several times from one part of the near-infrared spectrum to the
other and at different intensities. If a sample goes opaque at one part of the spectrum, the
quantitative measurement simply can be shifted from the second overtone down to the third or
fourth by going to shorter and shorter wavelengths. If the water band at 1900 nm is too strong
because the sample is 60 % water, then the water band at 1400 nm, which is 10 times weaker, can
be used. If necessary, the water band at 1100 nm, which is 100 times weaker, can be used. Thus,
we can choose the intensity by choosing the overtone and going to higher or lower overtones,
depending on whether the absorption is too strong or too weak.
Weak Is Better
The very characteristic that makes near-infrared useful is that the absorption bands are
weak.
This weakness of band intensity is, in fact, a virtue -- a very large virtue.
In the
159
mid-infrared, absorption spectrometry must be done with layers that are very thin, for example,
slices 10 - 30 microns thick. A considerable amount of work has been done telling how to run
samples in the mid-infrared. It is not particularly convenient. In the near-infrared region of the
spectrum, intensities are 10 to 100 times less, so to compensate, samples are 0.1 to 1 mm thick,
which is much more convenient. In certain cases, even greater thicknesses may be used. In fact,
the lowering in intensity is all compensated for by the change in thickness, and at the same time,
the system is instrumentally and experimentally more convenient.
Achieving quantitation additivity is necessary, so that the sum of the Absorbances of two
or more overlapping bands will equal the total Absorbance at that point in the spectrum. Additivity
requires reasonable linearity. It should also be pointed out that variations in the index of refraction
produce anomalous dispersion. The contribution of anomalous dispersion is linearly proportional
to the strength of the absorption band. The absorption can be controlled and made stronger by
putting in more of the sample or making the layer thicker. In the near-infrared, because of the
weak absorption, the issue of anomalous dispersion is sidestepped for all practical purposes. This
is not the case in the mid-infrared. In summary, it can be stated that because the bands are
intrinsically weaker in the near-infrared, no big variations occur in the refractive index.
Therefore, there is not a large superimposed variation of reflection on top of the absorption that
one is trying to measure. This simplifies the mathematics tremendously.
The near-infrared has still other virtues. If a sample is not homogeneous, as the case may
be in a slurry or in a powder, local variation can be dealt with normally by taking a large area.
Experimentally, we hope that by taking a large enough area of the sample the local variations will
average out and the average property of the sample will be measured. It is impossible to do this,
if any part of the sample is opaque. This refers specifically to the case of powders measured in
the diffuse reflectance mode. Because of the intrinsically weak absorptions in the near-infrared,
complete opacity is not achieved; thus, the light can make numerous bounces back and forth off
the various different particles to produce an averaging for the non-homogeneity.
The broad bands often encountered in near-infrared may appear to be rather dull and
uninteresting and certainly not aesthetically pleasing to the eye when looking at the plot of a
spectrum. In the near-infrared, high resolution is usually not necessary. A few sharp bands exist
where resolution for qualitative purposes may be advantageous. This occurs for spectra of simple
chemicals but broad band features are predominant in the spectra of foods and commodities. For
quantitative purposes, it is more important to reproduce the wavelength and have it a little broader
than having ultimate resolution. In making quantitative measurements, there is an advantage to
not requiring such a narrow band. Whether using a grating monochromator or using a filter
instrument, as the band pass becomes narrower and narrower, the total amount of energy coming
through becomes less and less. What is important is the percentage of the light allowed by the
bandpass that may be affected by the concentration of the analyte. Since the analyte bands are
160
broad, it is possible to have a relatively large bandpass and still provide good sensitivity to change
in concentration of the analyte. This spectral feature of the near-infrared is another advantage
that makes the job easier instrumentally and allows us to achieve the high signal-to-noise ratio
required.
In the near-infrared, specifically in the diffuse reflectance mode, quantitative measurements
of absorbance can be obtained where the scattering is very strong and the absorption is very weak.
If, on the other hand, the absorption is strong compared to the scattering, then the reflectance
becomes a nonlinear function of the sample concentration. If in the same mixture, another
substance has an absorption at that same wavelength, then the problem is compounded. Nonlinearity makes additivity impossible. In the near-infrared region, this is not a problem, because
scattering is vastly more intense than in the mid-infrared and absorption is vastly weaker. Thus,
all of the advantages of the near-infrared, particularly when it is applied to diffuse reflectance or
diffuse transmittance, actually stem from its apparent weakness.
Additivity is the central core for the assumption of near-infrared quantitation because we
do not have the luxury of samples containing one chemical analyte producing an isolated band for
which a simple one-term Beers law expression can be used. Modern near-infrared work involves
trying to determine multiple components. Numerous contributors to the absorbance at that band
can exist at any given wavelength. The object is to single out the effect of one of those contributors.
This can be done mathematically, but it can be achieved if, and only if, the absorptions or the
signals from the different components of the mixture are combined with each other correctly, so
the accumulated background may be subtracted quantitatively. In order for additivity to be
achieved that allows quantitation from the mixture of absorbers, a measurement situation is
required in which there is no anomalous dispersion, no resolution error, no nonlinearity, and no
specular error. It is fortuitous that the absence of all of these experimental features, which if
present could wipe out additivity, exclusively occurs in the near-infrared.
DEVELOPING A QUANTITATIVE NEAR-IR METHOD
Near-IR spectroscopic analysis is dependent on development of an empirical linear equation
in which the concentration of the analyte is related to optical measurements, usually expressed as
absorbance or, in the case of reflectance measurements, log 1/reflectance. To gain experience in
this empirical approach, the analyst must have a set of samples for which known values have been
obtained by another method. From this set of knowns, a learning process takes place. From this
learning set, with sufficient experimentation and statistical treatment of optical data, a final
calibration results. This is a multiterm linear expression with appropriate coefficients that makes
appropriate analytical use of optical data. Table III shows some of the options by which calibration
samples and the final optical wavelength for the analytical equation are obtained.
161
Table m . Sample Selection and Wavelength Selection
Sample Selection Considerations
Range of the Analyte
Distribution throughout the Range of the Analyte, Range of the Secondary Analyte, Range in Variation in
the Background Matrix
Sample Selection from Optical Data without a Priori Analytical Knowledge, Selection of Samples for
Verification
Wavelength Search Teclmiques
Visualization
Log 1/R Plotted vs. Wavelength(Spectrum). Compare spectra known to represent extremes of the
samples for consideration and look for differences between these spectra. Key on those wavelengths
where differences are apparent.
First Difference Plot
Second Difference Plot
Regression
Visualization via Simple Linear Regression (one optical term at a time)
Log 1/R Correlation Plot
First Difference Correlation Plot
Second Difference Correlation Plot
Multiple Linear Regression
Step-Up (Where to Stop?)
Reverse Stepwise (Where to Stop?)
Combinations (How to Choose from Equally Good Combinations?)
Selecting a Data Base
Choosing samples for a learning set in correlation spectroscopy is the important first step
toward developing a near infrared (chemometric) analytical method.
The final objective is to
produce an empirical analytical equation including terms for selecting wavelengths that will be
"robust" upon future application to all anticipated quantitative tasks for the analyte of interest in
the matrix of practical consideration.
Consider the case where you have a manufacturing
production run of a particular polymer, and you need to know the average molecular weight, the
amount of cross linkage, the thickness of the stock, and the presence of voids in the stock. In
addition to the polymeric product, there may be some low molecular weight materials, spent
polymerization agent, antioxidant, or additives present, as well as inert fillers (particulates of
glass, graphite, or titanium dioxide). Faced with all of the above variables, you must decide if a
preliminary division of sample types can be made on the basis of different formulation specifications. The scattering characteristics of particles imposes a variable that must be dealt with in any
attempt to quantitate the desired information (molecular weight, thickness, etc.). You may have
to first try to limit a particular learning set to those samples with only one of the finer formulations.
162
This, in turn, could be limited further to either high or low quantities of additives. (During the
regression, the presumed subsets, high or low, are flagged by the use of an extra term in the
multiple linear regression. That term indicates to which of two subsets each sample belongs.
Regression results will show if the presumed subset selection is real and indicate any bias.)
After considering the variables involved, you need to assemble a collection of samples in
which 1) the range of the primary analyte is broad, 2) the population distribution throughout the
range is fairly even, 3) the secondary analyte, or at least the secondary component, also has a
reasonably wide range, 4) the chemical matrix variables are represented in an appropriate
distribution, and 5) the physical matrix variables are represented in range and in an appropriate
distribution.
In order to get a good calibration, the range must not be too narrow. Good manufacturing
produces uniform samples and a narrow range. Also, a narrow range produces a poor calibration.
You may need to deliberately broaden the range by scheduling a special limited production run.
The alternative is to accumulate all of the odd lots from production over an extended period of
time. If absolutely necessary, spiked samples might be used to expand the range. Refer to Table
III regarding sample selection and wavelength selection.
A good deal of laboratory analyses by conventional means usually are needed to select a
calibration subset from a group of samples accumulated from routine production. Having accurate
reference data on the learning set is essential. The precision (usually decided by blind duplicate
determination) of the reference data serves as the performance target for correlation-based
(chemometric) near infrared analysis methods.
This precision (standard error of difference) between duplicates is calculated by Standard
Error = ,/E(Dif)2/(2n). Selection of a subset of samples suitable in range and distribution for one
of the constituents (analytes) does not guarantee an equally well distributed broad range for the
other analytes of interest, constituents 2, 3, etc. Typically, you choose the subset with respect to
the primary analyte and incorporate a reasonable matrix variability in hopes that the second and
third analyte distributions are adequate. The original subset used for constituent 1 should at least
be inspected to see if additional samples should be added to extend the range with respect to
constituent 2, or if samples should be selectively deleted to balance distribution with respect to
that analyte before regression to obtain its calibration. Sample selection for most of the calibrations
of the past few years has been done in this way. A common mistake among novices in
correlation-based methods, such as near infrared analysis, is to use a brute force approach and
assume that they have a collection of 100 samples available, all of these samples without regard
to the analyte distribution should be included in the learning set. This presents a problem, because
if a Gaussian distribution occurs along the analyte range in frequency of occurrence, then results
in future analyses will be excellent only if the middle portion of the range is used. The levering
effect of a few samples at either end of the range is counteracted by the huge mass of samples at
163
the center. The hazard is that, when analyzing
TableIV. Relative Weights of 19 Samples
samples that are either higher or lower than the
mean by a significant amount, the regression line
produced will be weak with respect to samples at
either extreme. Other patterns in disproportion
of the distribution of samples throughout the
19 Samples Selected:
analyte range could cause similar difficulties.
You can readily realize that in order to develop
a method, you have to have laboratory facilities
or a good analytical budget to support the method
development, because many reference analyses
45.00
7.00
39.00
2.00
4.00
34.00
18.00
43.00
40.00
20.00
21.00
13.00
11.00
44.00
6.00
16.00
15.00
12.00
22.00
Relative Weight:
0.635811
-0.012880
-0.012814
0.008733
-0.003472
0.003335
-0.002018
-0.873392E-03
0.927747E-03
0.795698E-03
-0.657216E-03
-0.697789E-03
0.611978E-03
-0.580781E-03
-0.418739E-03
-0.347178E-03
-0.252797E-03
0.194112E-03
0.148186E-03
are required, usually in duplicate, to establish the
expected precision.
Much has been said about representing all anticipated variables of the future within the
learning set, so that they can be handled statistically. The coefficients as well as the wavelengths
chosen can minimize problems associated with the variations. Another way to ensure that all the
variables are incorporated is to select samples based on optical orthogonallity. To accomplish
this, spectroscopic data are obtained on a large set of samples first (prior to obtaining lab reference
data), and a suitable program is used to select the samples in descending order of orthogonality.
Software can be used for a calibration involving row reduction applied to spectroscopic data,
which will order members of the set in decreasing orthogonality with respect to the preceding
members of the set. Table IV shows the ordered relative weight of 19 samples. The result is that
maximum optically observable differences are reflected in the set. Since this is done without any
prior knowledge of analyte composition by a reference method, it is independent of the variability
of just one constituent and actually reflects the total variability of both the matrix and the principal
analyte, secondary analyte, tertiary analyte, etc. The theory here is that nearly all of the optical
variables that your analysis instrument will see is accounted for in the regression model (i.e.
training set). Then other members of the set will have some representation within the optical
array of the training set at least within that population. This approach is particularly advantageous
if a limited number of samples is available. It can save considerable time and money in analytical
expenses, and it provides a guidance mechanism that has a greater probability of producing a more
robust calibration than would be obtained merely by chance.
SELECTING APPROPRIATE TERMS FOR THE ANALYTICAL EQUATION
Wavelength Selection
Visualization of wavelengths due to a particular analyte is the usual way that spectroscopic
v
terms are found. If the absorbing species is known and its wavelength of maximum absorbance
164
is well established, then the task is quite simple. If we then add, in the analytical wavelength
vicinity, absorption of some matrix material and then superimpose some noise on the spectrum,
the visualization technique produces results that are less obvious and, in fact, could be slightly
misleading. The time-tested technique of plotting the spectrum of a sample that is high with respect
to the analyte of interest and comparing it to a spectrum that is low in the same material is used
to look for changes in absorbance related to that difference.
In most absorption spectroscopic methods, we anticipate a change that is readily observable
in a plot. However, in near infrared absorption techniques, we are dealing with changes of just
a few milliabsorbance units rather than tenths of an absorbance unit, as might be expected when
working in other regions of the spectrum. Literally, the changes that one should observe may
actually be obscured within the width of the pen used to plot the spectrum. Therefore, we need
to examine digital data to see actual differences. This, of course, is somewhat tedious. Unlike
classical absorption spectroscopy performed in dilute solutions, when dealing with solids or dealing
with mixtures of liquids all of which have a prominent absorption spectrum, we have a closed,
highly interdependent system, which in no way resembles infinite dilution (one of the requirements
of applying Beer's Law). For example, with a highly interdependent system, if we consider a
three-component mixture in which one component is constant, then when the second component
decreases, the third component must increase. Thus, merely comparing digital readings at the
wavelength of maximum absorbance of the species to be measured (the analyte) is dangerous,
because the baseline is floating when concentrated liquids or solids are scanned. Actually, since
in such a closed system, a direct relationship occurs with an optical term and an inverse relationship
occurs with a different optical term, they are typically both used but with opposite signs on their
corresponding coefficients.
For spectra where the change in absorbance, or log l/R, is small as a function of
wavelength, i.e., where there are broad peaks rather than well defined, sharp peaks, it is sometimes
useful to have a derivative plot (actually a first difference plot) vs. wavelength. Thus, we look
for a slope at any point of the spectrum. Similarly, a second difference can be plotted; in the case
of three points along the spectrum, we correct the center point to a baseline drawn between adjacent
points on either side and plot the difference of the center point from the baseline vs. wavelength.
This commonly has been referred to as a second derivative. It is in actuality very similar to the
classical baseline method.
Visualization techniques may or may not be useful in identifying the primary wavelengths
for developing correlation equations. Since this is an empirical technique, the statistical technique
of regression also is used. In the simplest form, a simple linear regression is performed one
wavelength at a time vs. the concentration of the analyte. Obviously, this requires a whole learning
set of pre-analyzed samples with spectroscopic data for each member of the set. The correlation,
positive or negative with respect to the analyte variation at any wavelength, is plotted. The series
165
of these then produces a correlation spectrum. Simple regression (one term at a time) also may
be performed on first differences or second differences spectra. In the correlation spectrum,
positive peaks indicating a high positive correlation in excess of 0.9 or, alternately, a high negative
correlation in excess of 0.9, would point out good candidates for a first try in a multiple linear
regression equation.
S.tgl2r._U_~ From the simple linear regression, we then could choose the one wavelength
that has the highest correlation with the analyte. We then could add another wavelength to the
first wavelength to get a pair of wavelengths for a multiple linear regression with two variables
and subsequently from the remaining members of the set choose a third wavelength, a fourth, etc.
As additional optical data are added to the expression, we would anticipate that the correlation
coefficient should increase, that the standard error of calibration (SEC) should decrease, and that
other statistical indicators should change also. Refer to Table V where steps in a typical calibration
process are outlined. In the step-up, multiple linear regression, when an improvement in the
statistical indicators does not accompany an additional incorporation of another term into the
expression, then the step-up procedure is stopped. A step-up procedure is useful when many,
many wavelengths are available.
Reverse Stepwise. When a somewhat limited number of wavelengths exist, such as 25 or
30, and if we have the computational capability, we can start with all wavelengths in the expression
and then after the regression is performed, use a statistical term such as the student t-test for each
individual wavelength term, discard the one that has the t-test closest to zero, and perform the
regression all over again with one less datum. This procedure is repeated again and again, and
after each step, we examine the correlation coefficient, which may decrease slightly, and also the
standard error of calibration, which may increase slightly. If available, we also observe the F of
regression, which may change as the elimination procedure goes on. A drastic increase in the
standard error of calibration typically indicates that one too many variables have been removed.
Therefore, we can back up to the previous equation which probably would give us the right number
of wavelengths in the expression.
Coml~ination. A powerful multiple linear regression technique is used to identify the set
of wavelengths by combination rather than by step-up or reverse-stepwise procedure. In the case
of starting with 19 different wavelengths, if we choose a combination of three from a field of 19,
there are 969 different combinations that have to be tried. With modern computational power,
an all-possible-combinations search is realistic. In this type of procedure, a statistical indicator
such as the F of regression can be used to determine which combinations are saved and which are
discarded. One program of this type chooses five sets of three that give the best statistical value.
Combinations of four or five may be considerably more valuable, but the dividends do not always
justify the extra computational time.
166
Table V. Successive Steps in Calibration and Validation
Criteria of Successful RegressionResults
Correlation Coefficient (It should be large)
Standard Error of Estimate (Calibration) - should approach lab error
t-Test for Individual WavelengthTerms (look for large positive or negative)
F of Regression (look for large value)
Comparison of Standard Error to Analyte Range
Equation Selection
Minimum Number of Terms (terms with no information still contribute noise)
Indicator Wavelengths Should Make Sense Spectroscopically
Reference Wavelengths Should Be Included
Consider WavelengthsThat Should Be Included Based on Chemical Information
Testing of the Equation for Performance
Terms for Judging Performance SEP (Standard Error of Performance - This is distinguished from
the Standard Error of Calibration in that all members of the validation set are ordinarily not part of
the original regression database.
Validation Sample Selection
Cross Validation when Samples are Limited - Tests for robusmess of the calibration
Global Methods (applicable to scanning data)
Principal Component Regression (PCR)
Partial Least Squares (PLS)
Neural Networks
-
Establishin~ the Calibration
v
Multiple wavelengths chosen by a multiple linear regression procedure should not
necessarily be accepted without a careful look.
If several sets of different wavelengths give
approximately the same statistical results for a given set of data, we can choose to exclude a set,
for example, incorporating a moisture wavelength, if we want to avoid keying on differences in
moisture. Thus, even though this is an empirical technique, a little spectroscopic sense is certainly
valuable.
"Don't let the computer do your thinking for you." When considering the criteria of
successful regressions, a correlation coefficient in excess of 0.9, preferably approaching 0.999,
certainly is useful. However, the F of regression or standard error of calibration is perhaps of
greater value in judging success. The standard error of calibration (SEC) indicates the standard
error of difference between calculating the analyte value from the empirical equation produced
and the analyte true value from some reference method.
The object is to have the standard error of calibration approach as near as possible to the
standard error of difference of blind duplicates done by the reference method. We also should
compare the standard error of calibration to the analyte range. A standard error of 0.2 percent
167
may be very small over an analyte range of 10 percent, but would be considered very large over
an analyte range of 1 percent. When trying to determine initial feasibility of the technique, we
often artificially expand the range. When doing this, an error representing 5 percent of the range
would indicate a useful method. However upon application of the method, if the range is perhaps
only half as great as that used in the feasibility study, then the relative error within the range is
much larger.
Selecting the analytical equation depends first on having a good learning set from which
the wavelengths are chosen carefully and from which good coefficients will be produced. To be
successful analytically, the equation must be robust, i.e., it must be applied successfully to future
samples. To try to accomplish this, any variables that we anticipate ever seeing in the future, if
possible, should be incorporated into the original database. The use of unnecessary terms should
be avoided. If an equation with an adequate correlation coefficient and a reasonably low standard
error of calibration can be obtained with two or three wavelengths, there is little reason to use 17
or 18. A greater number of terms in the expression undoubtedly will produce a higher correlation
coefficient and a somewhat smaller standard error of calibration, but upon applying this equation
to future samples that were not in the original database, we may expect an equation less robust
than one with a minimum number of terms. The hazard is one of overfitting.
For a spectroscopist, the wavelengths chosen should make sense, whenever possible.
Typically, a wavelength representing an absorption maximum for the analyte would be incorporated into the expression, and it would be accompanied by a positive coefficient and would show
a highly positive t-test for that particular term. In a limited system, when the quantity of one of
the materials in the concentrate decreases, the other one increases. In such a case, it is possible,
also, at the wavelength of maximum absorbance of the secondary material in a mixture, to have
a highly negative correlation, to use a negative coefficient for that term, and to have a negative
t-test.
Furthermore, in the case of granular solids in which considerable light scattering occurs,
or in the case of diffusely transmitting liquids or solids, some nonindicating wavelengths or neutral
wavelengths are needed to allow some normalization to the scatter effect that shifts the baseline.
Any equation chosen should be examined with respect to spectroscopic intelligence and also
examined, in some cases, to deliberately exclude certain wavelengths, which might tend to bring
into play the variability of a component that should be ignored so the calibration will not be
dependent upon it.
When using derivative techniques, the multiple linear regression methods, listed in Table
VI, may be similar to those discussed under step-up or step-down. However, each term in the
expression is actually a composite of two terms in the case of a first difference, or three terms,
in the case of a second difference. If one of the terms incorporates a ratio of first differences,
then actually four optical readings are required to produce the ratio of first differences. Similarly,
168
Table Vl. Typical Computational Algorithms
a ratio of second differences requires six optical readings to be
(a) % = Z + a log l/R1 + b log 1/R2 + c log 1/R3 + ...
incorporated in the one ratio that
appears in the final equation.
Thus, a three-term equation involving three second-derivative
ratios actually would require
(b) % = Z + a(log R2- log R1) + b(log R4- log R3) + ...
(first difference)
(c)% = Z + a
(logR2-1ogR~)
(logR6-10gR5]+..
logR4 log
+b
logR8 logR7/ "
(pair: ratio first differences)
(d) % = Z + a(2 log R2- log R1 - log R3) +
b(2 log R5 - log R4- log R6) + ...
(second difference)
(e) % = Z + a
(2log R2-log R1-log R3 )
log R5 log R4 log R6
/ ' 2 log R8- log R7 - log R9 )
b ~ 2 1 o g R l l logR10 logR12
(trio: ratio of second differences)
g
+
measurement at 18 separate
wavelengths. One important
variable, which is part of the
calibration when the difference
equations are used, is the increment between adjacent wavelengths. This has been referred
to as the "gap." Some elaborate
schemes actually will vary the
gap of difference and optimize
Note that RI, R2, R3, R4 . . . . Rn represent reflectance in order for that. In some cases, a differby wavelength (Ref. 6).
ent increment will be used in the
numerator than in the denominator within the same ratio. On the other hand, certain operators
habitually will use the increment of their choice in all the methods that they develop. Obviously
then, the fine tuning of an analytical expression of this type is somewhat dependent on the choice
of the operator.
Transferring calibrations is certainly of interest when you want to use a calibration
produced by someone else or if you want to use the same calibration at several different locations
within the same company, and the calibrations are maintained by someone at the central laboratory.
Certain designs of instruments allow the possibility of matching the optical hardware so that
calibration transfer can be done without a great deal of mathematical intervention. This obviously
involves reproducing the transmission profile of interference filters used in the same model of the
instrument. In general, the best way to transfer a calibration is to transfer not only the numbers
from one instrument to another, but also the samples. Thus, a set of samples that have been
measured on one instrument can be taken physically and run on a second instrument, and the
differences in the analytical results for the very same samples can be measured. From this,
departure from the regression slope established by the first instrument is referred to as a skew,
and any offset is referred to as a bias. If the optical matching of all of the different filters involved
in the calibration is good, little or no skew will occur. In such a case, the transferred calibration
probably can be accomplished with just a change in the bias setting.
169
In the case of solid samples for which the diffuse reflectance technique is used, an even
finer match can be made by grinding two portions of each sample. In this way, a grinder and an
instrument at one location would involve a specific sample preparation and specific optical readings
on the individual samples. When the other portion of a divided sample is ground at a second
location with its local grinder, optical measurements then are performed on the second instrument.
By pairing each instrument with its grinder, even better calibration transfer and matching can be
accomplished.
Global Techniaues
Spectral data from scanning instruments may be treated as if they were discrete wavelengths
or with global techniques. Among global transforms, the three mentioned here include principal
component analysis (PCA), partial least squares (PLS), and Fourier transform (FT). Global
transform may be used for reduction of data (presumably without loss of information). With this
process, there is no wavelength selection, no spectrochemical consideration, and a threshold setting
or factor analysis is the result. Global calibrations utilize all wavelength information with no
wavelength selection. The sequence of a typical global calibration is shown in the block diagram.
Original
data set
transform.
Fransform
data set
reduction _~educed regression Calibration
"[lata set
constants
.IPrediction
-}values
The two global calibration methods mentioned here include principal component regression
(PCR) partial least square regression (PLSR). Both of these start with a PC transform. The PC
transform establishes the order of explaining variance in the data. The process involves orthogonal
axes and the weights are expressed in eigenvectors. The first principal c o m p o ~ t accounts for
perhaps more than 90% of the variance. Each successive principal component accounts for less
than the previous one.
Independent of PCR, principal component analysis provides useful information. If the
magnitude of the eigenvector of the first PC is constant for all wavelengths, then it is likely that
there is a non-chemical cause of the variance described by that PC. If wavelength specificity is
apparent from plotting the eigenvector of the second PC vs. wavelength, then a chemical source
of variance is likely. The wavelength pattern revealed may be either positive or negative. If
wavelengths of 1445 and 1940 nm have eigenvectors far removed from zero, it is likely that the
chemical source of variance in the sample set is water. PC analysis is useful to find out if the
wavelengths of interest from the chemistry of the analyte or those chosen by another calibration
procedure, have eigenvectors that exceed the eigenvectors of the first, second, or other low
principal component at those particular wavelengths of interest. The sources of variance that are
170
E2
P1
independent of the analyte variance represent sample
o.~
"noise" that must be exceeded by the "signal" from
El*
the absorption relative to the analyte.
The principal components that result from
applying the transform to the data produces new
variables P = CllE1 +C12E12 and P = C21E1
+C22E2. Weight selection produces the relation-
Direction of
first largest
ce
shipsCll + C12 = 1 andC21 + C22 =1. Thevector
representation is shown in Figure 23. The direction
of the first (largest) variance is shown with the two
highest magnitude opposed vector and the orthogonal
lower magnitude opposed vector representing the
direction of the second largest variance.
Regression of principal components pro\ Directionof second duced a calibration and the values predicted from
N~argest variance
near-IR analysis are compared to the lab values as
Fig. 23. Vectors of new variables P1 and P2.
with other methods previously discussed. Similarly,
a validation set is used to determine SEP.
Partial least squares regression assumes measured variables X,Y = f(latent variable) +
residuals. The PLS model attempts to describe the variance in the prediction set of data using
"latent variables". These are analogous to principle components used in PCR techniques. The
PLC latent variables are linear combinations of the prediction. PLS calculates the latent variables
in the order of importance and only the number of latent variables that are useful in the model are
retained.
The PLS algorithm involves iterative estimation of one term in the model by successively
estimating another term. Calibration for analysis by PLS consists of three steps: 1)Determination
of the optimum number of PLS factors to produce a useful model, 2)Use validation set determining
the predicted residual error sum of squares (PRESS) to accomplish 1), and 3)Use the weight or
loading spectrum for "chemical" information. Validation in this case is an integral step and
discards unuseful latent variables in the process of establishing the model and calibration. PLS
is considered to be a compromise between least square regression and principle component
regression. In some instances, one of these global methods has produced a more robust model.
One prominent practitioner of chemometrics for spectroscopic calibration suggests that in practice,
one should set the goal for the needed or desired standard error of performance. Try the least
complicated approach first. If that approach is within the goal, refine the expression, recheck the
validation and go about the next task. More sophisticated calibration procedures may be tried as
the next alternative.
171
Data Pretreatmenl
The most common form of data treatment of classical vibrational spectra is the division of
single beam measurement of the sample by the background (blank) data to produce a transmittance
spectrum or alternately, a log 1/R or absorbance spectrum. Baseline correction may be used as
an aid for visual comparison of overlaid spectra. When the data is noisy, Savitsky-Golay or other
smoothing algorithms may be used. This may be done not only for cosmetic display purposes,
but in fact prior to quantitative determination calculations.
Derivative spectra may be obtained (see 1st difference, 2nd difference, etc. examples in
Table VI) as a form of compensation for baseline shifts particularly for scattering effects and as
a bandsharpening procedure to bring out small changes in slope that occur in the Absorbance or
log 1/R plot. As mentioned under the wavelength selection part of developing a method, the result
of regressing the 1st or 2nd difference of each wavelength to produce a correlation plot vs.
wavelength has been used for that purpose. For quantitative method development this may not
be necessary. Baseline shift among members of a group of samples the same general type is
routinely compensated for without any data pretreatment by allowing the multiple linear regression
procedure to incorporate a single wavelength or a composited wavelength optical term into the
analytical expression resulting from the calibration procedure.
One form of normalization used when one has a collection of spectra or data points from
spectra to be compared is referred to as mean center and unit variance according to the steps:
1)Mean center by calculating the average absorbance and subtracting it from the spectrum
2)Establish a unit variance by dividing all absorbances by the standard deviation
The spectra that result from these operations are plotted above and below zero on the y axis that
intersects any point coincident with the mean absorbance. Because all spectra being compared
are plotted around zero, the bias has been removed. A high contrast in absorbance produces large
standard deviation that results in a large divisor. A correspondingly low contrast spectrum by
comparison is awarded a handicap by this operation so that the different contrasts produced for
spectra with generally high absorbances are normalized to those of spectra collected at low
absorbances. Another normalization process that is more statistical in its approach is the
multiplicative scatter correction that involves the following steps:
1) Mean centering the spectrum
2) Curve fitting the spectrum to the average spectrum
3) Dividing the spectrum by the curve fit value
Normalization data pretreatments are also useful under appropriate circumstances but not
in all cases. In many situations, the optics of the sample are such that there is nothing to be gained
from normalizing the data. In still other cases, a loss of information may result because the plots
all become relative and the absolute optically defined percent Transmittance is lost. Note that the
effect of scattering that would automatically be revealed as a vertical offset effect, a difference in
172
slope, and a difference in contrast between the Absorbance values at absorbing and non-absorbing
wavelengths would be lost. Additionally, the presence of a large particulate population of a size
smaller than the wavelengths being used would also be apparent from the spectra that are not
normalized. In such a case the portion of the spectrum affected would be avoided for quantitative
method development. Knowledge of these effects of scattering allows the analyst to consider the
affect on quantitation. For sorting and identification purposes, the granulation may be an important
item that is sought. Optical sieving is a practical way to measure uniformity or deviation from a
mean expected particle size.
Qualitative (Discriminan0 Analysis from Quantitative Near-IR Data
Qualitative determination is possible with quantitative near infrared data using a technique
described by statisticians as discriminant analysis. Objectively establishing the identity of a
granular solid or liquid is often quite important in a manufacturing setting. Work in discriminant
analysis using near infrared began in the pharmaceutical industry initially with discrete wavelength
data (7). Chain of custody record keeping is important in the pharmaceutical industry. When a
drum of material arrives from a supplier it is important to establish whether the powder present
is what it is reported to be or whether it is some other substance. The same is true for liquids.
Scanning the mid-infrared spectrum was used in the past, however, for ease of operation scanning
the near infrared spectrum often with a fiber optic probe is a more convenient approach. Once
this has been done, test data are compared with the spectroscopic data on record of established
standards for the substances (from individual suppliers and the production facility where it is to
be used). A statistically based, multiterm, pattern recognition procedure may be employed or a
spectral matching process may be used. Some FT-NIR instruments equipped with probes were
brought on the market specifically for the field of sample identification.
A discriminant analysis involves using a multidimensional function consisting of weighted
combinations of discriminators. This function may incorporate the raw absorbance data at selected
wavelengths. These wavelengths would have been chosen statistically in order of their value of
discrimination capability. Similarly multiterm expressions can be constructed for discrimination
where each term is a composite of several optically based factors. Such composited values could
be based on principal components or another composited term such as a canonical variable. The
number of terms used in discriminant analysis is determined by the calibration procedure that also
generates appropriate weighting terms (coefficients).
To illustrate the use of multiterm discriminate functions let us consider one case in which
the raw absorbance data is used. The Mahalanobis distance is expressed as a unit distance where
the unit is one standard deviation (8,9). A training set is used where materials of types A, B, and
C would be divided, and when the sample is run to collect optical data, the identity of A, B, or
C is keyed in with the optical data. When data are collected at many wavelengths for a group of
perhaps 20 specimens of each group, a consensus (cluster) of optical data values will be produced
173
at each wavelength for each type of compound.
Thus, a locus in multidimensional space will be
4
established for the optical response typical of A
4
at each wavelength, and correspondingly different values will occur for at least some of the
wavelengths for compounds of type C and com-
v.-
9 o
9
9
9
9
9
o 9
oe0
pounds of type B. The quantitative difference in
these optical values is the basis of discrimination.
1722 nm
Among numerous near-IR data points a Mahalanobis distance discriminant analysis search procedure identified wavelengths in descending order of their contributions as discriminators. In
O
the example cited between three dimensional
CO
ellipsoids for groups A, B, and C, the Mahalano- r
bis distances were: between A and B = 15,
9
9
9
9
r
9
between A and C = 27, and between B and C =
12. Raw optical data for the three top discriminating wavelengths are plotted in Figure 24
9
Ooo
1722 nm
where the success of sorting three different products (A, B, C) from quantitative data at select
wavelengths is obvious for the example shown.
9
9
When the discriminant function is established,
the program is applied to unknown materials and ~,
9
9
the Mahalanobis distances are calculated (for the
data of the new compound) from the centroid of
the multidimensional data of A and from the
centroid of the multidimensional data for B and
C. In the example cited, three optical terms are
used and it is easy to visualize an ellipsoidal three
dimensional graph.
The mathematical process
could just as well handle may terms in hyperspace. Each added term increases the size of the
cluster. Applications of the discriminant analysis
equation to unknown samples resulting in the
Mahalanobis distances (listed in Table VII) from
groups A, B, and C respectively. A low Mahalanobis distance indicates a potential hit (Note
underlined values). When the Mahalanobis dis-
9
9
9
o o
2348 nm
Fig. 24. Log 1/R values for selected wavelengths.
Table VII. Mahalanobis Distances of Unknows from
Each Group Shown by Closest Distances
Sample #
7
34
18
8
33
17
GroupA
1.05
27.6
15.1
1.54
26.2
15.7
GroupB
14.9
12.9
2.04
15.6
11.5
0.737
GroupC
26.8
2.39
12.5
27.3
2.02
11.3
174
tance from the other components is high, that
f:ANI
/
0
0
I
-CAN2
9
4
la'
r
8 ~176
~
,s
0 0
/./
/"
t.....,
.
;I,
i
from A, B, and C.
Composited optical data may be used
0
---~_
"q~'./4
"e
O-"
9
9
.
identity is excluded. When the Mahalanobis
distance is beyond a certain distance from any of
the group, then it is declared as being excluded
s s
,,0
"
-(.,'AN3
t
t
C'AI~'2
9,,~e e
"o
r.
9
.
also. One such form is that of canonical variables (CV). CAN1 is a value calculated from a
weighted combination of several optical terms
(Absorbances). CAN2 employs the same wavelengths and optical values but different coeffi-
cients used describe a qualitative characteristic
that is orthogonal to the preceding term. A series
of CV's are produced but usually not all are
Fig. 25. Sorting of four groups based on 3 canonical required to perform the desired discriminant
CAN3
9
variables,
-(:ANI
analysis. The graph in Figure 25 shows excellent
sorting of 4 groups in 3 dimensional space where each coordinate represents CAN1, CAN2, and
CAN3. This method of qualitative identification based on quantitative data is very useful and it
is an automatable approach. Not only is this technique used to perform qualitative analysis but
this technique may be used to develop smart near infrared systems as an automatic sorting
mechanism so that only the correct quantitative expression can be applied to a sample submitted
for analysis and an incorrect result is avoided.
Another approach involves spectral matching of scanning data. Typically, when data is
taken at many wavelength points, all points are processed in some type of spectral matching
scheme. One of the original spectral matching schemes, previously used in other regions of the
spectrum, was a comparison of the cosine at each point along the spectroscopic curve of an
unknown with the cosine functions of the stored scan of the known reference spectrum. Another
approach has been to establish a spectral library for particular materials that are involved and to
collect the spectrum from samples presumed to have the same identity as a member of the collection
in the library. The spectral data obtained from the test sample is then filtered mathematically by
applying a mean center correction, normalization, and compensation for baseline tilt. Euclidian
distances are calculated for each point in the spectrum and a statistical expression is used to test
the match of the two spectra or eliminate the standard being compared. Small Euclidian distances
indicate a probable hit. A large Euclidian distance eliminates the compared compound from further
consideration.
175
Strate~ for Succgssful Mgthod Development and Examples
There is no substitute for knowledge of the chemical system at hand and control of the
information going into the calibration and minimization of the variables. There is no doubt that
many a good near-infrared calibration has been performed by a statistician who was given
laboratory and spectroscopic data and told to proceed using a prepared chemometric routine. In
many cases the relationships have been obvious and a straightforward procedure has shown that
relationship so clearly that any deviation from it was recognized immediately as incorrect. In
other cases, however, a good calibration and developing a sound method have been very elusive.
In many of those difficult cases, a relationship cannot be found either because none exist or it is
severely obscure. After nearly three decades of encountering near-infrared applications and
attempted applications, this chemist is nevertheless still surprised to find old problems at last being
solved by a new strategy. Software can do only so much. For the most part, sophisticated
treatments reduce the uncertainty in theoretical threshold represented by the 1st principle
component or other contribution to variance greater than the variance caused by the analyte. As
each veil is removed, the image (correlation) becomes more clear. The applications laboratories
of several industrial, government, and instrument companies operate at the mercy of samples and
laboratory reference data supplied by sources over which they have no control. Often they have
little knowledge of the sample material itself and may not be fully aware of the significance of the
laboratory data with reference to that material and the desired method.
In the field of correlation spectroscopy, which we encounter with chemometric methods,
we are dealing with obvious relationships in which a large difference in strong absorption bands
and subtle relationships occur. In these obvious relationships, uncertainties "noise" does not
prevent revealing the relationship between the absorbance of a strong band and the presence of
the grouping responsible for that strong band, even in the presence of a complex matrix material.
Earlier successes resulted for easy-to-solve problems. These did not require a more comprehensive strategy to obtain successful method development. When dealing with more subtle optical
differences, the subtlety of the chemical differences also may come into play. Under these
circumstances, the chemist developing the method needs to maximize control of the situation and
minimize the introduction of systematic biases and random items, which confuse the issue and
obscure the subtle relationship that he or she is trying to discover. When dealing with these
difficult problems involving subtleties performing an iterative sorting process may be necessary
to eliminate the confusing issues.
Because OH absorptions are strong, any determination of moisture or polyols is expected
to work. The production of any condensation polymer that results in elimination of terminal OH
groups can be revealed readily at any stage of the process. For strong bands such as OH, the
correlation of spectroscopic data with laboratory data is overwhelming, and the presence of bad
numbers among lab data or spectroscopic data will not interfere with discovering the relationship.
176
PROTEIN
FRACTIONS
In the instrument business, it is well known that
d~e
if you want to impress a potential customer with
the reliability of an instrument, polyol data is
used because it almost never fails. Such molecular properties as unsaturation or the ratio of
methyl to methylene groups are somewhat less
obvious than those involving OH but nevertheless
we expect them to work provided that the rela-
g
.J
,:, '
~,'~ ' ' a:, '
- tionship that we seek is not obscured by the
Fig. 26. NIR soybean and wheat spectra showing presence of too many other materials in the
differences in protein bands,
sample matrix. When dealing with mixtures of
'
'
CARBOHYDRATE
FRACTIONS.
~oI
/
9~o, ~
~/
i 421~.
4 + ~ - " ~ . ......
c/q+, . . . . ~
0.6. r / ' ~ ] " ~ " / ~,,.,
- ~ " c~,
"
starc..~
~
// o . , . ~ ,
$~h
A
O.=
1.~8
/
/
|0~IA
I
.
.~,,
.//~
: ~ ~/
4
o
"'
proteins from multiple sources, such as soy pro-
v
'
tein in the presence of wheat protein (Figure 26),
the relationship is more subtle, because the differences may be small. Also, when looking for
\ a small amount of cellulosic polysaccharide in the
presence of starch (Figure 27) the subtleties
~o,I.,o..
require minimizing the variables in the strategy
,.~
being used for calibration and method develop-
'2:4
~r,
Fig. 27. NIR starch and cellulose spectra showing
differences in carbohydrate bands,
ment.
An example of measurement of obvious
and more subtle properties is found in a series of
intermediate product flour milling streams.
These streams have wide variations in chemical content and physical and optical characteristics
for quality control. Determination of non-endosperm is difficult because the signals are low and
the multipliers high. This analysis cannot tolerate large differences in chemical matrix composition
or scattering. It was done successfully only with custom calibrations for each test point stream
for an on-line analyzer in the flour mill. Determination of protein reliably in different flour steams
by near-IR required an iterative sorting process among more than 20 millstreams in the Kansas
State University pilot flour mill. To successfully measure the NH near-infrared bands in the
presence of all the other variations among the flour millstreams required a sorting process and a
categorization of each type of material that would affect the background and require different
calibration coefficients to produce a reliable protein analysis. This was a major task involving
knowledge of the milling process, analysis of each stream for components other than those being
analyzed on a routine basis; and an observation of the light scattering characteristics, which were
dependent on the particle size that occurred in the natural production of these different mill streams.
177
Various spectroscopic and statistical sorting techniques produced five different groupings.
For the purpose of protein analysis it was ultimately possible to merge these individual groups
into two major groups and one minor group. Advanced information concerning the milling process
itself did not provide the proper sorting nor did optical data related to light scattering or particle
size alone serve this purpose. Trial and error grouping and regrouping were needed.
In contrast to the major challenge of protein calibration that required data collected over
multiple sessions of pilot mill operations, the moisture calibration was less complicated. For the
protein determination, it was necessary to reduce the magnitude of the major contributor to
variance. The different moisture calibrations produced from the same five groups were evaluated
separately. After doing so, unlike the protein which required three groupings, all five subgroups
of milling streams could be merged into one overall moisture calibration, which did not suffer by
introducing an unacceptable standard error of performance.
Determining moisture in processed meat is economically useful because once a certain
stage of the processing is reached, by law the moisture level can not be altered. In this case,
success was achieved only by carefully controlling the time of sample weighing for reference data
and optical measurement to within less than one hour of each other. Similarly, simultaneous NIR
and reference data sample handling timing was critical for calibrating moisture in flour dough
because the water extracted in anhydrous methanol prior to GC determination to obtain reference
data was reduced by fermentation in only a few minutes.
When trying to establish a near-IR spectroscopic relationship based on some "functionality" rather than the actual chemical content of the material, we need to ask the question of what
property could be measured to reveal the spectroscopic relationship to functionality characteristics
rather than trying to work with a traditional causal chemical structural feature having a less
meaningful effect. One important specification in the buying and selling of fats and oils is the
solid fat index. The percent of material that is solid is determined at several different preestablished
temperatures. This requires a time consuming process. Regression of the properties at several
temperatures versus near infrared data is impractical. Regression of the percent solid at one
temperature may or may not be of value. Chemical consideration of what causes an oil to solidify
and become a fat is focused on two chemical characteristics.
One is the number of double bonds present in the fatty acids that make up the lipid. This
property is typically described by the iodine value that originally was based on titration of the
double bonds by the disappearance of iodine by an addition reaction. In the modem era, iodine
value often is determined by gas chromatographic analysis of fatty acid methyl esters, and these
data are convened to iodine values. Iodine values determined for the calibration set were used to
establish the appropriate wavelengths and calibration coefficients to obtain this property directly.
Measurement of unsaturation alone did not solve the property described by solid fat index. Solid
fat index is dependent also on the molecular weight, and therefore, high performance liquid
178
chromatographic (HPLC) data were obtained on the calibration set to produce a mean carbon
number. This contrived reference data incorporated both carbon chain length of the fatty acids
and unsaturation. From chromatograms of the fats and oils used for the calibration, the relative
contributions of C12, C14, C16, and C18 fatty acid chains were determined. With reverse phase
HPLC, fatty acids with two double bonds contributed to the population of the chromatographic
peak of a single double-bond compound with two less carbons, e.g., the peak area of C 14:1 would
be enhanced by C16:2 and C18:3. Thus, the mean carbon number determined on fats and oils by
reverse phase HPLC in fact did incorporate two causes of either liquidity or solid formation. A
high population of the C 16 and C 14 chromatographic peaks would most assuredly predict a liquid
at room temperature. This would be true either because there were in fact several C 14 fatty acids,
or there were C16 fatty acids with two double bonds, or C 18 fatty acids with three double bonds.
Either of these three types of fatty acids would contribute to the liquid state of the lipid. The result
was that the quality (functionality) of a food ingredient could be practically characterized from
near-IR data rather than the traditional time-consuming solid fat index that required multiple values
to produce a profile.
SOLVING ANALYTICAL PROBLEMS IN FOOD, BEVERAGE, AND AGRICULTURE:
EXAMPLES AND CONSIDERATIONS FOR SUCCESS
Take command of the situation; know your samples; seek information concerning
morphology; learn of expected variability from all sources that will affect your samples; be
responsible for reference data (preferably perform difficult procedures in your own lab); be aware
of the precision of the reference method; consider the measurements required to produce reference
data and ask yourself if there should be a molecular link between the property represented by the
lab reference data and a spectroscopic observation; be responsible for sample procurement and
handling; know the origin, history, and care or treatment of each sample; be prepared to exclude
samples or data that confuse the discovery process or obscure the relationship being sought. In
summary, filter the data, filter the samples, try to find the relationship in the absence of these
confusing factors. This strategy cannot be put into a computer program. Various statistical tools
will reveal an outlier. In practical terms casting out outliers may be fatal because the calibration
produced will work very well on samples in the middle range (when you do not really need it) but
will fail miserably to analyze those samples that are outside the middle range when the analytical
technique is needed the most.
One food and agriculture laboratory known to the author where the best possible
combination of resources is available is at the Quality Assessment Research Unit of the Russell
Research Center U.S. Department of Agriculture in Athens, Georgia. Table VIII shows a list of
particularly challenging food and agriculture related near-IR spectroscopic calibration triumphs.
Easy problems have been solved previously. Some of the calibration successes listed in the table
179
Table VIII. Project Areas of the Qualitative Assessment Research Unit Involving Near-IR
Reference Method and Comments
Problem
Spectral Re#on
1. Dietary Fiber
NIR 1100-2500 nm AOAC enzymatic assay, key was
and 850-1700 nm
minimizing error in reference method
2. Temperature to which
NIR 1100-2500 nm
thermocouple and controlled storage
3. Premature browning
NIR 400-2500 nm
color and enzyme level
4. Flax Quality
NIR 1100-2500 nm
subjective, scale created on residual
aliphatic CH Raman, py-GC-MS
5. Rice Flavor and Texture
NIR 850-2500 nm
sensory panel flavor and texture
6. Rice composition and quality
850-2500 nm (full and standard reference methods and new
reduced data sets)
ones developed for amylose/amylopectin, NMR, py-GC-MS
7. End use quality for wheat
400-1100 nm
poultry had been chilled
determine composition and milling
and baking quality on whole grain
involves use of NIR, FTIR, Raman
and NMR
8. Grain moisture online
sensing
NIR 400-2500 nm and ability to measure moisture content
9. Interpretation of NIR
spectrum
NIR, Raman, FT-IR, use of 2D correlation spectroscopy to
NMR
let one region help interpret the other
10. Multi-region models
Raman and NIR
10-33 GHz
and sense blended lots
small slices of more than one spectral
region in model
*Franklin E. BartonII (Phys. Org. Chem.), David S. Himmilsbach(Phys. Org. Chem.), DannyE. Akin (Biologist,
Microscopis0, DouglasD. Archibald (Anal. Chem., Chemomatrician), SandraKays (Anal. Chem.)
180
are used for regulation and commerce, some for labeling, and some in breeding programs to
predict quality. Use of near-IR wheat protein screening of early generation breeder samples at the
Kansas State Agricultural Experiment Station by the author enabled the breeders to raise the wheat
protein level by 2.5 % absolute in 10 years. One achievement, cited in Table VIII that particularly
caught the attention of many of us who deal with food analysis is the USDA near-IR dietary fiber
work. This method has been much needed for a long time and many other teams of researchers
who have attempted this, obtained unsatisfactory results and gave up. Persistence, diligence and
strategy has prevailed. What is particularly advantageous about the team approach taken at this
facility is that not only is near-infrared instrumentation available and persons to operate that
equipment, but at this site there is also the capability of obtaining mid-infrared spectra, Raman
spectra, and NMR data on various materials under the supervision of professional classical
spectroscopists and physical organic chemists. Wet methods and improved automated procedures
for obtaining reference data such as that required for dietary fiber are also available at this facility.
The precision and accuracy of reference methods are under constant scrutiny and when necessary,
improvements are made to tighten up procedures and reduce uncertainty ("noise") in the reference
data. The involvement of a biologist, who is a world class microscopist, is available at this location
and a research model microspectrometer with ultraviolet and fluorescence capability is available
for looking at morphological detail. FT-IR and Raman microspectrometers compliment the data
obtained by other means and the spatial resolution achievable makes it possible to look at a whole
section of material from the microscope field and obtain the spectrum of only the subsample of
interest. Last, but not least, are the services and valuable contribution of a highly experienced
chemomatrician. Examination of the list of achievements, most of them from relatively recent
efforts, shows the progress that can be made when a team approach is used.
INSTRUMENTATION FOR NEAR-IR SPECTROSCOPY
Perhaps 80% of the near-IR analyzers in use are interference filter instruments equipped
with a quartz tungsten halogen source and a PbS detector that requires a phase sensitive amplifier
and has a digital readout. Among filter instruments, the method of mechanical filter changing
differs as does the scheme for referencing at each filter wavelength to a blank transmittance or
reflectance standard. The detector optical geometry also varies with the instrument and depends
somewhat on the nature of the sample. For solid sampling where diffuse reflectance is used, the
location of the detector determines the ability to collect radiation from a particular angle and
direction unless optical averaging is done with sample motion or with an integrating sphere. In
some instruments, multiple detectors at different locations are used. Three generations of
workhorse filter instruments have supported wheat protein screening at Kansas State University
from 1976 to the present time.
181
Another type of near-IR instrument that Table IX. Discrete Wavelength Instrument Vendors
uses discrete regions of the spectrum is the dis- Interferencefilters:
Dickey-John
crete source instrument (Table IX). This is used
Infrared Engineering
in the very near infrared region in the range of
Perten Insmunents
850-1050 nm where silicon photodiodes work
Oxford Instruments (Foss Food Technology)
Bran and Luebbe (formerly Teclmicon)
well as detectors. Each discrete source is a
Klett
near-infrared version of a light emitting diode
Zeltrex
(LED). Each LED selected for a particular
wavelength also has a very small interference Discretesource LED's + filter:
filter attached to it in series to trim the optical
Zeltex (formerlyTrebor)
emission from the source. These multiple source
Futrex
Katrina Inc.
discrete wavelength instruments are used primarily in a transmission mode. In this region of the4.00
spectrum where absorption bands are weaker,
thicker samples commonly are used. Just as the 3"75
1.
information found in the 1600-2500 nm region is 3.50
duplicated in the 1100-1600 region, the same
information is duplicated again in the very near- 3.25
nm
infrared region below 1050 nm where silicon3.00 1 soybeans, 2 rye, 3 barley, 4 w h e a t
850 ....... 890
930
970
1010
1050
detectors work well. The discrete sources are
fired in sequence and in a brief time span data at
each appropriate wavelength are collected. The Fig. 28. Very near-IR spectrum of an oilseed and
cereals.
plot shown in Figure 28 illustrates that a measurable difference in the whole seeds of three cereal grains and that of a high protein oilseed can be
observed. These differences are the basis of such transmission instruments. Specialized instru-
ments in this spectral region ratio fat to water or water to fat with reasonable success.
Grating M0n0chrom.ator~
Scanning instruments until recently were mostly grating monochromator instruments of
two basic types. The slow scan stepping motor driven sine bar grating rotation type accumulates
signal and reference data completely at each wavelength before stepping to the next separate
wavelength. Typically a signal reading, reference reading, signal reading, and dark current
reading would be collected in sequence at each wavelength (grating angle). The resulting double
beam (in time) function produces a quotient. The log 1/R or log 1/T is calculated at each point
and the spectrum results from plotting each successive absorbance value.
Alternatively a fast scan vibrating grating system was used that accumulated reference
intensities at all wavelengths before scanning the spectrum of the sample. These systems were
equipped with automatic wavelength calibration check and correction by periodically inserting a
182
wavelength standard into the optical path. Successful grating monochromators produced for use
in the near-infrared are all characterized by having a very large grating with an f= 1.8-2.0.
Holographic gratings blazed specifically for the near-infrared region and having an f number lower
than those usually found on other optical instruments have provided success in the near-IR
instrumentation field. The introduction of grating monochromator instruments into the near-infrared field, exclusively held by filter instruments, allowed collection of adjacent wavelengths and
the use of various plotting or global chemometric quantitative functions. The grating
monochromator instruments, unlike the original filter instruments, provided a classical spectroscopic look at the spectral features that resulted rather than being limited to a purely statistical
approach.
Great changes have been made among the scanning near-infrared instruments, particularly
within the last decade. Random wavelength access is readily available from any instrument that
has scanning capability, but recently it has been possible to increase the duty cycle. Very rapid
response photodiode detectors such as indium, gallium, arsenide (InGaAs) have been introduced
to enhance the speed of data acquisition. Electronic wavelength switching as a means of scanning
the spectrum or for random wavelength access provides opportunities for increased speed in the
scanning or monitoring process. Two areas of electronic wavelength switching discussed here
include the grating polychromator diode array and the acousto-optic tunable filter spectrometer
(TFS). These scanning electronic wavelength switching instruments do not require moving parts,
which has an advantage for industrial use.
Diode Array~
In the diode array instruments, the rays exiting a dispersive device (placed after the sample)
fall simultaneously upon each of the elements of a detector array. Data from each element in the
array are polled with what could be described as electronic wavelength switching. This is in
contrast to the classical grating monochromator, where rotation of the grating is necessary to aim
a particular ray through the monochromator exit slit before traversing the sample and hitting the
detector. Diode arrays have become commonplace in the last two decades in the UV and the
visible region of the spectrum, where the technology of silicon photodiode arrays has been highly
developed. Ever since Hewlett-Packard introduced a diode array UV detector for high performance liquid chromatography and a similar laboratory benchtop UV spectrometer, a drastic change
has come about in UV and subsequently fluorescence instrumentation. In the very near-infrared
region (850-1050 nm) a silicon photodiode array can be used.
The engineering for silicon photodiodes and instrumentation in this region is advanced and
silicon arrays are produced at a relatively low cost. However, at the longer, more commonly used
wavelengths within the near-infrared region, the arrays are considerably more expensive.
Germanium as a choice to extend the wavelength range comes at the expense of having a higher
than desirable noise. InGaAs arrays work quite well at ambient temperature but cut off
183
at a 1700 nm limit. An extended range InGaAs detector that reaches 2400 nm requires
cooling and such a detector array is still expensive and each element is limited to small dimensions.
Two technical advances have made possible modern diode arrays for spectroscopic instruments.
The development of excellent photovoltaic devices and the use of memory chips to store
correction coefficients to compensate for the difference in sensitivities of each different element
in the diode array. Silicon photodiode array instruments in the near-infrared have been used for
specialized devices such as octane analyzers for gasoline. More recently, InGaAs array instruments have been produced for the more commonly used region of the near-infrared spectrum
which makes it possible to use the same wavelengths used previously in other types of near-infrared
instruments.
Several vendors of the diode array type instrument are listed in Table X along with the
vendors of other types of scanning instruments. In particular the diode array instruments of LT
Industries and those of Perten Instruments require comment. In the instruments of both of these
companies, white light going through the sample proceeds through the entrance slit of a
polychromator that has a concave grating with a low f number. Located on the Roland circle with
respect to the grating, is an array of either 256 or 512 InGaAs diodes. The diodes are spaced
appropriately, and in one instance, the array has diodes of 100 ~tm in height by 30 l.tm in width
that are spaced on 50 ~m centers. Typically the optical range covered by an array of this type is
800-1750 nm. At the high end, the sensitivity is somewhat reduced beyond 1700 nm. Such an
array may be operated with thermal electric cooling although ambient temperature operation may
be used. The LT Industries power scan instrument operates with fiber optics and uses multiplexing
in order to perform analysis with remote site access at several different sites with the same basic
optical instrument. The Perten Instruments version is a table top device which features a 5 inch
diameter optical stage to handle large heterogeneous samples. Various specialized sampling
devices have been designed.
The use of diode array instrumentation in the near-IR region of the spectrum for the kinds
of sampling commonly done in that region has two limitations. The cost of the near infrared diode
arrays is an issue in some regions of the spectrum. Beyond 1700 nm, a specially doped InGaAs
array operates well only with thermoelectric cooling to provide a high D*. Addition of this and
controlling electronics contributes to the cost. InGaAs arrays operating at room temperature have
an excellent D rating but only in a limited range from 800 to 1700 nm. Ge arrays are available
and operate throughout the region, but unless a particularly pure version of Ge is used, they tend
to operate with higher noise that decreases their D* and desirability.
A diode array instrument requires that white light is fed into the sample and after exiting
the sample, this radiation enters the entrance slit of the grating polychromator. In some cases
having white light enter the sample is undesirable to have. If a photochemical degradation of the
sample occurs, this would be an undesirable method for spectroscopic analysis. In the case of
184
Table X. Scanning Instrument Types and Developers
Electronic wavelen~h switching" Diode arrav eratin~ t~olvchromator
University of Washington silicon range photodiode (CaUis et al.)
Perkin-Elmer (silicon range octane instrument)
LT Industries InGaAs range (Powerscan)
Perten Instruments InGaAs parallel array
Buehler Insmmaents (futtwe model)
Zeiss Instruments
Electronic Wavelength switching: Acousto-Optic Tunable Filter Random Access Spectrometer
Kansas State University research Model (Wetzel/Eilert)
Brimrose Insmmaents of North America
Fiber Tech
Rosemont Analytical Inc. (industrial PbS system)
Bran and Luebbe (octane analyzer)
Various one-of-a-kind homemade systems
Interferometer: Fourier Transform Spectrometers
Bomem/Hartman & Braum (Canada)
Bruker (Germany)
Nicolet (Wisc.)
Bio-Rad Digilab (Mass.)
ATI (Wisc.)
Midac (Calif.)
Mattsen (Wisc.)
Buehler (Switzerland)
Bran and Luebbe (Germany)
Gratine Monochromator:
USDA (Beltsville) research model (Norris/Massey) Perstorp
Foss NIR Systems
Guided Wave
LT Industries
Analytical Spectral Devices
Bran and Luebbe (formerly Technicon)
v
slurries where scattering can be produced by either nonabsorbing or absorbing particles, the total
amount of radiation exiting the sample with the correct trajectory to enter the polychromator will
be limited. Predispersion eliminates the short wavelength rays that are scattered more, leaving
only the potentially absorbed rays to be collected and intercepted by the detector. Another factor
to be considered is that the elements in a diode array can be placed physically only just so close
together.
Therefore, gaps inevitably are present between the actual elements of the array.
Radiation that falls into the crack is lost completely. Interpolation of signals from adjacent pixels
probably would be representative of the lost radiation but that cannot contribute to the overall
intensity in the region lost.
In addition to the complete instruments just described, there are
modules described as "a spectrometer on a circuit board" that that can be plugged into a PC. For
specific information on the various diode array instruments, the reader is referred to the companies
listed in Table X.
185
Ar
Tunable Filter Spectrometer
Another method of achieving near-IR spectroscopic scanning and random wavelength
access with no moving parts is the acousto-optic tunable filter spectrometer. Acousto-optic tunable
filter spectrometers allow the taking of random wavelength accessed data without the need to
sweep through all data points. However, scanning by acquisition of data at sequential points is
an option. The main feature of this approach is that no moving parts are required, which makes
an industrial monitor based on this technology attractive. It has been the goal of the author of this
chapter and his coworkers to develop near infrared acousto-optic tunable filter process monitors
for on-line application. Based on early work conducted in our laboratory, a patent application
was filed and subsequently granted for a quantitative instrument based on this technology (10).
Quantitative data on corn oil in freon as well as numerous spectra were made public at the 1987
American Chemical Society meeting in Denver (11). Acousto-optic tunable filters (AOTF' s) were
proposed in 1969 by Harris (12). There is some similarity of an AOTF to an acousto-optic
modulator that is common in everyday use such as in a laser printer. In the modulator, ultrasonic
energy fed to an acousto-optic crystal through which the light must pass alternately transmits or
blocks the light. In tunable filter acousto-optic devices, a piezoelectric transducer bonded to an
acousto-optic crystal (usually tellurium dioxide) is used to insert ultrasonic energy into the crystal.
A particular, tuned, optical frequency passes through the filter corresponding to the RF (ultrasonic)
tuning frequency applied (the crystal is designed to be tuned within a particular wavelength range).
The optical frequency response curve is dependent on the design of the solid state device. An
ultrasonic absorber is placed at the opposite side of the crystal. A pulsed mid-IR industrial stack
monitor based on a terinary mixture IR-transmitting crystal was introduced by Westinghouse in
the late 1980's. Pulsing that device was necessary because the thermal conductivity characteristics
of the crystal required allowing time for cooling between pulses. The original 1986 continuous
wave version of the acousto-optic tunable filter spectrometer (TFS) instrument in our laboratory
was built around a custom designed TeO2 crystal using mostly components scavenged from other
near infrared instruments. What we developed over a period of years (13) was a high quality,
very rapid instrument. In optical characteristics, overall performance, and quantitative analysis
it is very competitive with both the commercial grating monochromator instruments and FF-NIR
instruments (14). A particular TeO2 AOTF device and frequency synthesizer based instrument
performed well with a thermoelectrically cooled PbS detector, a phase sensitive detector amplifier,
and appropriate software (15). With this advanced intermediate version, quantitative data were
collected to fully test the instrument, and experiments were performed to show wavelength
reproducibility as well as linearity. See Figure 29 in which the spectrum of 100% toluene is
compared with a 1" 1 mixture of toluene and benzene from which the benzene spectrum has been
subtracted.
186
An effort to take advantage of the speed
of the electronic wavelength switching was done
by incorporating an Epitaxx Inc. (Los Angeles,
CA) phosphorous doped InGaAs detector. This
thermoelectrically cooled photovoltaic detector
had a rapid operating speed and the phosphorous
doped version extended the wavelength range to
2400 nm, considerably beyond the 1700 nm
cutoff of a conventional InGaAs detector. The
research model we produced incorporated Glan
Thompson polarizers that gave a high efficiency
of polarization. They transmit in the region of Fig. 29. Spectral subtraction showing KSU TFS instrument performance. Spectrum 1 is a toluene (neat) ~ interest and provide good rejection when
trum. Spectrum2 is the result of subtracting a benzene
crossed. A blocker was added to give geometric spectrum from the spectrumof a 1"1 mixture of toluene
restriction that did not rely completely on the and benzene.
tuning efficiency and the polarizer efficiency to
eliminate untuned radiation and the tuned ray not
being used. This instrument was tested as a flow
through monitor (Figure 30) by using HPLC
pumps and a gradient programmer to produce
calibrations on binary mixtures (16). Spectral
subtraction was used to prove wavelength reproduction and linearity of response similar to that
shown in Figure 29. Software digitally controlled wavelength reproduction is extremely good.
It is an advantage over moving grating systems
and comparable to FT-NIR wavelength repro- Fig. 30. Scansat 1 second intervals with KSU Acoustoduction. One feature of an acousto-optic TFS is opticTFS instrumentof flowingliquid from 100%hexane
to 100% benzene.
the use of random wavelength access to provide
a relatively high duty cycle compared to grating monochromators and other scanning instruments.
This feature was partially responsible for the excellent quantitative success.
The InGaAs detector equipped instrument was capable of performing 480 analyses using
a two-wavelength expression in three seconds. This amounted to an analysis time of 8.3
milliseconds. An invited talk at the Pittsburgh Conference on Analytical Chemistry and Applied
Spectroscopy on this subject was entitled "Fastest Gun in the West" (17). Because an acousto-optic
TFS is completely software controlled, isophotonic data accumulation may be used to enhance the
signal-to-noise ratio at wavelengths where there is shortage of radiation intensity. A longer
187
accumulation time was programmed to enhance the signal-to-noise where needed. Other wavelengths where there was plenty of signal, either due to the fact that the instrument was transmissive
and sensitive in that region, or that there were no strong absorbers in that region could be sampled
for a much shorter period of time. This software-controlled, interactive, data acquisition enhanced
the quantitative performance in select cases where low signal was a problem.
Commercialization of a quantitative acousto-optic TFS instrument based on the working
experimental version was begun through cooperation of the Kansas effort with the Elmsford, New
York division of Bran + Luebbe Analyzing Technolgies. That company introduced an instrument
under the name InfraAlyzer AOTS~ at the 1990 New York meeting of the Pittsburgh Conference
on Analytical Chemistry and Applied Spectroscopy. After closing the American facility, the
parent office of Bran + Luebbe (Nordestadt Germany) reintroduced essentially the same
instrument at a later date under the name InfraPrime| as a high cost end processing monitoring
device for measuring octane number in the petroleum industry and for selected industrial
monitoring in the European chemical industry. That version (limited to InGaAs range) used the
two tuned rays in a double beam mode for monitoring and referencing.
More recently, Brimrose of America Inc. (Baltimore, MD), a well established OEM
supplier of optical devices started producing AOTF devices as an OEM supplier and subsequently
aggressively built an instrument company based on acousto-optic TFS. They offer an industrial
version for industrial monitoring, a table top instrument, and various specialized instruments.
With the very rapid expansion of this company and its product line, including dedicated instruments
for analyzing pharmaceutical tablets at the rate of 25 per second and a rapid fire single seed sorting
device for high oil, low oil, and reject categories, it is apparent that acousto-optic TFS instruments
are at last challenging the market previously dominated by grating monochromator and interference
filter instruments. A special light weight, low power consumption acousto-optic TFS instrument
the size of a man's wallet custom designed by Brimrose for NASA is scheduled for a space flight
near the turn of the century.
Fourier Transform (Interferom~try)
Fourier Transform Spectroscopy (FTS) has been the method of choice in the mid-infrared
region for decades. Throughput of an interferometer instrument is greater because, unlike a
grating monochromator, it has no entrance or exit slits. In the mid-infrared region there are two
distinct advantages including the multiplex advantage and the throughput advantage. These same
advantages would be expected for FI'-NIR, but the detector sensitivity and source intensity are
relatively high in this region of the spectrum, so the IR throughput is less of an issue. Also the
limitation of the dynamic range of the A/D converter results in leveling off of the signal-to-noise
ratio at a moderate throughput level, negating any increased throughput beyond that level. The
advantages that we expect to get from FT-NIR in comparison to a grating monochromator include
resolution, sensitivity, and precision of both the wavelength and intensity readings. High
188
resolution, band shape precision, and particularly wavelength precision may be advantages when
dealing with narrow band measurements. In dealing with food and agricultural commodities,
narrow band spectra are not too often encountered. For application to petroleum products, such
as octane measuring instruments, narrow band data are more of an issue. This is true because a
slight shift in the wavelength could cause a large change in the quantitation.
In an interferometer, a beam of radiation from the source entering the interferometer
encounters a beam splitter. This optical device reflects approximately half the incident radiation
and transmits the other half. In one pathway, a second mirror is encountered, and the radiation
coming off the second mirror rejoins the rays coming straight through the beamsplitter to proceed
onward to the sample target. Because the second mirror is moving, the pathway to and from the
movable mirror will be variable as a function of time. At different mirror positions, a difference
in the two path lengths produces interference. Data accumulated during the time of oscillation
are subjected to fast Fourier transformation. The optical frequency resulting from the interferometry of a broad range radiation beam is a cosine function of the difference between the two
mirror paths.
In the mid-infrared, the requirement for high spectroscopic resolution was a very good
reason to go to FI'-IR. Formerly, two reasons existed for not introducing the Fourier transform
technique commercially in the near-infrared. First, the cost of FT instruments during the
developmental period of modem near-infrared was high. Secondly, with radiation from all the
wavelengths simultaneously hitting the detector, the ability to discriminate between small intensity
differences was regarded as a limitation. The linear dynamic range of the A/D converter overtakes
the throughput limitation when we strive to increase the signal in a near infrared interferometer
type of instrument useful for application to quantitative analysis. At the present time, the cost of
interferometers has decreased and because of advances in electronics, adding more bits to the A/D
converter is no longer economically prohibitive. Thus the range of the A/D converter presently
can accommodate small differences between large signals.
The author of this chapter had the opportunity to evaluate a commercial FT-NIR
instrument, the Bomen MB-155 in 1991, when the A/D range was yet a presumed detriment. It
performed for quantitative purposes in a comparable fashion to a commercial grating monochromator instrument and to a homemade acousto-optic (TFS) near infrared instrument (14). More
recently an evaluation was made at KSU of a Nicolet Magna bench FT-NIR equipped with a fiber
optic probe (18). Quantitative results with homogeneous solid samples were excellent. For
heterogeneous samples, averaging of replicate probing was required with the geometry and
dimensions of the probe that was used. In the opinion of this author, it is safe to say that FT-NIR
is here to stay in quantitative as well as qualitative applications. Quantitatively it performed well
for five-component liquid test mixtures, for emulsions, and for granular solids. Spectral
subtraction with FT-NIR is useful.
In Table X are listed numerous vendors of FT-NIR
189
instruments. Many of these companies are recognized as long time mid-infrared manufacturers.
Most of these use a Michelson type of interferometer. Some do use the comer cube optical
configuration. An alternative to this is a polarization interferometer designed specifically for near
infrared for qualitative analysis industrial I.D. inspection. During just the past decade alone the
diode array instruments, the acousto optic TFS instruments, and the FT-NIR instruments have
become competitive in the near infrared field. Currently several alternatives are available for
routine and specialized use.
OBTAINING AND SORTING OUT INFORMATION ABOUT NEAR-IR AND IT'S
APPLICATIONS
In the past three decades, reliable written information about near-IR applications has not
necessarily always been available. For regular attendees of national meetings in North America
including the Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, the
Federation of Analytical Chemistry and Spectroscopy Societies, Eastern Analytical Symposium,
and the International Diffuse Reflectance Spectroscopy Conference (Chambersburg), this has not
been a problem. In the process of teaching numerous short courses at various cities under the
sponsorship of the Society for Applied Spectroscopy, Eastern Analytical Symposium, the
American Association of Cereal Chemists, the American Chemical Society, and other groups, we
have accumulated a collection of information on topics such as the theory, spectral features,
fundamentals, strategies for developing a method, the optics of sampling, instrumentation for
filters, instrumentation for scanning instruments, smart systems, and applications. Much of that
material is included in the text of this chapter. Handout materials that have been developed over
the years for use in these short courses have been distributed to participants in the short courses.
Through the Council for Near Infrared Spectroscopy, similar material was distributed to teachers
of courses of instrumental analysis, in chemistry departments in the United States and Canada,
who responded to the offer to furnish them.
In response to an invitation to develop a near-infrared course early in the development of
this field, this author had the good sense and good fortune in recruiting a prominent spectroscopist,
Dr. Tomas Hirschfeld, from outside of the field. We met in advance of this course at his location
at Lawrence Livermore Laboratory in Livermore, California to plan the topics, the order, and the
coverage. Subsequently, we repeated the offering of our course in a condensed version at an
Eastern Analytical Symposium a half year later. Not only was Hirschfeld the headliner for the
courses, but every lecture he gave revealed the results of his recent thoughts about the near-infrared
field that at the time was new to him. When working with an intellectual giant like Dr. Hirschfeld
it is difficult to sort out the ideas or feelings about a subject that are your own and the notions that
have been transplanted to your mind as a result of lectures, discussions, conversations, and
gleanings from extemporaneous informal lectures. He used to show the sketch of a bee and point
190
out that "near-IR analysis like the bee should not
work". The aerodynamics of a bee are not suited
to flying "but it flies anyhow".
.5
Near-infrared
also follows Beer's law as illustrated in Figure
31 by varying levels of benzene in a nonabsorbing
solvent in the near-IR region. If anyone still has
.3
doubts about the comparatively weak absorptions
in the near-IR region, refer to Figure 32 which
.
t2oo
T _ _ ~16oo. . . . . . ." .
~o'oo
.
"2~o
shows a spectrum of benzene from the ultraviolet
region of the spectrum to the far-IR. The near-IR
,,.,e,.~.~ ~,~
and very near-IR regions are enlarged to show
Fig. 31. Benzeneat different concentrationsin a solvent that absorption does indeed occur in these retransparent in the near-IR,
gions.
In the early days most information was advanced by marketing people from the most active
instrument companies. Quantitative near-infrared spectroscopic analysis was simply not a
recognized field in analytical or spectroscopic circles. Everyone was forced to acknowledge the
fact that near-infrared worked for quantitative purposes but understanding why it worked was not
obvious. Some of the explanations developed for the short course were used in the cover feature
article in the A pages of Analytical Chemistry years ago (6). This material has served as a primer
for those unfamiliar with the field. In that article, an attempt was made to make near-infrared
more palatable to practicing analytical chemists and classical spectroscopists. Since that time other
review articles and books have appeared (19-26). In general those monographs, actually written
by two or three authors tend to have greater continuity than edited books that included contributions
of many authors. On the other hand, in these books, each individual subject is treated by a specialist
in that field.
Initially when the production of new methods was done by applications chemists working
for instrument companies and by some of their industrial customers, very little was actually
published. In that frenzied period, we were concerned about the calibration of the day or month
that could produce an instrument sale. If it was in fact published, it was usually out of date by
that time or soon thereafter. At one time the only way to know the latest advances was to have
attended every meeting where oral presentations were made on the subject. Our handout material
for early short courses consisted of mostly photocopies of abstracts of presentations from meetings.
In this way at least the analyte and the matrix material of the application was identified and the
group of people working was also given so that if necessary they could be contacted directly.
Actually in the near-infrared field we have gone from a shortage of published information to a
condition of having more published information than we can sort out or keep up with.
4
Fig. 32. Benzene spectrum from ultraviolet to far-IR regions. Near-IR (right) and very near-IR regions (left) are enlarged to show absorptions.
(Note: enlargements are not to the same scale.)
192
Expecting to use cookbook applications does not necessarily work.
As an analytical
chemist, upon finding a method published in a journal article having to do with a chromatographic
or spectroscopic procedure, I would presume that if we followed the printed procedure to the letter
that we would have a reasonable expectation of success. Such is not the case with published
methods in near-infrared. To use a UV vitamin determination method, some means of separation
or some work-up procedure prescribed would have to be followed carefully. When it was followed,
we could expect that interferences would have been dealt with by a prior separation procedure or
by a chromatographic procedure prior to quantitation with the ultraviolet spectroscopic measurement. The advantage of no required separation, "user friendliness", and speed touted for
near-infrared is also a serious disadvantage for those expecting to read the literature and
immediately apply the same calibration equation coefficients, etc. and expect to get good
quantitative results. Various universal calibrations have been developed. Some of these have
actually worked well provided that their applicability was well defined and adhered to.
The primary value of the reporting of a particular method development in the literature is
that the reader knows that at least someone was successful with that particular analyte in that
particular matrix. When an author or speaker reports only a correlation and no validation, this is
a case of "buyer beware". This word of caution does not mean that none of the information given
is of any value whatsoever. When discrete regions of the spectrum have been chosen by another
worker, their identity is certainly of value. Also, coefficients supplied may be examined for their
sign and relative magnitude and the success in terms of the statistical parameters provides a
benchmark for the next worker.
Numerous reports are found on calibration transfer with the same model of the same
instrument and perhaps within the same corporation or organization method development does not
want to be duplicated and it is worth a considerable effort to assure calibration transfer. Perhaps
no organization has had more experience with this in terms of numbers of instruments than the
Federal Grain Inspection Service facility in Kansas City, MO. Calibration maintenance is an issue
where a good deal of serious near-IR work is done on a routine basis. In fact, large companies
such as Cargil and Raison Purina that have many units in the field analyzing a greater variety of
analytes in various sample matrices have a more complicated task. There is a good deal published
on the chemometrics associated with near-infrared and a number of software packages are available
from third party suppliers. This represents important progress in the field. Now that near-infrared
has become a legitimate part of the analytical and spectroscopic community, a number of articles
AppliedSpectroscopyas well as journals primarily devoted to
analytical chemistry. In addition the Journal ofNear-lnfraredSpectroscopy is devoted exclusively
are published in journals such as
to this subject. A considerable amount of literature is available in joumals associated with various
fields in which near-infrared is applied. Now that more is being written and said about
193
near-infrared there is no shortage of information but careful thought and discrimination is needed
now to sort out the useful from the confusing.
Early commercialization of cheometric based near-infrared instruments for wheat enabled
quantitative near-IR spectroscopy to reach the prominent position it now enjoys in the opinion of
the author. When only three near-IR instrument manufacturers coexisted (mid 1970's), DickeyJohn, Neotec, and Technicon, the following engineers made that commercialization possible:
Dave Funk and Hugh Schoen (Dickey-John), Bob Rosenthal, Don Webster, Ron Moen, and Issac
Landa (Neotec), and Bob Rachlis, Ed Stark, John Judge and Lee Pearlman (Technicon).
REFERENCES:
1. Fuller, M.P. and Griffiths, R.P., "Diffuse Reflectance Measurements by Fourier Transform
Spectroscopy", Anal. Chem., 50, 1906, 1978.
2. Kaye, W., "Near-Infrared Spectroscopy (a review) - II. Instrumentation and Technique",
Spectrochim. Acta, 7, 181-204, 1955.
3. Whetzel, D.B., "Near-Infrared Spectrophotometry", Appl. Spectrosc. Rev., 2(1), 1-67, 1968.
4. Goulden, J.D.S., "Diffuse Reflection Spectra of Dairy Products in the Near Infra-red
Region, J. Dairy Sci., 24, 242-251, 1957.
5. Massey, D.R. and Norris, K.H., "Spectral Reflectance and Transmittance Properties of
Grain in the Visible and Near Infrared", Trans. Amer. Soc. Eng., 8(4), 589-600, 1965.
6. Wetzel, D.L., "Near-Infrared Reflectance Analysis - Sleeper Among Spectroscopic Techniques",Anal. Chem., 55, l165A-1171A, 1983.
7. Rose, J.J., "Compound Identification Using High Speed NIRA", the Pittsburgh Conference
on Analytical Chemistry and Applied Spectroscopy, Atlantic City, NJ, 1983, paper #187.
8. Mark, H.L. and Tunnell, D., "Quantitative Near-Infrared Reflectance Analysis Using
Mahalanobis Distances", Anal. Chem., 57, 1449-1456, 1985.
9. Mark, H., Principles and Practice of Spectroscopic Calibration, John Wiley & Sons, New
York, 1991.
10. Kemeny, G.J. and Wetzel, D.L., "Optical Analysis Method and Apparatus Having
Programmable Rapid Random Wavelength Access", U.S. Patent 4,883,963, 1989.
11. Wetzel, D.L., Kemeny, G.J., and Eilert, A.J., "Using an Acousto-Optic Tunable Filter in
Near-Infrared Spectroscopy", American Chemical Society Meeting, Denver, CO, 1987.
12. Harris, S.E. and Wallace, R.W., "Acousto-Optic Tunable Filter", J. Opt. Soc. Amer., 59,
744, 1969.
13. Eilert, A.J., "Acousto-Optic Tunable Filter Spectroscopic Instrumentation for Quantitative
Near-IR Analysis of Organic Materials", PhD. Dissertation, Kansas State University, 1995.
14. Wetzel, D.L. and Eilert, A.J., "Quantitative FT-NIR of Thermochemical Properties of
Industrial Organic Fluids", Proc. SPIE-Int. Soc. Opt. Eng., 1575, 523-524, 1992.
15. Wetzel, D.L. and Eilert, A.J., "Quantitative Analysis with a High Duty Cycle Solid State
Random Wavelength Access Near-Infrared System", the Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, New Orleans, LA, 1992, paper #1135.
194
16. Wetzel, D.L. and Eilert, A.J., "Fast Liquid Flow Composition Monitoring with an
Acousto-Optic Tunable Filter NIR Spectrometer", the Pittsburgh Conference on Analytical
Chemistry and Applied Spectroscopy, Chicago, IL, 1991, paper #478.
17. Wetzel, D.L., "Fastest Gun in the West!", the Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, New Orleans, LA, 1992, paper #1135.
18. Wetzel, D.L. and Sweat, J.A., "FT-NIR Comparative Performance for Complex Samples",
Mikrochim. Acta [Suppl.] 14, 325-327, 1997.
19. Weyer, L.G., "Near-Infrared Spectroscopy of Organic Substances", Appl. Spectrosc. Rev.,
21(1&2), 1-43, 1985.
20. Martin, K.A., "Recent Advances in Near-Infrared Reflectance Spectroscopy", Appl. Spectrosc. Rev., 27(4), 325-383, 1992.
21. Burns, D.A. and Ciurczak, E.W., Handbook of Near-lnfrared Analysis, Marcel Dekker,
New York, 1992.
22. Williams, P.C. and Norris, K.H., Near-lnfrared Technology in the Agricultural and Food
Industries (2rid ed), The American Association of Cereal Chemists, 1998, in press.
23. Murray, I. and Cowe, I.A., Making Light Work: Advances in Near Infrared Spectroscopy,
VCH, New York, 1992.
24. Marten, G.C., Shenk, J.S., and Barton, F.E.II, Near Infrared Reflectance Spectroscopy
(N1RS): Analysis of Forage Quality, U.S. Department of Agriculture, Agriculture Handbook No. 643, 1989.
25. Davies, A.M.C. and Williams, P., Near Infrared Spectroscopy: The Future Waves, NIR
Publications, West Sussex, 1996.
26. Osborne, B.G., Fearn, T., and Hindle, P.H., Practical NIR Spectroscopy with Applications
in Food and Beverage Analysis, 2nd ed., John Wiley and Sons, New York, 1993.
Contribution no. 98-137-B Kansas Agricultural Experiment Station, Manhattan.
D. Wetzel and G. Charalambous (Editors)
Instrumental Methods in Food and Beverage Analysis
9 1998 Elsevier Science B.V. All rights reserved
195
Analysis of Fatty Acids
J. M. King and D. B. Min
Department of Food Science and Technology, The Ohio State University, 122
Vivian Hall, 2121 Fyffe Road, Columbus, OH, 43210
I. Introduction
There has been a great interest by consumers for nutritious foods. This
interest has caused the food industry to work on decreasing the caloric content
of foods while increasing the nutritiousness of their products. These products
are high fiber breads, cereals and other bakery products and low fat or non-fat
foods where fats and oils are replaced by fat mimetics.
Fat substitutions in the formulation of food products and the health claims
associated with the increased health benefits have caused the FDA to review
once again their labeling policies for foods. One of the areas covered by the
FDA is fat content, with specific rules relating to fatty acid content. These rules
are in the Code of Federal Regulations (CFR) 21, Part 101.9, "Nutrition Labeling
of Food" (1). The information requires that fatty acids be calculated as
triglycerides in three categories: saturated, polyunsaturated and
monounsaturated. Polyunsaturated fatty acids are defined as the cis,cismethylene-interrupted type, saturated fatty acids are defined as the sum of all
fatty acids containing no double bonds and monounsaturated fatty acids are
defined as those of the cis-monounsaturated type.
The analysis of saturated, monounsaturated and polyunsaturated fatty
acids by gas chromatography will be covered in detail and will focus on
improvements made during the last ten years.
196
I1. Lipid Extraction
Not much has changed since the Folch et al (2) and Bligh and Dyer (3)
methods of lipid extraction. Most scientists have used chloroform/methanol in
the ratio of 2:1 to extract lipids when further analysis of fatty acids is required.
Both methods involve the use of an aqueous salt solution to produce a biphasic
system. The non-lipid materials are in the aqueous phase and the lipid
materials are in the chloroform phase. Folch et al. (2) reported that the ratios of
chloroform to methanol to water must be 8:4:3 by volume, respectively. Several
researchers (4, 5, 6) have used various extraction procedures and have found
that the chloroform/methanol procedure worked best for extracting all classes of
lipids. One experiment where methylene chloride was substituted for
chloroform showed no statistical difference between the two solvents for
extracting total fat, fatty acids and sterols from various foods (4).
The solvent chloroform/methanol is widely used for the extraction of
lipids, but there are special cases where the method of using this solvent is
varied or where other solvents are required for complete extraction of lipids (7,
8, 9). Soxhlet extractions have sometimes been used to remove lipids from
seeds (10, 11, 12). Christie (9) reported that water saturated n-butanol was
recommended for the extraction of lipids from cereals or wheat flour due to the
lipids which are bound by the starch. Hammond (7) recommended a three part
extraction for wheat flour involving acid hydrolysis for total lipids and a separate
sequence extraction for non-starch and starch lipids. The non-starch lipids are
extracted with cold water saturated n-butanol and the remaining residue is then
washed with methanol and extracted with hot water saturated n-butanol.
Christie (13) also suggested that an ethanol/diethyl ether mixture may be used
for lipoproteins and isopropanol/hexane solvent for lipid extraction from animal
tissues. Christie recommended the use of a modified Bligh and Dyer (3)
method for large extractions where the filtering step is utilized before the
addition of the aqueous salt solution and a modified Folch et al. (2) method
where the methanol is added first followed by the chloroform for a more
complete lipid extraction. Christie stated that plant tissues must be extracted
initially with isopropanol to deactivate enzymes. These special extraction
procedures are important when complete analysis of all lipid classes is
197
required. The mixture of chloroform/methanol is adequate for fatty acid analysis
mainly from triglycerides and extraction is not required for salad and cooking
oils (8). Sometimes further purification of the lipids is performed, more recently
with supercritical fluid chromatography (14), but most often with thin layer
chromatography (15, 16, 17) and/or silicic acid column chromatography (18, 19,
20).
III. Fatty Acid Derivatization
Many of the experiments in derivatization of fatty acids to esters are
related to making the process of derivatization simplier and more rapid while
retaining accuracy (21, 22, 23, 24, 25). There is also a concern for artifacts
produced during derivatization which show up during gas chromatographic
analysis as unidentified peaks (9, 26). These peaks may be due to the use of
old or contaminated solvents or the inclusion of antioxidants such as BHT which
itself can be derivatized. Christie (13) stated recently that when used at proper
levels, which depends on lipid concentration, there should be no problem with
BHT artifacts since BHT can be lost during the evaporation process and usually
elutes at the solvent front.
First to consider is the need for saponification of the lipids in order to
release the fatty acids. Saponification is typically done using KOH or NaOH in
methanol or ethanol (13). This will help separate the fatty acids from nonsaponifiables. Mild saponification conditions must be used to prevent changes
in double bonds of polyunsaturated fatty acids.
Lipids can be transesterified directly without hydrolysis. The main
derivatization methods are acid catalyzed esterification of free fatty
acids/transesterification of bound fatty acids or base catalyzed
transesterification. The base catalyzed transesterification does not derivatize
free fatty acids so samples and solvents must be anhydrous to prevent
hydrolysis of triglycerides (7, 13). Typical base catalyzed derivatization
reagents are 0.5M sodium methoxide in anhydrous methanol or 0.2M
potassium hydroxide in methanol. The acid catalyzed derivatization requires
5% HCL in methanol, 1% sulfuric acid in methanol or 12% boron trifluoride in
198
methanol as reagents, with the later causing many problems related to artifact
formation and loss of polunsaturated fatty acids (7, 13, 27). Christie (13) stated
that boron trichloride could be used in place of boron trifluoride as a more
stable reagent. Internal standards such as heptadecanoic acid (C17:0) for
C8:0 to C22:0 fatty acids and isocaproic acid (C7:0) for volatile fatty acids (C2:0
to C7:0) are added prior to the derivatization step (7).
Other reagents are available for transesterification to shorten the
derivatization time. One of these is tetramethylammonium hydroxide in
methanol which transesterifies triglyceride fatty acids to methyl esters (22, 25).
This reagent minimizes ihe losses of C14:0 and C15:0 fatty acids and
unsaturated C18 isomers in the analysis of plant oils, margarine, lard and
animal tissues (25). Williams and Macgee (28) used trimethylphenylammonium
hydroxide to form salts with the fatty acids which are then pyrolized during
injection to form methyl esters. Garces and Mancha (21) used a complex
mixture of reagents to digest plant tissues and form methyl esters all in one step.
The reagent contained methanol, heptane, 2,2-dimethylpropane and sulfuric
acid with or without either benzene or toluene. The method was good for
samples containing a large amount of triacylglycerols or water. A nonpolar
solvent such as toluene can help solubilize the nonpolar lipids.
Diazomethane is used for esterification for volatile short chain fatty acids
in dairy products (7, 13). Because this reagent is very hazardous it is not
recommended for transesterifying hydrolyzed lipids. Christie (13) recommends
the Christopherson and Glass (29) method using sodium methoxide in
methanol. Hammond (7) recommends the use of potassium hydroxide in
methanol. Highly volatile fatty acid samples have also been analyzed as butyl
esters (30). Sometimes a combination of esterification methods are used for
samples containing both free fatty acids and bound fatty acids (31). Fatty acids
can be analyzed without derivatization, but most researchers prefer to prepare
fatty acid methyl esters (32).
199
IV. Gas Chromatographic Analysis
The next step is to choose the type of gas chromatographic system for
fatty acid analysis. This includes the selection of the column phase, the
temperature of analysis, the type of injection, the type of detector and whether
supplemental analysis will be necessary.
There are two kinds of columns: packed and capillary (7, 13). There was
another column type used for a short period of time called support coated open
tubular (SCOT) which was essentially a wide bore column of 0.5 to 1.25 mm
internal diameter (i.d.) with the liquid phase coated on a fine powdered support
(9). Packed columns can be made of glass or steel, are 2 to 4 mm i.d. and 1.5 to
2.5 m long. They are packed with a solid support such as deactivated
diatomaceous earth coated with liquid stationary phase. The support must be
inert to prevent interaction with the sample. Capillary columns also termed
WCOT (wall coated open tubular) can be steel or glass and most recently are
made of flexible fused silica. They are 0.25 or 0.32 mm i.d. and can be up to
100 m long. The inner wall is coated with liquid phase. There are also wide
bore WCOT columns which have an i.d. of 0.53 or 0.75 mm (7, 33, 34).
The liquid stationary phases are either nonpolar or polar. The nonpolar
phases are silicone polymers such as OV-1, SE-30 and SP-2100 and also
Apiezon high molecular weight hydrocarbons which are not used as often as
the silicone phases (7, 13). The retention times of unsaturated fatty acids on
nonpolar phases are less than those of their saturated counterparts, whereas
on polar phases the reverse is true. Polar phases are made of polyesters.
Christie (13) describes four types with varying polarities. The highest polar
phases are polar substituted alkylpolysiloxanes such as Silar 10C, SP-2340
and OV-275. The highly polar phases include polyethyleneglycol succinate
(EGS), polydiethyleneglycol succinate (DEGS), EGSS-X (methyl silicone
copolymer of EGS), and CP-Sil 84. Polyethyleneglycol adipate (PEGA),
polybutanediol succinate (BDS) and EGSS-Y (similar to EGSS-X, but with more
methyl silicone) are examples of medium polar phases. The low polar phases
include polyneopentylglycol succinate (NPGS), EGSP-Z (copolymer of EGS
with phenyl silicone), Carbowax 20M and Silar 5CP (50% phenyl and 50%
cyanopropyl silicones). Highly pure polar cyanopropyl polysiloxanes have very
200
high temperature stability up to 250~ (7). Stationary liquid phases are
sometimes bonded or crosslinked to capillary columns especially Carbowax
20M (35, 36, 37). Carbowax 20M is also available in a modified form with 2nitroterephthalic acid called FFAP. Hammond (7) reported that a majority of
fatty acid methyl esters can be analyzed on this column. Some times a mixture
of liquid phases is used to improve the resolution. A combination of SP 2310
and SP 2300 helped to resolve C18:3 from C20 fatty acids (38). The amount of
liquid phase affects separation in packed columns, but not in capillary columns.
Levels of liquid phase used in packed columns typically ranges from 5 to 20%
by weight (28, 39, 40, 41,42, 43, 44). Capillary columns are coated most often
at 0.20 and 0.25 micron thickness levels (37, 45, 46, 47, 48).
The common detectors are flame ionization (FID), electron capture
(ECD), photoionization (PID)and thermal conductivity (TCD) (7,13). The FID
detector is the one most widely used for fatty acid analysis. It should be
operated at a temperature of about 20~ to 50~ above the final column
temperature. The FID detector should give a linear relationship between its
response and the carbon number of the fatty acids.
The injection technique is important to obtain accurate results (45).
Samples may be injected directly onto the column, which is the common
method for packed columns, or through an injection port. Split or splitless
technique may be employed using an injection port (13, 48). Any of the
techniques may be used for capillary columns. If the sample is introduced
directly onto the capillary column, "cold trapping" is used (13, 46). The sample
must be injected at a column temperature near the boiling point of the solvent
used. Splitless injection requires that the sample be injected at a column
temperature below that of the solvent's boiling point. This causes the solvent to
form a film of temporary stationary phase which concentrates the sample (13).
Split injection can be used to prevent column overload, but can give inaccurate
results due to variations in sample volume, injector and detector temperatures
and the way the syringe is handled. "Cold trapping" can minimize some of
these problems (13). Christie (13) recommends a "hot needle" technique to
prevent evaporation of the sample before complete insertion of the syringe
needle into the injector. The sample is drawn all the way into the barrel of the
syringe and the needle is preheated for a few seconds in the injector prior to
201
pressing the plunger. The newest type of injection involves a temperature
controlled injection port (13, 47). This prevents discrimination between fatty
acid methyl esters of low and high volatility.
The gas chromatograph oven temperature can be controlled.
Temperature programs are either run isocratically or at various rates. This
depends on the type of sample analyzed and the information required. Short
chain fatty acids can be separated isothermally, but longer chain fatty acids and
complex mixtures usually require temperature programming to adequately
separate peaks (13, 35, 48, 49). The final oven temperature must not exceed
the temperature range of the column since high temperatures could damage the
liquid phase.
The complete gas chromatographic system must be optimized to obtain
accurate results. This topic was discussed in depth by Craske and Bannon
(50). They described various subjects to optimize the gas chromatographic
conditions and obtain accurate results. They discussed the use of a computor
integrator and theorectical FID relative response factors to correct peak areas.
Theorectical relative response factors have been proven to be accurate for all
even numbered fatty acid methyl esters from C4:0 to C22:0 (50,51). Primary
standards of saturated fatty acid methyl esters and saturated triglycerides
should be employed to optimize the chromatographic system and the overall
method, including derivatization. The easiest way to prevent errors during
injection is to inject directly onto the column with a rapid depression of the
plunger. Craske and Bannon (50) mentioned techniques to optimize a split
injection procedure, which was discussed earlier. The detector can be
optimized by adjusting the flow rates of the gases until a linear relationship is
found between chain length and retention time for a complex mixture of fatty
acid methyl esters. Craske (47) performed a collaborative study to separate
instrumental errors from chemical sample preparation errors. Olsson et al. (52)
developed a multivariate method for optimizing the analysis of fatty acid methyl
esters. Accurate quantifications of fatty acids can be calculated by equivalent
chain length and relative response factors only when the sample preparation
and gas chromatographic conditions are optimized (51, 53, 54, 55). Equivalent
chain length values were shown to vary with temperature gradient in linear
202
temperature programmed gas chromatography, but for each temperature
gradient the equivalent chain length factors can be predicted (54).
V. Analysis of Fatty Acids in Foods
Fatty acids can be analyzed without derivatization either on packed
columns or capillary columns (7, 23, 34, 56, 57). When free fatty acids are
analyzed as is, Hammond (7) recommends that they first be separated by thin
layer chromatography. A typical liquid phase for free fatty acid analysis is FFAP
(free fatty acid phase), which is an acidic form of Carbowax 20M (7, 32). The
acid helps prevent tailing and hydrogen bonding on the column. Acid treated
liquid phase columns have had problems with peak ghosting. This problem
was solved by using formic acid vapor to compete for the adsorptive sites on the
column (58,59). Underivatized fatty acid analysis is used more for analyzing
volatile fatty acids of biological samples than for food samples (32).
Volatile short chain fatty acids of dairy products require careful
derivatization and gas chromatographic analysis (13, 30, 40, 56, 57). These
fatty acids can be esterified with diazomethane and injected directly onto the
column (24) or they can be derivatized to butyl esters (30, 50). Even n-propyl
esters have been recommended (7). Increasing the molecular weight through
derivatization will prevent the loss of volatile fatty acids (9). On column
injection should be used to prevent discrimination among fatty acids (13, 32).
Many different phases are used to analyze dairy fatty acids including StabilwaxDA and Nukol on wide bore columns (56, 57) and capillary columns with CPSil84, an example of which is shown in Figure 1 (13). Short chain fatty acids of
coconut oil have been analyzed with an FFAP packed column after
diazomethane derivatization (31).
Lipids from animal tissues have been analyzed mainly by capillary
cloumns on various phases such as OV-275, Carbowax 20M and SP 2330 (60,
61, 62, 63). A wide bore capillary column was also used with OV-275 on Gas
Chrom R (64). Most of the lipid samples were saponified (61, 62, 63) prior to the
formation of fatty acid methyl esters with boron trifluride in methanol. Sawaya et
al. (60) used potassium hydroxide in methanol for direct transesterification to
203
la:1
mlk~
111 0
i
"
L
,
_3"'--'
,
J_
5
. . . .
9
,
__.
,,
9
2o
i~
18~
~J.__
_
_
i L._
9
25
ume~
Figure I.
Milk fatty acids (methyl esters separated on a fused silica
column coated with CP-SiL 84 T M The oven was held at 30~ for 3
min, then raised at 8~ per rain to 160~ and was held at this point
for a further I0 rain.
(Reprinted with permission from reference 13.)
detect pork in processed meat. The presence of C20:2 was a positive indicator
of pork in a canned meat. Other supporting methods including immunoassy
were required due to the variation of C20:2 in pure pork meat. The major fatty
acids found in meat regardless of the source are palmitic, stearic and oleic
acids (64). Other fatty acids found in meat are shown in Table 1 (64).
Lipids of seed and nut oils have been analyzed using capillary columns
more often than packed columns in recent years and saponification prior to
derivatization was employed (11, 12, 65, 66, 67, 68). Sosulski and Gadan (10)
described a glass column packed with GP 3% SP 2310 and 2% SP 2300 liquid
phases on Chromosorb W AW support for the analysis of chickpea lipids. They
Table 1. Fatty Acid Composition of Lean Meata (Reprinted with permission from reference 64.)
Fatty acid
Myristic (14:O)
Palmitic (16:O)
Stearic (18:O)
Arachidic (20:O)
Total safurafed
Palmitoleic (16:l)
Oleic (18:l)
Eicosamonoenoic (20:l)
Docosamonoenoic (22:l)
Total momnsaturaftxi
Beef
(Wb
2.9
23.3
13.6
0.4
40.24-2.5
4.3
37.8
0.2
0.1
42.4tl-2.7
3.8
Linoleic (18:2M)
0.1
Eicosadienoic ( 2 0 : 2 6 )
1.2
Linolenic (18:3w3)
0.2
Eicosatrienoic ( 2 0 . 3 ~ 9 )
Eicosalrienoic (20.3M)
0.4
1.4
Arachidonic f20:4M61
Docosatelraenoic (22 4 6 )
Eicosapentaenoic (20 5w3)
06
Docosapenlaenoic (22 5w3)
10
Docosahexaenoic (22 6w3)
01
[email protected]
8 8+/-22
2.0
0.22+/-0.06
Sheep
(n=6)
2.3
21 .o
15.5
0.9
39.7+/-1.8
Goat
(n=6)
1.6
19.5
15.1
0.4
36.6+//-2.4
N
bffab
(n=6)
Sawbardeer
(n=7)
Horse
(n=6)
Kangaroo
(k7)
(rts)
0.4
16.6
14.0
0.3
31.34-3.3
0.3
16.1
11.4
0.2
28.0+/4.3
0.7
18.1
9.3
0.2
14.6
12.3
0.6
27.7+/-2.8
11
21 5
10 7
2.1
24.6
3.1
10.7
1.5
11.6
0.2
13.84-3.2
13.3+/-4.0
15.3
0.1
2.4
0.1
0.9
27.2
0.5
3.7
28.14-4.5
2.4
37.9
3.0
37.4
0.1
40.3+/-1.2
40.5+/-4 5
0.2
26.94-4.6
0.7
0.5
0.1
10.3+/-1.4
6.3
0.1
1.4
0.5
0.3
2.4
0.1
0.8
0.9
0.2
13.0+/4.5
14.7
0.1
2.8
0.3
1.4
5.2
0.2
1.9
1.9
0.1
28 .&/-5.6
0.1
1.9
2.4
0.1
31.4+/-5.1
2.0
0.5
42.74-4.1
2.4
0.26+/-0.04
2.8
0.36+/-0.15
3.2
0.91+/-0.27
3.6
1.12+/-0.27
5.1
1.52+14l.3t
5.4
0.1
1.6
0.3
0.2
1.4
8.0
a As g 100 g-l total fatty acids (4-s.d. where appropriate).
b Number of muscle samples analyzed.
c Ratio of linoleic acid metabolies to linolenic acid metabolites.
d Ratio of PUFA to saturated fatty acids.
A - indicates that the component was not detected (limit of detection about 0.02 g 100 g-l total fatty acids).
0.9
6.9
0.8
1.1
18.1
0.2
0.1
19.544.5
19.5
0.3
3.6
0.2
1.3
87
0.4
1.4
1.9
0.6
37.9+/-5.4
4.0
1.37+/-0.32
p19
33 3+/-19
26
31 2
06
34 4+/-26
176
06
06
01
05
43
05
01
05
01
24 9+/-28
18 0
0 75+/-012
0
JA
205
used sodium methoxide in methanol to form fatty acid methyl esters. Various
transesterification methods including basic methanol reagents (67, 68), acidic
methanol reagents (39, 42, 43, 66) or boron trifluoride-methanol (12, 67) for
derivatization have been reported. Senter et al. (11) used boron trichloride in
methanol to form fatty acid methyl esters of walnut lipids. Kallio et al. (12)
compared supercritical fluid extraction and Soxhlet extraction with diethyl ether
to extract oil from turnip rapeseeds. Soxhlet method yielded 15% more oil than
the supercritical fluid extraction method. The predominant fatty acids found in
seed and nut lipids are oleic and linoleic acids. Artz and Saver (69) worked on
improving the analysis of free fatty acids using supercritical fluid extraction and
chromatography. Although the % relative error for the analysis of wheat flour
replicate samples was less than or equal to 10%, the extraction efficiency of the
method was greater than 99%. Sahasrabudhe (19) reported various solvent
systems for the extraction of lipids from oats. The amount of each lipid class
extracted varied with the type of solvent used. Solvents containing diethyl ether
or alcohol extracted the most free fatty acids, n-Hexane alone or with diethyl
ether extracted the highest amount of triglycerides. The choice of solvent
therefore depends on the type of lipids in the samples.
Fruits have been analyzed for their fatty acid composition. Highbush
blueberries and Schinus terebenthifolius berries both contained large amounts
of linoleic, oleic and palmitic acids (41, 70). Blueberries may be a good source
of essential fatty acids because they also contain a large amount of linolenic
acid. The blueberries were extracted with chloroform-methanol (2:1, v/v).
Higher amounts of essential fatty acids may be found if another solvent was
used to extract the lipids. Moneam and Ghoneim (70) reported that the largest
amount of fatty acids from schinus terebenthifolius berries could be extracted
with light petroleum. This solvent worked better than diethyl ether or
chloroform-methanol (2:1, v/v). Mango seeds were analyzed for lipid content to
determine if they can be used as a source of fat rather than being a waste
product of jelly production (20). The oil was extracted with n-hexane in a
Soxhlet apparatus, separated into lipid classes by silicic acid chromatography
and the fatty acids of the triglycerides were derivatized with sodium methoxide.
These fatty acid methyl esters were analyzed on a packed glass column with
6% BDS on Anakrom ABS support. Palmitic, stearic and oleic acids were the
206
main fatty acids. The large amount of stearic acid explains why the mango fat is
solid at room temperature. They reported that mango fat resembled cocoa
butter.
Many lipid samples contain isomers of fatty acids, especially fish oils (15,
16, 17, 35). Although separation of geometric and positional isomers is better
on capillary columns than packed columns, these isomers can not be
completely separated by gas chromatography alone (7, 13, 32). Chemical
methods in conjunction with silver nitrate thin layer chromatography
(argentation-TLC) to separate fatty acid methyl esters as idolactones and
methoxybromomercuric adducts have been used (71,72). Argentation-TLC
involves the use of a silica gel TLC plates treated with AgNO3. Ratnayake and
Beare-Rogers (72) used cloroform with 0.75% ethanol to develop their plates
and extracted the separated bands with hexane-chloroform (1:1 , v/v). Auxiliary
methods such as Fourier transform infared spectroscopy, nuclear magnetic
resonance spectroscopy, mass spectroscopy and high performance liquid
chromatography have been used to identify the isomers (32, 73, 74, 75).
Calvey et al. (76) analyzed hydrogenated soybean oil by supercritical fluid
chromatography with Fourier Transform infared spectroscopy. This method
allowed the use of one column to analyze both free fatty acids and triglycerides
which cannot be done with gas chromatography, but it was not possible to
obtain complete resolution of cis and trans isomers of fatty acids. High
performance liquid chromatography is usually utilized to separate triglycerides
prior to fatty acid analysis by other methods (77, 78, 79, 80). Christie (13)
reviewed the use of supportive methods for the analysis of fatty acids.
Hammond (7) explained in depth the use of thin layer chromatography and high
performance liquid chromatography for the analysis of lipids. Typical polar
phases for the analysis of isomers in gas chromatography are Carbowax 20M,
Silar 10C, OV 275, SP 2340, CP-Sil 88 and CPS2 (7, 13, 32).
Ackman et al. (81) developed a method for the analysis of the content of
eicosapentaenoic acid (EPA, 20:5n-3) and docosahexaenoic acid (DHA,
C22:6n-3) in fish oil products. This method was later evaluated by a
collaborative study by Joseph and Ackman (35) and was adopted first action as
an AOCS-AOAC method. This method used capillary column gas
207
chromatography on bonded Carbowax 20M for the analysis of fatty acids and
esters of fish oils, as shown in Figure 2 (35).
9
0
0
c
--
0
.
.
c
--
c
~
ml
0
Q
c
&
o
Q
m
Time (min)
Figure 2.
Temperature-programmed GC separation of menhaden oil fatty acid
methyl esters on flexible fused silica column coated with bonded Carbowax
20M.
(Reprinted with permission from reference 35.)
Quantification of EPA and DHA was determined with C23:0 as an internal
standard. EPA and DHA which are omega-3 fatty acids are essential to the
human diet for developing biological membranes in the retina and central
nervous system (82). Polyenoic fatty acids are used to produce EPA and DHA,
therefore, Ando et al. (17) focused on the separation of C16:3(n-4) and C16:4(n1) fatty acids using a series of steps. After the formation of methyl esters, the
unsaturated fatty acids were concentrated by a urea adduct method and
recovered by ether extraction. Then argentation-TLC was used to separate the
methyl esters by their degree of unsaturation. Further fractionation was made
on reversed phase thin layer chromatography plates to separate the methyl
esters by chain length. Finally, argentation-TLC was used again to separate
C16:3(n-4) from C16:4(n-1). Mass spectroscopy was used to confirm the
identification of gas chromatographic peaks. Nuclear magnetic resonance
(NMR) has also been utilized to analyze omega-3 fatty acids (83). 1H NMR can
208
be used to determine quantities of omega-3 fatty acids, while positions of the
fatty acids on the triacylglycerols can be distinguished with 13C NMR.
The analytical methods for trans-fatty acids have been developed due to
the possible negative health effects associated with these fatty acids (72, 73,
84). Although trans isomers of fatty acids are not found in nature, food
processing such as hydrogenation of oils produces trans isomers (82). The use
of SP 2340 fused capillary columns for the separation of trans fatty acid isomers
in margarine has been reported (72, 84). There were problems of overlap
which could not be solved by gas chromatographic analysis alone even when
75m long capillary columns were used as shown in Figure 3 (84).
0
Q
,m
~
?
m
c/222
L
TIME (MINUTESI
Figure 3.
Chromatographic trace of fatty acid methyl esters of a margarine
sample analyzed on a 75m glass capillary GLC column coated with SP-2340.
(Reprinted with permission from reference 84.)
209
McDonald et al. (73) analyzed trans-diene isomers in hydrogenated soybean
oil using packed and capillary columns with OV 275 on Chromosorb P and CP
Sil88, respectively. They obtained similar results for total content of trans
isomers on both columns. They also found that the capillary column provided
better, but still not complete separations of the isomers and used nuclear
magnetic resonance spectroscopy for identification. Fourier transform infared
spectroscopy is a simple, valuable method for the determination of trans
isomers (13, 75, 85, 86). A sharp peak at the 967 cm "1 area is indicative of
trans double bonds. Ulberth and Haider (85) were able to determine trans
unsaturation in edible fats using this method alone. Silver ion high
performance liquid chromatography was utilized to isolate trans-monoenoic
fatty acids which were then quantified by gas chromatography (87). This
method may be more reliable than others for low levels of trans fatty acids.
The most complex samples for fatty acid analysis are processed foods.
They can contain a mixture of all the above mentioned food sources. Very few
papers were found that discussed the fatty acids analyses of general food
products (4, 88, 89). Chloroform-methanol (2:1, v/v) was used as the lipid
extracting solvent, saponification with 0.5N HCI in methanol and derivatization
with boron trifluoride in methanol. Smith et al. (88) analyzed fried foods using a
packed stainless steel column with 15% OV-275 on Chromosorb W support was
used. Table 2 shows the fatty acid analyses of various products (88). Only
cheese snacks contained fatty acids below C14:0 which were most likely due to
the milk fat in the cheese. Corn and cheese snacks contained small amounts of
C18:2 tt, C18:2ct and C20:1, but donuts, french fries, chicken and fish samples
did not. Only fried fish contained C20:4 and C22:6 fatty acids which shows the
long chain fatty acids associated with fish oils. The largest amounts of
polyunsaturated fatty acids were found in corn snacks where mostly palm and
sunflower oils were used. Monounsaturated fatty acid content was highest in
french fries which are typically cooked in mixtures of beef fat and cottenseed oil.
Cheese snacks had the highest amount of saturated fatty acids, due to the
presence of cheese and milk. Significant amounts of trans unsaturated fatty
acids were found in all of the samples analyzed.
Slover and Lanza (89) analyzed various food products including Crisco
shortening, McDonald's fillet of fish sandwich with cheese, McDonald's Egg
Table 2. Fatty Acid Composition Ranges of Lipids Extracted from Various Deep-Fat Fried Foods
(Reprinted with permission from reference 88.)
@lo0 g of lipid
Fatty Acids
Corn snacks (qa Cheese snacks (6)
c1o:o
c12:o
C14:O
C16:O
C16:lc
C18:O
C18:lf
C18:lc
C18:211
C18:2ct
C18:26
C18:3w3
C20:l
C20:4
C22:6
Total
saturated
Monounsaturated
Polyunsaturated
trans-Unsaturated
0.1 - 0.8
9.0 - 24.8
0.2 - 0.7
2.4 - 4.2
0.0 - 16.2
17.8- 39.0
0.0 - 1.8
1.4 - 5.5
20.7 - 51.6
0.3 - 2.4
0.2 - 0.8
0.1 - 6.1
0.1 - 46.6
0.4 - 18.9
9.4 - 21.9
0.0 - 0.7
2.7 - 4.8
0.6 - 21.2
7.7 - 39.2
0.0 - 2.1
0.0 - 5.6
2.1 - 41.2
0.0 - 1.0
0.0 - 0.6
Donuts (4)
French fries (6) Chidten (5)
Fish (4)
0.2 - 2.5
11.5 - 20.9
0.3 - 3.2
10.5 - 17.5
9.5 - 32.8
29.1 - 29.8
0.7 - 3.4
13.6- 24.8
1.1 - 4.3
13.9 - 18.5
5.2 - 32.6
29.5 - 31.3
0.3 - 0.6
15.9 - 20.0
3.2-6.2
5.4 - 9.0
7.3 - 15.7
32.9 - 35.3
0.6 - 3.2
13.3- 24.7
1.1 -4.1
12.0 - 17.7
5.5 - 28.3
25.0 - 31.3
5.5 - 8.9
0.7 - 0.9
2.6 - 3.4
0.4 - 0.6
9.8 - 19.5
0.3 - 1.2
3.4 - 5.1
0.0 - 1.6
0.1 -1.0
0.1 -1.5
12.8 - 28.1
18.8 - 55.9
26.3 - 55.4
0.8 - 22.0
aNumber of examples analyzed.
14.3 - 84.4
8.7 - 61.1
2.1 - 42.4
1.o - 28.1
23.3- 43.3
44.4 - 65.9
6.6 - 10.3
10.1 - 34.3
29.5 - 49.2
43.1 - 66.5
3.1 -4.0
6.3 - 34.1
23.8 - 30.7
49.1 - 56.9
10.9 - 21.9
7.7 - 16.4
27.0 - 48.2
42.6 - 57.9
4.5 - 6.2
5.8 - 29.9
211
McMuffin and three types of baby food including beef liver, lamb broth and
chicken stew. They used a 100m long glass capillary column coated with
SP2340. Temperature programming gave a better separation of peaks than
isothermal analysis. They were able to separate the various cis and trans
isomers of 18:2w6, but not the geometric and positional isomers of C18:1. Thin
layer chromatography was used to help identify the isomers. Reference and
internal standards were used to quantify fatty acids in the products. Palmitic,
stearic, oleic and linoleic acids were the main fatty acids in all of the analyzed
foods.
Chen et al. (4) analyzed many food products and found that mayonnaise
had the largest amount of palmitic, oleic and linoleic acids out of all the food
products tested. Sausage contained the largest amount of stearic acid and beef
stew had the lowest amounts of all fatty acids.
VI. Standard Methods for Fatty Acid Analysis
The methods recommended by the FDA (1) are those of the Association of
Official Analytical Chemists (AOAC), 13th ed. (1980). No specific method was
listed for the analysis of saturated fatty acids, but for the cis,cis-methylene
interrupted polyunsaturated fatty acids they require the AOAC methods 28.071 to
28.074. This is an enzymatic-spectrometric method. The oil sample is
saponified with KOH, then 1.0M borate buffer and water is added. The mixture is
neutralized with 0.5N HCL and then enzyme solution containing lipoxidase is
added. Blanks with inactivated enzyme are used to zero the spectrophotometer.
The samples are measured at 234 nm as trilinolein, which is used as a standard.
Athnasios et al. (90) developed a gas chromatographic method to
analyze cis,cis-methylene interrupted polyunsaturated fatty acids in fats and
oils. They used a fused silica SP2340 capillary column to determine margarine
fatty acids. They first saponified the separated oil with methanolic NaOH and
then esterified the fatty acids with boron trifluoride. Heptane was used as the
extracting solvent and a split ratio of 1:100 was used for injection into the gas
chromatograph. The areas for 9,12-cis,cis-C 18:2 and 9,12,15-cis,cis,cis-C 18:3
212
methyl esters were combined for total cis-polyunsaturated fatty acid content in
the oil. They reported the fatty acids from C14:0 to C24:0 and obtained similar
results using either a lipoxygenase method or the gas chromatography method.
Although lauric acid (C12:0) is not shown, the method can be slightly modified
to determine the FDA (1) defined saturated and polyunsaturated fatty acids all in
one step.
The current AOAC (91) and AOCS (92) methods are the essentially the
same for general analysis of fatty acids as methyl esters using gas
chromatography. Both involve the use of columns packed with a polar polyester
type liquid phase on acid wash, silanized diatomaceous earth. The methods
employ boron trifluoride for transesterification. Sheppard (8) wrote a manual for
fatty acid analysis of lipids, for the FDA. Sheppard's analytical methods are
very similar to those of the AOAC and AOCS, but information for the extraction
of the lipids from foods was added to complete the methods. The sample is
basically homogenized and extracted with chloroform-methanol (2:1, v/v),
filtered, and washed with water in a separatory funnel. The lower chloroform
phase is collected, dried and the residue redissolved in petroleum ether, then
the fatty acids are converted to methyl esters with boron trifluoride as in the
other methods. The FSIS has recommeded both the AOAC, 15th ed., 1990,
960.39 and the Lipid Manual as references for total fat and fatty acid analysis,
respectively (93).
References
1. Code of Federal Regulations, Title 21 CFR, Part 101, Food Labeling, U.S.
Government Printing Office, Washington D.C. (1993).
2. J. Folch, M. Lees, and G. H. Sloane Stanley, A simple method for the
isolation and purification of total lipides from animal tissues, J. Biol. Chem.
226:497 (1957).
3. E.G. Bligh and W. J. Dyer, A rapid method of total lipid extraction and
purification, Candian J. of Biochem. and Phys. 37:911 (1959).
213
4. I.S. Chen, C. S. J. Shen, and A. J. Sheppard, Comparison of methylene
chloride and chloroform for the extraction of fats from food products,
JAOCS 58:599 (1981).
5. A.J. Sheppard, W. D. Hubbard, and A. R. Prosser, Evaluation of eight
extraction methods and their effects upon total fat and gas liquid
chromatographic fatty acid composition analyses of food products,
JAOCS 51:416 (1974).
6. W.D. Hubbard, A. J. Sheppard, D. R. Newkirk, A. R. Prosser, and T.
Osgood, Comparison of various methods for the extraction of total
lipids, fatty acids, cholesterol and other sterols from food products, JAOCS
54:81 (1977).
7. E.W. Hammond, Chromatography for the analysis of lipids, CRC
Press, Boca Raton, Florida, 1993.
8. A.J. Sheppard, Lipid Manual, U.S. Food and Drug Administration,
Wm. C. Brown Publishers, 1992.
9. W.W. Christie, Lipid Analysis (2nd ed.), Pergamon Press, Elmsford,
New York, 1982.
10. F.W. Sosulski and H. M. Gadan, Variations in lipid composition
among chickpea cultivars, JAOCS 65:369.
11. S.D. Senter, R. J. Horvat, and W. R. Forbus, GLC-MS analysis of
fatty acids from five black walnut cultivars, J. Food Science
47:1753 (1982).
12. H. Kallio, A. Johansson, and P. Oksman, Composition and
development of turnip rapeseed (brassica campestris) oil
214
triacylglycerols at different stages of maturation, J. Agric. Food
Chem. 39:1752 (1991).
13. W.W. Christie, Gas chromatography and lipids, The Oily Press Ltd., Ayr,
Scotland, 1989.
14. H. Kallio, T. Vauhkonen, and R. R. Linko, Thin-layer silver ion
chromatography and supercritical fluid chromatography of baltic
herring (clupea harengus membras) triacylglycerols, J. Agric. Food
Chem. 39:1573 (1991).
15. A.K.M. Aminullah Bhuiyan, W. M. N. Ratnayake, and R. G. Ackman,
Stabilit of lipids and polyunsaturated fatty acids during smoking
of atlantic mackeral (scomber scombrus I.), JAOCS 63:324 (1986).
16. A. Beltran and A. Moral, Changes in fatty acid composition of fresh and
frozen sardine (sardina pilchardus w.) during smoking, Food
Chem. 42:99 (1991 ).
17. Y. Ando, T. Ota, and T. Takagi, Japanese sardine oil as a source of
16:3(n-4) and 16:4(n-1) fatty acids, JAOCS 66:1323 (1989).
18. J. Hemavathy and J. V. Prabhakar, Lipid composition of rice (oryza sativa
I.) bran, JAOCS 64:1016 (1987).
19. M.R. Sahasrabudhe, Lipid composition of oats (avena satvia I.), JAOCS
56:80 (1979).
20. M.A. Ali, M. A. Gafur, M. S. Rahman, and G. M. Ahmed, Variations in fat
content and lipid class composition in ten different mango varieties,
JAOCS 62:520 (1985).
215
21. R. Garces and M. Mancha, One-step lipid extraction and fatty acid methyl
esters preparation from fresh plant tissues, Anal. Biochem. 211:139
(1993).
22. P.C. Fourie and D. S. Basson, Application of a rapid transesterification
method for indentification of individual fatty acids by gas chromatography
on three different nut oils, JAOCS 67:18 (1990).
23. D.P. Schwartz and D. G. Gadjeva, A simple method for the quantitative
isolation and resolution of unesterified fatty acids from fats and oils,
JAOCS 65:378 (1988).
24. J. D. Craske, C D. Bannon, and L. M. Norman, Limitations of ambient
temperature methods for the methanolysis of triacylglycerols in the
analysis of fatty acid methyl esters with high accuracy and reliability,
JAOCS 65:262 (1988).
25. R. Misir, B. Laarveld, and R. Blair, Evaluation of a rapid method for
preparation of fatty acid methyl esters for analysis by gas liquid
chromatography, J. Chrom. 331:141 (1985).
26. C.F. Moffat, A. S. McGill, and R. S. Anderson, The production of artifacts
during preparation of fatty acid methyl esters from fish oils, food products
and pathological samples, J. High Res. Chrom. 14:322 (1991).
27. N.C. Shantha, E. A. Decker, and B. Hennig, Comparison of methylation
methods for the quantitation of conjugated linoleic acid isomers, J. AOAC
Int. 76:644 (1993).
28. M.G. Williams and J. Macgee, Rapid determination of free fatty acids in
vegetable oils by gas liquid chromatography, JAOCS 60:1507 (1983).
216
29. S.W. Christopherson and R. L. Glass, Preparation of milk fat methyl
esters by alcoholysis in an essentially nonalcoholic solution, J. Dairy Sci.
52:1289 (1969).
30. J. K. Ha and R. C. Lindsay, Method for the quantitative analysis of volatile
free and total branched chain fatty acids in cheese and milk fat, J. Dairy
Sci. 73:1988 (1990).
31. N. Gopalakrishnan, C. S. Narayanan, A G. Matthew, and C. Arumughan,
Lipid composition of coconut cake oil, JAOCS 64:539 (1987).
32. H. Traitler, Recent advances in capillary gas chromatography applied to
lipid analysis, Prog. Lipid Res. 26:257 (1987).
33. D.L. Auld, M. K. Heikkinen, D. A. Erickson, J. L. Sernyk, and J. E. Romero,
Rapeseed mutants with reduced levels of polyunsaturated fatty acids and
increased levels of oleic acid, Crop Sci. 32:657 (1992).
34. W.E. May and D. J. Hume, An automated gas liquid chromatographic
method of measuring free fatty acids in canola, JAOCS 70:229 (1993).
35. J. D. Joseph and R. G. Ackman, Capillary column gas chromatographic
method for analysis of encapsulated fish oil ethyl esters: collaborative
study, J. AOAC Int. 75:488 (1992).
36. R.G. Ackman, Capillary GLC has finally 'arrived', JAOCS 66:293 (1989).
37. I. Norman, H. Traitler, and J. Aeschlimann, A fatty acid methyl ester
(FAMES) analyzer, J. High Res. Chrom. 12:308 (1989).
38. J.K. Daun, P. B. Mazur, and C. J. Marek, Use of gas liquid
chromatography for monitoring the fatty acid composition of canadian
rapeseed, JAOCS 60:1751 (1983).
217
39. M.L. Dahmer, P. D. Fleming, G. B. Collins, and D. F. Hildebrand, A rapid
screening technique for determining the lipid composition of soybean
seeds, JAOCS 66:543 (1989).
40. J.L. Iverson and A. J. Sheppard, Detection of adulteration in cow, goat,
and sheep cheeses utilizing gas liquid chromatographic fatty acid data, J.
Dairy Sci. 72:1707 (1989).
41. L.L. Wang, A. C. Peng, and A. Proctor, Varietal differences in lipid content
and fatty acid composition of highbush blueberries, JAOCS 67:499
(1990).
42. J. Hemavathy and J. V. Prabhakar, Lipid composition of calophyllum
inophyllum kernel, JAOCS 67:955 (1990).
43. J. Hemavathy, Lipid composition of murraya koenigii seed, JAOCS 68:651
(1991).
44. D. Firestone and W. Horwitz, IUPAC gas chromatographic method for
determination of fatty acid composition: collaborative study, JAOAC
62:709 (1979).
45. K. Eder, A. M. Reichlmayr-Lair, and M. Kirchgessner, Gas
chromatographic analysis of fatty acid methyl esters: avoiding
discrimination by programmed temperature vaporizing injection, J.
Chrom. 588:265 (1991).
46. F. Ulberth and M. Henninger, On column injection of fatty acid methyl
esters on to polar capillary columns without distortion of early eluting
peaks, J. High Res. Chrom. 15:54 (1992).
47. J. D. Craske, Separation of instrumental and chemical errors in the
analysis of oils by gas chromatography- a collaborative evaluation,
JAOCS 70:325 (1993).
218
48. J.L. Iverson and A. J. Sheppard, Programmed temperature gas
chromatographic analysis of esters of fatty acids, 13:505 (1975).
49. B. Freedman, W. F. Kwolek, and E. H. Pryde, Quantitation in the analysis
of transesterified soybean oil by capillary gas chromatography, JAOCS
63:1370 (1986).
50. J.D. Craske and C. D. Bannon, Gas liquid chromatography analysis of
the fatty acid composition of fats and oils: a total system for high accuracy,
JAOCS 64:1413 (1987).
51. C.D. Bannon, J. D. Craske, and A. E. Hilliker, Analysis of fatty acid methyl
esters with high accuracy and reliability. V. Validation of the theoretical
relative response factors of unsaturated esters in the flame ionization
detector, JAOCS 63:105 (1986).
52. U. Olsson, P. Kaufmann, and B. G. Herslof, Multivariate optimization of a
gas liquid chromatographic analysis of fatty acid methyl esters of
blackcurrant seed oil, J. Chrom. 505:385 (1990).
53. J. D. Craske and C. D. Bannon, Letter to the editor, JAOCS 65:11 (1988).
54. J. Krupcik and P. Bohov, Use of equivalent chain lengths for the
characterization of fatty acid methyl esters separated by linear
temperature programmed gas chromatography, J. Chrom. 346:33 (1985).
55. C.A. Wulff, A. H. Duthie, and S. Wulff, Functional analysis of gas
chromatographic data for C-4 through C-18:2 acids, JAOCS 1:34 (1980).
56. I.W. Choi and I. J. Jeon, Patterns of fatty acids released from milk fat by
residual lipase during storage of ultra-high temperature processed milk, J.
Dairy Sci. 76:78 (1993).
219
57. L. Ceccon, Quantitative determination of free volatile fatty acids from dairy
products on a Nukol capillary column, J. Chrom. 519:369 (1990).
58. A.H. Duthie, S. Wulff, and R. L. Clark, Formic acid trap for GC analysis of
fatty acids and their salts, J. Chrom. Sci. 21:185 (1983).
59. Research Note: Device for saturating gas chromatographic carrier gas
with formic acid for free fatty acid and barbituate analyses, J. Chrom. Sci.
18:273 (1980).
60. W.N. Sawaya, T. Saeed, M. Mameesh, E. EI-Rayes, A. Husain, S. Ali
and H. A. Rahman, Detection of pork in processed meat: Experimental
comparison of methodology, Food Chem. 37:201 (1990).
61. E.M. Gaydou, M. C. latrides, R. Ramananarivo, J. Artaud, and G. Peiffer,
Pattern recognition analysis of fatty acids. Application to beef fat tissue
classification, J. Agric. Food Chem. 32:651 (1984).
62. M.B. Solomon, G. P. Lynch, and D. S. Lough, Influence of dietary palm oil
supplementation on serum lipid metabolites, carcass characteristics, and
lipid composition of carcuss tissues of growing ram and ewe lambs, J.
Anim. Sci. 70:2746 (1992).
63. M.B. Solomon, G. P. Lynch, E. Paroczay, and S. Norton, Influence of
rapeseed meal, whole rapeseed, and soybean meal on fatty acid
composition and cholesterol content of muscle and adipose tissue from
ram lambs, J. Amin. Sci. 69:4055 (1991).
64. A.J. Sinclair, W. J. Slattery, and K. O'Dea, The analysis of
polyunsaturated fatty acids in meat by capillary gas liquid
chromatography, J. Sci. Food Agric. 33:771 (1982).
220
65. E.M. Gaydou, R. Raonizafinimanana, and J. P. Bianchini, Quantitative
analysis of fatty acids and sterols in malagasy rice bran oils, JAOCS 4:141
(~98o).
66. S.D. Senter and R. J. Horvat, Lipid constituents of black walnut kernels, J.
Food Sci. 44:266 (1979).
67. M. Demirbuker, L. G. Blomberg, N. U. Olsson, M. Bergqvist, B. G. Herslof,
and F. A. Jacobs, Characterization of triacylglycerols in the seeds of
aquilegia vulgaris by chromatographic and mass spectrometric methods,
Lipids 27:436 (1992).
68. L.S. Conte, O. Leoni, S. Palmieri, P. Capella, and G. Lercker, Half-seed
analysis: Rapid chromatographic determination of the main fatty acids of
sunflower seed, Plant Breeding 102:158 (1989).
69. W. E. Artz and R. M. Sauer, An improved micro-extraction cell for
supercritical fluid ectraction and chromatography of fatty acids, JOACS,
69:309 (1992).
70. N. M. A. Moneam and T. Ghoneim, Gas chromatographic analysis of total
fatty acids extracted from schinus terebenthifolious berries, J. Chrom.
361:391 (1986).
71. N. Vo Gaiday, A. B. Imbs, D. V. Kuklev, and N. A. Latyshev, Separation of
natural polyunsaturated fatty acids by means of iodolactonization, JAOCS
68:230 (1991 ).
72. W. M. N. Ratnayake and J. L. Beare-Rogers, Problems of analyzing C18
cis- and trans-fatty acids of margarine on the SP-2340 capillary column, J.
Chrom. Sci. 28:633 (1990).
73. R.E. McDonald, D. J. Armstrong, and G. P. Kreishman, Identification of
trans-diene isomers in hydrogenated soybean oil by gas chromatography,
221
silver nitrate thin layer chromatography, and 13C NMR spectroscopy, J.
Agric. Food Chem. 37:637 (1989).
74. M.M. Mossoba, R. E. McDonald, D. J. Armstrong, and S. W. Page,
Identification of minor C18 triene and conjugated diene isomers in
hydrogenated soybean oil and margarine by GC-MI-FT-IR spectroscopy J.
Chrom. Sci. 29:324 (1991).
75. R. Wood and T. Lee, High performance liquid chromatography of fatty
acids: quantitative analysis of saturated, monoenoic, polyenoic, and
geometrical isomers, J. Chrom. 254:237 (1983).
76. E. M. Calvey, R. E. McDonald, S. W. Page, M. M. Mossaba and L. T.
Taylor, Evaluation of SFC/FT-IR for examination of hydrogenated soybean
oil, J. Agric. Food Chem. 39:542 (1991).
77. Y. Ando and T. Takagi, Micro method for stereospecific analysis of triacylsn-glycerols by chiral phase high performance liquid chromatography,
JAOCS 70:1047 (1993).
78. P. Laakso and H. Kallio, Triacylglycerols of winter butterfat containing
configurational isomers of monoenoic fatty acyl residues. I. Disaturated
monoenoic triacylglycerols, JAOCS 70:1161 (1993).
79. P. Laakso and H. Kallio, Triacylglycerols of winter butterfat containing
configurational isomers of monoenoic fatty acyl residues. I1. Saturated
dimonoenoic triacylglycerols, JAOCS 70:1173 (1993).
80. A. Kemppinen and P. Kalo, Fractionation of the triacylglycerols of lipasemodified butter oil, JAOCS 70:1203 (1993).
81. R.G. Ackman, W. M. N. Ratnayake, and E. J. Macpherson, EPA and DHA
contents of encapsulated fish oil products, JAOCS 66:1162 (1989).
222
82. Diet and health, National Academy Press, Washington, D. C., 1989.
83. M. Aursand, J. R. Rainuzzo and H. Grasdalen, Quantitative high resolution
13C and 1H nuclear magnetic resonance of ~3 fatty acids from white
muscle of atlantic salmon (Salmo salar), JAOCS 70:971 (1993).
84. J. Sampugna, L. A. Pallansch, M. G. Enig, and M. Keeney, Rapid analysis
of trans fatty acids on SP-2340 glass capillary columns, J. Chrom.
249:245 (1982).
85.
F. Ulberth and H. J. Haider, Determination of low level unsaturation in fats
by Fourier transform infared spectroscopy, J. Food Sci. 57:1444 (1992).
86. M. M. Mossoba, R. E. McDonald and A. R. Prosser, Gas
chromatographic/matrix isolation/Fourier transform infared spectroscopic
determination of trans-monounsaturated and saturated fatty acid methyl
esters in partially hydrogenated menhaden oil, J. Agric. Food Chem.
41:1998 (1993).
87. T.G. Toschi, P. Capella, C. Holt and W. W. Christie, A comparison of silver
ion HPLC plus GC with Fourier transform IR spectroscopy for the
determination of trans double bonds in unsaturated fatty acids, J. Sci.
Food Agric. 61:261 (1993).
88. L.M. Smith, A. J. Clifford, R. K. Creveling, and C. L. Hamblin, Lipid content
and fatty acid profiles of various deep-fat fried foods, JAOCS 62:996
(1985).
89. H.T. Slover and E. Lanza, Quantitative analysis of food fatty acids by
capillary gas chromatography, JAOCS, 56:933 (1979).
90. A.K. Athnasios, E. J. Healy, A. F. Gross, and G. J. Templeman,
Determination of cis,cis-methylene interrupted polyunsaturated fatty acids
in fats and oils by capillary gas chromatography, JAOAC 69:65 (1986).
223
91. Association of Official Analytical Chemists, Official Methods of Analysis,
Section 963.22 (1990).
92. American Oil Chemists Society, AOCS Official Method Ce 1-62, Vol. 1
(1990).
93. Federal Register, Rules and Regulations, 58:663 (1993).
This Page Intentionally Left Blank
D. Wetzel and G. Charalambous (Editors)
Instrumental Methods in Food and Beverage Analysis
9 1998 Elsevier Science B.V. All rights reserved
225
Isolation of volatile flavor compounds from peanut butter using purge-and-trap techniques
Terri Drumm Boylston"b and Bryan T. Vinyard'
'Southern Regional Research Center, U.S. Department of Agriculture, Agricultural Research
Service, P.O. Box 19687, 1100 Robert E. Lee Boulevard, New Orleans, LA 70179
~Present Address: Department of Food Science and Human Nutrition, Washington State
University, Pullman, WA 99164-6376
INTRODUCTION
Roasted peanut flavor is composed of a complex blend of heterocyclic and other
volatile compounds formed during roasting. Thermal degradation reactions, including
Maillard reactions between carbohydrates, free amino acids and proteins, and lipid precursors
in the raw peanuts, contribute to the formation of desirable, roasted peanut flavor [ 1-2]. Lipid
oxidation reactions during storage or processing of the raw or roasted peanuts contribute to
the formation of numerous carbonyl compounds and the development of undesirable
cardboardy or painty flavors [3-5]. In recent years, the focus of flavor research has been on
the effects of processing and storage on the relative content of volatile flavor compounds and
the improvement of flavor quality [6].
Several hundred compounds have been isolated and identified in roasted peanut flavor
using solvent [7], distillation [8], vacuum degassing techniques [9-10], headspace analysis
methods with cryogenic [11] or adsorbent [ 12-13] trapping, and direct gas chromatography
methods [14-15]. With the exception of the direct gas chromatography method, these
methods use extremely large sample sizes (350 g to 70 kg) to identify the compounds present
at extremely low concentrations. Although the direct gas chromatography method uses a
small sample size (0.2 to 0.3 g), the use of packed column chromatography for the separation
of the volatile compounds does not provide the resolution and sensitivity needed for a
comprehensive study of the volatile compounds.
Headspace analysis techniques isolate the volatile flavor compounds in equilibrium
with the food. The volatile flavor compounds in the isolate are proportional to the contents of
the volatile flavor compounds in the headspace. These techniques do not result in a total or
exhaustive extraction of all volatile flavor compounds present, but rather result in an isolate
which is representative of the aroma perceived by the sense of smell [ 16-17]. The application
of concentration techniques, such as nitrogen purging, application of vacuum, cryogenic
trapping, and adsorbent trapping, to enhance the recovery of the volatile flavor compounds
during headspace analysis has been discussed extensively in the literature [ 12,16-26].
Numerous adsorbents are available for the collection of volatile compounds in the
headspace of foods. The characteristics of these adsorbents and the advantages and
limitations of their use vary widely [27]. Activated carbon was one of the first adsorbents
used for the isolation of volatile compounds. This adsorbent has a large adsorption capacity;
however, there are problems related to inconsistencies in quality and impurities in the
226
charcoal. Tenax-GC has been used widely for the isolation of volatile compounds from
foods because it shows excellent recovery of adsorbed compounds, good thermal stability, and
low affinity for water, although it is limited by its lower adsorption capacity [20,27].
Carbopack B/Carbosieve S-III have been suggested as replacements for the currently used
adsorbents for environmental research. The combination of these two adsorbents provides an
effective system for trapping and desorbing a wide range ofvolatiles [28].
Flavor research frequently entails a comprehensive approach, which involves sensory
analysis and the determination of chemical composition in addition to volatile flavor analysis.
In some cases, this approach reduces the amount of sample available for the analyses.
Therefore, the objective of this study is to develop a reproducible, quantitative headspace
analysis method for the isolation of volatile flavor compounds from roasted peanut butter.
Three purge temperatures, 50~ 70~ and 90~ three purge times, 1 hr, 2 hr, and 4 hr; and
two adsorbents, Carbopack B/Carbosieve S-III (CP/CS) and Tenax-GC (Tenax) will be
compared. Maximum recovery of the volatile flavor compounds allows the detection of the
compounds present in very low quantities. This method will be applied to future research in
which the effects of postharvest and processing treatments on the relative content of the
volatile flavor compounds will be determined.
MATERIALS AND METHODS
Peanuts (Arachis hypogea L. cv. Florunner) were grown at the Coastal Experimental
Station in Tifton, GA during the summers of 1988 and 1989. Medium and jumbo peanuts
(>7.15 nun, diameter), of good quality were roasted at 163~ at an air flow of 0.01 cu. fk/min
for 11 and 11.5 min, respectively, in a surface combustion roaster (Midland-Ross Corp.,
Toledo, OH), and combined in equal quantities. Peanut butters were prepared as described by
Sanders et al. [29]. The finished peanut butters had an L value (Hunter Lab Colorimeter) of
49.0.
Isolation of Volatile Flavor Compounds
The adsorbents: (1) Tenax-GC (300 mg, 60-80 mesh, Teklab, Inc., Baton Rouge, LA)
or (2) Carbopack B graphitized carbon black (400 rag, 60/80 mesh, Supelco, Inc., Bellefonte,
PA) and Carbosieve S-III carbon molecular sieve (200 mg, 60/80 mesh, Supelco, Inc.) were
packed into borosilicate glass tubes (84 mm long, 9 mm od, 1 mm wall) between two plugs of
silanized glass wool. The traps were rinsed with 10 ml methanol and conditioned in a stream
of nitrogen (15 ml/min) at 250~ for 2 hr.
Peanut butter (50 g) was uniformly and completely coated onto the walls of a 300-ml,
3-neck, round-bottom flask using a long-handled spatula. Tetradecane (internal standard, 100
lag) was added to the surface of the sample prior to purging. The peanut butter flowed to the
bottom of the flask during the isolations, regardless of purge temperature. The center neck of
the flask contained the trap, which was held in place using a Midi-Ace threaded adapter (Ace
Glass, Inc., Vineland, NJ) with a 5 psi vacuum applied to the system through the trap. The
samples were purged with nitrogen (50 ml/min) through a nitrogen inlet tube placed in the
side neck of the flask. The third neck of the flask was stoppered. Purge (water bath)
227
temperatures were set at 50 ~ 70 ~ and 90~ and volatiles were collected for 1, 2, or 4 hr.
Prior to elution of the volatiles from the trap, pentadecane (quantification standard, 20 ~tg)
was placed on the top of the adsorbent. The volatiles were eluted from the trap with 15 ml
hexane, using positive pressure, and concentrated to 100 ~tl under a stream of nitrogen.
Identification and Quantification of Volatile Flavor Compounds
The volatile flavor compounds were separated on a cross-linked, 5% phenylmethyl
silicone fused silica capillary column (HP-5, 50 rn, 0.32 mm od, 0.52 ~t film thickness,
Hewlett-Packard, Avondale, PA) installed in a gas chromatograph equipped with a flame
ionization detector (Model 5890A, Hewlett-Packard). The GC oven temperature was initially
held at 35~ for 15 min, then increased at a rate of 2~
to a final temperature of 250~
and held for 45 min. Injector and detector temperatures were set at 200~ and 250~
respectively. The extracts (2.5 ml) were injected using a split injection, with a column flow
rate of 1.1 ml/min, purge flow rate of 2.0 ml/min, split vent flow rate of 15.0 ml/min, and split
ratio of 14:1. Area counts vs nanogram quantities (20-500 ng) were plotted for the standards,
tetradecane and pentadecane. Peak areas of all volatile compounds were converted to
nanogram amounts by using the standard curves developed for the quantification standard,
pentadecane. Relative quantification of the volatile flavor compounds in the peanut butter
was based on the recovery of pentadecane.
Identification of the compounds was confirmed through the comparison of retention
times of standards and Samples, and mass spectrometry. A gas chromatograph-quadrupole
mass spectrometer (Model 4500, Finnigan-MAT, Cincinnati, OH) interfaced with an Incos
data system was used to confirm the identity of the volatile compounds. GC conditions were
the same as for the chromatographic analysis. The conditions for the mass spectrometer were
set as follows: ionizing voltage, 70 eV; emission current, 0.3 mA; electron multiplier voltage,
1800 kV; ion source temperature, 150~ ionization chamber pressure, 6.0 x 10"~ atm; and
scan range 33 to 250 m/z in 0.95 sec with a 0.05 sec hold. Identification of mass spectra was
based on matches with the library within the Incos data system and mass spectral data
published in the literature [8, 30-36].
Statistical Analysis
The experimental design consisted of a 3-way factorial in a completely randomized
design structure, with adsorbent (Tenax-GC and Carbopack B/Carbosieve S-III), purge
temperature (50 ~ 70 ~ 90~ and purge time (1, 2, 4 hr) as the main factors. Two crop years
(1988, 1989) served as the replications for the experiment; all treatment and replication
combinations were repeated 3 times. For each purge temperature, purge time and crop year
combination, the two adsorbents were paired.
The purge isolation conditions which resulted in the maximum recovery of total
volatile flavor compounds from peanut butter were determined based on the recoveries of the
48 most abundant compounds isolated using frequency tables and multivariate statistical
techniques. For each year and adsorbent combination, the mean recoveries for the 48 volatile
flavor compounds were ranked from highest (rank=-1) to lowest (rank=-9) with respect to the 9
purge temperature and time combinations. The data for each year and adsorbent were pooled
to form one table. For each rank, the number of volatile compounds with that given rank
228
were determined. Based on this table overall rank of recovery for each purge time and
temperature combination were determined (PROC FREQ) [37].
Paired t-tests on adsorbent differences for each volatile compound from both crop
years were performed to compare the recovery of the volatile flavor compounds. The t-tests
were performed on 3 sets of recovery data determined by the rank analysis: (1) the purge
temperature-time combinations for the maximum recovery (Rank=l), (2) the purge
temperature-time combinations for the three highest overall recoveries, and (3) the purge
temperature-time combination for the maximum overall recovery.
Principal component analyses (PROC PRINCOMP) and cluster analyses (PROC
VARCLUS) were performed on the data to identify a necessary and sufficient subset of the 48
variables that, when maximized, would also maximize the recovery of all 48 variables [37].
The multivariate analyses were performed on the subsets of data for each year and adsorbent
combination, with the data from all purge temperatures and purge times included within the
four subsets. The number of clusters allowed to form in the multivariate analysis was
specified as the number of principal components explaining a non-zero proportion of the
overall variation in the data. These analyses objectively identified and selected a set of 'near'
linearly independent variables or a key variable set. From these multivariate statistical
techniques, a subset of volatile flavor compounds was selected that represented the response
of all 48 volatile flavor compounds. Analysis of variance (PROC GLM) was used to
determine the effects of adsorbent and purge temperature and time on the recovery of the
volatile flavor compounds selected to make up the key variable set as representative of the
initial 48 volatile flavor compounds [37].
RESULTS AND DISCUSSION
Representative, reproducible chromatograms were obtained using purge-and-trap
techniques for the isolation of volatile flavor compounds from peanut butter. Pyrazines,
furans, sulfur compounds, aldehydes, ketones, alcohols, and other volatile compounds
identified in the peanut butters using this technique are listed in Table 1. The compounds
positively identified in this research have been identified previously in roasted peanuts [38 and
references within].
Isolation of Volatile Flavor Compounds
Headspace analysis results in the isolation of a representative sample of the volatile
flavor compounds above, and in equilibrium with, the food [ 16-17]. Although an exhaustive
extraction of the volatile flavor compounds is not possible using headspace analysis
techniques, through controlling the conditions of the isolation, recovery of the volatile flavor
compounds can be maximized. Artifact formation during the isolation was minimized by the
selection of a purge temperature below 132~ the critical minimum temperature for flavor
formation in peanuts via Maillard reactions [39], and the continuous nitrogen purge of the
samples to minimize oxidative degradation.
229
Table 1
Volatile flavor compounds identified in peanut butter using purge-and-trap headspace
analysis',
Kovats
Kovats
RI Compound
RI
Compound
Pyrazines:
Aldehydes:
809 Hexanal b
828 Methylpyrazine
858 2-Hexenal b
909 2,5-Dimethylpyrazine b
902 Heptanal b
910 2,6-Dimethylpyrazine b
958 2-Heptenal b
917 2,3-Dimethylpyrazine b
914 Ethylpyrazine b
1004 Octanal
1012 2,4-Heptadienal
1001 Trimethylpyrazineb
1059 2-OctenaJ b
997 2-Ethyl-6-methylpyrazine b
1105 Nonanal b
1000 2-Ethyl-5-methylpyrazine b
1161 2-Nonenal
1003 2-Ethyl-3-methylpyrazine b
1218 2,4-Nonadienal
1019 2-Methyl-6-vinylpyrazineb (t)
1208 Decanal
1022 2-Methyl-5-vinylpyrazine b (t)
1260 2-Decenal
1080 3-Ethyl-2,5-dimethylpyrazine b
1296 t, c-2,4-Decadienal
1085 2-Ethyl-3,5-dimethylpyrazine b
1320 t, t-2,4-Decadienal b
1090 5-Ethyl-2,3-dimethylpyrazine
962 Benzaldehyde b
1083 2,6-Diethylpyrazine b (t)
1047 Phenylacetaldehyde b
1087 2, 5-Diethylpyrazineb (t)
1101 Dimethyl-2-vinylpyrazineb (t)
1157 2,3-Diethyl-5-methylpyrazineb
Ketones:
800 2-Hexanone
1159 3,5-Diethyl-2-methylpyrazine b
795 3-Hexanone
1162 2,5-Dimethyl-3-propylpyrazine
891 2-Heptanone
1190 2-Methyl-5-(1-propenyl)pyrazine
993 2-Octanone b
1104 6,7-Dihydro-5H984 2,3-Octanedione (t)
cyclopentapyrazine (t)
1043 3-Octen-2-one
1188 6,7-Dihydro-2-methyl-5H1096 3,5-Octadien-2-one (t)
cyclopentapyrazine (t)
1094 2-Nonanone
1225 6,7-dihydro-2, 5-dimethyl-5H1141 3-Nonene-2-one b
cyclopentapyrazine (t)
1191 2-Decanone
1034 2-Hydroxy-3-methyl-2Furans:
cyclopentene-l-one (t)
965 5-Methyl-2-furfural b
978 2-Acetyl-3-hydroxy~ran (t)
991 2-Pentylfuran
1220 2-Vinylbenzofuran
230
.Table 1 (cont.)
Kovats
RI Compound
Sulfur Compounds:
761 Dimethyldisulflde
972 Dimethyltrisulfide b
790 3-Methylthiophene
795 2-Methylthiophene
1023 2-Acetylthiazole b
1236 Benzothiazole
1271 5-(2-Hydroxyethyl)-4methylthiazole (t)
Kovats
ILl
Compound
Alcohols:
874 1-Hexanol
811 2-Hexanol
806 3 -Hexanol
872 2-Hexen- 1-ol
972 1-Heptanol
1098 1,2-Heptanediol (t)
1072 1-Octanol
981 1-Octen-3-ol b
1173 1-Nonanol
1269 1-Decanol
Other Heterocyclic Compounds:
976 Pyranone (t)
1058 Trimethyl-2H-pyranoneb (t)
Miscellaneous Compounds:
1030 Limonene b
1112 3-Hydroxy-2-methyl-4H1037 Benzenemethanol b
pyranone (t)
1149 2,3-Dihydro-3,5-dihydroxy-61118 Benzeneethanol (t)
1221 2-Phenoxyethanol (t)
methyl-4H-pyran-4-one (t)
1222 Vinylphenol b
998 3-Methoxypyridine (t)
1017 2-Vinylpyridine (t)
1265 2-Phenyl-2-butenal (t)
1036 2-Acetylpyridine b
1177 Naphthalene
1277 Methylnapthalene
1013 1H-Pyrrole-2-carboxaldehyde (t)
1067 2-Acetylpyrroleb (t)
1182 1-(2-Furanylmethyl)-1H-pyrrole (t)
1238 3-Ethyl-4-methyl-lH-pyrrole-2,5-dione (t)
' Identification of volatile flavor compounds based on comparison of mass spectrum and
retention times of authentic standards except where labeled tentative (t), where no authentic
sample was available.
b Identified volatile flavor compounds included in statistical analyses to determine the
optimum conditions for purge-and-trap isolation. (Ten compounds were included in the
statistical analyses, but not identified.)
231
Internal standards were added to the peanut butter sample prior to purging
(tetradecane) and to the top of the trap prior to solvent elution (pentadecane). The recovery
of the tetradecane, as well as the volatile compounds in the sample, is dependent on purge
temperature and purge time. Addition of an internal standard with the sample to determine
the conditions necessary for maximum recovery of the volatile flavor compounds would not
be valid. Therefore, the internal standard added to the traps following purging and prior to
solvent elution, was used as the quantification standard. In the application of this method, the
addition of an internal standard to the sample prior to purging would be recommended.
Solvent elution results in lower recoveries of volatile compounds and the loss of highly
volatile compounds due to evaporation during concentration [27]. However, thermal
desorption contributes to potential artifact formation [21 ]. Therefore, the volatile compounds
were desorbed from the adsorbent by solvent elution, rather than by thermal desorption to
minimize thermal degradation of the volatile compounds.
Statistical Analysis
The major objective of this study was to determine the conditions for the maximum
recovery of volatile flavor compounds from peanut butter. Forty-eight of the major volatile
flavor compounds were included in the statistical analyses to determine the effects of purge
temperature, purge time, and adsorbent on their recovery. Of these, 38 volatile flavor
compounds were positively or tentatively identified. Direct comparison of each of the
treatment effects was complicated by the number of compounds of interest and the variable
effects of purge conditions on the recovery of the compound. Therefore, two alternative
statistical methods, frequency table analyses of the recovery ranks [40] and multivariate
statistical techniques [41] were used to determine the conditions for maximum recovery of
volatile flavor compounds from peanut butter.
Multivariate statistical techniques [37, 40-41] were used to select a smaller set of
compounds which can be used to determine the effects of the three isolation factors: purge
temperature, purge time, and adsorbent on the recovery of the volatile flavor compounds.
Principal component (PCA) and cluster analyses (CA) divided the 48 variables into groups
called clusters based on the similarities and differences between the values observed for them.
The response of the compounds to the different purge temperature and time combinations
within each cluster were highly correlated with the other volatile compounds identified within
that particular cluster. The variable showing the greatest correlation with its cluster was
chosen to best represent the properties of all variables contained within that cluster.
The initial solution obtained through PCA and CA resulted in the selection of 16 and
24 variables for the Tenax and CP/CS adsorbents, respectively. The principal component and
cluster analyses were then repeated on this selected data set, for each adsorbent and year
combination, to determine if further reduction in the number of variables could occur. From
this analysis, a reduced data set containing 6 dependent variables for each adsorbent x year
combination, resulting in 12 different volatile flavor compounds, were selected (Table 2).
Vinylphenol, 2,5-dimethylpyrazine, and trimethylpyrazine were identified as key components
for the first three clusters for each adsorbent and crop year combination. Other compounds
identified within the key variable set include acetylpyrrole, hexanal, phenylacetaldehyde, 3-
232
ethyl-2, 5-dimethylpyrazine,
and benzaldehyde.
pyranone,
2- ethyldimethylpyrazine,
ethylmethylpyrazine,
Table 2
Volatile flavor compounds in peanut butter identified through multivariate statistical
techniques as representative of all volatile flavor compounds with respect to purge condition
Cluster
Tenax-GC
1
2
3
4
5
6
CP/CS
1
2
3
4
5
6
1988
1989
Vinylphenol
2, 5-Dimethylpyrazine
Trimethylpyrazine
Acetylpyrrole
Hexanal
Phenylacetaldehyde
Vinylphenol
2, 5-Dimethylpyrazine
Trimethylpyrazine
Phenylacetaldehyde
Unknown-MW 155
Acetylpyrrole
Vinylphenol
2, 5-Dimethylpyrazine
Trimethylpyrazine
Dimethylvinylpyrazine
2-Ethyldimethylpyrazine
Benzaldehyde
Vinylphenol
2, 5-Dimethylpyrazine
3-Et hyl-2, 5- dimethylpyrazine
Pyranone
Ethylmethylpyrazine
Hexanal
The factors and interactions found to be significant by the analysis of variance for
these selected volatile flavor compounds are summarized in Table 3. Interactions between
purge temperature, purge time, and/or adsorbent were significant (P<0.05) for
trimethylpyrazine, 3-ethyl-2,5-dimethylpyrazine, dimethylvinylpyrazine, and 2,3-dihydro-3,5dihydroxy-6-methyl-4Hpyran-4-one. Thus, for these compounds, the factors do not act
independently and means cannot be pooled.
The recovery of individual flavor compounds is influenced by isolation conditions.
The effect of purge temperature, purge time, and adsorbent on the recovery of volatile flavor
compounds will be discussed in further detail as evaluated through rank and multivariate
statistical analyses.
Effects of Purge Temperature and Purge Time
Recovery of the volatile flavor compounds isolated in the purge-and-trap isolation
from peanut butter increases with an increase in purge temperature and purge time. Results of
both statistical analysis techniques indicated that maximum recovery occurred most frequently
at a purge temperature and time of 90~ for 4 hr, followed by purge conditions of 90~ for 2
hr and 70~ for 4 hr for both the Tenax and CP/CS adsorbents (Table 4). These purge
conditions resulted in isolates with characteristic roasted peanut flavor, as determined by an
informal aroma evaluation of the isolates by researchers who have experience in peanut flavor.
233
A comparison of the contents of the volatile flavor compounds, representative of the
clusters identified through multivariate statistical analysis, as a function of purge temperature
and purge time are shown in Tables 5 and 6.
Table 3
Results of analyses of variance on the recovery of key volatile flavor compounds identified
through multivariate statistical techniques
Compound
Vinylphenol
2,5-Dimethylpyrazine
3-Ethyl-2,5-dimethylpyrazine
Trimethylpyrazine
3-Ethyl-2,5-dimethylpyrazine
Acetylpyrrole
Phenylacetaldehyde
Dimethylvinylpyrazine
2,3-Dihydro-3,5-dihydroxy-6methyl-4Hpyran-4-one
Hexanal
Unknown-MW 155
2-Ethyl- 5-methylpyrazine
Benzaldehyde
Significant Effects (P<0.05)
Trap
Trap
Trap
Trap
Trap
Temp
Temp
Temp
Temp
Temp
Temp
Temp
Temp
Temp
Time
Time
Time
Time
Time
Trap*Temp*Time
Trap*Temp*Time
Time
Time
Trap*Time
Temp*Time
Temp
Temp
Temp
Time
Time
Time
The observed relationship between purge temperature and purge time and the recovery
of volatile flavor compounds would be expected due to the shift in equilibrium of the volatiles
between the food and the headspace above the food during the isolation. The volatility of
these compounds influenced the degree to which the purge temperature and time influenced
their recovery, with the more volatile compounds affected by purge conditions to a lesser
degree.
Breakthrough or partial loss of some highly volatile compounds is a potential problem
during headspace analysis with trapping on Tenax. However, these highly volatile compounds
usually do not have a significant impact on flavor of roasted peanuts [ 12]. In this study, the
recovery of the volatile flavor compounds was greatest as purge temperature or purge time
increased. Thus, it was assumed that no significant loss of volatile flavor compounds
occurred during the isolation.
The purge temperature and time combination which resulted in the greatest recovery
of volatile flavor compounds is at the maximum temperatures and times chosen for this
experiment. These conditions also contributed to the lowest mean coefficient of variability,
which is important in determining the effects of postharvest and storage conditions on the
relative contents of volatile flavor compounds. Further increases in the purge temperature and
time will more than likely increase the recovery of volatile flavor compounds through
234
enhancement of the volatilization of these compounds into the headspace. However,
these increases may also contribute to a greater potential for artifact formation during the
isolation.
Table 4
Summary of frequency table analyses of the
recovery ranks as a function of purge
temperature and purge time"
Purge Conditions
Rank
Temperature (~
Time (hr)
1
90
4
2
3
4
5
6
7
90
70
90
70
50
70
2
4
1
2
4
1
8
50
2
9
50
1
Data for 3 replications and two traps have
been pooled. Rank = 1 indicates highest
recovery of each volatile compound. Results
of rank analysis were verified with a Chisquare goodness of fit test (Conover, 1980).
Effects of Adsorbent
Tenax-GC has been used widely for the isolation of volatile compounds from foods
because it shows excellent recovery of adsorbed compounds, good thermal stability, and low
affinity for water, although it is limited by its lower adsorption capacity [20,27]. Carbopack
B/Carbosieve S-III have been suggested as replacements for the currently used adsorbents for
environmental research. The combination of these two adsorbents provides an effective
system for trapping and desorbing a wide range of volatiles [28].
Recovery of the volatile compounds was greater when Tenax was used as the
adsorbent in comparison to CP/CS, as shown by the rank analysis (Table 7) and analysis of
variance of the compounds selected through multivariate statistical techniques (Tables 6 and
8). This difference in recovery of the volatile flavor compounds may be attributed to
differences in the affinity of the polymer adsorbents for the specific volatile flavor compounds
or the release of the compounds from the respective polymers with solvent elution. Although
Mosesman et al. [28] noted a higher recovery of volatile compounds using Carbopack and
Carbosieve adsorbents, these researchers were applying these methods to the isolation of
Table 5
Influence of purge temperature and time on the recovery of key volatile flavor compounds
identified through multivariate statistical techniques."
Content (ng/g)
Temp. (T)
Time (hr)
50
70
VinvlDhenol: S.E. 141
1
47
188
2
87
299
4
117
556
Mean
83'
34Sb
2.5-DimethvlDmzine: S.E. 29
1
153
23 3
2
180
243
4
214
227
Mean
182b
23 5'b
Acetvlpp-ole: S.E. 5
1
2
7
2
4
11
4
5
19
Mean
3b
12b
Phenvlacetaldehvde: S.E. 22
1
28
51
68
2
40
4
43
82
Mean
37b
67'
2.3-Dihvdro-3.5-dihvdroxv-6-methvl-4H~vran-4-one*:
S.E. 9
- Sb
1
2b
2
4b
6b
4
5b
1Ob
90
Mean
60 1
762
1515
959'
279"
382"
729'
283
256
276
272'
223'
227'
239'
22
31
59
38'
10"
15"
28'
53
58
42
5 1'
44'
55'
5 6'
13b
Sb
46'
N
Hexanal: S.E. 5
1
2
4
W
24
27
26
26'
21
20
32
24'
9
13
9'
15
27
47
28b
33
51
99
61'
18'
29'
53d
62
77
103
81'
90
118
142
117b
126
144
167
146'
93'
113'
137d
12
19
27
19'
25
31
35
30b
39
43
52
45'
25'
3 1"
38d
Mean
Unknown - MW 155: S.E. 8
1
5
2
4
Mean
2-EthYl-5-methylpyrazine: S.E. 10
1
2
4
Mean
Benzaldehvde: S.E. 4
1
2
4
Mean
20d
15
19
27
21'
22d
29d
Contents of volatile flavor compounds are based on integrator response to pentadecane.
Data for 3 replications and 2 adsorbents (Tenax, CPKS) have been pooled for analysis
of variance. Interactions between temperature, time, and adsorbent effects were not
significant (p>O.O5). Temperature means with same superscripts (a-c) are not
significantly different (p>O.OS) for each compound. Time means with same superscripts
(d-f) are not significantly different for each compound.
Significant interactions between temperature and time dictate that each factor must
be considered separately.
Q\
Table 6
Influence of purge temperature, time, and adsorbent on the recovery of key volatile flavor
compounds identified through multivariate statistical techniques."
Content (ng/g)
50
70
90
Trimethylpwazine*: S.E. 25
Tenax
CP/CS
1
65'"
107&
121bX
2
8 Sb"
1 17bx
1 77'
4
76&
171"
178"
1
4SE"
7lCb
132-'
2
6SDb
1 1 6BCx
1 1 3BcDy
4
109'-
10SBCDy
1 76h
Mean
Dimethylvinylpyrrole': S.E. 4 (time). 6 [temp.)
1
20
24
45
3 Og
2
27
34
65
42'
4
34
56
76
55'
Tenax Temp Mean
26'"
SObX
7OU
CPICS Temp Mean
27Bx
26By
54*y
Contents of volatile flavor compounds are based on integrator response to pentadecane.
Data for 3 replications have been pooled for analysis of variance. Means with the same
superscript (a-d or A-E) for the purge temperature and time effects for each adsorbent are
not significantly different (p>O.O5). Means with the same superscript (x-y) for the adsorbent
effects for each purge temperature and time are not significantly different (p>O.OS).
* Significant interactions between temperature, time, and adsorbent dictate that each factor
must be considered separately.
Significant interactions between temperature and adsorbent dictate that these factors must
be considered separately. Time means with the same superscript (e-g) are not significantly
different (p>O.O5).
239
environmental contaminants, which are more volatile than the volatile flavor compounds
present in food. The superior ability of Tenax to trap volatiles with longer retention times, in
comparison to Porapak Q, has been demonstrated for plant volatiles [27].
Table 7
Paired t-test comparisons of recovery of volatile flavor
compounds from peanut using Tenax and CP/CS adsorbents"
Maximum recoveryb:
Mean:
Std. Error:
t statistic:
Prob > Itl:
- 14 ng/g
4 ng/g
-3.28
0.0015
Most frequent maximum recovery (90~ for 4 hr):
Mean:
-11 ng/g
Std. Error:
4 ng/g
t statistic"
-2.84
Prob > Itl:
0.0055
3 Most frequent maximum recoveries (90~ for 4 hr,
90~ for 2 hr, 70~ for 4 hr):
Mean:
- 19 ng/g
Std. Error:
3 ng/g
t statistic:
-5.80
0.0001
Prob > It[:
9Contents of volatile flavor compounds are based on
integrator response to pentadecane. Positive difference
value indicates greater recovery with the CP/CS adsorbent.
b Maximum recovery was achieved at a number of
temperature-time combinations depending upon the adsorbent,
compound, and year.
The identification of phenylacetaldehyde as a component of the key variable set when the
volatiles were trapped on Tenax, but not when trapped on CP-CS may indicate differences in
the affinity of these adsorbents for different flavor compounds.
Significantly more
phenylacetaldehyde was recovered with the Tenax adsorbent than with the CP/CS adsorbent.
Connick and French [42] also noted a lower recovery of phenylacetaldehyde with trapping on
Carbotrap in comparison to trapping on Tenax. The low recovery was attributed to aldol
condensation reactions with the basic graphitized carbon black adsorbent.
240
Table 8
Influence of adsorbent on the recovery of key volatile flavor compounds identified through
multivariate statistical techniques. ~
Content (ng/g)
Compound
Tenax
CP/CS
S.E.
Vinylphenol
530'
397'
115
2, 5-Dimethylpyrazine
242'
217'
23
Acetylpyrrole
23'
13b
4
Phenylacetaldehyde
85'
18b
11
2, 3-Dihydro-3,5-dihydroxy-6-methyl4Hpyran-4-one
12"
10"
4
Hexanal
27'
20'
4
Unknown - MW 155
38'
29"
6
2-Ethyl-5-methylpyrazine
122"
107"
8
Benzaldehyde
31"
32"
3
~ Contents of volatile flavor compounds are based on integrator response to pentadecane.
Data for 3 replications, 3 purge temperatures (50, 70, 90~ and 3 purge times (1, 2, 4 hr)
have been pooled by the analysis of variance to test for a trap (adsorbent) main effect.
Interactions between trap (adsorbent) and temperature and/or time effects were not significant
(p>0.05). Means within rows with the same superscript (a-b) are not significantly different
(p>0.05).
CONCLUSIONS
In this study, a headspace analysis method for the isolation of volatile flavor
compounds from peanut butter was developed. The headspace analysis method provides a
representative sample of the volatile flavor compounds present in equilibrium with the food.
The conditions for the maximum recovery of volatile flavor compounds from peanut butter
include trapping on Tenax-GC with a purge temperature of 90~ for 4 hr. The recovery of a
majority of volatile flavor compounds was greater when Tenax was used as an adsorbent
rather than CP/CS. Frequency table analysis of ranks and analysis of variance of a key
variable set of compounds selected through multivariate statistical techniques were used to
determine the conditions for the maximum recovery of volatile flavor compounds from peanut
butter. The headspace analysis method developed in this research will be applied to future
studies to determine the effects of postharvest and processing treatments on the flavor quality
of peanuts.
241
REFERENCES
Johnson, BR, Waller, GR, Foltz, RL. J Agric Food Chem 1971; 19:1025-1027.
Buckholz, LL, Jr, Daun, H. In: Teranishi R, Barrera-Benitez H, eds. Quality of
Selected Fruits and Vegetables in North America. Washington, DC: American
Chemical Society, 1981; 163-181.
Pattee, HE, Singleton, JA, Johns, EB. Agric Biol Chem 1971; 19: 134-137.
4
Forss, DA. Prog Chem Fats Other Lipids 1972; 13(4): 177-258.
Litman, I, Numrych, S. In: Supran MK, eds. Lipids as a Source of Flavor.
Washington, DC: American Chemical Society, 1978; 1-17.
Teranishi, R. (1989). In Teranishi, R, Buttery, RG, Shahidi, F, eds. Flavor Chemistry:
Trends and Developments. Washington, DC: American Chemical Society, 1989; 1-6.
Brown, DF, Senn, VJ, Dollear, FG. J Agric Food Chem 1973; 21:463-467.
Walradt, JP, Pittet, AO, Kinlin, TE, Muralidhara, R, et al. J Agric Food Chem 1971;
19: 972-979.
Mason, ME, Johnson, B. Hamming, M. J Agric Food Chem 1969; 14: 454-460.
10
Shu, CK, Waller, GR. J Food Sci 1971; 36: 579-583.
11
Ho, C-T, Lee, M-H, Chang, SS. JFood Sci 1981; 47:127-133.
12
Buckholz, LL, Jr, Withycombe, DA, Daun, H. J Agric Food Chem 1980; 26: 760-765.
13
Buckholz, LL, Jr, Daun, H, Stier, E, Trout, R. J Food Sci 1980; 45: 547-554.
14
Fore, SP, Dupuy, HP, Wadsworth, JI. Peanut Sci 1976; 3" 86-89.
15
Fore, SP, Fisher, GS, Legendre, MG, Wadsworth, JI. Peanut Sci 1979; 6:58-61.
16
Weurman, C. J Agric Food Chem 1969; 17: 370-384.
17
Dravnieks, A, O~Donnell, A. J Agric Food Chem 197 I; 19:1049-1056.
18
Jennings,WG, Filsoof, M. J Agric Food Chem 1977; 25: 440-445.
242
19
Buttery, RG, Seifert, RM, Ling, LC, Soderstrorn, EL, et al. J Agric Food Chem
1982; 30: 1208-1211.
20
Reineccius, GA, Anandaraman, S. In Lawrence JF, eds. Food Constituents and Food
Residues. New York: Marcel Dekker, 1984; 195-293.
21
Olafsdottir, G, Steinke, JA, Lindsay, RC. J Food Sci. 1985; 50:1431-1436.
22
Suzuki, J, Bailey, ME. J Agric Food Chem 1985; 33" 343-347.
23
MacLeod, G, Ames, JM. J Food Sci 1986; 51" 1427-1434.
24
Vercellotti, JR, Kuan, J-CW, Liu, RH, Legendre, MG, et al. J Agric Food Chem 1987;
35: 1030-1035.
25
Vercellotti, JR, St. Angelo, AJ, Legendre, MG, Sumrell, G, et al. J Food Comp Anal
1988; 1 239-249.
26
Risch, SJ, Reineccius, GA. In Parliment, TH, McGorrin, 1LI Ho, C.-T, eds. Thermal
Generation of. Aromas, Washington, DC: American Chemical Society, 1989, 42-50.
27
Cole, RA. J Sci Food Agric 1980; 31" 1242-1249.
28
Mosesman, NH, Betz, WR, Corman, SD. LC-GC 1988; 6: 328-336.
29
Sanders, TH, Vercellotti, JR, Blankenship, PD, Crippen, KL, et al. J. Food Sci.1989;
54:1066-1069.
30
Bondarovich, HA, Friedel, P, Krarnpl, V, Renner, JA, et al. J Agric Food Chem
1967; 15: 1093-1099.
31
Friedel, P, Krampl, V, Radford, T, Renner, JA, et al. J Agric Food Chem 1971; 19"
530-532.
32
Kinlin, TE, Muralidhara, R, Pittet, AO, Sanderson, A, et al. J Agric Food Chem 1972;
20" 1021-1028.
33
Pittet, AO, Muralidhara, R, Walradt, JP, Kinlin, T. J Agric Food Chem 1974; 22:
273-279.
34
Vitzthurn, OG, Werkhoff, P. J Agric Food Chem 1975; 23" 510-516.
35
Vernin, G, Petitjean, M. In Vernin G, eds. Chemistry of Heterocyclic Compounds in
Flavours and Aromas. Sussex, England: Ellis Horwood Limited, 1982; 305-342.
243
36
Ho, C-T, Jin, QZ, Lee, M-H, Chang, SS. J Agric Food Chem 1983; 31 1384-1386.
37
SAS Institute Inc. SAS/STAT User's Guide, Ver. 6, 4th Ed., Cary, NC, 1989.
38
Maarse, H, Visscher, CA. In Volatile Compounds in Food - Qualitative and
Quantitative Data, Vol. II. Zeist, The Netherlands: TNO-CIVO Food Analysis
Institute, 1989; 76.
39
Mason, ME, Waller, GR. J Agric Food Chem 1964; 12: 274-278.
40
Conover, WJ. Practical Nonparametric Statistics, 2nd. Ed., New York: John Wiley &
Sons, 1980.
41
Tabachnick, BG, Fidell, LS. Using Multivariate Statistics. New York: Harper & Row,
1983.
42
Connick, WJ, Jr, French, RC. J Agric Food Chem 1991; 39: 185-188.
ACKNOWLEDGMENTS
We thank J.R. Vercellotti for his input and valuable discussions, E.J. Williams for
providing the peanut samples, and E. St. Cyr, Jr. for his assistance in sample preparation.
This Page Intentionally Left Blank
D. Wetzel and G. Charalambous (Editors)
Instrumental Methods in Food and Beverage Analysis
9 1998 Elsevier Science B.V. All rights reserved
245
GC-MS(EI, PCI, NCI, SIM, ITMS) Data Bank Analysis of Flavors and
Fragrances. Kovats indices a
, , c9
G. Vernin *b , C. Lageot b, and C. Parkany~
bChimie des ArSmes-Oenologie, Associ~ au CNRS, URA 1411,
Facult~ des Sciences et Techniques de St-J~.rSme, Case 561,
F13397 Marseille C~dex 20, France
CDepartment of Chemistry, Florida Atlantic University, PO Box 3091,
Boca Raton FL 33431-0991, USA
1.
INTRODUCTION
Regardless of the method of isolation of volatile fragrant compounds (1, 2) in
essential oils and aromas, their analysis and identification would not be
conceivable without coupled GC-MS which is the best available method at the
present time. Thanks to this technique which was originally developed in 1967 and
subsequently improved upon, thousands of flavoring and fragrant compounds
have been identified and the information about them compiled in a number of
publications (3-12).
Upon bombardment with electrons, a molecule will undergo fragmentation to an
extent which will depend on its stability. The observed fragments and their relative
intensities are characteristic of each molecule.The obtained data are stored on
computer disks with a data acquisition system as an integral part of the instrument.
It is thus possible to store a large number of spectra which are subsequently
compared with a data bank linked to the above system, in order to facilitate their
identification. Computerization in flavor research has been introduced (13) and
numerous data banks have been developed in the last twenty years. The
respective references are given in Table 1.
In addition to a chromatogram reconstructed on the basis of the total ionization
current which represents a digital imprint of a mixture subjected to analysis, a
listing is obtained which can contain hundreds of spectra depending on the
complexity of the mixture.
aSome parts of this chapter have been previously published in Analusis, 20(7) 3438 (1992) and are being reproduced here with permission of the editor.
Author to whom correspondence should be addressed.
246
Table 1
Compilation of information on electron impact (El) mass spectra (reviews, data
bases and data banks).
Publications
Authors (Ref.)
Ryhage and Von Sydow (14)
Thomas and Whillhalm (15)
Monoterpenoid alcohols
Von Sydow (16)
Monoterpenoid aldehydes and ketones
Von Sydow (17)
Esters of monoterpenoid alcohols
Von Sydow (18)
Monoterpenes and derivatives
Von Sydow (19)
Sesquiterpenes
Hirose (20)
Hayashi et al. (21)
Yukawa and Ito (22)
Terpenes and related compounds
Moshonas and Lund (23)
800 volatile compounds (flavors, fragrances) Jennings and Shibamoto (24)
Diterpenes, terpenes and terpenoids
Enzell et al. (25-31)
Isoprenoids from tobacco
Enzell et aL (32)
Heterocycles
Vernin (7)
Natural and synthetic flavors and fragrances Tucker and Maciarello (33)
Monoterpenoid hydrocarbons
Mass spectra data banks and Library
search
(Type of compounds, origin)
Monoterpenes and derivatives (Sweden)
Monoterpenes and derivatives (Japan)
US Environmental Protection Agency
Collection of mass spectral data (USA)
Registry of mass spectral data (USA)
Compilation of mass spectral data (France)
EPA/NIH mass spectral data base (USA)
Computer GC-MS/Kovats indices
Library search
SPECMA data bank of volatile compounds
of flavors and fragrances (France)
(at the present time not commercialized)
Eight-peak index of mass spectra
(England)
NBS library compilation (USA)
Von Sydow (19)
Yukawa and Ito (22)
MSSS (34)
Middledisch (35)
Stenhagen et aL (36)
Cornu and Massot (37)
Heller and Milne (38, 39)
Craveiro et aL (40, 41)
Petitjean et aL (42-44)
Vernin etaL (45-47)
Mass Spectrometry Data Center
(48, 49)
Finnigan Mat (50)
247
Table 1 (Continued)
Publications
Authors (Ref.)
PBM data base (USA)(HP59943A)
Data base of essential oils
CD and DC Rom systems for mass
spectral data (Japan)
Compilation of mass spectra of volatile
compounds in foods (The Netherlands)
Wiley/NBS registry of mass spectral data
(USA)
MPI library of mass spectral data (Germany)
NIST/EPA/NIH mass spectra data base
Hewlett Packard (51)
Chien (52)
Wiley and Sons (53)
Bench-top/PBM version 3.0
Bench-top GC/MS
McLafferty (61)
Umano (62)
(USA)
De Brauw et al., TNO
(54, 55)
McLafferty (56)
McLafferty and Stauffer (57)
Henneberg et al. (58)
NBS(59, 60)
However, answers obtained from existing data banks are often incorrect or nonexistent. It is estimated that, on average, only 25% of answers are correct in the
case of essential oils and that this percentage is even lower in the case of food
aromas containing numerous heterocyclic compounds.
There are many reasons why these answers are incorrect.
a) Dissimilarity index comparing the spectrum of the unknown with that of a
reference compound is too high because of the parameters of the instrument
(quadrupole or magnetic) or the analytical conditions (saturated spectra, source,
temperature, etc.)
b) Several compounds give a single peak in the spectrum of a mixture. In those
cases where one of the products is identified, a computer program makes it
possible to substract the spectrum of this known substance and to obtain the
spectrum of the unknown.
c) Results for trace substances (the background noise is too prominent).
No answers can be obtained mostly in those cases where the data bank does
not contain the reference spectrum or the dissimilarity index exceeds a threshold
value.This means that it is necessary to make a visual comparison of the spectrum
of the unknown with the libraries of available spectra or with spectra published in
various publications. On the other hand, the data bank must contain several mass
spectra of the same compound, especially when they do not have the same base
peak. Quadrupole apparatus gives more important fragments at m/z 41 and 39
than the magnetic one. This fact must be taken into account.
This is a long and exacting operation. However, it can be shortened if the principal
fragmentation patterns of the molecules are known.
248
In the case of terpene, sesquiterpene and diterpene derivatives, fragmentation
processes have been described in several books (37, 63-65), reviews and other
publications (25-32). Mass spectra of heterocyclic compounds found in food flavors
and Maillard reaction have been compiled (7, 66). However, identification based
only on electron impact (El) mass spectra has a relatively limited usefulness.
Chromatographic data, selected ion monitoring (SIM) and gentle ionization
techniques leading to quasi molecu lar ions are needed, as well as the use of high
resolution mass spectrometry in GC/MS coupling (67).
2. CHROMATOGRAPHIC
DATA
Retention times (RT) and relative retention times with respect to a standard ((zR )
cannot be used because these quantities are not reproducible and vary greatly
with the temperature of the chromatographic column. Kovats indices (KI)
introduced by Kovats in 1958 (68) do not suffer from this disadvantage. At
isothermic temperature, Kovats uses the following logarithmic equation with linear
alkanes as reference compounds
K I = 1 0 0 n + 100(l~
"
I~
~
log t'R(n+l )" log t'R(n)
where t'R(x), t'R(n) and t'R(n+l) are the reduced retention times of the compound X
and the linear alkanes with n and n+l carbon atoms, respectively, which are
eluted just before and after compound X. Their reproducibility on polar columns
(Carbowax 20M, DB WAX, FFAP, BP 20, HP 20) and nonpolar columns (SE 30, SF
96, OV 1, OV 101, OV 117, DB 1, DB 5) is good under similar GC conditions on a
given column.
In linear programmed temperature, Van den Dool and Kratz (69) use the
following formula:
X M(n)
KI = 100 n + 100i(/.
'~
\ M(n+i) " M(n) /
/
where X, M(n ) and M(n+i ) are either the retention temperatures or the adjusted
retention times of straight-chain aliphatic esters.
The calculation of KI values starts with the injection of either a standard
hydrocarbon mixture (C6 to C22 ) for CW 20 M and (C6 - C30) for OV 101 (or linear
ethyl esters) under the same linear programming conditions. Their retention times
are recorded directly or manually by the computing integrator when peak X in the
sample appears between hydrocarbon (or aliphatic ethyl ester) peaks with n and
n+l carbon atoms of n-paraffin (or straight-chain of the acid moiety of ethyl ester).
249
The Kl(x ) is calculated from the following equation:
Kl(x )= 100n + 100
I tR(x)
tR(n+l)
" tR(n+l ) /
....
tR(n)
The software of the integrator can detect each peak and calculate each KI value
automatically and print it out with other GC data
The dependence of retention index towards temperature has been extensively
studied (70).
In the programmed temperature mode which is the usual case, variables such
as carrier gas flow-rate and program rate affect the measurement since they
determine the temperature range that the sample is exposed to prior elution.While
temperature has a relatively little effect on Kovats indices on nonpolar columns, it
can have quite marked effects on polar phases.
According to Shibamoto (71), theoretically, it is necessary to use isothermal
conditions (i.e. the Iogaritmic equation) to calculate the Kovats indices.
The use of Kovats indices was recommended by a number of authors (7, 42-45,
72-77).
In GC/MS (El, CI) high search performance is increased by combining retention
indices and mass spectra in mass library search of volatiles (7, 40, 45-47, 78).
They can be also used as computer-assisted correlation with aroma (79). Owing to
this importance, Kovats indices have been compiled in various reviews, books and
many other publications (See Table 2).
Andersen et aL (82-85)identified sesquiterpene hydrocarbons and heteroannular
dienes in various essential oils using retention data.
Test searches for ~decalactone and volatiles of jasmine absolute and men's
Cologne were carried out using a search file created for about 2.000 compounds
(77). Ramaswami et al. (90) reported a compilation of 60 sesquiterpene
hydrocarbons in alphabetical order as well as their Kovats indices on polar and
nonpolar columns. They noted that retention indices using fatty acid ethyl esters
calibration or linear hydrocarbons, however occurate and reproducible, do not by
themselves represent conclusive evidence for the identification, mass spectra
being the decisive tool.
Davies (91) tabulated some 900 Kovats indices of almost 400 monoterpene
and sesquiterpene derivatives on methyl silicone and/or Carbowax 20 M liquid
phases. Kovats indices of heterocyclic compounds have also been reported in
numerous papers dealing with food flavors, Maillard reaction, model systems (7)
and yeast extracts (93).
In coupled GC/MS one can also use a number of scans but a prerequisite for
this is to find at least ten compounds for which the Kovats indices are known and
which are uniformly distributed on the reconstructed chromatogram. A simple
program called MBASIC.SCAN1 using the linear relationship KI = a (scans) + b,
can be used to calculate all Kovats indices for the listing (94).
250
Table 2
Kovats indices: theory and compilation
Compounds
Aliphatic halides, alcohols aldehydes and
ketones
Aliphatic, alicyclic and aromatic compounds
Esters (linear temperature)
Sesquiterpene hydrocarbons
Theory of the retention index system
Retention indices
Calculation and application of the
retention indices
Review on the use of retention indices
Monoterpene hydrocarbons
Volatile compounds of flavors and
fragrances (1240 compounds)
Monoterpenes
Kovats indices as a preselection routine
in mass library search of volatiles in
essential oil analysis
Diterpene hydrocarbons
Retention index library
Dependence of retention index on
temperature
Use of retention index mass spectral
search for identification of the volatiles
in fragrances
Retention indices in essential oil analysis
(183 compounds)
Sesquiterpene hydrocarbons
(60 compounds)
Monoterpene and sesquiterpene
derivatives (a review, 400 compounds)
Natural and synthetic pyrazines
Yeast extracts
Authors (Ref.)
Kovats (68, 80)
Wehrli and Kovats (81)
Van Den Dool and Kratz (69)
Andersen et aL, (82-85)
Erdey et aL, (86)
Ettre (72)
Majlat et aL (87)
Haken (73)
Saeed et al. (74)
Jennings and Shibamoto (24)
Swigar and Silverstein (88)
Alencar et al. (75)
Perry and Weavers (76)
Sadtler research laboratories (89)
Albaiges and Guardino (70)
Yamada et al. (77)
Shibamoto (71)
Ramaswami et al. (90)
Davies (91)
Mihara and Masuda (92)
Ames and Elmore (93)
Table 3.
Comparison between experimental and calculated Kovats indices (KIA) of methyl esters on a nonpolar column*.
~~
Series
1
2
3
~
~
Products
Scans(S) KIA(exp)a
KIA (calc)b Linear equationsC
KIA(calc) =
Methyl n -heptanoate
Methyl n -0ctanoate
Methyl n -nonanoate
201
31 2
429
1007
1107
1207
1008
1105
1208
0.877(S)+ 831.6
Methyl n- nonanoate
Methyl n -decanoate
Methyl n- undecanoate
429
543
653
1207
1307
1407
1206
1308
1406
0.893(S)+ 823.4
Methyl n -undecanoate
Methyl n -dodecanoate
Methyl n dridecanoate
653
748
840
1407
1507
1607
1406
1508
1606
1.069 (S)+ 708.1
Methyl n -tridecanoate
Methyl n 4etradecanoate
Methyl n -pentadecanoate
840
937
1016
1607
1707
1807
1606
1717
1803
1.1 32 (S)+ 652.7
Methyl n -pentadecanoate
Methyl n -hexadecanoate
Methyl n -heptadecanoate
1016
1100
1178
1807
1907
2007
1806
1909
2006
1.234(S)+ 552
Table 3 (Continued)
Series
Products
6
Methyl n -heptadecanoate
Methyl n -0ctadecanoate
Methyl n -nonadecanoate
Methyl n -eicosanoate
t
4
vl
t4
Scans(S) KIA(exp)a KIA(calc)b
1178
1255
1389
2007
2107
2207
2307
2003
21 13
Linear equationsC
KIA (calc) =
1.430 (S)+ 319
2305
a Aliphatic methyl esters have been obtained by methylation of the acid fraction of Cisfus ladaniferus essential oil
from southeastern France (Esterel) (95).
Reported by Jennings and Shibamoto (1980).
Calculated from linear equations: KIA = a . Scans + b according to the M.BASIC. SCAN1 program.
253
To obtain the calculated indices with a better accuracy, it is necessary to work
with a series within 200 to 300 index units because the above relationship is not
linear. If linear alkanes or straight-chain methyl or ethyl esters are present in the
medium, the task is simplified and excellent accuracy is obtained as shown in an
example in Table 3.
3. SELECTED ION MONITORING (SlM)
An approach complementary to the use of Kovats indices, useful especially in
the case of complex mixtures, is the selection of a certain number of characteristic
fragments of a particular compound or a homologous series. This technique is
called selected ion monitoring (SlM). Table 4 shows some of these ions for different
groups of products. Quantitative determinations can be made from calibration
curves.
Hirvi and Honkanen (96) claimed that determination of small quantities of a
substance by MS is possible using the mass fragmentographic selected ion
monitoring (SIM technique) as GC detector. By monitoring ions of one or some few
specific masses of the whole spectrum, a thousand-fold increase in sensitivity
(picograms amounts) and in specificity can be obtained.
The quadrupole mass filter method is more practical, cheaper and easier to use
than the magnetic sector method. They applied this technique to the study of the
major volatiles of berries of the genera Vaccinium and Fragaria growing in
different areas in southern and western Finland. For example, furaneol can be
detected in the berries of strawberry varieties by selecting ions at m/z 142 and 43,
7-1actones at m/z 85, trans-2-hexenal at m/z 42 and 57 and ethyl esters at m/z 88,
43, 60, respectively.
Selected ions for acids and esters upon NCI/OH" were reported by Hendriks
and Bruins (97) (See Table 5).
Vernin et aL (98) used the SIM technique upon El to study the heavy fraction of
a juniper essential oil (See Table 6).
Frey (99) detected synthetic flavorant additives to some essential oils by
selected ion monitoring.
Monoterpenoid and sesquiterpenoid fractions of basil essential oils (PCI/NH3)
can be separated selecting the ions at m/z 137 and 205 (100) (See Figure 1 and
Table 7).
Also SIM (PCI/i-C4H10) makes it possible to separate the main sulfide
components of garlic essential oils by selecting ions at m/z 115, 121,147 and 179
(101,102) (See Figure 2).
Isopentyl alcohol and its esters which are important components of Syrah wines
can be selected by using the ions at m/z 55 and 70 upon El (103) (See Figure 3).
Differences between Kovats indices (DIK) on a polar column (Carbowax 20
M) and a nonpolar column (OV 101) can be determined for these selected ions as
well. These differences are characteristic of a particular group of compounds. This
can be seen from the average values reported in Table 4 (last column).
254
Table 4
Identification of different groups of organic compounds on the basis of ions
obtained by the electron impact.(EI ) method and the Kovats indices on polar and
nonpolar columns*
Compounds
Selected ions
(mass fragmentometry)
KID = (KIP- KIA)
Hydrocarbons
Straight-chain alkanes
Straight-chain alkenes
Monoterpenes (C 10H16)
Limonene
Sesquiterpenes (C 15H24)
57,43,71,85
56,70
93,136
68
93,161,204
0
50+/- 10
160 +/- 50
175
195 +/- 60
Ethers and acetals
Alkyl ethyl ethers
Methylthio alkyl ethers
Dimethyl acetals
Diethyl acetals
Benzyl ethers
Diethylene: glycol monoalkyl ethers
Ethylene: glycol acetals
59,45
47
75,103,47
103,75,47
91,107
45,59
73,45
50 +/- 20
180 +/- 20
200 +/- 10
140 +/- 10
400 +/- 20
600 +/- 20
315
44,57, (M +- 28)
43,58
41,42,55,69
300 +/- 20
310 +/- 30
375 +/- 30
81
425 +/- 30
43,61
43, 71
43,71
265 +/- 30
200 +/- 20
230 +/- 20
Carbonyl compounds
Saturated aliphatic aldehydes
Alkylmethyl ketones
Cis- and trans-aliphatic aldehydes
o~,~-Unsaturated compounds
2,4-Alkadienals
Esters
Alkyl acetates
Alkyl isobutyrates
Alkyl butyrates
255
Table 4 (Continued)
Compounds
DIK = (KIP- KIA)
Selected ions
(mass fragmentometry)
Methyl esters
Ethyl esters
cis-3-Hexenyl esters
trans-2-Hexenyl esters
Linalyl esters
Cyclohexyl esters
Citronellyl esters
Geranyl esters
13-Phenethyl esters
Benzyl esters
Alkyl benzoates
Alkyl salicylates
Alkyl cinnamates
Alkyl anisates
Alkyl anthranilates
7-Lactones
8-Lactones
Alcohols
Secondary aliphatic alcohols
Primary alcohols (aliphatic saturated)
1-Alken-3-ols
Acids
Fatty acids with a straight chain
* G. Vernin, unpublished results.
74,43,87
88,101
82,67
67,82
93,80
67,82
69,82
69,80,93
104
91,108
105,77
120,92
131,103
135
119
85
99
240
240
280
280
270
280
280
340
520
540
485
520
640
680
820
660
680
+/- 20
+/- 20
+/- 20
+/- 20
+/- 30
+/- 20
+/- 40
+/- 40
+/- 40
+/- 50
+/- 40
+/- 50
+/- 30
+/- 30
+/- 30
+/- 30
+/- 30
45
31,56
57
410 +/- 20
455 +/- 20
455 +/- 20
60,73
800 +/- 30
256
Table 5.
Selected ions in NCI/OH of acids and esters in Cannabis sativa essential oil (97)
Selected ions (RCOO')
Compounds
87 (C4H702)"
Butyric acid, isobutyrates, butyrates
73 (C3H502)"
Propionates
115 (CsH 1102) -
Hexanoic acid and hexanoates
101 (C5H902)"
Isovalerates and valerates
59 (C2H302)"
Acetates
129 (C7H 1302)"
Heptanoic acid and heptanoates
99 (C5H702)"
Tiglatesa
a In geranium essential oils (136)
Table 6
A SlM (GC-MS/EI) study of the heavy fraction of a Juniper essential oil (98)
Selected ions
Principal separated compounds
69
(E)-~-Elemene (1081), nerolidol (1654), cadinenes
(1316-1324), (z-cadinols (1825-1877), C 15H24 (1194)
farnesol (2001), spathulenol (1752), a ketone (1427)
J~-Elemene (1080), (z-humulene (1190)
7-Cadinene (1323), Ar-curcumene (1333), germacrene B
(1251), spathulenol (1753), cubenol (1682), C15H22 (1688)
93
119
121
136
154
105,161
189,204
105,161,220
(E)-(z-Cadinol (1877), (Z)-e~-cadinol (1825), 7-muurolene
(1192), (E)-~-elemene (1084)
e~-Terpineol (1209, terpinen-4-ol)-(E)-~-elemene (10741083) etc.
Terpinen-4-ol (1081 )
ec-Muurolene (1271), 8- and 7-cadinenes (1316-1326)
T. cadinol (1808), r
(1877), e~-ferulene (1362)
~-Elemene (1084), germacrene B (1248+1251)
calamenols (2004,2022,1938), spathulenol (1753),
caryophyllene (1592), humulene oxides (1608-1659)
* The corresponding scan numbers are given in parentheses
mlz 137
Monoterpenes
+
1,8-cineole
m/z 205
Sesquiterpenes
Ic'lJO
I
I
I
I
t
I
I
I
-I
1
I
Figure 1 . Comparison of the monoterpenoid (+ 1 ,&cineole)(m/z137) and sesquiterpenoid
(rn/z 205) fractionsof Yugoslavian (A) and Malagasy (B) basil oils obtained by mass
fragmentornetry in positive chemical ionization (1 00).
258
Table 7
Interpretation of compounds selected by ions at m/z 137 and 205 in Figure 1 (100)
Scans
KIP
Compounds
Selected ions m/z 137
221
264
314/317
332
390
454/460
525
1030
1057
1085
1100
1137
1182
122 7
(z-Pinene
Camphene
I~-Pine ne
Sabinene
Myrcene
1,8-Cineole
trans-~-Ocimene
Selected ions mlz 205
1024
1078
1096
1191
1597
1674
1800
1567
1590
1607
1674
1947
2000
2085
cis- (z-Bergamotene
13-Eleme ne
13-Caryophylle
ne
trans- 13-Farnesene
Humulene oxide
C15H24
T. Cadinol
too$
q
2000
1000
52 2 3
Li
2 20
1.35 21
n
L335
I500
500
2 3 2L
131. 2 2
60-
‘1.1
sflo
so0
I
2 3 2L
I .31 2 2
-
60LO
20-
m/z
(Ctl2
=
179
CtI-CHZ-S)2 S
Figure 2. Mass fragmentograms upon PCI (isobutane) of the main sulfide components of a Mexican
garlic essential oil. Selected ions were m/z 115, 121, 147, 179 between scan 0 and 200 (101, 102).
Figure 3. GC/MS(EI) reconstructed aromagram of a Syrah wine (A) and selected ions at m/z 55 (€3)
and 70 (C) for isopentyl alcohol and isopentyl esters (103)
Table 8.
Interpretation of isopentyl alcohol and isopentyl esters selected by ions at m/z 55 and 70
in a Syrah wine (See Figure 3) (103).
Scans
KIP
m/z
Compounds
364
402
579
620
87 1
1093
1185
1478
1614
1100
1180
1250
1280
1445
1547
1640
1840
1927
45,55,70
55,41,42,43,70
traces
70,43,57,55,85
70,43,56
45,433570
70,43,57,55,127
70,43,71,55
87,70,43,55
lsoamyl acetate
lsopentanol
lsoamyl butyrate
lsoamyl isovalerate
lsoamyl hexanoate
lsoamyl lactate
lsoamyl octanoate
lsoamyl decanoate
lsoamyl 3-hydroxybutyrate
262
The differences are a function of the nature of alkyl groups (straight or branched),
for the first members of a homologous series. In the series of linalyl esters,
citronellyl esters, geranyl esters, benzyl esters, and phenethyl esters, the average
value is that for the butyrates. DIK is a constant for straight chains. The highest
values are observed for the formates, lower values for the acetates, and the lowest
values for branched chains. Propionates, valerates, isovalerates, and pivalates are
characterized by an important fragment at m/z 57, tiglates by a fragment at m/z 83,
and hexanoates by a fragment at m/z 99. This means that characteristic fragments
and DIK values allow one to obtain information about a particular group of
compounds.
4. ION TRAP MASS SPECTROMETRY (ITMS)
Ion trap mass spectrometry (ITMS) can be classified as belonging to
quadrupole filters (104, 105). It is a three dimensional quadrupole mass
spectrometer with MS/MS instrumentation.
Among these advanced techniques, only one has become commercially
available. It is a technique called mass selective instability (MSI)(106). Figure 4
shows schematically the principle of the method (107).
Filament
Terminal electrode
Tension
(radiofrequency)
]
\
-$
5=
.,-.
l
=~
~_t~.
~ ~
E~=
~
~.
~ v
t._
Terminal electrode
Electron multiplier
Figure 4. Simplified scheme of ion trap mass spectrometry (107).
The electrodes (circular and terminal) are hyperbolic. Tension in the radiofrequency region (fixed frequency, variable amplitude) is applied to the central
electrode (the terminal electrodes are grounded). All the ions above a certain m/z
are on trajectories which are in the interior space of the electrodes (hence the
name "ion trap"). As the amplitude increases, the ion with a certain m/z become
unstable and are ejected in the direction of the terminal electrodes (107).
263
The sensitivity of the method is increased by addition of helium (p = 102
pascal). The collisions between the gas and the ions augment the efficiency of ion
trapping.
Applications
Adams (108) collected 500 ITMS spectra of volatile compounds in essential
oils. Subsequently, in 1991, he published (109) mass spectra of e~- and J3-cedrene,
thujopsene, cuparene, and widdrol which are the constituents of cedar wood
essential oil. Later, he and Weyerstal (110) compared the ITMS spectra of cis- and
trans-sabinene hydrates with those obtained by the electron impact (El) method.
ITMS spectra generally resemble the quadrupole mass spectra.
5. CHEMICAL IONIZATION TECHNIQUES
Because in many cases it is not possible to obtain molecular mass by the
electron impact (El) technique, since 1966 various authors started using gentler
ionization methods. The theory and application of these techniques were
developed in Harrison's book (111) and in articles published by several authors
(112- 118).
In the electron impact (El) method, the pressure of the source is so low that
molecules can exit without any collisions with neutral molecules. In a chemical
ionization chamber, the pressure is much higher (0.07 to 1.5 torr.). This means that
no ion can leave without a prior collision with neutral molecule(s). These collisions
between ions and molecules can lead to a series of chain reactions (e.g., in the
case of methane).
During this process, a certain amount of energy is transferred to the derived ion.
The result is fragmentation which is milder than with the electron impact (El)
method. In many cases, it is possible to obtain quasimolecular ions (M + H) + and
(M- H)+.
Chemical ionization leads to the formation of derived ions by reaction between
reactive ions and neutral molecules. This phenomenon can be obtained by
positive chemical ionization (PCI) (115) as well as by negative chemical ionization
(NCI) (116).
Positive chemical ionization (PCI)
The gaseous reagents used are mainly methane, isobutane, and ammonia. The
following ions are observed, respectively 9
- with methane: CH5 + (m/z 15), C2H5 + (m/z 29) and C3H5 + (m/z 41)
- with isobutane" C3H 7 (m/z 43), i-C4H 9 + (m/z 57)
- with ammonia: NH4 + (m/z 18), NH4 + . NH3 (m/z 35) and NH4+. 2 NH3 (m/z 52).
264
In the presence of a molecule M, the principal reaction is proton transfer:
f~ t
i-C4H9+
+ M
foH4t
~
i-C4H8
NH4+
+ MH +
NH 3
The reactivity depends on the acidity of the positive ions and the basicity of
neutral molecules. Heats of formation of protonated molecules have been reported
by Lias et al. (118). In addition to these transfer reactions, secondary reactions are
observed:
a) Transfer of a hydride ion from the molecule to the reactive cation (with methane
as the reagent and saturated aliphatic hydrocarbons and cycloalkanes as the
substrates).
b) Alkylation with methane and isobutane, especially with molecules containing
oxygen and nitrogen
c) Association involving hydrogen bonds in the case of ammonia.
NH4 + + M
~
MNH4 +
~ MH+NH3 ~
MH++NH 3
These reactions take place with alcohols, ethers, and esters, but especially with
hydrocarbons. Unusual positive ion reagents in CIMS has been described by
Vairamani et aL (119).
Negative chemical Ionization (NCI)
The hydroxide ion, OH" is most commonly used among the various ions utilized
for this purpose (H', NH2", OH', CH30", CH3S', CN', F', CI'). It is obtained by
electron impact from the mixtures N20/He/H 2 (1 : 1 : 1) or N20/He/CH 4 (1 : 1 : 1)
(120).
a) Proton transfer reactions.
These reactions take place with acids, alcohols, ketones, and alkenes but not with
saturated hydrocarbons.
OH- + RCOOH ~
RCOO" +
OH" + ROH
RO"
~
OH" + RCH2-CI-R'
O
.
~
H20
+ H20
R-C=',CI-R' +H20
HI
O"
265
b) Substitution and elimination reactions.
(A" + BC
X"
+
~
AB + C')
RCOO" + XR'
"I
RCOOR'
- ! ~
X" + R-O-R' -
i
[
R-COX + R'O"
RO"
=
+
XR'
XR + R'O"
H
I
OH" + R-CH-CH2OR ~
H20 + R-CH =CH 2 + R'O"
c) Reactions involving association via hydrogen bonds (CI" ion).
The associated molecules are in equilibrium with associated ions (cf. Eq. 1)"
XHCI"
,,
1
-
X
. HCI
2
~
X" + HCI
However, if the acidity of XH is too low, hydrogen bonds cannot be formed. If XH is
strongly acidic, the associated ion decomposes according to Eq. 2.
The orders of acidities of different molecules and the basicities of the negative
ions have been established (117).
Applications
The positive and negative ionization techniques are mostly used in analysis of
essential oils and aromas. The available information is summarized in Table 9.
Negative chemical ionization (NCI) of unsaturated terpenoid hydrocarbons
gives molecular ions with low intensity because protons are difficult to abstract.
Furthermore, the ion (M - 1) can react with the gaseous reagent nitrous oxide.
Saturated terpenes cannot be ionized by this technique.
The behavior of monoterpenes and some sesquiterpenes of Bulgarian rose oil
in positive chemical ionization (PCI) was studied by Hadjieva et al. (128). In the
case of monoterpenes, o~-humulene, I~-caryophyllene, and J3-bergamotene give an
important fragment at m/z 81. In this case, chemical ionization is not very useful
because El spectra still give the molecular ion. However, certain fragments are
characteristic of a given structure.
Table 9.
Some applications of the positive (PCI) and negative (NCI) chemical ionization methods in flavors and fragrances
Compounds
Methods
Authors (Ref.)
2-Norbornyl derivatives
Alcohols, esters and oxygenated
CI
PCl/i-C4H 0
NCVOHPCl/i-C4H 0
Jelus eta/. (121)
Bruins (122)
PCIA-C~H~O
PCI/CH4, NH3
Budzikiewicz and Busker (124)
Knight eta/. (125)
NCVOHNCVOH-
Houriet et a/. (126)
Hendriks and Bruins (127)
PCI/CH4
Hadjieva eta/. (128)
NCVOHPCl/i-C4H10, NH3
PCVNH3
Bos et a/. (129)
Williams and May (130)
Terpenoids
1,3-Bis (2-chloroethyl)-l -nitrosourea
(an anticancer agent)
Alkenes
Aldehydes, ketones and alcohols (terpenoids)
Alcohols
Valeriana officinalis essential oils
(elemol, valeranone, valeranal and a-kessyl
alcohol)
Bulgarian rose oil (Rosa darnascena Mi//.)
(cis- and trans - linalool oxides )
Essential oils
Aliphatic alcohols and esters
Cyclohexanone
Bruins (122)
Weinkam and Lin (123)
Tabet and Fraisse (131)
h,
o\
o\
Table 9 (Continued)
Compounds
Methods
Authors (Ref.)
Holly essential oil (Ruscus aculeatus)
PCI/NH3
Fellous and George (133a)
Cyclic ethers
NCI/OH-, NH2-
De Puy (133b)
Roman charnomilla
Mixtures (flavorings)
Cyclic monoalcohols (stereoisomeric)
NCVOHCI
PCl/i-C4H10, NH3
George (1133b)
Rapp et a/. (1 14)
Winkleretal. (134)
Hemp (Cannabis sativa L. )
Mentha spicata
NCI/OHNCI/OH-
Hendriks and Bruins (97)
Papageorgiou etal. (155)
Geranium essential oils
PCl/i-C4H10
Fraisse etal. (135)
NCVOHPCI
PCl/i-C4H10
PCI/CH4, NH3
Vernin eta/. (136)
Jalonen etal. (137)
Vernin etal. (100)
Sarris et a/. (138)
PCM-C4Hl o, CH4
PCl/i-C4Hlo
May and Williams (139)
Einborn e l a/. (140a)
Stereoisomeric norbornenyl compounds
Basil essential oils
Saturated and unsaturated cyclic and
aliphatic alcohols
Alkyl esters
Conjugated dienes
Dootlittle eta/. (140b)
Table 9 (Continued)
Compounds
Methods
Authors (Ref.)
Stereoisomeric norbornenols
Miscellaneous terpenoids
An ortho ester from the Mentha piperita
essential oil
Garlic essential oils
PCI
EI/CI
PCI/i-C4H10
Jalonen and Taskinen (141)
White (2)
Koepsel etal. (142)
PCl/i-C4H10
Vernin etal. (101, 102)
Eniantomeric menthols
NCVOHCI
Tabet (143)
Unsaturated alcohols
Terpenoid and nonterpenoid esters and
alcohols
Bicyclic sesquiterpene alcohols
Humulene diepoxides
lsoborneol and isobornyl acetate, and
essential oils
Munson etal. (144)
Lange and Schultze (145)
PCVNCI
PCVNCI
PCl/i-C4H10
Madhusudanan et a/. (146)
Lam and Deinzer (147)
Bruins (117)
NCVOH-
Bruins (117)
Table 9 (End)
Compounds
Methods
Authors (Ref.)
lsopulegol isomers
Terpenoid and nonterpenoid esters
Volatile esters and phenylpropanoids
Thermal degradation of Amadori intermediates
PCl/i-C4H10,NH3
CI
P C M - C ~ H ~NH3
O,
PCIA-C~H~O
Lange and Schultze (148)
Lange and Schultze (148b)
Lange and Schultze (154)
Vernin eta/. (149)
PCI
PCIA-C4Hl
PCl/i-C4H10
PCl/i-C4HI 0,NH3
PCI/CH4
NCI (N20/CH4)
Badjah etal. (150)
Vernin eta/. (98)
Schultze etal. (151)
Schultze eta/. (152)
Zupanc etal. (1992)
Carceles (161)
a-and P-Pinene derivatives
Juniper essential oils
Sesquiterpene hydrocarbons
Sesquiterpene alcohols
Non alcoholic beverages "Cokta"
Sulfur components of chive
(Allium Schoenoprasum)
270
Reactive gases i.e. (CH3)3CH, NO, CH3NH 2 add to double bonds of olefins
giving rise to chraracteristic fragmentation patterns (124).
A number of papers have been devoted to the NCI behavior (OH-, NH2" ) of
cyclic ethers by De Puy et aL (132). In this case the gas phase with amide and
hydroxide ions undergoes a ~-elimination which prevails with retro-aldolization
reactions.
PCI mass spectra of terpenoid alcohols (borneol, e~-terpineol, linalool,
citronellol) were reported by Knight et aL (125). The PCI/NH 3 give best results than
the PCI(CH4) mode. Tables with numerical values of mass spectral data obtained
using the NCI technique (OH-) were published (117). The values include the data
for 13 esters (formates, acetates, isovalerates), 13 terpenoid alcohols, 8
sesquiterpenoid alcohols, 2 phenols (anethole, eugenol), 4 oxides and 2 terpenoid
ketones (carvone, pulegone). This technique is of special interest for (M - H)-ions
from alcohols and esters because the PCI technique (isobutane) does not always
give the (M + H) + ions. NCI/OH" is particularly suitable for the identification of
esters of essential oils (127). Esters are cleaved by an apparent nucleophilic
displacement reaction with the formation of RCOO" ions. In the case of alcohols,
the RO ion is usually the base peak. More or less prominent loss of a molecule of
water from the (M - H)" ion in stereoisomeric carveols suggests their configuration
(117).
PCI of alkyl esters were studied by May and Williams (139):
1) CI(MS)in general produces (MH) + ions indicative of the molecular weight,
(RCH2COOH2) + and (RCH2CO) + ions are indicative of the acid moiety, and
(R) + ions indicative of the alcohol moiety.
2) When methane is used as reagent gas:
a) in general, mass spectra are more complex than when isobutane is used as
reagent gas.
b) for methyl and ethyl esters the (MH) + ion diminishes the MS.
c) in esters other than methyl and ethyl, the (RCH2COOH2) + ion dominates the
spectra and the (MH) + or (RCH2CO) + is the second major ion.
d) as the acid and alcohol moieties increase in chain length, the (MH) + ion
decreases and fragments from the alkyl chain (RCH2) +, (R) +, and (R - H) +
assume greater importance.
e) (RCH2CO2C2H5)+and (RCH2CO2C3H5) + are found with propyl and higher
esters.
f) low levels of (M-C2H5)+ and (M-C3H5)+ were found with all esters.
271
g) the presence of iso- radicals in either the acid or alkyl moiety increases the
contribution from the iso- radical. It also decreases the relative contribution from
the (RCH2COOH2) + ion.
3) When isobutane is used as reagent gas:
a) the (MH) + ion is the major ion and in general (RCH2COOH2) + is the next
major ion.
b) as the acid and alcohol moieties increase in chain length, the relative
intensities of the (RCH2)+, (R) + and (R-H) + ion increase.
c) the presence of iso- radicals increases the relative contribution from the isoradical.
in terpenoid esters, nucleophilic displacement with OH" ions gives carboxylate
ions making it possible to obtain information about the acid in the ester.
The PCI mass spectra of twenty alcohols (saturated and unsaturated, cyclic and
aliphatic) at difforent operating pressures and source temperatures using methane
or ammonia as reagent gas wore studied by Sarris et al (138). The best conditions
for determining the molecular woight of these alcohols wore low temperatures (i.e.
90~ associated with a high pressure of the gas (0.3 mbar). These conditions
favored the formation of the major ion (M -17) + and (M - 1)+ when methane was
used as reagent gas, and (M - 17) +, M+ and (M + 18) + when ammonia was the
reagent gas. For determining M.W., ammonia is the preforred reagent gas.
Similarly, Lange and Schultze (148b) insist on the importance of the proper choice
of experimental conditions with respect to the type of reagent gas, the reagent gas
pressure and the ion source temperature. They reported mass spectra of neryl and
geranyl acetates at two difforent (PCI/NH3) pressures (0.15 and 0.70 mbar),
respectively. At
the higher pressure, the base peak occurs at m/z 214
corresponding to the fragment (M + NH4) +.
In general, the PCI technique has been more widely adopted for analysis of
carbonyl campounds than the NCI method, and it has been used for sulfur
compounds in garlic oil (101,102).
In the NC! technique, correct analysis of unsaturated monoterpenes is
complicated because of the secondary reactions between (M - H)" ions and nitrous
oxide.
It is also wortwhile to mention Lange's and Schultze's works (145, 148, 154)
using PCI (with isobutane and ammonia) on terpenoid alcohols, isopulegols, and
sesquiterpene hydrocarbons and alcohols (151,152).
Upon PCI/i-C4H10 sesquiterpene hydrocarbons give an intense
quasimolecular ion (M + 1 ) + (151 ).
The extent of fragmentation is different for the various compounds, permitting a
certain structural classification within this group. Sesquiterpene alcohols and
272
esters may yield similar spectra with isobutane impeding unequivocal classification
of this class of compounds. With ammonia a typical set of ions (M + 1)+, (M + 18) +
and (M + 35) +, respectively, is produced. According to their specific mass numbers,
these ions unambiguously characterize the loss of hydrocarbons. The molecular
mass is obtained from the species (M + 1)+ and/or (M + 18) +.
Twenty-one sesquiterpene alcohols were investigated by Schultze et aL (152)
using isobutane and ammonia as the reagent gases.
The PCI/i-C4H10 mass spectra show intense fragment ions (M + H - H20) + at
m/z 205 for all substances. These compounds possess tendency to form different
further fragments. This behaviour reveals structural characteristics to a certain
degree.
Only three compounds display quasi molecular ions (M + H) + with relative
intensity > 10%. Therefore, isobutane is not well suited to confirm the molecular
mass and classification of this group of compounds is not unambiguous. With
ammonia a typical set of ions is formed: (M + H - H20) +, (M + NH 4 - H20) +, (M +
NH4) + and (M + NH 3 . NH4) +, respectively. These ions make it possible to confirm
the molecular mass and to classify this group of compounds.
A number of papers have been devoted to the volatile components of essential
oils" Bulgarian rose, Roman chamomilla, hemp (Cannabis safiva L.) Valeriana
officinafis L., geranium, basil, juniper, garlic, and chive, as well as to the various
flavorings (mixtures) and the products of the Maillard reaction (See Table 7).
As an example, let us examine the behavior of cis- and trans-linalyl oxides
present in many essential oils and natural flavors.
Positive chemical ionization (PCI) of these compounds with methane (128)
shows a difference between the intensities of the (M + H - H20) + ion which is 55%
for the cis-isomer and 30% for the trans- isomer. This can be explained as due to
different chemical stabilization of the above cation. The cation derived from the cisisomer is more stabilized mostly because of the ion-dipole interaction between the
positive charge and the vinyl group. In the case of the trans-isomer, the vinyl group
is more distant from the positive charge.
cis
trans
Figure 5. Configuration of the (M + H - H20) + ions based on the cis- and translinalyl oxides (128).
273
The same phenomenon is observed with the positive chemical ionization (PCI)
using isobutane (135). In the case of cis-isomer the respective ion (M + H - H20) +
represents 71% as compared to 34% for the trans- isomer (cf. Figures 6 to 9).
With negative chemical ionization (NCI/OH'), the ion (M - H - H20)" at m/z 151
is the base peak of the cis -isomer, it is 86% in the case of the trans-isomer. In both
cases, also the M - H)" ion at m/z 169 is present.
It can be assumed that the (M - H - H20 )" ion is formed in three steps :
a) Attack of the OH-ion on the proton with respect to the heterocyclic oxygen,
leading to the formation of a carbanion (cf. Scheme 1).
H20
9
,
--.--
H
H
.)
OH T,
I (-OH')
H
._ I
OH~--'~
H
(M-H-H20)
m/z 151 (86-100%
Scheme 1. Formation of the ( M - H - H20 )- ion by negative chemical ionization
(NCI, OH') of cis- and trans-linalyl oxides.
b) Elimination of the O H ion from the tertiary alcoholic group.
c) A new attack of the OH" ion on the two protons of the methylene group in the
ailylic position and the formation of the anion (M - H - H20)" stabilized by
conjugation.
274
Hem " CIS-LI~LOOL ~IDE (E.I.)
Fo~ute 5ru'te' C18 !!18 02
(P.M. = 178 )
Origine ' ( ; E ] ~ ] ~ ECYPT;L/EI~I]N et al. ,P~RF. COS'I.A]~I1ES,52,51-~l, 1983.
]Ka : 1878
IXp - 1428
I)I~ : 3.~
Refe~nce :
8:
8
:00
~0"
!11 i4r
i i j i i: i i i i ' i
~" i i I J i
0
20
~0
(;0
$0
100
1~0
'.~,0
i 1 I"I
J.~O
180
t i i t t
200
~20
2~0
Spectre en impact elec~:ronique ' 59(188)
43 (48), 68 (34), 94 (34), 41 (28), 111 (28), 55 (25), 67 (25),
81 (28), 83 (14), 53 (8), 79 (8), 77 (2), 137 (2), 155 (2)
2~0
93 (25)
MoR ' (Z)-LII~LOOL OXIDE (F)
Fot~ule b~'Le ' C18 H18 02
(P.M. = 178 )
Or i 9 i ne ' 3~I~OSA(RE~1011); UL~NI Met a 1., J. EssElfr. oIL RESEARCH,1991, 83--97.
II(a- 1888
IKp- 1448
DIK- 368
Reference8:
8
'~176
~i
i
20
i!'l !
,
~O
60
30
100
~
6. El m a s s
spectra
of
,,
I~.~)
Spec~,re en ~pac~ electronJque ' 5g(ll~)
43 (57), 94 (45), 41 (38), 55 (27), 68 (22),
155 (3), 137 (2)
Figure
.
i I i i i I t I i I i I I
i I i i i i'l
cis-linalool
93 (22),
oxides.
67 (15), 111 (14)
275
Horn' CIS-LIN~LOOL ~l])g (P.C.I./i~4H18)
Fomule brute'C!8 HI8 02
(P.M. = 178 )
Origine ' GE]~MI~ E~/~;UERMIN et aI.,P~E.CO~.~R~ES,SX, SI-GI, 1983.
lHa : !878
IKp : 1428
])IX : ~58
.qe,~e~ce :
8"
8
.J
+)"
ii7%$
!liL"
;0.
~'
I
0
i
20
1
10
i
" t
i
gO
%0
l
i '(
I
'-~0
:20
i"t
I
I'+O
~ i
i
~.gO
i
I~O
i
-~90
i
i
-~20
i
-'i'
...~0
2gO
Gg(188)
G7 (85), 153 (G8), 79 (42), 81 (42), 85 (37), 83 (28), 95 (17), 91 (14)
93 (11), 94 (11), G5 (8), 189 (8), 111 (8), 135 (8), 154 (8), 13l (2.)
Motto ' CIS-LIRALOOL OXIDE (N.C.I./OH-)
Fomula b~%e' C18 H18 02
(P.H. = 178 )
Origiae ' GXR~MILII E~PI';UERNIN e% al.,PRRF.CO~.AR~ES,52,51-fiI,1983.
IKa : 1878 IKp- 1428
DIK- 358
Re~erence8:
8
"-)0 ,
~oJ.
l
"i
4
~iiilii
0
20
I
!I
~I
9
i ~i i i i I i ;'i
10
$0
80
~.~)0
'-20
'.~0
'i i I i k i ~ i i
'.$0
'.%0
"00
220
X~,O
2~0
151(188)
57 (65), 87 (28), 66 (22), 81 (28), 71 (14), 69 (11), 133 (8), 169 (8)
85 (5), I~ (5), 111 (5), 135 (5), 149 iS)
Figure 7. PCI/-i-C4H 10 and NCI/OH" mass spectra of cis-linalool oxides.
276
I ~ ' ~-LIRALOOL ~IDE (E.I.)
Fomule bru~e ' C18 H18 02
(P.M. : 178 )
Ori ine ' ~F.]~/~]~ E~PT;UE~IN et aI.,PRRF.CO~I.~OffE~,S2,51-61,1983.
!.~a = 1888
] Xp : 1~.58
DI ]( = 378
Refe~nce :
8:
8
:oo
?
.,)
t",I!1i
,m
I
i
20
;
I ~
~O
(0
eO
~O0
ii
1~0
!
:
i
1~0
t
r
i
i
1 I
I~0
1SO
i
I
200
i
i
2~.0
J i
2~0
250
Spectre en impact electroni ue ' 59(188)
43 (45), 55 (37), 68 (37), 41 (31), 93 (28), 94 (28), 111 (28), 67 (Z2)
81 (28), 83 (14), 53 (8), 79 (8), ?7 (5), 119 (2), 137 (2), 155 (2)
Nm ' ( E ) - LIi~LOOL O~IDE (F)
Fomule brute' C18 H18 02
(P.M. = 178 )
Ori ine ' JI~I1ROSA(REUNION);UE]~IIII et al.,J.ESSEiIT.OIL RES~ROI,1991, 83-97.
l](a : 1878
IKp : 1418
DIK : 348
Reference 8:
8
20~
~7
~ " I I i I
0
20
~0
t
~0
L~
i I' I']
~0
i' t i Jl
I~O
'-20
'~.0
t~!
i I I i 1 I I F I
IgO
~80
8pecire en impact electronique ' 59(188)
43 (83), 94 (48), SS (3G), 41 (34), 93 (3B),
39 (15), 81 (15), 69 (11), 137 (3), 155 (3)
F i g u r e 8. El m a s s s p e c t r a of
trans-linalool
200
68 (24),
oxides.
-"-"0
2~0
2r
6? (2B), 111 (28)
277
Horn' TPa~IS-LI~LOOL ~IDE (P.C.I./i-C~H18)
Fomule brute" C18 H18 02
(P.M. = 178 )
Origine ' (;ERf~I~ E~;UE~HIH et al.,Pt~.CO~.n,qOl~ES,fi2,51-61,1,~
DI~ = 378
]:~e~nce =
8:
8
l~a = 1~88
I~p = 14E8
t
:!
,
".|g!,
oi
i'i
0
67 (88),
91 (11),
i
i j i J
.~
qO
95 (11),
~$g
~ i
*0
79 (@),
"it::ti
i
90
i
iO0
I
J
t20
~ 1
:~0
1 }
}
1~0
:~0
t
i
200
1
I
220
i
I
2,0
69( I88 )
81 (@), 85 (31), 153 (34), [83 (25), 65 (11), 71 (11)
93 (5), 94 (S), 189 (2), 111 ((2), 135 (2), 137 (2)
Horn ' TRA~-LIHALOOL OXIDE (M.C.I./OH-)
(P.M. = 178 )
Fomule brute ' C18 H18 02
Origine ' GERAHII~I E~PT;UERHIH et aI.,PARF.CO~.~IifflES,52,51%l, 1983.
l}(a : 1888
l~p = 1458
])l]( - 378
Reference 8:
B
B7
tO0
.o
'~0"
Ii
t
, 6~
!
'.~3
,'"1
I
t~
I
~!t
,[
"-3 ~ i i i i ( i i i I ~ i i
57(188)
151 (85), 66 (54), 71 (25), 133 (22), 69 (17), 169 (14), 85 (11), 111 (11)
99 (11), I]1 (8), 1117(S), 189 (5), 149 (S), 135 (S)
F i g u r e 9. P C I / i - C 4 H I o
and NCI/OH"
mass
spectra of trans-linalool
oxide.
278
One would not expect too much of a difference between the ions formed from
the two isomers, except for the easier approach of the OH" ion in the case of the
cis- isomer in the first step of the process.
It is surprising that in the NCI (OH') spectra of these two isowers published by
Bruins (117), the (M - H) ions are the base peak, even though the (M - H - H20 )"
ion represents only 3 - 5%.
As additional examples, the El and PCI/NH 3 mass spectra for 2-acetylfuran
(Figure 10), 5-methylfurfural (Figure 11 ), 2-acetylpyrrole (Figure 12) identified in the
thermal degradation of the Fruc-Valine Amadori intermediates (103) and several
alkylpyrazines (Figures 13 to 21) are given. These mass spectra have been
reconstructed from our SPECMA data bank written in Turbo Pascal (version 6)
using Pentium 90 (Intel processor)).
The NCI (N20/CH4) technique was recently applied in a study of sulfur
components of a chive extract (161) obtained from an aqueous crushed extract
treated by ultrasound at room temperature. The NCI spectra of sulfur components
do not give the quasi-molecular ion (M-H'). As for diallyl disulfide in garlic, the 1prop-enyl disulfides in chives give a very abundant ion at m/z 72.
In the PCI method, 2-acetylfuran and 5-methylfurfural give the (M + H) + ion and
an adduct cation at m/z 128, (M + NH4)+ which is the base peak for 2-acetylfuran.
This ion is very weak in the case of 2-acetylpyrrole (m/z 127).
All alkylpyrazines (PCI/NH3) give the quasimolecular ion (M + H) + as the base
peak. We have also observed a weak peak at m/z (NH4 + + 2 NH3) and an
important fragment at m/z (NH3) which is not reported in the reconstructed mass
spectra.
The El, PCI/i-C4H10 and NCI/OH" mass spectra of sulfur derivatives found in
garlic essential oils have been published elsewhere (101, 102). Allylmethyl
disulfide and diallyl disulfide, the most abundant constituents of these oils are of
interest. Their mass spectra are shown in Figures 22 to 24.
In the case of NCI, the quasi molecular ions (M - 1) are not observed. The most
intense peak at m/z 105 in diallyl disulfide corresponds to the loss of allyl alcohol
from the anionic molecule or of allyl carbene from the quasi molecu lar ion (See
Scheme 2).
The base peak at m/z 72 in the two mass spectra arises from the homolytic
cleavage of the S - S bond in the quasimolecular ions leading to an allylthio
radical and a radical anion which cyclizes to give the more stable mesomeric form.
in the PCI method, the quasimolecular ions are the base peaks.
279
M,om : 2-~CL"rYLFUP~N(E.I,)
Formule br~ate : C6 H6 02
I~a :
8S~5
(P.M. = 118 )
!~p : !SIS
DI}( :
6.?.1]
~nce
:
B'
B
i
,?9
2
,
i!
o'
0
i
i
20
i !i
I
qO
I ~ ~
i
(0
$0
i"
~00
'
-
:~0
~
:~0
'
i
I
:~0
i
:~0
Spectre en i~pact; electronique ' 9S(!BB)
111] (39), 39 (21]), 43 (17), 96 (5), 67 (4),
0
ZOO
i
I
i
229
i'
2~0
2;0
11 (3)
Morn : 2-RCETYLFURAR(P, C, I./RIO)
Formule 5ru'~e : C8 H6 02
(P,M. = 111] )
Origine : FRUC.-~JALIME(~ADORI):UERMIMet aI.,B,S,C.F.,4,681-694,1987.
IKa = 895
IKp : 1515
DIK = 6ZB
~s
=
B:
B
"I
~o
l
~i' i'l
0
20
I
1
I
~0
111 (%), 111] (IB),
i
60
1
,
80
95 (B)
I
,
"00
,
;
:.'0
',
~
i
i
:
!
,
I
;
I
,
i
::0
128(188)
F i g u r e 10. El a n d P C I / N H 3 m a s s s p e c t r a of 2 - a c e t y l f u r a n .
280
Morn' ~-tI,ETHYL~RFUR~L (E. I. )
F:r~ule 5rute ' C6 H6 02
(P.ff. : I'6 )
O~i~ine ' FRL~.~LiMF. (~,~Rl);'J~!ll et aI.,B.S.C.F., 4,681-694,1987.
DiX : 6-~8
Ref~,,'enc~ :
9:
8
IRa : 9~.B IR?- ~,~B
'~: ,~
~
i
i
I
i' i"i
i
i
i
i
i
r i
Spectre en ~pact electronique ' 118(188)
189 (87),
53 (q8),
57 (15),
39 (13),
51 (11),
i
~ !
i
41 (19),
'
i
~ :
81 (8),
188 (2)
Nora ' 5-tI~HYLFIER,,FURAL(P, C. I./~3)
For~le brute ' C6 H6 02
(P.M. : 118 )
Origine ' FH~.-UALINE(AIIAI)OHI);UERHIM et aI.,H.S.C.F.,4,681-694,1987.
Oil( : 641]
~ference :
8:
8
l]{a : 9q8
IKp : 1588
oo!
I
20"
o
i
0
i
20
i
i
i
~0
12H (?B), 52 (18)
i i
60
i
$0
i
I
'~OO
I
I
~.O
;~.0
:r
"80
200
220
2~0
2fO
111(1B9)
Figure 11. El and PCI/NH3 mass spectra of 5-methylfurfural.
281
M:~ : 2-AC~tL ,i~,,RC.LF. (E.I.)
Fomule bru'Le ' C6 H? H I 0 1
(P.M. = 189 )
Or !gtne9 ' F~Uf..--v,vi
"^' ~E (:~RDOI~I);~JERMiMet. aI . B.S.C.}'.
.
. 4,b81-694,1987
.
I](a : !858
iKp : 1938
l)ll( = 888
Rcqerence =
B:
8
'.0.}
~~
iI.
i
t i ~ ! ~ .' I ! i i i i 11
Spectre en hpac'l; elec'l:ronique ' 94(188)
1B9 (82), 6G (57), 39 (3?), 43 (15), 118 (6),
Nor, ' 2-ACk"~LI~, RROLE(P.C. I./MH3)
Fomule brute ' C6 H? ~! 1 0 1
53 (6),
67 (6)
(P.M. = 189 )
Ori9ine ' FRUC.~ALIME(~'IAI)ORI);UI~RHIMet aI.,B.S.C.F., 4, b81-G94,1987.
l](a = 1858
ioo
l](p : 1938
])l]( :
888
{
i
~s
=
8:
B
;
'~
:o
I
I
I
i
94 (2B), 127 (8),
Figure
12. E l a n d
t
~
I
i
t
i
!
:
t
'
i
f
~
t
~
}
t
t
118(188)
52 (7), 189 (S)
PCI/NH
3 mass
spectra
of 2-acetylpyrrole.
282
,%m : ME'!~YLI'YR~I~ (E.I.)
For~u]e bru~ : C5 H6 ~
(P.M. - g4 )
Origine ' FRUC.-UnLI~ (M~ILL~RD)' UERMIMet aI,B.S.C.F.,4,[email protected]%9~,I987.
II(a : 818
II(p : 126B
DII( : 458
Re~erence:
8"
8
i
l
"t
I0
4
5D
I~
')i
i
i
t
i
i
I
,
,
~
,
.
~
;
;
;
Spec'Lre en impact elecieonique ' 94(188)
67 (5B), 39 (2B), 53 (19), ,t2 (15), 43 (14),
~
~
,
i
95 (11),
i
41 (1B),
BB (4)
Morn : tI~HYLI'YR~IHE (P. C. I./NH3)
Fo~ule b~te : C5 H6 M2
(P.M. : 94 )
Ori9ine : FRUC.-~JALIMU~ADORI):UERMIMet aI.,B.S.C.F., 4,6BI-694,198?.
II(a : B18
:oo
II(p : 1268
Oil( : ' ] 5 8
Re~eeence:
8:
8
i
J
:~
!
I
i i I I i i I i i i i I ii
20
52 (6)
~0
60
$0
I00
I~0
i ~ i i ~ i i i i
1~0
1~0
150
200
-'20
~.0
~r
95(188)
F i g u r e 1 3 . El a n d P C I / N H 3 m a s s s p e c t r a of m e t h y l p y r a z i n e .
283
: 2,5--DINETIfi'~AZiME (E.I,)
Fomule b~'ute ' C6 H8 M2
(P,II, = 188 )
O,',!~ine : FHUC.-L:~LI,"IE(.,-11aDOHI);'JE~!Met al. ,H.S.C.F., u 681-694, I.o87.
IRa :
918
l}(p : IL?.B
])11( :
qlB
Reference:
8:
8
:OOT
!l'
t
I
.:,~
:~
~',
i"
i ~ 'I" ~ t
0
.~0
.tO
t
i I ~i i'
r
.~0
100
:20
i i
:~0
I i ', i ! i i
"t;O
Spec+.re en impact elec+.roniclue ' a,2(188)
1BB (97), 39 (33), B1 (12), 189 (7), 52 (5),
66 (2)
"gO
200
41 (4),
i I
2~0
~*'0
43 (3),
2~;0
BB (3)
Horn : Z,5--DlrlETHYL~P,RZIME (P,C.I./~3)
(P.M. : 188 )
For~ule brute: C6 H8 M2
Origine : FRUC.-~JAL]ME(AII~ORI);UERM]MeL aI.,B.S.C.F.,4,691-694,198?.
DIK = 418
P,ef'erence =
8:
8
I]{a = 91t]
II(p = 1321]
. .'rt'.
'-3"3 -
~0"-
20
I
0
I
20
I
i
'~r
1
1
~0
i
I
~0
i
i
i
;00
i20
i i ! t , i i i
"~0
L~O
;~0
200
i i
220
t i
2:0
2~0
109(180)
52 (3)
F i g u r e l 4. El a n d P C I / N H 3 m a s s
spectra
of 2 , 5 - d i m e t h y l p y r a z i n e .
284
:I1
,
I
1
I
0
?O
10
.:
-.,
j
I
I0
80
:QQ
:tP
150
IT0
110
:3L
121
110
160
Spectre en impact electronique : 18888(188)
42 (961, 48 (571, 39 (481, 67 (81, 189 (81, 41 (61, 66 (51, 52 (31
187 (31
Hm :
2,6-DIHETWLPI~INE (P.C.I.RR13)
Fomule brute : Cb H8 ti2
(P.H. = 189 1
Origine : FRUC.-VILIHE(#1ADOAI):VEININ et al.,B.S. C.F. ,4,681-b84,1987.
IKa = 895 IHp = 1325
DIK = 4 3
Reference = 8:
8
52 (41
189(lea 1
Figure 15. El and PCI/NH3 mass spectra of 2,6-dimethylpyrazine.
285
Horn: Z,3-DItlL31~LPY,.~IHE (P.C.I./HH3)
Fo~ule breie : CG H8 h?,
(P.M. = 188 )
Or;,gine ' FRUC,-UAL!HE(~'I~DORI);UE~IH ei aI.,B.S.C.F.,4,681-694, !987.
l~a : 988
IKp : 1348
Dll( = 4~8
ReFerence =
8:
8
J
52
~0"
I
~
I )-I
0
:,0
i ~' '~"|
1.0
60
i ~ i"','l
A.O
:,)0
')"~
~ ! "'i,
:.')
:~,3
:~0
i i i t 1 i '1"'
:*.0
2)0
:~0
~0
~0
189(188)
_
52 ('t8), 182 (14)
H~ ' 2-~THYL-3,S(or3, 6)-DIM~/LIX/Pd~ZI HE (P. C. I./HH3)
Fomule brute ' C8 HI2 H2
(P.M, = i36 )
Origine ' FI~C.-UALIHE(~I~I)ORI ),UEI~IIM eL aI.,B.S.C.F.,4,681-694,1987.
IKa : 18fi8
I](p : 1478
DI}( : 418
Reference =
8:
8
1 i
0
i' I
20
i
i'
IO
52 (281, 138 (18)
~ i "!
gO
~0
i
'i
IO0
i
i
:20
1 t
'.~,0
i
I
t
J,$O
I
I
200
i
i
~0
"1 I
210
I
I
250
137(188)
Figure 16. PCI/NH 3 mass spectra of
2-ethyl-3,5
i
igO
(or 3,6)-dimethylpyrazines
2,3-dimethylpyrazineand
286
Mo~ : TRIMETH%I~'R~I,NE(E,I.)
~,,'i~i~e FRUC.-/P.i.iHE<~:(:..~':U~')HIIIe! ~I 8.8 C.F.,a.,6BI-69a.,.,~?.
ll~p- 1~89
DIN = 480
Re.;'erence =
O:
8
iHa = 1888
i
"1
40"
I
>
i
i
i
i
I
,,.l
;
J
i
i
~
:
,
i. . . . .
Spectre on i~pac~ electronique ' 4Z(188)
122 (G4), 39 (21), 81 (15), 48 (12), 43 (G),
66 (Z)
~
;
'
-
$3 (6), 123 (S),
i
88 (2)
Non' TRIMPI'HYLPYR~IHE (P.C.I./NH3)
Fo~ule brute ' C? H18 N2
(P.M. : 122 )
Origine ' FRUC.-UALINE(~ADORI);UERM!M et al.,B.S.C.F., 4,581_694,1~?.
I ~ = 1888
I,~p = 1"!88
Dig = 488
Reference =
8'
9
)0
+0"
~' i i i ) t
0
20
,
~0
i i"i
~0
i i i I i i
30
!00
1 7. E l a n d
i" ~':
'.~0
'.~0
i ~ i' I ' i i... ~ ]
180
~00
~0
2+0
~0
123(169)
$2 (6)
Figure
',~0
!
PCI/NH3
mass
spectra
of
trimethylpyrazine.
287
For~uie 5r,~%e : C,7 HIB H2
IXa = 995
(P.PI. = 122 )
D]}( = 398
]Xp = 1385
~eFe,,'ence=
8
8:
.,I.&,.
|
:o"
!,i +) ,"~+
9
Spec~:t'e
.
en
9
,
i
~pac'l;
,
i
!
i
,
i
elec%,"onique
:
I
,
"
i
"
i
!
i
.
~
' 121(188)
122 (75), 39 (3G), 56 (24), 9,t (15), 42 (1']), 54 (11), 48 (18), 41 (7)
66 (4)
Ro~ ' Z-~L-5~L~B~I
hE(P. C. I./HH3)
For~le bru%e ' C7 H10 R2
(P.M. = 122 )
Origine ' FIIUC,~ALIHE(AIIADOR] );UERH]H e% aI.,B.S.C.F. ,4,681-694,19117.
IKa- 995
IXp- 1385
Dig- 398
ReFerence-
8'
8
~0"
-H
'i
I
'~
t"
i
i
~
i
52 (12), 121 (9)
!
i
i
'
I
i
:
i
J
~
i
i
1
t
i
t
i
I
i
~
123(188)
Figure 18. El and PCI/NH 3 mass spectra of 2-ethyl-5-methylpyrazine.
288
Nora : 2-1SOI]UTYL-5-t~,I~lX/R~ZI HE (E. I. )
Fo~ule brute : C9 HJ.4 H2
(P.M. = 4.58 )
Origine'FR~.-URLtNE (~..~DOi]I);~JERff!M eL aI,B.S.C.F.,4,~!-69~,1987.
~,00_
.,;~.
I
..,., -,
t
~
,
I
0
i
;
20
i" i
1t0
i
i
gO
1
i
~.0
I
i
I~O
"29
~,
:
l
i
I
I
I
i
'
Spectre en impact electroniclue '. 188(188)
39 (28), 135 (14), 41 (11), 42 (9), 158 (9), 189 (8),
88 (2)
i
i
,
66 (7), 149 (4)
Rm : Z-ISOBU~,L-5--I1D}h'Lh'Rt~]ME {P.C.I,/HH3)
Fo~ule brute: C9 H14 H2
(P.M, = 158 )
Otigine : FRUC.-~JAL]HE(~ADORI);UERN]Net a|. ,B.S.C.F.,4,681-694,1987.
IRa =
8
l](p = 1498
DIi( =
B
Reference =
8:
8
',09
~0"
~0
i
'~
~52
I
0
152 (1B),
t
:20
I
I
40
t
'
g0
i
]
80
52 (9), 188 (8)
2
i
:00
~
.
120
I
i
I.~..1
.
I
;E;0
i
I
180
i
I
~'.00
i
I
~.20
I
i
2,0
,~go
151(181])
F i g u r e 19. El a n d P C I / N H 3 m a s s s p e c t r a of 2 - i s o b u t y l - 5 - m e t h y l p y r a z i n e .
289
Morn ' 2- I SOBLITYL-H(or 6 ~'IL"]'HYLI'YR~IME (E. I. )
F•mu!e brute' C9 H14,42
(P.M. : 158 )
I
,~
'2
O~'ig~ne ' ~'RI'C
,.,~,.. ' ! T l
C.F ,4,GBI-594,:.BT.
IRa :
B
!X~ : [email protected]
DI]( :
8
Re~e,,'ence:
8:
8
: . ) ) ,,
i
'~
)i
i
i
1
L
i
I
i
i
I
'
j
',
i
i
I
I
I
i
i
I
Spectre en ~mpac~ e|ec+.ronique ' 188(188)
135 (13), 39 (13),
94 (3), B8 (2)
I
67 (18), ,t3 (9), 189 (B), 158 (5), 149 (4),
i
:
93 (4)
Norm ' 2-ISOBb'l~L-3(or 6)-METH%PYR~INE (P. C. I./NH3)
For~ule brute ' cg H14 M2
(P.M. : 1SB )
Ori9ine ' FROC.-~3ALINE(~L~DORI);VERNIN et al.,B.5. C.F., 4,681-694,1987.
I}(a :
8
I}(p : 15B8
DII{ :
B
Reference:
B:
8
"3) i
,~
07
i
i I ! i l' I i i i i !:i' i i i" I1~2 i I l"i
20
152 (11),
110
52 (9),
~;0
90
188 (9)
:00
:20
',~,0
:~0
1~0
i i i i t i
200
220
2~0
151(188)
Figure P.O. El and PCI/NH 3 mass spectra of 2-isobutyl-3 (or 6) rnethylpyrazine.
290
,,-.-~ov~;~,-,,,5(or
3,o,'-~,ur,~nzLr~,'~,"~',',r, ( E , | )
Fomule b r u t e ' C18 H16 H2
(P.M. = 164 )
Oeigine ' F~I~.-'JAL]~ (~r'~DORI);UE~.HiN et al. :8.S.C.F., ~.,~.91-69a., 1987.
I~a :
8
Z]~.~ : 1538
DiK :
8
?.es
:
8:
8
~,) -
i
.~,~
!! . . . .
0 '
i
i
1
i
'
,,
,
;
:,
,
,
i
z
,
I
i
i
t
i
i
I
i
-'r
0
Spectre en impact electeonique ' 122(181])
42 (19), 39 (17), 149 (11), 123 (!8), 121 (7), 164 (6), 163 (4), 187 (3)
188 (3), 66 (2), 67 (2), 94 (1), 95 (1)
M~'
2-1SOB~L-3,5Cor
3,6~DIM~L~R~IHE
Fomule brute ' C18 H16 H2
(P.M. :
(P,C,I,/HH3)
164 )
Oeigine ' L~UC.-~JALINE(~ADORI);UERMiN et aI.,B.S.C.F.,4,6B1-694,
] Ka 8
I Kp - 1538
DI ]( 8
Refez'ence B:
lgBT.
B
,.,),)
r
,~i
~
)
L
o
i
I
:.o
I
i
~o
E
i
~o
~
i
so
166 (11), 122 (8), 52 (3)
~
, ~
:oo
'
,
::o
'
:.,o
~
i
i ~
i
:~o
:~o
i
i
.'oo
i
]
=:o
i
i
~:o
i
:~o
165(188)
Figure 21. El and PCI/NH 3 mass spectra of 2-isobutyi-3,5 (or 3,6)
dimethylpyrazine.
291
Nora ' ALLYLtlET~L DISULFIDE (E,I,)
Fomule br~te : C4 FIB $2
(P,li, : !2B )
Orioine : GARLICE.O.;UZP,~IM et aI,,PLAMT.,fl~ . , I ~ 6 , 9 6 - ! B I ,
IXa :
8
!Rp = ~38
DI~ =
8
Re.~erence=
B:
q
..]
'
!,,I
i i i k i
i i
i
~ I i' i i
i
i i i
~~
en ~pacs elec~ronique ' 41(:[B8)
128 (51), 39 (24), 44 (17), 53 (13), 72 (6),
i
i i ~i-
i
?8 (6), 121 (3), 122 (3)
Morn ' ALLYI/1ETHYLDISULFIDE (P.C.I./i-C4HIB)
FoPmule brute : C4 H8 S2
(P.M. -- 128 )
Origine : GARLICE.O.;UERNIN et aI.,PLAIfrA MEI).,19B6,9fr1111.
I Xa =
8
I](p = 1238
DI]( =
8
Reference=
8:
'~176
~
I
i
I
l
I
I
I
!21] (2B), 73 (21]), 176 (B),
i
I
,
I
~
i
l
I
I
I
I
i
i
i
~
i
I
121(1BI])
55 (4)
Figure 22. El and PCI/i-C4H 10 mass spectra of allylmethyl disulfide.
292
Hm ' DI~LL~L DISULFIDE (E.I.)
Fo~ule brute ' C6 HiB $2
(P.M. : 1~.6 )
Ori9ine ~RLIC E.O. 'UE~IIN a ML:TZGF~,E.O. ~ W~.XES,S P R I ~ L~L~,1991, 99-138.
ira = IIZ8
I~(p = 1436
PI~( : 316
ib~erence =
8'
8
'
',,)o
.:' c,-
T~)-
l
~.).-
~o
I,!!
l
t
I
1
,
[
1
I
i
i
t
~
t
i
t
I
I
i
~0
Spectre en impactelec'Lronigue
'
41(188)
BI (14), 113 (7), 146 (5),
45 (19),
I
~00
85 (5),
73 (6),
;
i
i
220
i
~0
2~3
79 (5), 1B3 (3)
Nero ' DIRLLYL DISULFIDE (P.C.l./i-Cah118)
Fomule brute ' C5 H18 $2
(P.M. : 146 )
Origine ' G;~RLiC E.O. ;UE~IN ~ ML'~GEJLE.O.a ~ES, SPRIHGF~ UERL~G, 1991, 99-138.
]Ka : 1128 I~(p- 1436
DIX - 316
Res
:
8:
B
"-~)0
,:
:,0"
%0"
I
o
.
!:'3
I
0
i
~ "i
i
20
~0
60
il
i"i
i' i
80
!00
r
~ i
:20
lg
i
!
I~0
t
,,
]
I
:GO
147(188)
73 (19), 146 (15), 43 (11), 283 (11), 115 (11),
45 (7), 185 (7), 89 (7), 148 (7), 91 (3)
Figure
i
I
~.$0
i' ~ i
200
55 (11),
I
2-~0
i
~0
i
2~0
~ (11), 149 (11)
23. F:l and PCI/i-C4H 10 mass spectra of diallyl disulfide.
293
Num' t%LLYIXIETHYLDISULFIDE(N.C.I./OH-)
Fomule beute ' C4 H8 $2
(P.ti. = 128 )
Oeigine ' GARLICE.O.;UERNIN et aI.,PLANTA tlED.,1986,9fr181.
IRa =
8
Igp = 123B
DI R =
8
?,efeeence -
8:
7
o,j j,,~,j
0
79 (68),
I
1
20
qO
60
73 (58),
i
$0
I
100
44 (58),
I
i
i
IXO
t
I' I
~-~0
:60
i
I
i~O
t
;
i
~.00
i
~0
I
I
~0
~:0
72(188)
47 (24)
' DlfiLLYL DISULFIDE(N.C.l./~-)
Fomule brute' C6 H18 SZ
(P.M. = 146 )
Oeigine ' GARLIC E.O.;UERNIN & ~GE,q,E.O.& ~/~KES,SPRII4GER UERL~G,1991,99--I~.
b
IKa = 1128
IKp = 1436
DII = 316
Refeeence =
B'
B
'~176
-~I
~
73
i
0
;
20
i
i"'i
110
t
t
60
~0
........
I 4 1 i
%00
:20
i
i
~.'+0
i
~"~
160
i 'i
I~0
i
~00
i
i
I
::0
2'+0
t
_i
260
72(188)
47 (77), 73 (58), IB5 (45), 75 (36), 44 (31)
Figure 24. NCI/OH" mass spectra of allylmethyl disulfide and
diallyl disulfide.
294
CHz=CH-CHz-S-S-CH-CH= H2 ~
CH2=C"m-CH 2-S-S-CH=CH-CH z
-HOfH
CHz=CH-CH 2- St']- S_C'~H:CH.~'H ~ ~
,CH-CH-CH 2.,.CH2=CH-CH 2- S-S'-
CH2=CH-CH2-S-S:HO-
CH
CH
/%
CHz=CH-CH2-S~- S . ~ / CH ~
[(1L5) !
CHz,-CH-CH2-S:-
CH
CH2
-:CH2
CH2=CH-CH2-S
/%
+S
\/
-CH=CH-CHz:------,~ ;S-CH=CH-CHz:-
lci- il
I721
Scheme 2. Formation of ions at m/z : 145, 105, 73 and 72, of diallyl disulfide
NCI(OH') (101,102).
Miscellaneous
techniques
Various coupling techniques in mass spectrometry have also been described,
in particular:
- Capillary GC-MS and capillary GC-FTIR upon El and CI (156, 157)
- MS/MS (104)
- SIMS stable isotope (13C) mass spectrometry applied to o~- and 13-ionones
extracted from raspberry fruit, wine and raspberries (158).
- MDGC - IRMS (multidimensional gas chromatography (159). Isotope ratio mass
spectrometry (159).
All these techniques have considerably enlarged the potential of mass
spectrometry in the analysis of flavors and fragrances (160).
295
6.
CONCLUSIONS
Quasimolecular masses obtained by positive chemical ionization and negative
chemical ionization complement the electron impact data. These techniques make
it also possible to obtain information about the geometry of stereoisomers such as
carveols, isopulegols, and linalyl oxides.
Negative chemical ionization (NCI, OH') is better suited for analysis of
terpenoid alcohols and esters, whereas positive chemical ionization (PCI,
isobutane) is a technique used for carbonyl compounds, ethers, and sulfur
compounds.
Positive chemical ionization (ammonia) is also a good analytical method for
essential oils, especially for their monoterpene and sesquiterpene components.
Also the chromatographic data (e.g., Kovats indices) should be regarded as an
additional means of identification.
Acknowledgements
The authors wish to thank Mrs G. M.F. Vernin and Mrs. R.M. Zamkotsian for their
technical assistance, the Central Analitical Services of Lyons for the mass spectra
obtained by the chemical ionization techniques, and Pr. J. Metzger for fruitful
discussions.
7.
10a
REFERENCES
Maarse H, Belz R eds. Isolation, Separation and Identification of Volatile
Compounds in Aroma Research, Akademie-Verlag, Berlin: 1981.
Witte L. In: Linskens HF, Jackson JF, eds. Modern Methods of Plant Analysis
Springer Verlag, Berlin: 1986; 3: 134-145.
Masada Y, Analysis of Essential Oils by GC-MS, Wiley, New York: 1976.
Lawrence BM, Essential Oils, Allured Publ. Corp, Wheaton IL:1976-1978;
1979; 1980; 1980-1987.
Van Straten S ed. Volatile Compounds in Food, Central Institute for
Nutrition and Food Research TNO, Ultrechtseweg 48: PO Box 360, The
Netherlands,1977; 1979;
Walradt JP. In: Theimer ET, ed. Fragrance Chemistry: the Science of the
Sense and Smell, Academic Press, New York: 1982; 575-615.
Vernin G, ed. The Chemistry of Heterocyclic Flavouring and Aroma
Compounds, Ellis Horwood Publ. Chichester: England, 1982.
Srinivas SR. Atlas of Essential Oils, Wiley, New York: 1986.
Maarse H, Visscher CA, 6th ed. Volatile Compounds in Food. Qualitative
and Quantitative Data, TNO-CIVO Food Analysis Institute, Zeist: The
Netherlands, 1989; 1:2: 3:
Maarse H, Boelens MH. In: Bessiere Y, Thomas A.F. eds. Flavour Science
and Technology, The TNO Data Base, Volatile Compounds in Food: Past,
Present and Future, Wiley J. and Sons, New York: 1990; 275-278.
296
10b
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Maarse H, Visscher CA, eds. Willemsens LC, Boelens MH, co-eds. Volatile
Compounds in Food. Qualitative and Quantitative Data, An Essential Guide
for Trade and Industry, TNO Nutrition and Food Research, Ultrechtseweg 48
PO Box 360, Zeist: The Netherlands, 6th ed. 1989: 3rd Suppl. 1992.
Reineccius GA. Instrum Methods Qual Assur Foods 1991; 45: 67-81.
Mosandl A. J Chromatog 1992; 624: 267-292.
Schmid JP. In: Bessiere Y, Thomas AF eds. Flavour Science and
Technology, Wiley J. and Sons, New York: 1990; 239-254.
Ryhage R, Von Sydow E. Acta Chem Scand 1963; 17:2025-2035.
Thomas AF, Willhalm B. Helv Chim Acta 1964; 47: 475-488.
Von Sydow E. Acta Chem. Scand 1963; 17: 2504-2512.
Von Sydow E. Acta Chem. Scand 1964; 18:1099-1104.
Von Sydow E. Acta Chem. Scand 1965; 19: 2083-2088.
Von Sydow E, Anjou K, Carlson G. Archives of Mass Spectral Data,
Swedish Institute for Food Preservation Research Fack, Gothenburg, S.
1970.
Hirose I. Shitsuryo Bunseki 1967; 15: 162-168.
Hayashi S, Sato H, Hayashi N, Okude T, Matsuura T. J Sci Hiroshima 1967;
31 : 217-231.
Yukawa Y, Ito S. eds. Spectral Atlas of Terpenes and the Related
Compounds, Hirokawa Publ. Co. Tokyo: 1973.
Moshonas MG, Lund ED. The Flavour Ind 1970; 375-378.
Jennings W, Shibamoto T eds. Qualitative Analysis of Flavor and Fragrance
Volatiles by Glass Capillary - Gas Chromatography, Academic Press, New
York: 1980.
Enzell CR, Ryhage R. Ark Kemi 1965; 23: 367-399.
Enzell CR. Tetrahedron Letters 1966; 2135-2143.
Enzell CR. Ark Kemi 1967; 26: 425-434.
Enzell CR, Ryhage R. Ark Kemi; 1967; 27: 213-229.
Enzell CR, Wahlberg I. Acta Chem Scand 1969; 23: 871-891.
Enzell CR, Appleton RA, Wahlberg I. In: Waller GR, Dermer OC eds,
Terpenes and Terpenoids, Wiley Interscience. New York: 1972; 311-406.
Enzell CR, Wahlberg I. In: Waller GR, Dermer OC eds, Terpenes and
Terpenoids, Wiley Interscience. New York: 1980; Suppl 1:311-406.
Enzeil CR, Wahlberg I, Ryhage R. Mass Spectrom Rev. 1984; 3(3): 395-438.
Tucker AO, Maciarello MJ. In: Charalambous G. ed. Spices, Herbs and
Edible Fungi, Elsevier. Amsterdam: 1994; 34: 605-656.
Mass Spectral Research System. 1974. Users' Manual; US Environmental
Protection Agency, Office of Planning and Management. 1974.
Middledich BS. ed. A Guide to Collections of Mass Spectral Data. American
Society for Mass Spectrometry. 1974.
Stenhagen E, Abrahamson S, McLafferty FW. eds. Registry of Mass
Spectral Data, Wiley and Sons, New York: 1974.
Cornu A, Massot R. Compilation of Mass Spectral Data. Heyden and Sons
Ltd. London: 1966; 1967; 1971;1975.
Heller RS, Milne GWA. EPA/NIH Mass Spectral Data Base, US Government
Printing Office, Washington: DC 1978.
297
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
Heller SR, Milne GWA. EPA/NIH Mass Spectral Data Base, US Government
Printing Office, Washington: DC 1980; 1983; 1987.
Craveiro AA, Alencar JW. 178th ACS National Meeting, paper 70 (AGFD)
Div. Washington: DC 1979.
Craveiro AA, Rodrigues AS, Andrade CHS, Matos FJA, Alencar JW,
Machado MIL. 8th International Congress Essential Oils, Publ. Fedarom,
Cannes-Grasse: 1982; 241-254.
Petitjean M, Vernin G, Metzger J. Ind Aliment Agric 1981;9: 741-751.
Petitjean M. These Sciences, Universit~ d'Aix-Marseille 3,St-J~rSme: 1982.
Petitjean M. Vernin G. Metzger J. In:.Charalambous G, Inglett G eds.
Instrumental Analysis of Foods: Recent Progress. Academic Press. New
York: 1983; 1: 97-124.
Vernin G. Petitjean M. Metzger J. Parf Cosm Ar6mes 1983; 51: 43-52.
Vernin G. Petitjean M. Poite JC, Metzger J, Fraisse D, Suon KN. In: Vernin
G, Chanon M. eds. Computer Aids to Chemistry. Ellis Horwood Publ. Ltd.
Chichester: 1986; 7: 294-333.
Vernin G, Petitjean M, Metzger J, Fraisse D, Suon KN, Scharff C. In:
Sandra P, Bicchi C. eds. Capillary Gas Chromatography in Essential Oils
Analysis. Huethig Verlag, Heidelberg: 1987; 287-328.
Mass Spectrometry Data Centre: Eight-peak Index of Mass Spectra. The
Royal Society of Chemistry Nottingham: 1983.
Mass Spectrometry Data Centre: Eight-peak Index of Mass Spectra. The
Royal Society of Chemistry (3rd ed.) Nottingham: 1986.
Finnigan Mat, Library Compilation (38.000 entries) San Jose CA : 1984.
Hewlett Packard. HP 59943A/PBM Data base. Copyright Hewlett Packard
and Co: 1988.
Chien MJ. In: Lawrence BM, Mookherjee BD, Willis BM. eds. Flavors and
Fragrances: A World Perspective. Elsevier. Amsterdam: 1988; 923.
Wiley and Sons, CD Rom System, DC Rom System for Mass Spectral Data
New York: 1988.
De Brauw N, Bouman J, Tas AC, La Vos GF. Compilation of Mass of Volatile
Compounds in Foods. Central Institute for Nutrition and Food Research,
TNO Zeist, The Netherlands: 1981.
De Brauw N, Bouman J, Tas AC, La Vos GF, Gromberg LG. Compilation of
Mass Volatile Compounds in Foods. Central Institute for Nutrition and Food
Research, TNO Zeist, The Netherlands: 1988.
McLafferty FW. Wiley/NBS Registry of Mass Spectral Data. 4th ed. Wiley J.
New York: 1988.
McLafferty FW. Stauffer DB. eds. The Wiley/NBS Registry of Mass Spectral
Data. Wiley J. and Sons. New York 1988; idem. J. Wiley Interscience Publ.
New York: 1989.
Henneberg D, Weimann B, Joppek W. MPI Library of Mass Spectral Data.
Max Plank Institut fur Kohlenforschung. M01heim Germany: 1989.
National Institute of Standards and Technology. NIST/EPA/NIH Mass
Spectra Data Base. PC Version 4.0, NIST, 62.500 entries Gaithersburg Ma:
1992.
298
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75a
75b
76
77
78
79
80
81
82
83
84
85
86
87
National Library of Standards. Mass Spectra. US Department of Commerce
National Bureau of Standards NBS. National Standard References Data
Systems NSRDS, EPA/NIH Mass Spectral Data Base. 1992: 1:- 4.
Mc Lafferty FW. Wiley Registry, Benchtop/PBM Version 3.0, Palisade Corp.
Newfield, New York: 1992.
Umano K. Fragrance J 1992; 20(1): 27-32.
Budzikiewicz H, Djerassi C, Williams D.H. Mass Spectrometry of Organic
Compounds Holden-Day Inc: 1967.
Budzikiewicz H. Mass Spectrom Rev 1991; 10(6): 519-522.
Merrit L. Mac Even CN. eds. In: Practical Spectroscopic Series. Part A.
Mass Spectrometry. Dekker M Inc. New York: 1979: 3.
Porter ON, Baldas J. Mass Spectrometry of Heterocyclic Compounds Wiley
Interscience, New York: 1971.
Lange G, Schuitze W. In: Brunke J ed. Progress in Essential Oils Research
De Gruyter and Co, Berlin: 1986; 597-617.
Kovats E. Helv Chim Acta 1958; 41: 1915-1932.
Van den Dool H, Kratz PD. J Chromatog 1963; 11: 463-471.
Albeiges J, Guardino X. J High Resolut Chromatogr Chromatogr Commun
1985; 8(6): 301-302.
Shibamoto T. In: Sandra P, Bicchi C. eds. Capillary Gas Chromatography in
Essential Oils Analysis, Huethig Verlag, Heidelberg: 1987; 259-274.
Ettre LS. Chromatographia 1973; 6(11): 489-495.
Haken JK. Adv Chromatogr New York 1976; 14: 367-407.
Saeed T, Redan G, Sabdra P. J High Resolut Chromatogr Chromatogr
Commun 1979; 2: 75-76.
Alencar JW, Craveiro AA, Matos FJA. J Nat Prod 1984; 47: 890-892.
Alencar JW, Craveiro AA, Matos FJA, Machado MIL. Quimica Nova 1990;
13(4): 282-284.
Perry NB, Weavers RT. J Chromatogr 1984; 284: 478-481.
Yamada H, Harada K, Nakamura S, Mihara S. J S C C J 1987; 21 (1): 30-35
Libbey L.M. J E O R 1991;3: 193-194.
Teranishi R, Buttery RG, Stern DJ, Takeoka G. In: Bessiere Y and Thomas
AF eds. Flavour Science and Technology Wiley J and Sons New York:
1990; 267-278.
Kovats E. In: Giddings JC, Keller RA eds. Advances in Chromatography
Dekker M New York: 1965; 119-127.
Wehrli A, Kovats E. Helv Chim Acta 1959; 42: 2709-2736.
Andersen NH, Falcone MS. J Chromatogr 1969; 44:52-59
Andersen NH, Syrdal DD. Phytochemistry 1970; 9: 1325-1340.
Andersen NH, Syrdal DD, Lawrence BM, Terhune SJ, Hogg JW.
Phytochemistry 1973; 12: 827-833.
Andersen NH, Bissonette P, Liu CB, Shunk B, Ohta Y, Tseng CLW, Moore B,
Huneck S. Phytochemistry 1977; 16(11): 1731-1751.
Erdey L, Takacs J, Szalanczy. J Chromatogr 1970; 46: 29-32.
Majlat P, ErdSs Z, Takacs J. J Chromatogr 1991; 91" 89-103.
299
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
Swigar AA, Silverstein RM. In: Aldrich Chem Co, Monoterpenes, Infrared,
Mass 1H NMR and 13C NMR Spectra and Kovats Indices, Milwaukee Wl
1981.
Sadtler Research Laboratories. The Sadtler Standard Gas
Chromatography Retention Index Library Division of Bio-Rad Laboratories
Inc. Philadelphia: 1985.
Ramaswami SK, Briscese P, Garguillo RJ, Von Geldren T. In: Lawrence BM,
Mookherjee BD, Willis BJ. eds. Flavors and Fragrances" A World
Perspective, Elsevier, Amsterdam: 1988; 951-980.
Davies NW. J. Chromatogr 1990; 503: 1-24.
Mihara S, Masuda H. J Chromatogr 1987; 402: 309-317. J Agric Food
Chem 1988; 36: 1242-1247.
Ames JM, Elmore JS. Flavour and Fragrance J 1992; 7: 89-103.
Boniface C, Vernin G, Metzger J. Analusis 1987; 15: 564-568.
Guy I. DEA, Facult~ des Sciences Marseille S-J~r6me, 1993. Contrat
ANVAR J 92 07063UJJ.
Hirvi T, Honkanen E, Instrumental Analysis of Foods Academic Press Inc.
New York: 1983; 85-95.
Hendriks H, Bruins AP, Biomed. Mass Spectrom 1983; 10(6): 377-381.
Vernin G, Metzger J, Suon KN, Fraisse D, Ghiglione C, Hammoud A,
Parkanyi C. Lebensm wiss u Technol 1990; 23: 25-33.
Frey C. In: Lawrence BM, Mookherjee BD, Willis BJ, eds. Dev Food Sci
Flavors and Fragrances: A World Perspective, Elsevier, Amsterdam: 1988;
18: 517-524.
Vernin G, Metzger J, Fraisse D, Suon KN, Scharff D. Perfumer and Flavorist
1984; 9:71-86.
Vernin G, Metzger J, Fraisse D, Scharff C. Planta Medica 1986; 2: 96-101.
Vernin G, Metzger J. In: Linskens HF, Jackson JF. eds. Modern Methods of
Plant Analysis New Series Essential Oils and Waxes. Springer Verlag
Berlin: 1991 ; 12:99-130.
Vernin G, Boniface C, Metzger J, Fraisse D, Doang D, Alamercery S. in:
Charalambous ed. Frontiers of Flavors Elsevier, Amsterdam: 1988; 655685.
Busch KL, Glish GL, McLuckey SA. Mass Spectrometry/Mass
Spectrometry Techniques and Applications of Tandem Mass
Spectrometry VCH Publ. New York: 1988; 21.
HObschmann HJ, Schubert R. In: Brunke EJ ed. Progress in Essential Oil
Research De Gruyter and Co, Berlin 30: 1986; 643-648.
Stafford GC Jr, Kelly PE, Syka JEP, Reynolds WE, Todd JFJ. Int J Mass
Spectrom Ion Proc 1984; 60-85.
Lauris JN, Cooks RG, Syka JEP, Kelley PE, Stafford GC, Todd JFJ. Anal
Chem 1987; 59(13): 1637-1685.
Adams RP. Identification of Essential Oils by Ion Trap Mass Spectrometry
Academic Press Inc. San Diego: 1989.
Adams RP. In: Linskens HF, Jackson JF. Modern Methods of Plant Analysis
New Series Essential Oils and Waxes. Springer Verlag, Berlin: 1991; 12:
159-173.
300
110
111
Adams RP, Weyerstahl P. J E O R 1992; 4: 197-200.
Harrison AG. Chemical Ionization Mass Spectrometry CRC Press Inc. Boca
Raton FLO: 1983.
112
Munson MSB, Field FH. J Amer Chem Soc 1966; 88(12): 2621-2630.
Richter WJ, Schwarz H. Angew Chemie Int Ed Engl 1978; 17: 424-439.
113
114
Rapp U, Hoehn M, Kapitzke C, Dielman G. In: Baltes W ed. Recent
Developments in Food Analysis Verlag Chemie Gmbh, Weinheim New
York: 1982; 162-169.
115
Westmore JB, Auladdin MM. Mass Spectrom Rev 1986; 5: 381-465.
116
Budzikiewicz H. Mass Spectrom Rev 1986; 5: 345-380.
117
Bruins AP. In: Sandra P, Bicchi C. Capillary Gas Chromatography in
Essential Oil Analysis, Huethig Verlag, Heidelberg: 1987; 329-357.
Lias SG, Liebman JF, Levin RD, J Phys Chem 1984; 13: 695-808.
118
Vairamani M, Mirza UA, Srinivas R. Mass Spectrom Rev 1990; 9: 235-258.
119
Smith ALC, FH Field. J. Amer Chem Soc 1977; 99(20): 6471-6483.
120
Jelus BL, Dalrymple DL, Michnowicz J, Munson B. 1978; 13: 163-166.
121
122
Bruins AP. Anal Chem 1979; 51:967-972.
123
Weinkam RJ, Lin HS. Anal Chem 1979; 51(7): 972-975.
124
Budzikiewicz H, Busker E. Tetrahedron 1980; 36: 255-266.
Knight DW, Knight ME, Rossiter M, Jones A, Games DE. Adv Mass
125
Spectrom 1980; 8B: 1116-1126.
Hourriet R, Stahl D, Winkler. Environm Health Perspect 1980; 36: 63-68.
126
Hendriks H, Bruins AP. J Chromatogr 1980; 190: 321-330.
127
Hadjieva P, Sandra P, Stoianova-lvanova B, Verzele M. Riv Ital EPPOS
128
1980; 62: 367-372. 1lth IUPAC Int Symp on Natural Products. In: Marekov
N, Ognyanov l eds. Izd BAN, Sofia: 1978; 2: 464-468.
Bos R, Bruins AP, Hendriks H. In: Kubeczka KH ed. Atherische 5le Analytik
129
Physiologie Zusammensetung, George Thieme, Stuttgart: 1982; 25-32.
Williams AA, May HV. J Inst Brew 1981; 87: 372-375.
130
131
Tabet JC, Fraisse D. Org. Mass Spectrom 1981; 16: 45-47.
De Puy CH, Beedle EC, Bierbaum VM. J Amer Chem Soc 1982; 104: 6483132
6488.
133a Fellous R, George G. Parf Cosm Ar6mes 1981; 41: 43-45.
133b George G. Parf Cosm Ar6mes 1982; 43: 37-40.
Winkler FJ, G(Jlacar OF, Buchs A. Anal Chem Symp Ser Recent Dev Mass
134
Spectrom Biochem Med Env Research 1983; 8: 307-317.
Fraisse D, Scharff C, Vernin G, Metzger J, 9th Int Congr Essential Oils
135
Singapore: 1983; Essential Oils Technical Paper 1983; Vol 3: 100-120.
Vernin G, Metzger J. Parf Cosm Ar6mes 1983; 52" 51-61.
136
Jalonen J, Taskinen J, Gildewell C. Int J Mass Spectrom Ion Phys 1983; 46:
137
243-246.
Sarris J, Etievant PX, Le Quere JL, Adda J. In: Adda J. ed. Progress in
138
Flavour Research 1984 Proc 4th Weurmann Flavour Research Symposium
Elsevier, Amsterdam: 1985.
May HV, Williams AA. In: Adda J ed. Progress Flavour Research 1984 Proc
139
4th Weurmann Flavour Research Symposium Elsevier, Amsterdam: 1985.
301
140a
Einhorn J, Virelizier H, Gemal AL, Tabet JC. Tetrahedron Letters 1985; 26:
1445-1448.
140b Doolittle RE, Tumlinson JH, Proveaux A. Anal Chem 1985; 57: 1625-1630.
141
Jalonen J, Taskinen J. J Chem Soc Perkin Trans II 1985; 1833-1838.
142
Koepsel M, Krempel A, Surburg H. In: Brunke EJ ed. Progress in Essential
Oil Research De Gruyter W Publ Berlin: 1986; 241-248.
143
Tabet JC. Tetrahedron 1987; 43: 3413-3420.
144
Munson B, Feng TM, Ward HD, Murray RK. Org Mass Spectrom 1987; 22:
606-609.
145
Lange G, Schultze W. Flavour Fragrance J 1987; 2: 63-73.
146
Madhusudanan KP, Srivastava R, Kulsrhestha DK, Fraisse D. Org Mass
Spectrom 1987; 22: 368-372.
147
Lam KC, Deimzer ML. J Agric Food Chem 1987; 35: 1987.
148a Lange G, Schultze W. In: Schreier P. ed. Bioflavour 87 De Gruyter W Publ
Berlin: 1988; 115-122.
148b Lange G, Schultze W. In: Schreier P ed. Bioflavour 87 De Gruyter W Publ
Berlin: 1988; 105-113.
149
Vernin G, Metzger J, Obretenov T, Suon KN, Fraisse D. Proc 10th Int Congr
Essential Oils Fragrances and Flavors Washington DC: 16-20 Nov 1986. In:
Lawrence BM, Mookherjee BD, Willis BJ. eds. Flavors and Fragrances: A
World Perspective Elsevier, Amsterdam: 1988; 999-1028.
Badjah Hadj Ahmed AY, Meklati BY. Flavour and Fragrance J. 1988; 3: 13150
18.
151
Schultze W, Lange G, Schmaus G. Flavour and Fragrance J. 1992; 7: 5564.
152
Schultze W, Lange G, Schmauss G. J E O R 1992; 4(5): 497-510.
153
Zupanc M, Prosek M, Dusan M. J High Resolut Chromatogr 1992; 15(8)
510-513.
154
Lange G, Schultze W. Org Mass Spectrom 1992; 27(4): 481-488.
155
Papageorgiou VP, Arguriadou N, Kokkini S, Mellidis AS. Chem Chron
1983; 12: 27-35.
156
Smith SL. J Chromatogr Sci 1984; 22: 143-148.
157
Herres W, Kubeczka KH, Schultze W. In: Brunke J ed. Progress in Essential
Oil Research De Gruyter W and Co Berlin, New York: 1986; 507-528.
158
Braunsdorf R, Hener U, Lehmann D, Mosandl A. Dtsch Lebensm Rundsch
1991; 87(9): 277-280.
159
Nitz S, Weinreich B, Drawert F. J High Resolut Chromatogr 1992; 15(6):
387-391.
160
Torrent A. Analusis Magazine 1992; 20(7): 66-70.
161
Carceles M.F, Thesis Sciences, Nice-Sophia Antipolis University 1995.
This Page Intentionally Left Blank
D. Wetzel and G. Charalambous(Editors)
Instrumental Methodsin Food and BeverageAnalysis
9 1998 Elsevier Science B.V. All rights reserved
303
GAS CHROMATOGRAPHIC TECHNOLOGY IN ANALYSIS OF
DISTILLED SPIRITS
Kevin MacNamara I and Andreas Hoffmann 2
'Irish Distillers Limited, Bow Street Distillery, Smithfield, Dublin 7, Ireland
2Gerstel GmbH, AktienstraBe 232-234, 45473 Mtilheim an der Ruhr, Germany
INTRODUCTION
Aged distilled spirits constitute a complex mixture of some hundreds of flavour compounds in
an ethanol-water matrix. These flavour compounds, or congeners, originate from the original
raw materials and the subsequent processes of mashing, fermentation, distillation and ageing,
which together produce the final product. Complex interdistillate reactions between congeners
are also an important part of the flavour scenario, and this is particularly important during the
ageing phase in oak barrels.
Tables I and 2 outline the principle congener classes found in whiskey, cognac and rum together
with an indication of compound numbers by class [1]. The naturally occurring relative
concentrations of these compounds can vary from high mg/l to low ng/1. Each congener in turn
has an odour intensity and flavour contribution which is determined not only by its concentration,
but also its unique sensory threshold value. For this reason the most abundant congeners may be
the most amenable to analysis, but the resultant quantitative information may have little relationship
to the perceived flavour of the spirit. On the other hand this type of information does describe the
gross structure of the spirit and is an important data base for ensuring process continuity and
product authenticity.
Chemical Class
Hydrocarbons
Hydroxyl compounds
Carbonyl compounds
Carboxylic acids
Esters
Lactones
Acetals / Esters
O-Heterocyclics
N-Heterocyclics
Miscellaneous
TOTAL
Whiskey
Cognac
Rum
7
53
24
35
79
4
21
7
28
11
15
104
93
43
195
10
43
23
10
10
38
88
63
30
115
7
76
41
23
16
546
497
269
Reproduced with permission from [11
Table 1. Number of volatile compounds, identified in whiskey, cognac and rum.
304
Chemical Class
Aliphatic compounds
Aromatic compounds
Terpenoid compounds
Heterocyclic compounds
TOTAL
Whiskey
Cognac
Rum
187
42
5
35
382
79
52
33
309
88
36
64
269
546
497
Reproduced with permission from [1]
Table 2. Number of volatile compounds, identified in whiskey, cognac and rum.
In an interesting study Salo [2] prepared a synthetic whiskey using purified individual congeners
in accurately known amounts, and with levels based on gas chromatographic data obtained from
a real whiskey. Odour threshold values were used to determine the relative odour intensity of
individual aroma compounds, and aroma fractions of alcohols, esters, acids, carbonyl compounds
and mixtures of these compounds. The results clearly showed that the contribution of the most
abundant congeners to the total odour intensity is quite low. The alcohol fraction accounted for
70% and the acid fraction 11% of the total congener concentration, yet both together only
contributed 7.5% of the total odour intensity.
In the light of these facts a commercial producer of distilled spirits needs a particular analytical
strategy to service and satisfy various needs, which might include:
- Rapid, high-throughput direct analysis of the abundant congeners for production process
continuity and authenticity information.
- Various sample preparation techniques for enriching trace congeners from the ethanolwater matrix. These can usually be divided into simple enrichment procedures or more
elaborate investigative schemes giving enriched compound classes that also facilitate sensory
information.
- Use of the trace congener information provided by investigative research to devise further
high-throughput analytical techniques that make use of advances in modem GC technology
and detection systems.
The previous discussion constitutes a basis for the thematic content of this chapter which is to
review historical and present chromatographic approaches to distilled spirits analysis, and to
explore the potential and application of modem GC technology in this area. The term "GC
technology" has been chosen to emphasise an instrumental dimension in the sense that very
successful analytical schemes can be constructed if the gas chromatograph is regarded as something
more than an oven and a column. Cold capillary programmable injection technology, and static
and dynamic headspace, and large volume liquid injections that exploit the cold injector can
produce systems for efficient routine analysis of spirits as is, or after minimum sample preparation.
Newer detection modes and technologies can be exploited for similar benefits resulting from
sensitivity and specificity enhancements. Capillary columns of such different coatings, dimensions
305
and phase ratios are now available that sample preparation schemes can be devised to make
maximum use of these variables. The aim should be to tailer this technology to the nature and
relative concentrations of the congeners and on this basis the following areas will be reviewed
and explored.
Direct Injection: Review of various approaches using custom packed columns and modem capillary
columns to profile the most abundant congeners without sample preparation. Use of specialised
capillary injection techniques, especially in conjunction with mass spectrometric target selected
ion monitoring, to extend the range of congeners that can be directly determined in the natural
matrix. Introduction to a cooled injection system with programmable vaporisation as a universal
capillary injection system.
Matrix Removal: Review of many of the approaches developed for this purpose ranging from
simple solvent extraction, to multi-step combined extractive and instrumental schemes, which
can give enriched subtractions with selective congener types. This gives an opportunity in turn to
suit column selection (coating type and phase ratio) and detection strategy to individual congener
groups.
Large VolumeInjection: This is an important concept and technology for the future and has the
potential to provide a natural link between data obtained from research investigations, and the
need to produce rapid routine methods. If the compound contents of 500 pls of a simple extract
can be transferred to the column, it offers the possibility of low-cost routine analysis of trace
congeners. The fact that usually only 1-2 lal of complex extracts are used for analysis also represents
an expensive waste of resources. Therefore compounds indicated as potential flavour contributors
from classical multi-step sample preparation routes could be monitored by automated large volume
injection of relatively simple sample extracts. The success of this technique relies on programmable
injection technology with simultaneous solvent venting and retention of analytes of interest in the
injector liner. The same cooling and venting properties of the programmable injector can be
similarly exploited for large volume injection from either headspace or thermal desorption devices.
In fact the design of these units can be greatly simplified when their operation involves a
programmable injector.
Two Dimensional Chromatography: The separating power of single highly efficient capillary
columns is in many cases not sufficient for complex subfractions and extracts obtained from
distilled spirits. Very interesting selective sulphur and nitrogen traces are available but mass
spectrometric identification of the compounds is often very difficult. The vastly increased separating
power of serially coupled columns offers an efficient solution to this problem. Commercial apparatus
is now available which allows two oven systems to be easily configured for computer controlled
2-dimensional separations, and quickly revert back to two independent gas chromatographs when
not being used for this purpose.
There are many different approaches for the determination of abundant and trace congeners in
distilled spirits. In many cases the apparatus and techniques are unique to the particular group or
laboratory, and it can be difficult for other workers to reproduce the exact scheme adopted. This
chapter also contains some procedures favoured by one of the present authors (K MacN), but a
306
difference is that they are rooted as much as possible in a dialogue with a specialist instrument
manufacturer, and are therefore based on commercially available systems. Finally much of the
discussion and many of the examples centre around whiskey, but the methods and technology
should be applicable to all distilled spirits.
DIRECT INJECTION
The most abundant compounds present in distilled spirits are the fusel alcohols and fatty acid
esters, together with acetaldehyde and its acetal. Fusel alcohols are produced in fermentation
from amino acids via decarboxylation and deamination, while the esters are formed in yeast cells.
Acetaldehyde is the most abundant carbonyl compound and its reaction with the dominant ethanol
produces diethyl acetal, or 1,1-diethoxyethane, at the mg/1 level. Table 3 gives typical amounts of
these compounds found in malt and light blended whiskey. Full malt whiskeys are produced in
traditional pot stills and tend to have upper range levels of these compounds. Lighter whiskeys
are blends of pot still product and whiskey produced in column stills. Therefore depending on the
degree of blending commercial whiskeys can have these major congeners in a concentration
range of single figure mg/1 to greater than 103 mg/1. Rapid separation and quantification of these
compounds is important for reasons ranging from production consistency to market place
authenticity and many different approaches have evolved, usually with the universal flame ionization
detector.
Compound
Acetaldehyde
Ethyl Acetate
Diethyl Acetal
Propanol
iso-Butanol
Amyl Alcohols
Ethyl Caprylate
Ethyl Caprate
Ethyl Laurate
Malt Whiskey
200 mg/l
280
"
49
"
350
"
990
"
1600 "
22
"
32
"
12 "
Blended Whiskey
76 mg/1
29
"
5 "
570
"
450
"
420
"
3
"
10 "
8 "
Table 3. Typical amounts of major congeners found in different types of whiskeys.
Packed Columns. Before the introduction of modern fused silica capillary columns
chromatographers relied on packed columns with their limitations of both low plate number, and
low mass sensitivity at the detector due to solute diffusion in the packing. On the other hand they
have the advantage of low cost, are easy to operate, and can be produced with many specific
phases for different separation problems. Open tubular capillary columns offer much higher
separating power but the availability of columns with unique phases is more limited than with
packed columns. Since the phase selectivity has a major effect on separation certain packed
phases have retained their applicability in distilled spirits analysis.
307
Duncan and Philp [3] used two gas chromatographic methods to directly determine a range of
major congeners in whiskey. Using 5% polyethylene glycol 1500 on a 60-80 mesh support in a
10 ft. copper column they separated the major fusel alcohols and esters. Their second method
involved using the same phase in a 27 ft. column but also incorporating in the column a small
length in which dionyl sebacate was the stationary phase. This allowed the additional separation
of ethyl acetate, diethyl acetal and isoamyl acetate. Brunelle [4] evaluated various stationary
phases and produced a collaborative study which resulted in the adoption of an official final
action of the AOAC. Kahn and Blessinger [5] reported some difficulties in this method and from
their own investigations proposed two alternate methods. One of these allowed determination of
ethyl acetate and fusel alcohols and was adopted as a first action alternate method. The second
method allowed quantification of acetaldehyde and acetal as well as ethyl acetate and the fusel
alcohols, but did not get approval until gas chromatographic instrumentation, in particular oven
temperature control and programming, improved. The current AOAC official methods (16th
Edition, 1995) recommend the following phases and conditions for packed column analysis of
higher alcohols and ethyl acetate in distilled spirits.
23% Carbowax 1500 on Chromosorb W, 60-80 mesh, acid washed. Oven 70~
isothermal.
2% Glycerol and 2% 1,2,6-Hexanetriol on Gaschrom R, 100-120 mesh, non-acid
washed. Oven 80~ isothermal.
Figure I shows a separation of whiskey congeners on this second phase, but using oven temperature
programming for optimum separation.
Analysis Conditions:
Oven: 40~
5~
to 80~
Inj.: 100~ Det.: FID, 150~
Carriergas: 30 ml/min N2
4
Compounds:
1. Acetaldehyde
2. Ethyl Acetate
3. Acetal
4. Ethanol
5. n-Propanol
5. Isobutanol
7. 3-Pentanol
8. 2-Methyl-l-Butanol
9. 3-Methyl-1-Butanol
7
9
1
3
Figurel. Separation of whiskey congeners, packed column.
308
Cabezudo and co-workers perfected the mixed phase approach to congener separation [6]. A
fortran program was devised to find the best combined phases from two to four single phases for
a general purpose congener separation, or best combined phases for separation of specific groups
of congeners. Analyses were carried out isothermally at 50~ and allowed elution and separation
of up to 20 compounds. Di Corcia and co-workers pioneered the investigation and introduction
of modified graphitized carbon black for GC analysis of distilled spirits [7]. They first used
Carbopack B modified with Carbowax 20 M, and trimesic acid as acidic deactivating agent, but
later found that this latter treatment was not necessary if the carbon surface was initially acid
washed. Different Carbowax loadings were used for different congener groups and these columns
also allowed elution of free fatty acids.
Martin [8] described a single procedure, using 80-120 mesh Carbopack B as solid support and
5% by weight Carbowax 20M as liquid phase, which allowed separation of acetic acid and the
isomers of amyl alcohol. These columns are presently available in Silocosteel, which is stainless
steel coated with a deactivated fused silica inner layer, and the manufacturers claim improved
inertness, durability and flexibility compared to traditional glass packed columns [9]. Figure 2
show separation of whiskey and rum congeners on these columns.
Analysis Conditions:
Oven: 70~ 4~
to 150~
Inj.: 200~ Det.: FID, 250~
Carriergas: 20 ml/min N2
0.5 ~1 direct injection
9
_IL
5
Reproduced with permission from [9]
._...~
Analysis Conditions:
Oven: 65~
4~
to 150~
Inj.: 200~ Det.: FID, 250~
Carriergas: 20 ml/min N2
0.5 lal direct injection
2
Compounds:
1. Acetaldehyde
2. Methanol
3. Ethanol
4. Ethyl Acetate
5. n-Propanol
6. Isobutanol
7. AceticAcid
8. activeAmyl Alcohol
9. IsoamylAlcohol
7
9
6 8A
11
Compounds:
.
1. Acetaldehyde
2. Methanol
3. Acetone
4. Ethyl Formate
5. Ethanol
6. Ethyl Acetate
7. n-Propanol
8. sec-Butanol
9. Isobutanol
10. active Amyl Alcohol
11. Isoamyl Alcohol
12. n-Amyl Alcohol
Reproduced with permission from [9]
Figure 2. Separation of whiskey (top) and rum congeners on 2m x 2mm 5% Carbowax 20M
80/120 CarboBlack B column.
309
Finally a phase with a specific selectivity has recently been described [ 10] for monitoring compounds
such as 2-propanol and acetoin in spirits and distilled wines. F i g u r e s 3 and 4 compare separation
of a standard solution of congeners on this column compared to the standard Carbowax 1500
phase. 2-Propanol and acetoin which can be markers for adulteration or oxidation changes coelute
with the dominant ethanol on the standard phase.
EtOH 5
Analysis Conditions:
Oven: 55~
60~
to 145~
Inj.: 200~ Det.: FID, 200~
Carriergas: 15 ml/min N2
1 pl direct injection
3
12
,7
Comoounds;
1. Acetaldehyde
2. Methanol
3. 2-Propanol
4. n-Propanol
5. Ethyl Acetate
6. 2-Butanol
7. Isobutanol
8. n-Butanol
9. Acetoin
10. 2-Methyl- 1-Butanol
11. 3-Methyl- 1-Butanol
12. 4-Methyl-2-Pentanol (int. Std.)
13. Ethyl Butyrate
Reprinted from the Journal of Chromatographic Science with permission of Preston Publications, a division of
Preston Industries
F i g u r e 3. Chromatogram of a standard solution using a 2m x 2mm MFE-Vinicol column.
EtOH+3+9
52
Analysis .Conditions:
Oven: 85~ isothermal
lnj." 200~ Det.: FID, 200~
Carriergas: 15 ml/min N2
1 lal direct injection
13
10+11
Compounds:
1. Acetaldehyde
2. Methanol
3. 2-Propanol
4. n-Propanol
5. Ethyl Acetate
6. 2-Butanol
7. Isobutanol
8. n-Butanol
9. Acetoin
10. 2-Methyl- l-Butanol
11. 3-Methyl- 1-Butanol
12. 4-Methyl-2-Pentanol (int. Std.)
13. Ethyl Butyrate
Reprinted from the Journal of Chromatographic Science with permission of Preston Publications, a division of
Preston Industries
F i g u r e 4. Chromatogram of a standard solution using a 4m x 2mm Carbowax 1500 column.
310
Capillary Columns. The open tubular design of capillary columns, where a phase without support
packing is deposited as a thin film on the inner wall of a low internal diameter column offers
significant performance advantages, but also places greater demands on the entire chromatographic
operation. The practical differences between packed and open tubular column separation results
from reduced intra brand broadening for individual solute molecules in the capillary and the much
longer column lengths allowed by the open tubular design. These factors have been summarised
by Jennings [ 11 ] as follows :
A packed column will offer a range of flow paths to solute molecules giving a spread of
residence times in the mobile phase. The open tubular design offers a single flow path with
more uniform movement in and out of the mobile phase.
Solute molecules also experience a similar wider residence time spread in the stationary
phase due to its much higher concentration in a packed column and its non-uniform
film thickness.
Packing materials are inefficient at heat transfer and so a range of temperatures will exist
across any transverse section of the packed column. Solute volatiles are affected and this
again leads to dispersion of individual molecules of the same solute.
These factors therefore express themselves as compounds that elute with wide peak widths after
short residence times in packed columns and mass sensitivity at the detector will be low for later
eluting compounds. Co-elution and overlapping cannot be easily avoided. For this reason specific
phase selection is very important for packed columns as the separation factor has to be optimised
to compensate for these inherent disadvantages.
The open tubular design brings its own difficulties, principally in terms of the need for increased
GC hardware sophistication, and especially the problem of transferring the sample compounds in
a narrow band to the small diameter capillary column. The earliest columns were glass and required
special deactivation procedures with delicate handling and manipulation. The trend at that stage
was also to modify packed column instruments and the technique was largely confined to specialist
laboratories. The introduction of robust, flexible, fused silica columns and chromatographs with
dedicated capillary pneumatics, injectors and detectors, and especially more precise oven
temperature control (Hewlett-Packard 5880A) finally made the technology much more accessible.
Grob et. al. were among the first to investigate direct analysis of distilled spirit congeners on
glass capillary columns [ 12]. They used a column coated with Carbowax 400 and injected with
splitting and an oven starting temperature of 25~ This phase was not bonded to the column wall
in the sense that it was not non-extractable and so the first 60 cm of the column was left without
phase to avoid its removal and subsequent redeposition problems, which condensation at the low
initial temperature could cause. De Nijs and de Zeeuw reported a chemically bonded Carbowax
on fused silica which they termed CP Wax 57 CB [ 13]. This column was resistant to washing with
polar solvents such as methanol and even water, and had an upper temperature limit of 220~
MacNamara used this column with split injection to profile the major congeners in distilled spirits
[ 14] and Figures 5 and 6 shows typical separations for a standard mix of congeners and a whiskey.
311
12113
"
l'l
I I
[
I
A:a~s40Cc;dmiti2~s;
~C/min t~ 200~
lnj.: 200~ Det.: FID, 200~
Carriergas:40cm/s H2
I!
4
tel
I
18
~ ) _
i~
J
9
23
24
F i g u r e 5. Split capillary separation of a standard congener mixture in 40% v/v ethanol on a
50m x 0.25ram x 0.25pm CP-Wax 57 CB column.
7 8
"~
12 j13
[
I
I
I
11 J
[
"
[
t
II
11
l[
1[
I
,L
Compounds:
1. Acetaldehyde
2. MethylAcetate
3. EthylAcetate
4. DiethylAcetal
5. Methanol
6. Butanol-2
7. Propanol
8. Isobutanoi
9. IsoamylAcetate
10. Butanol-I
11. 4-Methyl-2-Pentanol (int. Std.)
12. 2-Methyl-1-Butanol
13. 3-Methyl-1-Butanol
161
li
14. Ethyl Caproate
15. Ethyl Lactate
16. Ethyl Caprylate
17. Furfural
18. Ethyl Caprate
19. Phenyl Ethyl Acetate
20. Ethyl Laurate
21. 2-Phenyl Alcohol
22. Lauryl Alcohol
23. Ethyl Myristate
24. MyristylAlcohol
25. Ethyl Palmitate
20 21
Figure 6. Split capillary separation of a commercial whiskey on a 50m x 0.25mm x 0.25pm
CP-Wax 57 CB column.
312
CP Wax 57 is a unique phase with a unique selectivity, and major advantages are separation of the
following pairs:
ethyl acetate -diethyl acetal
isobutanol- isoamyl acetate
isomers of amyl alcohol
Diethyl acetal and isoamyl acetate have quite low sensory thresholds, but in addition exhibit
azeotropic behavior with ethanol and water which effect their behavior during distillation.
Table 4 gives composition and boiling point data for the ternary azeotropes of these two compounds
with ethanol and water, and shows that the net boiling points of the compounds have been reduced
to below the boiling point of the common ethanol-water azeotrope.
A-Component,
B.P. (~
B-Component,
B.P. (~
C-Component,
B.P. (~
._ Aze0tropic Data
B.P. Wt. Wt' W t .
(~ (%A) (%B) (%C)
Water,
100
Ethanol,
78.3
DiethylAcetal,
103.6
77.8
Water,
100
Ethanol,
78.3
Isoamyl Acetate,
142.0
69.0
11.4 27.6 61.0
not stated
Table 4. Ternary azeotropes of ethanol~water with diethyl acetal and isoamyl acetate.
Distillation fluctuations or non-equilibrium conditions can therefore affect the levels of these
compounds in a spirit, and even though their absolute concentrations will be quite low, perceived
aroma can be influenced through their high odour intensities. The information in Figure 5 and 6
traces is also a balance which is determined by the injection split ratio. The split ratio is used to
meter the proportion of the injection volume delivered to the column and is a trade-off between
resolution before and just after the ethanol peak, and useful sensitivity for later eluting compounds.
If the split ratio is too low resolution and peak shape for pre-ethanols suffers due to a reverse
solvent effect [ 12] if it is too high detection of the late eluters becomes more difficult. The late
eluting compounds in these traces, where the split injection amounts are much less than could
have been delivered to a packed column, highlight previous comments on increased detector
mass sensitivity.
This column is also very stable and our experience is to obtain 2 years daily use without any
deterioration in performance. A baseline separation for the amyl isomers is not achieved but this
is not necessary for accurate ratioing due to the 70/30 proportions usually found in distilled
spirits. A more polar phase such as Carbowax 400 with much lower viscosity will baseline separate 2- and 3-methyl- 1-butanol (Figure 7) but is a non-bonded phase and has an upper temperature
limit of only 100~ A disadvantage is that CP Wax 57 does not elute symmetrically free fatty
313
acids and an alternative approach is necessary for these compounds. Masuda et. al. [ 15] injected
whiskey directly with splitting to a relatively apolar 5% phenylmethyl silicone capillary and achieved
separation of a combination of alcohols and esters together with acetic, octanoic, decanoic and
dodecanoic acids.
Compounds:
Analysis Conditions:
3
Oven: 50~ isothermal
Inj.: 150~ Det.: FID, 200~
Carriergas: 90kPa H2
0.1 lal split injection (100 ml/min)
5
4
6
1.
2.
3.
4.
5.
6.
7.
8.
Ethyl Acetate
Methanol
Ethanol
Isoamyl Acetate
2-Methyl Propanol
1-Butanol
2-Methyl- 1-Butanol
3-Methyl-1-Butanol
Reproduced with )ermisslon from Chrompack B.V., Middleburg, The Netherlands
Figure 7. Separation
o f a testmix on a 50m x 0.32mm x 0.21um Carbowax 400 column.
Direct splitless injection of distilled spirits is also possible and can give additional and complimentary
data to that produced by split injection of the same sample [ 16]. Figure 8 shows a direct splitless
injection of a whiskey on an FFAP phase, which is a standard Carbowax polymer modified with
nitroteraphthalic acid. The splitless injection transfers much more matrix and congeners to the
column and gives increased peak areas for late eluting and minor congeners, together with
symmetrical peaks for free fatty acids. The resolution of peaks around the solvent has been
"
_
6
~
Compounds:
1. Ethyl Caprylate
2. Acetic Acid
3. Furfural
4. Ethyl Caprate
5. Ethyl Laurate
6. 2-Phenyl-Alcohol
7. Ethyl Myristate
8. Caprylic Acid
9. Ethyl Palmitate
10. Capric Acids
"
Analysis.Conditions_:
Oven: 60~
3~
to
220~
lnj." IYFV,40~ 12~ to
Det." FID, 200~
220~
Carriergas." 40cm/s H2
'
Figure 8. Splitless injection o f a whiskey on a 60m x 0.25ram x 0.25pro FFAP column.
314
destroyed but this information is available from a corresponding split injection of the same sample.
Therefore each injection mode gives unique and complimentary information which builds a
comprehensive profile of the most abundant congeners. This approach was applied to quantify
thirty three compounds in 14 samples of each of two different malt whiskeys [ 17]. Using linear
discriminant analysis techniques a very good differentiation of the two types could be obtained
(Figure 9) and it is also clear that the intra variation in one of the whiskeys is more pronounced.
Further statistical testing showed that five compounds alone could account for 98.5% of the
variation. Four of these compounds were quantified from the splitless part of the analyses.
o
1.0-
,~
o
r
o
o o
<,o
Malt B
0.5
-0.0Malt A
-0.5-
-1.0-1'.0
9
-0'.0
9
110
Figure 9. Differentiation of whiskeys using direct injection congener data.
Figure 10 shows Chernoff faces for the two whiskey sets constructed on these five compounds.
Figure I0. Chernofffaces for malt A (top) and malt B, constructed on different levels of five
compounds.
315
One problem which detracts from the usefulness of direct splitless injection is the variability of
FFAP columns from different commercial sources. This is related to different procedures for
modifying the Carbowax phase with nitroteraphthalic acid to induce acidity. The treatment probably
produces an ester linkage which can be more easily hydrolysed by certain solvents and conditions.
A result of this is that the acidity can be removed with repeated injections and the phase will
slowly evolve into standard Carbowax, with increasingly deteriorating acid peak shape. Figure
11 show this phenomenon in its early stages. After about 50 injections the column is loosing
acidity and is eluting acids faster than a new column. The pairs furfural and acetic acid, and C~6 ~ester and C~0-acid, have actually inverted, and the C~0-ester and Ca-acid pair are about to.
Analysis Conditions;
Oven: 60~
3~
to 220~
Inj. PTV, 40~ 12~ to 220~
Det. FID, 200~
Carriergas: 40cm/s H2
I lal splitless injection (2min)
Compounds:
1. Furfural
2. Acetic Acid
3. C~0-Ester
4. Ca-Acid
5. Ct6 t-Ester
6. Clo-Acid
Analysis Conditions:
Oven: 60~
3~
to 220~
lnj.: PTV, 40~ 12~ to 220~
Det.: FID, 200~
Carriergas: 40cm/s H2
1 pi splitless injection (2min)
Compounds:
1. Furfural
2. Acetic Acid
3. CIo-Ester
4. Ca-Acid
5. Cl6:t-Ester
6. CI0-Acid
Figure 11. Splitless
injection o f a whiskey on a new (top) and a deteriorating 60m x 0.25mm x
0.251am FFAP column.
316
After further use acid peak shape begins to deteriorate and the column must be changed. This
behavior and its extent differs with similar columns from different manufacturers. With columns
that allow a reasonable number of injections, direct splitless analysis gives rapid useful information
that must be balanced against costs of higher column usage.
Splitless injection is a complicated process involving slow transfer of compounds of interest to
the capillary column for refocusing and separation. In conventional hot splitless injections a pressure
wave is created by the explosive vaporisation of the sample, giving a non-homogenous vapour
cloud which is a recipe for discrimination. The same effect distributes the sample and any involatiles
it may contain to every comer of the injection liner. Compounds of interest can be lost through
the septum purge and involatiles have a greater chance of reaching the column entrance. Cold
programmable injection (PTV), where the sample is deposited cold in a glass liner which is then
linearly programmed to the desired final temperature, is aesthetically and technically superior to
conventional flash vaporisation. Both discrimination and decomposition of labile substances have
been shown by various authors to be dramatically reduced [ 18-20].
When a sample is deposited cold in a programmable injector its compound content can be
uniformly removed by programmed heating and any involatiles tend to remain relatively undispersed
in a section of the liner. One such injector (Gerstel CIS-3) also has a septumless head which both
simultaneously avoids septum particle problems, and the need for a septum purge flow which
gives discrimination and general loss of compounds. Table 5 gives reproducibility of absolute
peak areas for 6 replicate autosampler runs of ppm solutions of C13-C20hydrocarbons in hexane
using cold split and splitless injection. The numbers given are expressed as % relative standard
deviation from the 6 runs.
Mode
Cl3
El 4
Ci5
C16
Ci7
C18
C19
C20
split
1.12
0.98
0.90
0.58
0.83
0.82
0.72
0.65
splitless
0.86
1.06
1.04
1.52
0.68
0.64
1.02
0.73
Table 5. Reproducibility of absolute peak areas for cold split and splitless PTV injection.
There is no essential difference in the reproducibility of the split mode compared to the splitless
mode and both offer a reliable and reproducible method for capillary sample introduction.
One important variable in splitless injection is the splitless or transfer time and cold injectors
have a distinct advantage here. They have liners with smaller intemal diameters than conventional
injectors to provide a low thermal mass and allow rapid heating. This in turn allows a higher
carrier gas velocity in the smaller i.d. liner and means transfer of the compounds to the column is
faster and occurs at lower temperature. This can be even further enhanced by using pressure
programming (a higher inlet pressure) during the splitless transfer. Figure 12 shows the effect of
liner diameter on the splitless transfer of the C30 hydrocarbon in hexane. 75% is transferred to the
column at a temperature of 210~ for the 1.2 mm i.d. liner, and 290~ is required for the
3.4 mm i.d. liner [21 ].
317
9
100-
+ _ _
+
~,,
,300
!.2 mm
-250
75-
o
-2oo
o
o
g~
-150
50-
100
E
25-50
0
~
0
0
1'0
2()
'
3()
4'0
Splitless time (sec)
5i)
60
Reproduced with permission from [211
Figure 12. Effect of liner diameter on PTV splitless transfer Dashed line: actual temperature in
the liner
Cold programmable injection is the method of choice as a universal capillary injection technique
and avoids the different disadvantages of hot split/splitless and cold on-column approaches. It
can also be used for large volume injection applications and adapted to headspace and thermal
desorption devices, and these areas will be investigated later.
An additional major advantage of capillary columns is that their small volumetric flow
requirements allow the use of low-cost benchtop mass spectrometers as GC detector. The column
can be directly interfaced to the ion source and the latest PC-based control and data reduction
technology make this technique very accessible. When such a GC-MS is used in selected ion
monitoring mode (SIM), the range of compounds that can be detected and quantified by direct
splitless injection of spirits is profitably increased. In the SIM mode only characteristic ions from
selected compounds are monitored continuously, rather than scanning all ions over a mass range.
In the former case the time spent detecting the ion current at a particular mass is a much higher
percentage of the total cycle time than in scanning mode. This is manifested as an apparent
increased instrument sensitivity but is in fact due to a much higher signal-to-noise ratio.
The application of this approach means that higher esters can be directly quantified in light
whiskeys where their level is at low or sub-ppm level due to a high blend ratio. The same approach
can be used for similar level trace phenolics that contribute to peatiness in full malt whiskeys. In
practice operating procedures are established which involve programming the instrument
acquisition parameters to monitor specific ions from the known compounds in specific retention
time windows. Ions to be monitored must be carefully chosen as there is always the danger of
interference from the same ions from other compounds. To make the best use of this approach a
dedicated software package [22] is recommended for target compound analysis.
318
In general the three principles inherent to target compound analysis are:
- Presence and integration of all the target ion masses.
- All ions must co-elute within a retention time window.
- Target ion ratios must fall within a calibrated range.
A compound is determined to be present if the characteristic ions (a reference ion and up to
2 qualifier ions) are detected co-elulting in a specified retention time window and they meet the
ion ratio tests (Figure 13). The specified retention time windows for locating the characteristic
ions can be defined in terms of absolute or relative retention time, or retention time relative to an
internal standard. The ion co-elution test is then performed and this usually employs a small
absolute time window to test for co-elution.
~
~
~
L
Figure 13. Target concepts.
tep 3"
S
t
e
p
2:
Step 1"
Qualifier ions co-elute with
reference ion
Reference ion in retention
time window
Peaks integrated in
extracted ion window
Reproduced with permission from [22]
This window can be as small as one scan (or 0.025 mins) and is quite a strenuous test. When this
test fails, even after optimum adjustment of instrument parameters, the entire analytical
methodology needs to be re-evaluated. The final test is the ion ratio test, and the area ratio of the
qualifier ions to the reference ion must fall within a target ion ratio limit, which is defined as
acceptable variance of the ion ratio from the calibrated ratio. The calibrated/expected ion ratio
for each qualifier can be automatically determined using the ion ratio from a calibration run.
When compound parameters and calibration data are correctly programmed, samples can be
automatically run and processed with comprehensive report generation. Figure 14 shows a selected
ion monitoring TIC for 7 compounds in a very light blend after 1 lal direct splitless injection, with
319
decanol-3 as internal standard. Table 6 gives the corresponding report with amounts based on
four levels of calibration. The reference and qualifier ions for each compound are also shown
together with ion retention time and confirmatory ion ratios. Direct injection and the high SIM
S/N ensures very good accuracy and precision.
Analysis Conditions:
Oven: 60~
3~
to 220~
lnj.: PTV, 40~ 12~ to 220~
Carriergas: 40cm/s H2
1 lal splitless injection (2min)
60(0X)0-
Compound.s:
1. Ethyl Caprylate
2. Decanol-3 (ISTD)
3. Ethyl Caprate
4. Phenyl Ethyl Acetate
5. Ethyl Laurate
6. 2-Phenyl Alcohol
7. Ethyl Myristate
8. Ethyl Palmitate
MSD
2
5
400000
20(K)(O
1
8
!
20
2'5
3'0
3'5
40
4'5
A____
go
Figure 14. Selected ion monitoring chromatogram for 7 compounds, very light blend.
Compound
Ethyl Caprylate
Decanol-3 (ISTD)
EthylCaprate
Phenyl Ethyl
Acetate
Ethyl Laurate
2-Phenyl Alcohol
Ethyl Myristate
Ethyl Palmitate
RT
19.64
19.64
19.64
26.81
26.81
26.82
28.29
28.29
28.29
35.77
35.76
35.77
36.55
36.55
36.55
39.52
39.52
39.52
43.91
43.91
43.91
50.62
50.62
50.62
Mass
Area
Amt
(mg/l)
88
101
127
59
69
111
88
101
115
104
91
65
88
101
157
91
92
122
88
101
157
88
101
157
5273036
1785649
1173132
11103872
9561073
1784033
13988259
5563892
1022386
1306376
286197
173271
12581317
5884328
1337508
7174672
3732274
1506499
1252651
641310
145820
2542767
1369236
270671
1.878
Table 6. Corresponding report to Figure 14.
Ion Target
Range
Ion Ratio
found
100.00
27.15 - 40.73
17.58 - 26.38
3.200
68.81 - 103.21
12.69 - 19.03
5.228
31.91 - 47.87
5.72 - 8.58
0.316
17.79 - 26.69
10.26- 15.38
3.709
37.20- 55.80
8.30- 12.46
1.628
41.97 - 62.95
16.79- 25.19
0.583
41.54 - 62.32
9.22 - 13.82
1.605
43.51 - 65.27
8.31 - 12.47
33.86
22.24
100.00
86.10
16.06
100.00
39.77
7.30
100.00
21.90
13.26
100.00
46.77
10.63
100.00
52.02
20.99
100.00
51.19
11.64
100.00
53.84
10.64
320
Specialised Direct Injection Techniques. Hagman and Roeraade described an approach using
precolumn backflush to selectively profile the pre-ethanols in alcoholic beverages [23]. In this
configuration (Figure 15) a short packed precolumn was coupled to a capillary column via an
effluent splitter. After injection the volatile pre-ethanol congeners passed to the main capillary
column and the remaining less volatile compounds were backflushed from the pre-column using
a 10 port rotary valve. By keeping part of the pre-column in the hot injector, and the rest in the
GC oven, the authors achieved an acceptable balance between rapid sample evaporation and
preseparation in the precolumn. The preseparation could also be influenced by the split ratio,
since this also influences the flow through the precolumn. Shiomi [24] used a two-oven multiHeated Inlet
Carrier Gas In
Vi
_
l/acked
Precolumn
GC Oven
V2
Splitter
Column
Reproduced with permission from [23]
Figure 15. Configuration for precolumn-backflush.
dimensional system and two wide bore columns of opposing polarity to achieve baseline separation
of diethyl acetal from the pre-ethanol group of compounds. The acetal co-eluted with ethyl acetate
on a wide bore Carbowax precolumn, and this segment was heartcut to an apolar widebore
column for the increased resolution. By using a single capillary column, with lower internal diameter
to give more plates per meter, acetal and ethyl acetate can usually be adequately separated.
MATRIX REMOVAL
The range and number of flavour contributing trace compounds in distilled spirits that are not
readily amenable to direct injection is very substantial. Sample preparation involving matrix removal is a prerequisite for further analysis, and strategies must be carefully chosen to minimise cost
and effort. This is a vast area and this section only attempts to describe general approaches with
an emphasis on strategies that can be further complimented by injection or chromatographic
techniques and/or provide additional sensory information.
Since many of the contributing trace congeners are already known from previous research
efforts, solvent extraction gives sufficient enrichment to allow detection. An apparatus for
321
continuous extraction (Figure 16) has been described by Rapp and Mandery [25].
Figure 16. Apparatus for continuous extraction with Freon.
Spirits are reduced to 15% v/v and a total volume of 250 mls at this strength can be extracted by
50 ml of Freon 11 (fluorotrichloromethane)containing 10% dichloromethane. This mixture has
an azeotropic boiling point of 26~ and so the extraction proceeds virtually at room temperature.
In experiments with model solutions these authors found that after 20 hours a very good enrichment
of most trace congeners can be obtained. The solvent can also be removed and recovered at low
temperatures and the residual solvent can be substituted for ethanol to give extracts which retain
the flavour complexity of the original distillate. The apparatus requirements are also quite simple
and 6-12 continuous extractions can be performed overnight.
A simpler but less intensive approach using a higher boiling Freon has also been described
[26, 27]. 10 mls of spirit at 20% can be batch extracted with 100 I.tl of Kaltron (1,1,2-trichlorotrifluoroethane) by simple shaking, and the Kaltron extract can then be removed directly for GC
injection. Liddle outlined a similar approach using iso-octane as solvent and extracting by vortex
mixing [28]. These approaches are simple and cost effective but can never give the degree of
enrichment achieved with an overnight continuous extraction of larger quantities. But if the
simple batch extraction is also regarded as a procedure for transferring compounds into a different solvent, then the technique of large volume injection can be used to compensate for the
lower extraction efficiency and provide greater analytical automation. This idea will be investigated
in a later section.
322
For more detailed information physical techniques with and without solvent extraction can be
used to enrich and preparatively segment distilled spirits. The most comprehensive approach in
this area was described by Ter Heide and colleagues [29, 30] who perfected a separation scheme
(Figurel7) for distilled beverages that allows maximum sensory and analytical information. 500
liters of the distilled beverage under investigation is batch extracted with a specific solvent mix to
give a flavour complex. This was then subjected to many sub-separations, including distillation
and preparative chromatography, and at each stage the isolated fractions and components were
examined by an expert sensory panel. Not only were many new compounds identified by this
work, but also their relative contribution to the overall beverage flavour could be investigated.
[ Distilled Beverage i
I
Solvent Extraction
Ethanol / Water
.
.
.
-!
.
l
Flavour Complex
I
NaOH
]
I
!
[
Na-Salts of Acidic Compounds ]
Neutral Flavour Complex
]
I
I
Distillation
Regeneration
Fusel Components[
pH=l
I
[
Acids
[
pH=9.5
I
]
i
Phenols
]
NonVolatiles
Fusel-free Flavour Complex
I
Short Path Distillation
]
[ Vo,atileFlavourCompouns 1
I
Several Chromatographic Methods
I
I
I
I
!
I
Many Fractions
Reproduced with permission from [1]
Figure 17. Comprehensive separation schemefor distilled beverages.
323
When the resources of an intemational flavour company are not available, then less comprehensive
approaches have to be developed to give similar results. Figurel8 shows an automated fractional
vacuum distillation apparatus which can be used to fractionate up to four liters of a distilled spirit.
Distilling at 80 mbar and recirculation of the sample through a thin-film evaporator ensures low
temperature fractionation and minimum contact time of the solutes with heated surfaces. The
separation results from the combined effects of solute volatility and azeotropic behavior and
|
Head Condenser
Reflux Divider
Distillate R eceivers
Separation Column
Motor
J
L
i
i
'Pre-Hc
~___ _
~
.
.
.
o
--
Cold Traps
Thin Film
Evaporator
Vacuum Pump ~. ,~,
Circulation Pu
. . . . . . . . .
i
..I
Reproduced with permission from Normschliff, Wertheim, Germany
Figure 18. Apparatus for vacuum fraction distillation.
324
gives concentrated fractions of volatile and fusel compounds. The liquid nitrogen traps ensure
complete recovery and since there is no solvent involvement, the concentrated fractions can be
readily investigated for aroma and taste. Since the fractionation is also strongly influenced by
volatility, the fractions themselves are suitable for matching with different injection techniques
and different capillary columns. The first fraction will contain only the volatile top notes of the
distillate and is recovered exclusively from the cold traps. This sample is best analysed on a highly
retentive column and Figures 19 and 20 show traces on such a column compared to a regular
capillary column.
10000
Analysis Co..n.dition.s"
Oven: 40~
5~
to 240~
Inj.." PTV, 40~ 12~ to 220~
Carriergas: 24 psi He
1 lal split injection (1/30)
5000
0
3'0
1'0
....
9FID, 250~
4'0
Figure 19. Volatile top notes of the distillate, separated on a standard 60m x 0.25mm x 0.251am
Carbowax 20M column.
10000-
Analysis Conditions:
Oven: 40~
2~
to 80~ 5~
220~
lnj.: lrl~, 40~ 12~ to 220~
Det.: FID, 250~
Carriergas: 7 psi He
1 Ial split injection (1/30)
F--7
I
i
I
I
5000
to
J
i
I
I
0 TI~
0
fo
2o
3o
40
Figure 20. Volatile top notes of the distillate, separated on a highly retentive 30m x 0.32mm x
5.01am 5%Phenyl-Methylsilicone column.
325
Taniguchi recently described an approach using centrifugal partition chromatography to separate
aroma substances in whiskey new distillates [31]. This is a liquid liquid partition between two
immisable solvents and the centrifugal field retains the stationary phase solvent more firmly in the
separation columns. Using the less polar solvent as mobile phase allowed faster elution of less
polar esters and acetates (Figure21) and in this way they could preparatively separate slightly
more polar sulfur compounds of interest.
!
Ethyl e s t e
I
0
i
!,
~.,--
,
9
'
....
.
1.7
.,
0
:
S
t
.
(1)
'
[
!
"
I
.3. o..J..
Acetates
'
9 ,
(4)
,~s
_
i
(2)
~
0
9
]
~x
'O
~
4J
(3)
i
:
:
,' ~ ~
~ Fatty Acids !:
o
o.
.
.
I
!
i
i
|
s
i
:
9
'
EIO
. . . . .
i
t. . . . .
'
FrGr I
Fr. 37
Fr. 38
' -
;:
.
o ~ ~ oEt i
OEt
t
Fr. 36
i
,,
o
Fr. 40
FrGr2 -
9
,0
' o>__) ii
E,
t,o |i
J
Acctals
t
:
i
i
i
i
i
0
:t
#
:
: Fr. 44
'
Ft. 47
'
t
FrGr 4-----l--~FrGr 5
I
Reproduced with permission from [31]
21. Compounds eluted in earlier fractions by centrifugal partition chromatography
(FrGr: fraction group).
Figure
Their initial starting material was a concentrated solvent extract of a whiskey and after separation
and solvent substitution for ethanol, the fractions could be investigated sensorially. Compounds
more polar than ethyl lactate remained in the stationary phase and were recovered from the same.
LARGE VOLUME INJECTION
Figure 22 shows specific sulfur traces, with solvent venting, for increasing injection volume of a
simple Kaltron extract of a whiskey, compared to a normal non-solvent splitting injection. The
traces serve as an introduction to this rapidly developing, technologically important area of capillary
gas chromatography. Except for a loss of very early eluting compounds during solvent splitting,
the recovery of the majority of the sulfur species correlates with the injection volume. The two
most important advantages of this technique are reduced sample preparation time and greater
levels of sensitivity. Substantial investigation of the variables and parameters for successful
implementation have been described by various authors and commercial chromatographic units
are available for complete automation of procedures.
326
1.0e+06 ::
Analysis Conditions:
Oven: 40~
5~
to 250~
lnj.: PTV, 60~ 10~ to 250~
Carriergas: 14 kPa He
5 lal splitless injection (lmin)
8oooooi
600000
SCD
400000
2ooooo
0
lO
20
1.0e+06-
30
40
Analysis Conditions:
Oven: 40~
5~
to 250~
Inj.: PTV,-20~ 10~ to 250~
SCD
Carriergas: 14 kPa He
40 IA injection, solvent venting with 201.d/min injection
speed, then splitless (lmin)
80(0)00
600000
400000
200(0s
0
10
20
30
40
1.0e+06:
Analysis Conditions:
Oven. 40~
5~
to 250~
Inj." PTV,-20~ 10~ to 250~
SCD
Carriergas." 14 kPa He
100 ~1 injection, solvent venting with 201al/min injection
speed, then splitless (lmin)
80(0)00
600000
400000 84
20(0)0
0
10
20
30
40
Figure 22. Specific sulfur traces of a Kaltron extract, 5, 40 and 100 pl injected (from top) on
15m x 0.53mm x 1.Opm Supelcowax 10.
327
Large volume injection into capillary columns can be classified into two categories, depending on
whether the solvent is separated from the solutes in the injector or in the chromatographic system
[32]. In the latter case the large volume is passed directly to the column and can give rise to the
well known problems of solvent flooding, phase damage and solute peak deformations [33]. Use
of a retention gap, i.e short length of deactivated uncoated capillary tubing, between the injector
and the main column, can refocus broadened deformed peaks [34]. For general industrial
practitioneers this approach can be impractical as additional leak-tight capillary to capillary
couplings are required, and the retention gaps will need frequent changing if they become dirty or
deactivated. When the solvent is removed using the temperature programmable features of a cold
injector, three distinct approaches can be defined:
- Solvent venting or splitting. This was the approach used by Vogt and co-workers who were
the initial pioneers in this field [35,36]. The sample is introduced slowly with the split vent
open, and the injector at a suitable temperature below the solvent boiling point. After a
certain period the split valve is closed and the solute contents of the large volume are
transferred in the splitless mode to the column. The solutes are enriched out of the gas
phase.
- PTV vapour overflow. This is a technique which was developed by Grob [37] and involves
injecting the sample in the splitless mode, but at a relatively high injector temperature and
with a substantial septum purge flow. The solvent flash evaporates and escapes with the
septum purge flow, but low volatility solutes are retained in the liner due to the solvent
evaporative cooling effect.
-
Solid phase extraction (SPE) in the liner. In this case the sample as a liquid is passed
through an adsorbent-packed liner by a high flow of carrier gas. The solutes are retained in
the packing material and after drying the liner is heated to transfer the compounds to the
column [21,38]. The solutes are enriched out of the liquid phase. This technique can only
be applied to aqueous samples and a special valve configuration is necessary to prevent
water from reaching the analytical column.
Both the PTV vapour overflow technique and the SPE/FFV approach have practical disadvantages;
the former is limited to low volatility solutes and the later to aqueous samples. For these reasons
the solvent venting technique represents the best practical approach and it is in this area that most
development work has been done. Some early results in this area were from the publications of
Herraiz [39] and Villen [40]. Herraiz and co-workers investigated the efficiencies of several
packing materials when they act as a substitute for low PTV initial temperatures. Test mixes of
various compounds were prepared in Freon 11 and 2 lal injections were introduced onto the
different packed liners with solvent venting at an initial temperature of 30~ Tenax gave the best
overall performance in terms of recovery of volatile and semi-volatile compounds. Chromosorb
101 gave similar results but its maximum allowable upper temperature of 220~ was
disadvantageous for desorption of less volatile compounds. Villen and co-workers compared the
normal off-line external concentration of an extract with internal concentration on the adsorbent
bed of a PTV liner in the solvent split mode. Again using Tenax as adsorbent the data showed that
the internal PTV concentration gave more precise and accurate results. Grob has pointed out that
328
the usual off-line concentration can involve significant loss of volatile components due to coevaporation with the solvent [41]. The volumes injected in this study ranged from 2.5 to 25 lal.
A much more fundamental study of solvent elimination was carried out by Staniewski and
Rijks in 1992 [42]. They thoroughly investigated the optimization of a PTV injector for large
volume injection with solvent splitting, as a function of liner temperature and design, solvent
type, speed of introduction, and purge gas flow and purge time during venting. The sequence of
events occurring during the entire process is shown in Figure 23. The solvent elimination rate
determines the speed of sample introduction and these processes must be matched in order to
achieve a steady state in the liner, so that the mass flow of liquid solvent entering equals the mass
flow of solvent vapour being eliminated.
Cold
Sample
Injection & Transfer
Solvent "
E l i m i n a t i o n ,I . . . . . . .
Separation
~ Liner
9
!
J
9 j
.:
.w
J
-
'~,,.
iColu . n a n ~
iTemperature
J
On
io.
!On
Purge Status
[
_
9 Inlet Pressure
Needle
J
Injection ~ . .
....
Time
Pentane 10.4
Dichloromethane |0.7
Acetone ~ 3 . 3
Chloroform 111.5
Methanol ~ 4 . 8
Tetrahydro Furane 111.5
Diisopropyi Ether |1.1
Hexane 111.2
Ethyl Acetate ~2.6
Cyclohexane 112.2
Acetonitrile ~ 4 . 9
Water ~
1,4-Dioxane ~ 7 . 1
0
59.6
1'0
2b
Reproduced with permission from [42]
Volume (ml)
Reproduced with permission from [42]
Figure 23. Sequence of events during sample
Figure 24. Saturated vapour volumes of 11~l
introduction, solvent elimination and sample
transfer.
of different solvents at 20~
Figure 24 shows the saturated vapour volumes for unit volume of different solvents at 20~
and
these values are the minimum volume of gas required to remove the solvent as vapour from the
liner. Therefore more polar solvents will need a relatively longer time to eliminate. The maximum
injection rate which is equal to the solvent elimination rate, can be increased by decreasing the
pressure in the liner and increasing the total gas flow rate in the liner. Newer instrumentation with
electronic pressure control allow the necessary automation of these parameters. A further advantage
is that low programmed pressure during solvent elimination results in much less solvent transfer
to the column, even for very large volumes. Based on the above considerations a model was
proposed to calculate the best sample introduction rates, liner temperatures and purge flows for
optimum recover3' of test compounds in various solvents. They injected 150 lal of such a test mix
in hexane to a liner packed with glass wool and achieved 90% recovery of components with
volatilities lower or similar to a heptadecane.
329
For the traces in Figure 22 from the Kaltron whiskey extract the following calculations were
applied based on the recommendations of Staniewksi and Rijks.
M 9 Pv * s
d*R*T--*
I
M
Pv
s
d
R
T
p,,
p,
Po
Pi
-I
= max. solvent venting speed (lal/min)
= molecular weight of solvent (g/tool)
= vapor pressure of solvent at inlet temperature (bar)
= split flow (ml/min)
= density of solvent (g/mol)
= gas constant (0.08312 l*bar/K*mol)
= temperature of PTV inlet (K)
= absolute pressure at split outlet (bar)
= absolute inlet pressure (bar)
For Kaltron (1,1,2-trichloro-2,2,1-trifluoro ethane), with an inlet temperature o f - 10~ a split
flow of 100 ml/min and an inlet pressure of 0 ("stop flow") the following maximum sample
introduction speed can be reached'
187 9 0.0868
9 100
1.58,0.08312
9 263
9
1
1
= 47.0 pl/min Kaltron
It could be naturally assumed that the boiling point of a solvent is the dominant contributory
factor when calculating its elimination rate as a vapour. However the following comparison for
water (bp. 100~ and isooctane (bp. 99~
show the necessity of utilizing all the equation
parameters.
114 9 0.2803
9 100
0.69,0.08312
9 333
18
9 0.1956
9 100
1 * 0.08312
9 333
1
,
1
9
1
1
= 167 pl/min isooctane
= 12.7 pl/min water
For practically similar boiling points there is more than an order of magnitude difference in the
solvent venting rate at the same temperature.
MOiler and co-workers determined pollutants in aqueous samples by a similar approach but with
a liner packed with Tenax TA [43]. They proposed the calculation of breakthrough volumes for
the compounds to be enriched on the liner, and achieved this by connecting the PTV with packed
liner directly to the detector without any separation column. In this way the PTV insert can be
regarded as a short packed column and the specific breakthrough volume is defined as the volume
of carrier gas, relative to 1 gram of adsorbent, needed to release the substance at the end of the
column. This data coupled with similar data for the elimination behavior of water allowed them to
inject 500 lal of aqueous standard solutions of pesticides and nitroaromatics with good recovery
of analytes of interest. Mol and co-workers investigated the implications of liner dimensions on
large volume solvent splitting [21 ]. Larger internal diameter liners can contain more solvent, and
so require less fine control when introducing the solvent. On the other hand smaller internal
diameter liners are more efficient during splitless transfer to the column and induce less breakdown
of thermolabile substances.
330
Figures 25 and 26 depict the schematics for a commercial PTV system which allows automated
solvent venting (Gerstel, Miilheim an der Ruhr, Germany). All parameters including injection
volume, injection speed and all PTV settings are controlled from Chemstation software. Electronic
pressure control can be programmed for minimum column flow during venting to give minimum
transfer of solvent to the column. The latest auxiliary gas control technology also controls the
split purge flow, which can be automatically economised after the splitless transfer step of
components for the chromatographic run. The large volume injector uses a standard HewlettPackard autosampler tray for automated sequence runs, and when this sampler is turned off the
PTV automatically reverts to a standard cold split/splitless injector for low volume sample
introduction.
Figure 25. Schematics for Gerstel PTV in solvent venting/stop flow mode.
331
Figure 26. Schematics for Gerstel PTV in splitless mode.
The Gerstel large volume sampler is in fact a multipurpose sampler, in that it can be easily upgraded
to a headspace injector or sampler. Static headspace injection is a valuable technique for
determination of trace volatile compounds in foods and beverages. An initial drawback was the
need to inject the large volume headspace vapour with splitting to produce sharp initial bands,
and this in turn limited the sensitivity of the technique. A further aspect for consideration is that
the process for transferring headspace vapour to a chromatographic column can involve significant
dilution of the solutes with carrier gas. Takeoka and Jennings [44] attempted to solve this problem
with a retractable on-column injector equipped with a fused silica needle syringe to allow gas
sample injection directly into capillary columns. A dewar flask with coolant refocused the solutes
332
in the first loop of the analytical column, while the sample matrix passed through. An important
advantage of this instrumental configuration is the much more inert fused silica sample transfer
path to the capillary column. Kolb [45] and Wylie [46] both used automated multi-sample headspace
analysers but with whole column cryofocusing to allow splitless injection of the vapour. This
apparatus allows multiple headspace injection, which involves several rapid headspace injections
before the start of a chromatographic run. In this way the injections are superimposed by the
cryofocusing and sensitivity can be dramatically increased. Barcarolo and colleagues [47] avoided
cryogenic cooling and reconcentrated solutes by using a pre-column coated with graphitised
carbon. This led from the headspace injector to a 3-way press-fit connector, linked to the analytical
column and a vent line. Better peak shapes were obtained from splitless injections of ground
coffee headspace, but recovery of highly volatile substances was not quantitative.
Use of the cooling functions of a PTV injector offers an elegant solution for the cryofocusing
necessary for large volume headspace injection. In addition the liner can be packed with various
adsorbents in order to act as a short pre-column with contributory retention capability. This
approach was investigated by Poy and Corbelli [48] with additional evaluation of the trapping
efficiences of various adsorbents. A carrier gas line and sample transfer tube pneumatically
connected the headspace sampler to the GC. A probe end at the tip of the sample transfer tube
entered through the septum of the PTV inlet port.
The Gerstel headspace sampler also uses the Gerstel PTV injector and has been designed to
incorporate all the best features necessary for optimum headspace results. In particular the
headspace transfer line to the PTV injector has been replaced by a gas tight autosampler syringe
to provide a totally inert path for the sample. This system has a vial preheating module which
allows each sample to be heated at the same temperature for the same length of time. For injection
the sample is drawn from the headspace vial into the heated syringe and for multiple sampling
from a single vial the syringe can be filled before each injection with a fixed volume of inert gas.
To optimise the system for different application the following parameters can be individually
programmed from the software :
Sample preheating temperature.
-
-
Sample preheating time.
- Internal delay for multiple injections.
- Vial filling with inert gas before each injection.
- PTV injection temperature profile.
-
-
Needle penetration depth into the vial.
Syringe filling speed during rinsing and sampling.
- Injection depth into the PTV for different liner types.
Similar to the large volume liquid injector the headspace sampler uses the standard HewlettPackard sample tray for up to one hundred 2 ml vials. Figure 27 shows 1,3 and 5 ml headspace
injections of 10 pg/1 dimethyl sulfide and dimethyl disulfide in 20% aqueous ethanol with specific
sulfur detection. For this analysis the PTV liner at -30~ was packed with a little Carbotrap
during the headspace injections and a small split flow is established. After the single or multiple
333
injections the system then switches automatically to splitless operation and the PTV is heated to
transfer solutes of interest to the column. Using this technique the sensitivity increase over
corresponding liquid injection can be 103, depending on the volatility of the solutes.
Analysis Condition.s:
Headspace: Turret (Vial) 60~ Syringe 60~
Oven: 60~
2~
to 80~ 5~
to 220~
Inj.: PTV,-50~ 10~ to 250~
SCD
Carriergas: 20 cm/s He
headspace transfer into PTV liner packed with Carbotrap at -50~ in
split mode, then splitless transfer of compounds to column (2min)
340000 84
300000-
260(0
Dimethyl D i s u l f i d e . . _ . 1 x 1ml injected
220000
.
0
2
4
6
8
10
.
_
.
.
.
.
.
12
_
-
_
_
14
16
340000
3(X)0(O-
Dimethyl Sulfide
Dimethyl Disulfide
260000
3 x I ml injected
220000
9
0
2
4
6
8
10
12
14
-
,
.
,
16
Dimethyl Sulfide
340000-
Dimethyl Disulfide
300000-
260000
5 x 1ml injected
220000
0
2
4
6
8
10
12
14
91'6 ....
Figure 27. Specific sulfur traces of a headspace standard, l Oppb each dimethyl sulfide and
dimethyl disulfide in 20% v/v ethanol, on 25m x 0.32mm x 5.011rn 5% Phenyl Methylsilicone
column.
334
Both detection limits and the functionality of the PTV can be further enhanced by incorporating
its operation with a thermodesorption unit (Gerstel TDS-2) which itself can be temperature
programmed. In this approach an adsorption tube containing the solutes of interest from a large
volume purge and trap sampling is placed in the thermodesorption unit, which is connected directly
to the PTV. The thermodesorption oven is then heated in a controlled programmed manner to
desorb the components which are cyofocused in a cooled PTV liner [49-51 ]. Sample transfer to
the column can be subsequently carried out in normal split or splitless mode of operation, with
heating of the PTV. An interesting variant of this approach is to place the actual material for
analysis (usually a solid) in an empty thermodesorption tube and purge the volatiles directly to the
PTV for enrichment under cryofocusing. In this dynamic headspace approach the sample is kept
at low heat but purged for an extended period to transfer the maximum amount of purgeable
volatiles to the cooled PTV. Figures 28 and 29 show the nitrogen traces from a normal and a high
colour speciality malt after this instrumental gas phase extraction and enrichment. The extra
roasting involved in producing the high colour malt has generated a much higher proportion of
flavour active nitrogen compounds.
800000Analysis Conditions:
Thermodesorption: 20~ 20~
to 80~
Oven." 40~
5~
to 170~ 15~
to 300~
lnj.: PTV,-150~ 12~ to 280~
NPD
Carriergas." 100 kPa He
thermal extraction transfer into PTV liner at -150~ in split mode,
then splitless transfer of compounds to column (2min)
i
600000
400000i
2oooooi
0
i0
20
30
40
50
Figure 28. GC-NPD trace of a normal malt after thermal extraction and PTV enrichment.
80(0)00
600000
4ooooo!
2oooooi
oi~
0
10
20
30
40
50
Figure 29. GC-NPD trace of a high colour malt after thermal extraction and PTV enrichment.
335
However Figures 30 and 31 show similar GC nitrogen traces of the extracts of the corresponding
whiskeys from the same two malts, and using the same chromatographic separation conditions.
Increased levels of nitrogen compounds are also evident in the whiskey from the high colour
malt, but these are different more volatile compounds than those from the high colour malt itself.
The additional conversion of starch to sugar inherent in this malt is most likely contributing to
production of volatile nitrogen compounds via Maillard reaction during distillation of the fermented
wash.
500000Analysis Condition.s:
Oven: 40~
5~
to 170~ 15~
Inj.: PTV, 60~ 12~ to 280~
Carriergas: 100 kPa He
llal split injection (1/30)
400000
_
30(0)00
to 300~
NPD
200000 ~
10(0)00
k._au__a___
~
1
0
10
20
30
40
Figure 30. GC-NPD trace of a whiskey extract from the normal malt.
50
500(0)0
40(0)00300000
200000
looooo
0
10
20
30
40
50
Figure 31. GC-NPD trace of a whiskey extract from the high colour malt.
MULTIDIMENSIONAL GAS CHROMATOGRAPHY
It has been clearly pointed out by various authors [52, 53] that the resolution afforded by modem
efficient capillary columns is really insufficient for separation of the component content of many
real-life samples. A normal capillary column contains about 100,000 plates, but Giddings [52]
calculates that 500 million theoretical plates would be required for 0.99 separation probability of
a 100 component mixture. Viewed in a different way Smits [53] estimates that the total number of
336
peaks that can be separated on the normal capillary column is only 0.1% of all known volatile
compounds. By serially coupling capillary columns of different selectivities this basic resolution
disadvantage can be dramatically lessened, and the subsequent appearance of a hidden world of
compounds vividly highlights the limitation of one dimensional chromatography.
Many compounds still remain unidentified in distilled spirits and these most likely are trace
compounds that remain undetected due to resolution difficulties in one dimensional capillary
chromatography. It is also probable that these compounds could be important sensory contributors
and progress in understanding perceived taste and aroma of spirits will require their separation
and identification. Sample manipulation and fractionation of spirits can also produce enriched,
highly contributory sub-fractions, but these are usually also enriched in many matrix compounds,
and the basic resolution problem still remains. In our laboratory (K MacN) we have been working
with a capillary two-dimensional hyphenated system for some years and in our experience two
principle aspects are necessary for successful implementation of this approach in an industrial
environment.
- The transfer of compounds should preferably be by the "valveless" technique, which uses
pneumatic flow switching and balancing of gas pressures [54]. However if this pressure
balancing is a manual time-consuming, trial and error procedure and not an automated
user-friendly operation, it will not gain acceptance in routine laboratories.
- The construction of the hyphenated system should incorporate as much as possible existing
GC and GC-MS hardware, and allow maximum flexibility and interconversion between use
of the components as stand-alone units or linked to form the full system.
In the context of the latter point Figure 32 shows the stepwise build-up of a powerful 2-dimensional capillary system (Gerstel MCS-A) from an existing Hewlett-Packard GC-MSD and
Chemstation. In the first step a second GC is added which functions as an additional oven to
house a pre-column, and contains the column coupling device, the pressure and flow pneumatics
for switching, and a heated interface through which the main column passes to join the precolumn in the first oven. The heated interface can be further upgraded to a cryotrap (CTS) for
refocusing of volatile compounds, and the CTS itself can be upgraded to a cryo-enrichment
device (CTE) which allows coupling of different diameter columns with each retaining its optimum
gas velocity. Finally a PTV injector with an optional multipurpose sampler (large volume or
headspace) can be added to either of the two GC ovens. When added to the first oven the additional second separation dimension becomes possible and an example using initial large volume
headspace injection is shown later.
Modem GC and GC-MS Chemstation software is both multi-instrument and multi-tasking
and this attribute can be used to good advantage for maximum flexibility of such a hyphenated
system. This means that when not being used for 2-dimensional work the total system is in effect
a collection of separate units, all working independently and simultaneously, and controlled from
the Chemstation. The first oven could be running a GC-ECD analysis on an apolar column, and
the second oven operating in GC-MSD on a polar column. If PTV injectors (Gerstel CIS-3) are
fitted to one of both ovens then large volume injection with solvent elimination, or headspace or
thermal desorption with PTV cryofocusing is also possible. Control of all these devices is also via
the Chemstation under MS Windows.
337
Existing GC/MSD with PTV
I:PTV
2: Existing GC
3: Main Column
4: MSD
5: PC
4
2_
r
4
5
,
V
7I
7
Add GC/PTV with column switching
hardware and pneumatics
6: New GC
7: Column Switching Device
8: Monitor Detector
9: Precolumn
10: Heated Interface
Upgrade interface to allow both
heating and cooling
1O: Cryotrap
Upgrade cryotrap with enrichment
option
11: Enrichment Device
4
5
f
Add multipurpose sampler for large
volume liquid and headspace injection
or thermal desorption system to either
PTV injector
12: Multi Purpose Sampler or
Thermal Desorption System
F i g u r e 32. Modular build-up of a 2-dimensional GC/MSD system.
338
Conversion to a two-dimensional unit, and optional deconversion back to independent operation,
must be simple and rapid to maintain the overall flexibility of such a system. This is achieved by
means of a unique column switching device which couples both columns, and computer controlled
pneumatics for automatic pressure equilibration after transfer of components. The miniature
switching device remains as a permanent fixture in the first oven and the specialised pneumatics
are also factory installed, so that only additional column connections and Chemstation parameter
entries are required during conversion. Column connections in the switching device are with
Graphpack technology which assures leak-free coupling.
Figure 33 shows the switching principle in the column switching device. The zero condition is
defined as an uninterrupted carrier gas flow through both columns and this implies total transfer.
Compounds of no interest are prevented reaching the main column by a countercurrent venting
flow through the switching device supplied by an electronic mass flow controller. This must be
greater than the carrier gas flow and functions both to vent unwanted components and supply the
original zero condition carrier gas flow in the second column. Therefore this venting flow is
normally activated and must be switched off for transfer of compounds to the second column.
Figure 33. Switching principle in column switching device.
The venting flow line exists from the switching device to an electronic equilibration proportional
valve with a build-in pressure sensor (Figure 34). This unit continually reads the pressure at the
column switching device and automatically re-equilibrates during transfer of compounds without
any operator need for repetitive pressure manipulations. The only Chemstation parameter entries
required are the initial zero condition pressure and the desired compound transfer times.
339
Figure 34. Flow scheme of Gerstel switching pneumatics.
Figure 35 shows the additional pneumatics required when transfer operations with columns of
different dimensions (0.53 mm pre-column to 0.25 mm main column) are required. Here again an
electronic proportional valve is used but without a mass flow controller so that the action of the
valve is to automatically and continuously effect a carrier gas split flow to achieve a pressure
reduction before the main column to allow its optimum carrier gas velocity. This allows a high
flow through a high capacity megabore precolumn and a low optimum flow through a narrow
bore main column which could be connected to an MS. Compounds are transferred at the precolumn
flow rate and trapped in the cryointerface. The pressure reducing split flow is then temporarily
turned off for temperature programmed transfer of these compounds to the main column.
Figure 35. Flow scheme of Gerstel pneumatics to couple columns of different diameters (cryotrap
enrichment option).
340
MacNamara used such a system with automatic pressure equilibration to investigate medium
boiling sulfur compounds in whiskey [55]. The compounds were first isolated and enriched from
whiskey using a combination of fractional distillation and preparative GC. This process however
also resulted in a complex matrix of non-sulfur compounds which would have made mass
spectrometric identification very difficult. Therefore the compounds of interest were located on a
polar precolumn of a two-dimensional configuration and cut in groups to an apolar main column.
By establishing a sulfur trace with retention times for each of these cuts on the main column
(Figure 36), and then repeating each cut to an MSD, clean mass spectra for all the compounds of
interest could be recorded. This procedure led to the identification of a number of new sulfur
compounds [55, 31 ].
Figure 36. Polar to apolar cutting of whiskey sulfur compounds with alternate FID and SCD
detection on pre- and main column.
341
Van Ingen and co-workers used a two-dimensional approach to detect and quantify the restricted
compound ethyl carbamate at the ppb level in alcoholic beverages [56]. The sample was extracted
with dichloromethane and after concentration and injection the ethyl carbamate was heart-cut
from an apolar to a polar column. An FID detector could be used after the second column and the
much increased resolution on the second column prevented false positive results. MacNamara
and Hoffmann also determined ethyl carbamate in whiskey using a two-dimensional approach but
with large volume direct injection and selected ion monitoring after the second column [57].
20 lal of spirit could be injected into a PTV liner with solvent venting and the majority of the ethyl
carbamate content was retained for transfer to the chromatographic system. This combination of
large volume injection and selected ion monitoring resulted in reproducible quantification at the
5 lag/l level with three ion target confirmation of positives.
When headspace injection is used with PTV for large volume vapour sampling significant
sensitivity advantages can be achieved. Figure 37 shows the FID and specific sulfur traces of the
volatile fraction of a whiskey on a thick film polar capillary after injection of 5 ml of headspace
vapour to a cooled PTV injector.
Figure 37. 5 ml headspace injection of whiskey volatiles, precolumn chromatograms, FID(top) and sulfur trace on 50m x 0.53mm x 1.Ol~mCarbowax 20M.
342
When the same sample is injected as a liquid the same trace outline is obtained but with areas less
by a factor of 100. The indicated cut region, which contains two sulfur compounds, is then
transferred with cryofocusing to a thick film apolar column in the second oven and Figure 38
shows the corresponding FID and sulfur trace of the cut. Both sulfur compounds are recovered
with increased resolution for spectral investigation, and the complexity of the FID trace after
cutting highlights the hidden complexity of the sample [58].
12000
FID-trace
10(~
8000-
II l[ II
Sulfur
Compoundllll
60004000-
9
0
5
10
15
~
.
.
.
-
-
-
-
-
n
20
400000SCD-trace
300000- Analysis Conditions:
Oven: -50~
cut), 70~
to 60~
2~
to 80~ 5~
to 220~
Det.: FID and SCD
20(0)00- Carriergas: 20 cm/s He
10(0)O
0
5
f0
15
20
Figure 38. FID (top) and sulfur trace on 25m x 0.32mm x 5.01am 5%Phenyl Methylsilicone
column (main column).
Multidimensional capillary GC can also be used to preparatively isolate pure compounds from
complex mixtures. In Figure 39 the first trace (A) shows the separation of a concentrate of yeast
extract volatiles, and peaks numbered 1,2 and 3 could not be identified by GC-MS. These unknown
compounds were then isolated as pure substances by repetitive injection, cutting, and trapping in
a 2-column system (B and C). Regions of interest are transferred from the pre-column to the main
column, and after separation on the main column, the single compounds are further cut and
isolated in cooled glass traps [59]. After 55 injection cycles sufficient amounts of pure components
were available to allow structure elucidation by 1H-NMR, IR and MS.
343
1
jL
A: Analytical Separation
Figure 39. Isolation of cis- and trans-2,4,5-trimethyl-5-hydroxy-3-thiazoline and 2-isobutyl4,5-dimethyl-3-thiazoline from yeast extracts.
344
REFERENCES
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
D. de Rijke and R. ter Heide; Flavour Compounds in Rum, Cognac and Whiskey; Flavour of Distilled
Beverages, Ellis Horwood Limited (1983).
P. Salo, L. Nykanen and H. Suomalainen; Odour Thresholds and Relative Intensities of Volatile Aroma
Compounds in an Artificial Beverage Imitating Whiskey;J. Food Sci. (1972), 37, 394-398.
R. Duncan and J. Philp; Methods for the Analysis of Scotch Whiskey; J. Sci. Fd. Agric. (1966), 17,
208-214.
R.L. Brunelle; Collaborate Study of the Quantitative Determination of Fusel Oil and Ethyl Acetate by Gas
Liquid Chromatography; J. Assoc. Anal. Chem (1968), 5"/, 915-921.
J. Kahn and E. Blessinger; Collaborate Study of the Quantitative Gas Liquid Chromatographic
Determination of Fusel Oil and Other Components in Whiskey; J. Assoc. Anal. Chem. (1972), 55,
549-556.
M. Cabezudo, E. Corostiza, M. Herraiz, J. Fernandez-Biarge, J. Garcia-Dominguez and M. Molera;
Isothermal Selective Separation of Alcoholic and Acetic Fermentation Products; J. Chr. Sci. (1978), 16,
61-67.
A. DiCorcia, R. Samperi and C. Severini; Modified Graphitized Carbon Blackfor the Gas Chromatographic
Analysis of Alcoholic Beverages; J. Chromatography (1979), 170, 245-248.
G. Martin, J. Burggraff, R. Dyer and P. Buscemi; Gas Liquid Chromatographic Determination of Congeners
in Alcoholic Products with Confirmation by Gas Chromatography/Mass Spectrometry; J Assoc. Anal.
Chem. (1981), 64, 186-190.
Restek Corporation; The Restek Advantage (1994), Vol. 5, No. 4.
L. Gonzalez and R. Gonzalez-Lara; Selective Separation of Alcoholic Beverages and Distilled Wines by
Gas Chromatography; J. Chr. Sci. (1994), 32, 195-198.
W. Jennings; Modem Methods in Food Analysis; AVI Publishing Company, Inc. (1984), Chapter 13.
K. Grob, H. Neukom and H. Kaderli; Higher Alcohols in Alcoholic Beverages by Direct Analysis on Glass
Capillary Columns; J. High Res. Chr. (1978), 1, 98.
R. de Nijs and J. de Zeeuw; A Chemically Bonded Polyethylene Glycol Stationary Phase for Gas
Chromatography; J. High Res. Chr. (1982), 5, 501.
K. MacNamara; Rapid Determination of Major Congeners in Distilled Spirits by Direct Analysis on Bonded
Capillary Columns; J. High Res. Chr. (1984), 7, 641.
M. Masuda, M. Yamamoto and Y. Asakura; Direct Gas Chromatographic Analysis of Fusel Alcohols,
Fatty Acids and Esters of Distilled Alcoholic Beverages; J. Food Sci. (1985), 50, 264-265.
K. MacNamara and N. Burke; Determination of Free Fatty Acids and Other Major Congeners in Distilled
Spirits by Direct Analysis on a Bonded FFAP Capillary Column; J. High Res. Chr. (1985), 8, 853-855.
Irish Distillers Group Ltd., Internal Report.
G. Reglero, M. Herraiz and M. Cabezudo; Sampling in Capillary Gas Chromatography: A Comparison
between the Split and the PTV-Split Procedures; Chromatographia (1986), 7-12, 333-336.
K. Eder, A. Reichlmayr and M. Kirchgessner; Gas Chromatographic Analysis of Fatty Acid Methyl Esters:
Avoiding Discrimination by Programmed Temperature Vaporizing Injection; J. Chromatography (1991),
588, 265-272.
J. Requeiro, I. Diaz, E David and P. Sandra; Possibilities of PTV Injection for the Analysis of Triglycerides;
J. High Res. Chr. (1994), 17, 180-183.
H. Mol, H.-G. Janssen, C. Cramers and U. Brinkman; Large VolumeSample Introduction Using Temperature
Programmable Injectors: Implications of Liner Diameter; J. High Res. Chr. (1995), 18, 19-27.
Target 3.2 Operators Manual; Thru-Put Systems, Inc., P.O. Box 617290, Orlando, Florida 32861-7290,
USA.
G. Hagman and J. Roeraade; Rapid Determination of Volatiles in Alcoholic Beverages by PrecolumnBackflush Gas Chromatography; J. High Res. Chr. (1989), 13, 99-103.
K. Shiomi; Determination of Acetaldehyde, Acetal and Other Volatile Congeners in Alcoholic Beverages
Using Multidimensional Gas Chromatography; J. High Res. Chr. (1991), 14, 136-137.
H. Mandery; Gaschromatographisch-Massenspektrometrische Untersuchung fliichtiger lnhaltsstoffe des
Traubenmost- und Weinaromas: Auswirkung der Siissung auf die Aromazusammensetzung; Dissertation
Univ. Karlsruhe, Germany (1983).
345
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
A. Rapp, I. Yavas and H. Hastrich; Einfache und schnelle Anreicherung (,,Kaltronmethode") von Aroma
stoffen des Weines und deren quantitative Bestimmung mittels Kapillargaschromatographie; Deutsche
Lebensmittel-Rundschau (1994), 90, 171-174.
A. Rapp, H. Hastrich, I. Yavas and H. Ullemeyer; Zur einfachen, schnellen Anreicherung (,,Kaltronmethode")
und quantitativen Bestimmung von fliichtigen lnhaltsstoffen aus Spirituosen; B ranntweinwirtschaft (1994),
1, 286-289.
E Liddle and A. Bossard; The Analysis of Wines and Spirits by Capillary Gas Chromatography; Alcoholic
Beverages, Elsevier Applied Science Publishers (1984), Chapter 7.
R. ter Heide, E de Valois, J. Visser, P. Jaegers and R. Timmer; Concentration and Identification of Trace
Constituents in Alcoholic Beverages; Analysis of Food and Beverages, G. Charalambous (ed.), New York
(1978), 249-287.
R. ter Heide; Instrumental Analysis of Alcoholic Beverages; Proceed. Syrup. on Flavour Research of Alcoholic
Beverages, L. Nyk/inen and E Lekkonen (ed.), (1984), 149-165.
T. Taniguchi, N. Myajima and H. Komura;An Application of Centrifugal Counter-Current Chromatography
on Flavour Chemistry: Separation of Aroma Substances in Whiskey New Distilled.
J. Hinshaw; Very-Large-VolumeInjection; LC-GC International (1994), Vol 7, No. 10.
K Grob; Solvent Effects in Capillary Gas Chromatography; Proceed. 5th Int. Symp. on Capillary
Chromatography, Elsevier Scientific Publishing Company (1983), 254-265.
K. Grob; Classical Split and Splitless Injection in Capillary GC; Hiithig Verlag, Heidelberg, Germany
(1986).
W. Vogt, K. Jacob and H. Obwexer; Sampling Methods in Capillary Column GLC Allowing Injections of
up to 250/~l; J. Chromatography (1979), 174, 437-439.
W. Vogt, K. Jacob, A.-B. Ohnesorge and H. Obwexer; Capillary GC Injection System for Large Sample
Volumes; J. Chromatography (1979), 186, 197-205.
K. Grob; PTV-Vapor Overflow - Principles of a Solvent Evaporation Techniquefor Introducing Large
Volumes in Capillary GC; J. High Res. Chr. (1990), 13, 540-546.
H. Mol, H.-G. Janssen, C. Cramers and U. Brinkman; Use of a Temperature Programmed Injector with a
Packed Liner for Direct Water Analysis and On-Line Revesed Phase LC-GC; J. High Res. Chr. (1993), 16,
459-463.
M. Herraiz, G. Reglero, E. Loyola and T. Herraiz; Sampling of Volatile Components Using a PTV in the
Solvent Split Mode; J. High Res. Chr. (1987), 10, 598-602.
J. Villen, T. Herraiz, G. Reglero and M. Herraiz; Experiments with the PTV in the Solvent Split Mode for
Concentration of Volatiles; J. High Res. Chr. (1989), 12, 633-635.
K. Grob; On-Column Injection in Capillary Gas Chromatography; HiJthig Verlag, Heidelberg, Germany
(1987).
J. Staniewski and J. Rijks; Solvent Elimination Rate in Temperature Programmed Injections of Large
Sample Volumesin Capillary Gas Chromatography; J. Chromatography (1992), 623, 105-113.
S. Miiller, J. Efer and W. Engewald; Gas Chromatographic Water Analysis by Direct Injection of Large
Sample Volumes in an Adsorbent-Packed PTV Liner; Chromatographia (1994), 38, 694-700.
G. Takeoka and W. Jennings; Developments in the Analysis of Headspace Volatiles: On-Column Injections
into Fused Silica Capillaries and Split Injections with a Low-Temperature Bonded PEG Stationary Phase;
J. Chr. Sci. (1984), 22, 177-184.
B. Kolb and B. Liebhardt; Cryofocussing in the Combination of Gas Chromatography with Equilibrium
Headspace Sampling; Chromatographia (1986), 21, 305-311.
P. Wylie; Headspace Analysis with Cryogenic Focussing: A Procedure for Increasing the Sensitivity of
Automated Capillary Headspace Analysis; Chromatographia (1986), 21, 251-258.
R. Barcarolo and P. Casson; Volatile Enrichment on Automatict Static Headspace Using a Precolumn and
Stopping the Carrier Gas Flow During Injection; J. High Res. Chr. (1995), 18, 5-8.
F. Poy and L. Corbelli; Automatic Headspace and Programmed Temperature Vaporizer (PTV) Operated in
Cryo-Enrichment Mode in High Resolution Gas Chromatography; J. Chr. Sci. (1985), 23, 114-119.
A. Hoffmann and R. Bremer; Design, Performance and Applicability of a Multi-Functional Thermal
Desorption System for Trace Analysis in Capillao, GC; Proceed. 16th Int. Symp. on Capillary
Chromatography, E Sandra and G. Devos (ed.), Riva del Garda, Italy (1994), 1165-1175.
346
50
51
52
53
54
55
56
57
58
59
A. Hoffmann and W.R. Sponholz; Direct Thermal Analysis of Solids - A Fast Method for the
Determination of Halogenated Anisols and Phenols in Cork; Proceed. 16th Int. Symp. on Capillary
Chromatography, P. Sandra and G. Devos (ed.), Riva del Garda, Italy (1994), 1011-1020.
H. W6rmann and A. Hoffmann; Trace Analysis of PAH's and PCB's in Soil through On-Line Direct
Thermal Desorption; Proceed. 16th Int. Syrup. on Capillary Chromatography, P. Sandra and G. Devos
(ed.), Riva del Garda, Italy (1994), 416-424.
J. Giddings; Multidimensional Chromatograph), Techniques and Applications; Marcel Dekker, Inc. (1990).
R. Smits; Hyphenation - An Overview; LC-GC International (1994), Vol. 7, No. 9, 505-508.
D. Deans; A New Technique in Heart-Cutting in Gas Chromatography; Chromatographia (1968), 1,
18-22.
K. MacNamara; Investigation of Medium Boiling Sulfur Compounds in Whiskey; Elaboration et
Connaissance des Spiritueux, Lavoisier-Tec et Doc, R. Cantagrel (ed.), Paris (1993).
R. van Ingen, L. Nijssen, E. van den Berg and K. Maarse; Determination ofEthyl Carbamate in Alcoholic
Beverages by Two-Dimensional Gas Chromatograph)'; J. High Res. Chr. (1987), 10, 151-152.
K. MacNamara and A. Hoffmann; Use of a GC/MSD Multi Column Switching System for Direct
Determination of Ethyl Carbamate in Alcoholic Beverages with Three Ion Confirmation of Positives;
Proceed. 13th Int. Symp. on Capillary Chromatography, P. Sandra (ed.), Riva del Garda, Italy (1991),
1662-1672.
Irish Distillers Group Ltd., Internal Report.
P. Werkhoff, W. Bretschneider and S. Brennecke; Multidimensional Preparative Capillary Gas
Chromatography in Flavour Research; Gerstel Aktuell (1991), No. 12, Gerstel GmbH, Aktienstrasse
232-234, 45473 Miilheim an der Ruhr, Germany.
D. Wetzel and G. Charalambous (Editors)
Instrumental Methods in Food and Beverage Analysis
9 1998 Elsevier Science B.V. All rights reserved
347
Analytical methods for color and pungency of chiles (capsicums).
M.M. Wall and P.W. Bosland
Department of Agronomy and Horticulture, New Mexico State University,
Box 30003, Las Cruces, New Mexico, United States
INTRODUCTION
Worldwide, there are thousands of chile varieties varying widely in their size,
shape, color, flavor, and pungency level. The uses for these chiles are equally as
diverse as the fruit types in the Capsicumgenus. Chile fruit are primarily consumed as
a fresh vegetable or dehydrated for use as a spice. However, the range of food products
that contain chile or its chemical constituents is broad, and includes ethnic prepared
foods, meats, salad dressings, mayonnaise, dairy products, beverages, candies, baked
goods, snack foods, breadmgs and batters, salsas and hot sauces. Chile extracts are also
used in pharmaceutical and cosmetic products. Quality standards differ for these varied
products, especially in the food industry. However, almost all fresh or dried chile is
evaluated for color and pungency at several stages, from harvest and preprocessing, to
f'mal product formulation.
The yellow, orange and red chile colors originate from the carotenoid pigments
produced in the fruit during ripening. Over 25 different pigments have been identified
in chile fruits (1). These pigments include the green chlorophylls (a and b); the yelloworange lutein, zeaxanthin, violaxanthin, antheraxanthin, [3-cryptoxanthin and [3carotenes; and the red pigments, capsanthin, capsorubin and cryptocapsin, that are
distinctive to Capsicums. Chile carotenoids are important not only as food colorants,
but also for their immense nutritive value. The capsanthin, capsombin and cryptocapsin
pigments are valued mostly as natural colorants, whereas B-carotene, a-carotene, ycarotene and 13-cryptoxanthin have provitamin A activity. These provitamin A
carotenoids are essential to human nutrition, and the oxygenated carotenoids
(xanthophylls) have been studied as anti-cancer agents (2).
Besides color (carotenoids), another important quality attribute of Capsicum is
pungency (heat). Some have argued that pungency is one of the five main taste
sensories, along with bitter, sweet, sour, and salty. Physiologically, the senses
responsible for our perception of flavor can be divided into three anatomical systems.
In the oral cavity, the classical gustatory pathways through the tongue and soft palate
are responsible for our sensitivity to the four basic tastes, sweet, sour, salty, and bitter.
In the nasal passages, the olfactory receptors provide sensitivity to a wide variety of
volatile compounds, producing the sensations we normally assign to smell. In addition
to these two systems, the trigemmal nerves in both the oral and nasal cavities provide
348
sensitivity to thermal, tactile, irritation, and pare sensations (3). The trigeminal
innervation is also chemically sensitive to compounds that are pungent, and hence
provide an important part of our appreciation of flavor as a whole. Because the
capsaicmoids are potent stimuli of the oral trigemmal nerves they are a desirable
attribute of many foods. In most parts of the world, pungency increases the acceptance
of the insipid basic nutrient foods.
The word "pungency" can be confusing. Some prefer "hot flavor", heat, fiery,
or spicy to that of pungency. In this chapter, the sensory response will be identified as
pungency and the substance responsible for pungency by its chemical name, capsaicm,
dihydrocapsaicin, etc. The capsaicmoids can be analyzed or estimated by their physical
or chemical characteristics, but the pungency of a chile product can only be validated
through a correlation with the perceived heat associated with oral consumption. This
has become very relevant in recent years because of the complexity of the food items
containing capsaicmoids.
There is a preference for specific levels of pungency in internationally traded
chile products, and with paprika, the absence of pungency is important. Red chile and
paprika are dehydrated and sold as whole pods, or ground into powder. The dried red
powder is classified into five groups based on pungency level: non-pungent or paprika
(0 to 700 scoville heat units), mildly pungent (700 to 3,000), moderately pungent
(3,000 to 25,000), highly pungent (25,000 to 70,000) and very highly pungent
(>80,000). The very highly pungent powder is mainly grown in Asia. Paprika may be
obtained from any one of many types of C. annuum. However, in the United States,
it is considered a product, not a pod-type. The Hungarian word for Capsicum is
"paprika." Thus, Hungarian paprika may be pungent or non-pungent, depending on the
cultivar. Other areas where a knowledge of individual capsaicmoid content is useful
are taxonomic, breeding, medicinal, and biosynthetic research purposes.
In this chapter, analytical methods for determining chile color and pungency are
described. Early methods estimated the total concentrations of either pigment or
pungency compounds using sensory evaluations or spectrophotometry. These
techniques were primarily used for quality control checks in the food industry.
However, as the food industry has evolved and become more competitive, precise
methods to analyze quality components are necessary. High performance liquid
chromatography (HPLC) facilitates the accurate separation and quantification of chile
quality components. HPLC is rapidly replacing older methods of compositional
analysis by the food and medicinal industries, especially in the area of pungency
analysis.
CHILE COLOR ANALYSIS
Chile color can be evaluated from 3 different perspectives: surface color,
extractable color, and carotenoid profiles. Surface color is a measurement of the visual
color perceived by the viewer. It is sometimes referred to as reflective color. Surface
349
color varies according to cultivar, growing conditions, dehydration and storage
conditions, and the coarseness of ground samples. Surface color measurements are
important when dehydrated chile is to be used as a retail spice or as a coating on foods.
Extractable color is a measurement of total pigment content. Extractable color analyses
are useful when chile is added as an ingredient or colorant in oil-based foods,
cosmetics, or pharmaceuticals. Extractable color and surface color measurements are
standard quality evaluations in the spice industry. Analytical methods that separate and
quantify individual chile carotenoids, providing pigment profiles, are used mostly for
research and development. HPLC is the most accurate method and is being used
increasingly by oleoresm, drug, and vitamin manufacturers for routine analysis.
Measuring Surface Color
Surface color measurements are used to specify colors perceived by the human
eye. Verbal descriptions of colors can be difficult and confusing, because two people
may describe the same color in very different terms. The perception of color varies
according to the sensitivity of an individual's eyes, the size of the object being viewed,
the light source for illumination, the background color and contrast, and the angle at
which an object is viewed. Quantifying colors, or expressing colors numerically,
facilitates color communication and standardization. Visual color can be quantified
using a colorimeter (color difference meter or chromameter). Several colorimeters are
available including the Gardner color difference meter, the Hunter colorimeter and the
Minolta chromameter. The older colorimeter models are typically large instruments that
are confined to laboratory use. Several portable models are now available such as the
Hunter Miniscan (Reston, VA), the X-Rite 918 (Grandville, MI) and the Mmolta CR300 series (Ramsey, NJ) (4). These instruments enable the capsicum industry to
quantify, and therefore set standards for, surface color.
The Commission Intemationale de rEclairage [(CIE) International Commission
on Illumination] established a tristimulus color system commonly used for surface
color measurements (5). The tristimulus values, XYZ, were based on the theory that
the eye possesses receptors for three primary colors (red, green, blue), and that all other
colors are perceived as mixtures of these primary colors. The XYZ values were
determined from color-matching functions that corresponded to the eye's sensitivity at
various wavelengths of the visible specmun. The CIE developed the Yxy color space
from the tristimulus values. The Yxy color space (a numerical expression of color) was
improved to the L'a'b" system (5). The L'a'b" (CIELAB) color space is the most widely
used system today. This system is based on a 3-dimensional color space with 3
coordinates (L'a*b'). In the CIELAB system, color is represented spherically.
The elements of perceived color are lighmess, hue, and chroma and they are
determined from the L'a'b" coordinates. The L~ coordinate measures the value or
lighmess of a color and is located on the central (vertical) axis of the CIELAB color
space. This axis is achromatic and ranges from black (0) at the bottom to white (100)
350
at the top. The a* and b* values are chromaticity coordinates and indicate directions
away from the center of the color sphere. The a* coordinate denotes red when positive
and green when negative, and b" denotes yellow when positive and blue when negative.
Hue angle (h ~ and chroma (C) can be determined from the a* and b* coordinates. The
hue and chroma color aspects are easier to conceptualize than a* and b* values (6). Hue
sets the kind of color (red, yellow, blue, green, etc.) and equals the arctangent b'/a*. At
any horizontal cross-section of the color sphere, all hues are represented in a 360
degree circle (the color wheel). A sample with a hue angle of 0 ~ is purplish-red, 90 ~
is yellow, 180 ~ is bluish-green and 270 ~ is blue (4). Paprika samples typically have
hue angles between 30 ~ and 45 ~, which is the range of red to orange on the color
wheel. Chroma is a measure of color saturation or purity. A sample with a high chroma
is more vivid than one with a low chroma value, even though both samples may have
the same hue. Colors located near the central axis of the color space have low chromas
that indicate dull, achromatic colors, with more gray. Colors located near the periphery
of the color space are vivid. Chroma (C) is calculated from the square root of the sum
of @)2 and (b*)z
The capsicum industry can use these surface measurements to compare the
quality of lots or to set specifications for their products. Quality control technicians
may determine the optimum hue, chroma, and value for their product and communicate
this information to suppliers. For example, a sample with a hue angle of 30 ~ and a
chroma of 50 would be reddish-orange and bright. However, a sample with a hue angle
of 45 ~ and a chroma of 30 would be a dull, orange color. An understanding of the
elements of color makes communicating and comparing surface color standards less
complicated and more consistent for producers and processors.
Measuring Extractable Color-Total Pigment Content
The market value of paprika depends largely on its color intensity, because
paprika and its oleoresm are used principally as natural coloring agents. Therefore, the
spice industry requires a simple, reproducible method for measuring the total content
of red and yellow pigments in dehydrated capsicums. In an early method for estimating
the red pigments in paprika, a visual comparison between a sample extract and a
standard solution of potassium dichromate and cobaltous chloride was made by
panelists (7). The Essential Oil Association of America (EOA) adopted a standardized
color matching method to determine color values of capsicum oleoresins. This visual
matching method was subjective and was eventually replaced by spectrophotometric
methods (8).
Current procedures for measuring extractable color (total pigments) in
dehydrated capsicums and oleoresms were developed and approved by the Association
of Official Analytical Chemists (9, 10) and the American Spice Trade Association (11).
Extractable color is measured using a spectrophotometer and is designated in ASTA
units. Generally, the higher the ASTA color value, the greater the effect on the
351
brightness or richness of the final product. Paprika with 200 ASTA color units would
give a brighter red to a finished product than an equivalent amount with 100 ASTA
color units.
ASTA method 20.1 is the procedure used by the capsicum industry in the United
States. The technique is simple, does not require complex equipment, and is relatively
inexpensive. To analyze for color, dried capsicums are first ground to pass through a
1-mm sieve. Between 70 and 100 mg of the sample is weighed and transferred to a
100-mQ volumetric flask. Acetone (100 m0 is added to the flask, and the flask is
stoppered, shaken, and allowed to stand for 16 hours in the dark. The color intensity
of the extract is measured using a spectrophotometer set at a 460 nm wavelength and
calibrated with an acetone blank. A portion of the sample extract is transferred to a
spectrophotometer cuvette, and the absorbance is measured and recorded. The
absorbance of the extract should be between 0.30 and 0.70.
A standard glass filter (Standard Reference Material 930d, National Institute of
Standards and Technology, USA) is used in ASTA method 20.1 to account for
instrument variability. The previous ASTA method (20.0) used a chemical color
solution (potassium dichromate/coboltous chloride) for standardizing the instrument.
The absorbance of the standard glass filter is measured at 465 nm, and an instrument
correction factor (If) is calculated by dividing the absorbance reported by the NIST, by
the absorbance recorded on the instrument being used. The absorbance of the glass
filter should be determined each time the instrument is turned on. ASTA units are
calculated for the sample extract using the formula:
ASTA color - Absorbance of the sample extract x 16.4 x If
Sample weight in grams
The ASTA/AOAC method measures total pigment content without
differentiating between the concentrations of red and yellow pigments. In a method
described by Baranyai and Szabolcs (12), the total pigment concentrations can be
partitioned into contributions by the red and yellow components. The method is based
on the reduction of the red pigments with sodium borohydride to produce yellow
pigments. An increase in absorption at the maxima for the yellow pigments is
determined to provide a more exact measurement of total pigment concentration.
In the Baranyai and Szabolcs reduction method, paprika powder (0.5g) is
extracted with 100 mQ of benzene on a shaker for 30 minutes. A 10 mQ portion of the
extract is diluted to 50 mQ. The diluted extract (10 mQ) is mixed with 10 mQ of 96%
ethyl alcohol and divided into two test tubes. Sodium borohydride is added to one of
the test tubes, and after 40 minutes, sodium hydroxide is added and the solution turns
yellow. The solution is filtered and the absorbance is measured at 455 nm using a
spectrophotometer. The absorbance of the unreduced solution in the second test tube
is measured at 510 nm, (the absorption maxima of the major red pigments, capsanthin
and capsorubm) and compared to the absorbance of the reduced solution at 510 nm.
352
Using this method, pigment concentrations can be calculated as follows: red
carotenoids (mg/g) = 1.5 As10 x D and total carotenoids (mg/g) = A455x D. Ill these
calculations, As~0and A455are the absorbances measured at the respective wavelengths,
and D is the dilution factor. The concentration of yellow pigments can be determined
from the difference between total pigment concentration and red pigment
concentration.
Carotenoid Biosynthesis
The diverse and brilliant colors of chile fruit originate from the carotenoid
pigments present in the thylakoid membranes of the chromoplasts. In plants,
carotenoids are synthesized in both the chloroplasts of photosynthetic tissues and the
chromoplasts of flowers, fruit and roots. Chemically, carotenoids are lipid-soluble,
symmetrical hydrocarbons with a series of conjugated double bonds. The double bond
structure is responsible for the absorption of visible light. Carotenoids function as
accessory pigments for photosynthesis, but more importantly, as photoprotectants in
the plant. The primary function of 13-carotene and other carotenoids is to protect the
chloroplasts from photo-oxidative damage. In flowers and fruit, carotenoids are
important for the attraction of pollinators and seed dispersers.
Carotenoids are tetraterpenoid (C40) compounds synthesized from eight
isoprenoid units. Three molecules of the primary metabolite, acetyl-coenzyme A, form
mevalonic acid that is a precursor to the C5 compound, isopentenyl pyrophosphate.
Isopentenyl pyrophosphate is the precursor to plant terpenoids, and through a sequence
of chain elongations forms geranyl pyrophosphate (C10), famesyl pyrophosphate (C~5)
and geranyl-geranyl pyrophosphate (C20). Two molecules of geranyl-geranyl pyrophosphate form the intermediate, prephytoene pyrophosphate, and then phytoene (C40).
Pytoene is desaturated through a series of reactions into lycopene (C40H56), the direct
precursor to carotenoids. Lycopene undergoes cyclization reactions to form the cyclic
carotenes (i.e. [3-carotene) (13-16).
The enzymes catalyzing the conversion of isopenteny! pyrophosphate into
phytoene have been studied and identified, because they are not membrane-bound.
They include isopentenyl pyrophosphate isomerase, geranyl geranyl pyrophosphate
synthase (GGPS) and phytoene synthase. GGPS catalyzes three condensation reactions
in the conversion of isopentenyl pyrophosphate to geranyl geranyl pyrophosphate (13).
GGPS has been purified from Capsicumchromoplasts, and cDNA encoding for GGPS
activity has been isolated (17, 18). Phytoene synthase catalyzes the reactions forming
prephytoene pyrophosphate and phytoene and also has been isolated from Capsicum
(19, 20). The enzymes involved in the subsequent desaturation of phytoene to
lycopene, cyclization of lycopene to carotenes, and further hydroxylation steps are
tightly membrane-bound, making study more difficult. However, some of these
enzymes have been isolated from Capsicumchromoplasts. Phytoene desaturase (PDS)
catalyzes four desaturation steps from the colorless phytoene to the orange-red
353
lycopene, and lycopene cyclase forms the cyclic carotenoids (13). Phytoene desaturase
has been isolated from Capsicum (21) and lycopene cyclase activity has been
demonstrated in Capsicum chromoplasts (22).
Following formation of the cyclic carotenes, 13-carotene is hydrolyzed into the
xanthophylls, cryptoxanthin, and zeaxanthin. Xanthophylls are oxygenated derivatives
of carotenes. Zeaxanthin undergoes epoxidation to form antheraxanthin (zeaxanthin
monoepoxide) and violaxanthin (zeaxanthin diepoxide). These xanthophyll epoxides
are the direct precursors to the unique red carotenoids of chile fruits. Red chile fruits
synthesize the cyclopentenyl keto-carotenoids, capsanthin, capsorubm and cryptocapsm, in the chromoplasts. Cryptocapsin is formed from cryptoxanthm through the
intermediate, cryptoxanthm-5,6-epoxide. Antheraxanthm undergoes rearrangement to
form capsanthin, the most abundant red keto-carotenoid in chile fruits, and a small
amount of capsorubm. Violaxanthin is the primary pathway for capsorubin synthesis.
(23). The enzymes involved in xanthophyll biosynthesis remain unknown, but the
enzyme catalyzing the conversion of antheraxanthin and violaxanthin to capanthin and
capsorubm has been purified and cDNA has been isolated (24).
Separation and Quantification of Carotenoids
A strong interest in understanding carotenoid biosynthesis led to the
development of techniques to separate and quantify these pigments. Capsanthin, the
major pigmem in red Capsicumfruits, was isolated by Zeichmeister and Cholnoky (25)
in 1927 through extraction with petroleum ether. Carotenoids can be partitioned into
pigment classes (hydrocarbons vs. oxygenated xanthophylls) based on differing
polarities. This involves phase separations with immiscible solvents and column
chromatography. In most cases, thin-layer chromatography (TLC) follows the initial
partitioning steps (26).
Carotenoids are unstable when exposed to light, oxygen, or high temperatures.
Therefore, careful handling in the laboratory is required to prevent the oxidation and
degradation of the pigments. All procedures should be conducted in subdued light and
at low temperatures. Some extraction steps should be carried out in a nitrogen
atmosphere. Carotenoids are extracted with organic solvents, although no standard
method exists. Acetone, ethanol, methanol, and hexane are common extraction
solvents. For flesh fruit, ethanol or acetone serves as a dehydration agent and an
extraction solvent.
Thin-layer Chromatography
Working with bell peppers, Buckle and Rahman (27) developed a system using
column chromatography and thin-layer chromatography to separate and quantify
changes in pigments during the ripening of Capsicums. At the time, the method was
simple, rapid and inexpensive relative to the previous column chromatography methods
354
(26). Cellulose was used as the solid phase on the thin-layer plates, and various solvent
systems were tested as the mobile phase. Cellulose was chosen for its neutral
properties. Relative to aluminum oxide, magnesium oxide or silica adsorbents,
cellulose plates reduce the possibility of pigment isomerization. Pigment extracts of
immature and mature green fruit were chromatographed with a light petroleumacetone-propanol (90:10:0.25, v/v) solvent system. Extracts from partially and fully
ripe fruits were chromatographed in a hexane-propanol (99.9:0.1, v/v) mobile phase.
TLC bands containing a mixture of pigments were separated on columns containing
magnesium oxide-Hyflo Super-Cel (1:1 w/w) and rechromatographed on thin-layer
plates. The criteria for the identification of individual pigments were R~ values, band
position and color, diagnostic chemical tests, absorption spectra, and cochromatography with authentic standards. During the maturation and ripening of bell peppers
26 pigments were separated and identified. Rahman and Buckle (28) followed with a
study of five Capsicumcultivars at four stages of maturity.
Thin-layer chromatography of chile carotenoids has been described and used
successfully by several authors (26-30). However, TLC is a time-consuming method
and often requires further chromatography of unseparated fractions. Problems with
reproducibility, accurate quantification, and isomerization of the pigments are likely.
Although HPLC has replaced TLC for the separation and quantification of carotenoids,
TLC remains useful as a tool for identifying HPLC peaks.
HigJa Performance (Pressure) Liquid C~omatography Analysis
High performance liquid chromatography (HPLC) is the most effective and
accurate tool for the separation and quantification of carotenoids. The earliest report
describing HPLC methods for separating paprika carotenoids was in 1982 (31), but
many reports of improved methods have been published since then (32-43). Method
development and peak identification are easier to achieve with the newer instruments
equipped with diode array detectors and accompanying software. Also, the use of
microbore columns can decrease run times, reduce solvent use and waste, and improve
sensitivity. The quantitative determination of individual carotenoids by HPLC analysis
revolves sample preparation, extraction, saponification (optional), HPLC separation,
peak identification and quantification.
HPLC methods have been developed for the separation of either saponified or
unsaponified pigments. Paprika naturally contains carotenoids esterified with fatty
acids, as well as free (tmesterified) carotenoids. Saponification removes the fatty acids,
leaving only free pigments. HPLC analysis of saponified extracts separates free
pigments and provides information on the types of carotenoids present in a sample.
When unsaponified pigments are analyzed, the naturally occurring compounds
separated are hypophasic or "free" xanthophylls (zeaxanthin, capsombin, capsanthm,
violaxanthin, antheraxanthin), carotenoid monoesters (primarily capsanthin, capsorubin
and zeaxanthin monoesters), epiphasic carotenoids (cryptocapsin, cryptoxanthin, 13-
355
carotene), and carotenoid diesters (26, 44). The ratios between free and esterified
pigments, and between monoester and diester carotenoids, can be determined using
HPLC analysis of unsaponified samples (37).
Extraction Methods
Fresh chile samples are typically deseeded and destemmed, cut into small (1 cm)
pieces and stored a t - 2 0 C under nitrogen until extracted (40). Samples are
homogenized in acetone and extracted until all the color has been removed. Extracts
are combined and ethyl ether is added. A NaCI solution is added to separate the phases.
The pigments are contained in the ether fraction. After treatment with anhydrous
sodimn sulfate to remove the water, the volume of the ether fraction is reduced using
a rotary evaporator (40). Other methods use methanol with calcium carbonate or
magnesium carbonate for extraction of pigments from flesh chiles (1, 45). Samples are
extracted 3 times with methanol (allowing 18 to 24 hours between extractions) and
once with ethanol. Extracts are combined, diluted with ethanol, washed with water,
dried over anhydrous sodium sulfate, and evaporated to reduce the volume.
Samples of dried, ground paprika are extracted with acetone, mechanically
shaken for 10-30 minutes and left in the dark at room temperature for 4 to 16 hours (34,
37, 38). Addition of 0.5% BHT to the extracts minimizes oxidation (38). Extraction
with chloroform- 2- propanol- acetone (2:1:1 v/v/v) for 20 minutes at room temperature
has also been reported (35). An aliquot of the extract (about 5 m0 is evaporated to
dryness under nitrogen, redissolved in the HPLC eluent, and passed through a 0.45 ~tm
filter into HPLC vials.
Saponification
Saponification of chile extracts converts the carotenol fatty acid esters into
hydroxy-carotenoids. The diester carotenoids are especially sensitive to alkaline
conditions and undergo hydrolysis quickly (37). Saponification allows for HPLC
separation of pigments free from fatty acid groups and a simpler chromatogram than
with unsaponified carotenoids. Paprika or oleoresin samples are extracted in ethanol
(38), methanol (32) or acetone (40) with 2% BHT. In one method, a 60% aqueous
potassium hydroxide (KOH) solution is added and the samples are heated at 60C for
25 minutes in a nitrogen atmosphere (38). Other methods use 20% or 30% methanolic
KOH without heating (1, 3 l, 40), and a method using sodimn methoxide has been
proposed as slower and less destructive to labile carotenoids than the potassium
hydroxide method (37). After saponification, water is added, and the carotenoids are
exhaustively extracted and washed with hexane. The hexane fractions are combined,
washed with water until flee of alkaline, and dried over anhydrous sodium sulfate (38).
Solutions are evaporated to dryness, dissolved in the HPLC injection solvent and
passed through a microfilter for analysis (44).
356
Equioment and Columns
_
_
The basic components of an HPLC system are an injector, a solvent delivery
system, a colunm, a detector, an integrator and recorder, or a computer. Guard columns
protect the main column during extended use and are usually included. A single pump
is required for isocratic methods, whereas a ternary pump is needed when a gradient
is imposed. Newer instruments usually have ternary solvent delivery systems.
Carotenoids can be detected with a UV-vis variable wavelength detector or a
photodiode-array detector. A conventional UV-vis detector monitors elutmg
compounds at their maximum wavelength, and is designed to measure the absorbance
at a single point in the specmma at one time. The solvent flow must be stopped for
spectral scanning (if possible), and the peak can elute faster than the time required for
the spectral scan. The diode-array detector is designed to continuously scan the
absorbance of eluting compounds over the entire spectrum (or a selected range). The
diode-array detector facilitates the detection of compounds with different maximum
wavelengths, the identification unknown peaks based on their absorption spectra, the
confirmation of a peak's identity, the determination of peak purity, and the
quantification of non-separated peaks. The diode-array detector is therefore a powerful
tool for carotenoid analysis.
Carotenoids can be separated with either normal or reverse phase chromatography, but reverse phase methods are preferred because of lower column
equilibrium times and less pigment transformations during analysis (40). Several
methods are similar, but no standard method exists. Most methods for separating
unsaponified carotenoids include either an isocratic or gradient mobile phase with C~8
reverse-phase columns (Table 1). Some methods have two C~s columns in series (1) or
a C~8 and a C8 column in series (37, 38). Columns sizes range from 125 to 250 mm in
length with 3.4 to 4.6 mm internal diameters and 5-10 ~tm particle size packings. In C18
columns, porous silica particles form the column's support matrix and octadecyl silane
(ODS) fimctional groups are bonded to the matrix. Reverse phase columns separate
molecules based on their hydrophobic properties. The more polar molecules elute
quickly and very hydrophobic molecules interact strongly with the colmnn and elute
at later retention times. In carotenoid analysis of paprika extracts, the "free"
carotenoids elute first, followed by the carotenoid monoesters, 13-carotene, and finally
the carotenoid diesters (36).
Identification_ and Quantification
Individual carotenoids can be identified by their spectral properties, absorbing
principally in the 400-500 nm wavelengths. Most carotenoids exhibit 3 absorption
maxima, with 1 major peak and 2 minor peaks. The exact wavelengths of the 3 maxima
vary among the carotenoids, and therefore can be used for identification. Spectral
properties should be examined in different solvent systems, because a shift in maxima
Table 1 Summary of HPLC methods for carotenoid separation in Capsicum using reverse-phase chromatography
Reference
Column
Biacs et al.,
1989, 1993
Chromsil C,,,10 pm or
Nucleosil ODS, 5 pm,
250 x 4.6 mm
Bureau & Bushway,
1986
Partisil 5 ODS
250 x 4.6 mm
Fisher & Kocis,
1987
Gregory et d.,1987
Ittah et d.,1993
Levy et d.,
1995
Matus et d.,
199 1
Mejiaet d.,1988
Solvents
(v v v)
acetonitrile-2-propanolwater (39:57:4)
Gradient
Time
min
Flow
ml/m
Detection
Wavelength
lsocratic
30-35
1-1.2
438 nm
acetonitrile-tetrahydrofuran- Isocratic
water (85:12.5:2.5)
---
2
470 nm
Zorbax C , ,
250 x 4 6 mm
A:acetone-water (75:25).
B:acetone-methanol (75:25)
Linear steps
60
I
428,460,
480,510 nm
Waters Resolve C,,
I50 x 3 9 mm
methanol-ethyl acetate
Linear
20
1.8
475 nm
Merck RP-I 8, 5 pm,
250 x 3 4 mm, in series
with RP-8, 5 pm,
125 x 3 4 mm
A acetonitrile-2-propanol
(40 60)
B water (1 4%)
B: 14% to
0% in 40
min.
40
0.8
260-540 nm
Chromsil C,,, 6 pm.
250 x 4 6 mm (2 in
series)
A methanol-water (88 12)
Linear steps
45
1.5
430, 450,480,
400,340 nm
15
1
460 nm
Waters Nova-Pak
I50 x 3 9 mm
B methanol
C acetone-methanol (50 50)
I acetonitnle-methanoltetrahydrofuran (58 35 7)
I1 acetonitriletetrah drofuran-water
(85
2 5)
--_
115
Mineuez-Mosauera
& H&nero-Mehdez,
1993,1994
S herisorb ODS 2, 5 pm,
290 x 4 mm
1:acetone-water (75:25).
11:tetrahydrofuran-water
(52 48)
I: Linear
11-Isocratic
17
I:l 5
11.1
450 nm
w
cn
-4
358
will occur. The degree of spectral shift can be used for identification.
Several diagnostic chemical tests are available to identify different carotenoids
by their functional end groups (1, 36, 41). Acid treatment converts those carotenoids
with the 5,6-epoxide group (violaxanthin, antheraxanthm, capsanthin-5,6-epoxide,
cryptoxanthin-5,6-epoxide) to furanoid oxides, resulting in shifted retention times and
absorption maxima. The reduction of the red, highly-conjugated ketones (capsanthin,
capsorubin, cryptocapsin, capsanthin-5,6-epoxide) with sodium borohydride produces
yellow, unconjugated alcohols and therefore, marked changes in retention times and
absorption maxima.
Carotenoids can be identified by their HPLC retention times or the 1~ values and
band colors on thin-layer chromatography plates. Cochromatography with authentic
standards is necessary for identification and quantification. Standards of a-carotene
and 13-carotene are commercially available from Sigma (St. Louis, MO, U.S.), whereas
others (lutein, zeaxanthm, cryptoxanthin, capsorubin, capsanthin) can be requested
from Hoffman-LaRoche (Nufley, NJ, U.S. and Basel, Switzerland). Standards can also
be prepared from natural sources. An internal standard such as 13-apo-8'-carotenal or
canthaxanthin should be used to monitor extraction efficiency and for quantitation (44).
Problems Associated with HPLC .Analysis
Although carotenoid analysis has been advanced with the use of HPLC,
limitations exist to the current methodology (46). An awareness of the potential
problems that may occur during analysis can improve the accuracy and repeatability
of the technique. Stereoisomers and degradation products may form during the
extraction process, because carotenoids are sensitive to heat, light, and oxygen. These
artifacts will be more readily detected with HPLC, and the chromatograms may be
misinterpreted. Also, reactions between the carotenoids and the mobile phase and
injection solvents are possible, producing artifacts. Carotenoids vary in solubility in
different solvents, and this must be considered when selecting chromatography
solvents. In most situations, the injection solvent should be identical to the mobile
phase to reduce peak distortions. Another problem area is the interaction of metal
surfaces (stainless steel fittings, filters, frits, and columns) with the pigments, causing
artifacts and irregular peaks during chromatography (46). The use of metal-free
columns, peek tubing, and Teflon frits can overcome this problem. Future
improvements in HPLC columns, detectors, and computer software are likely. As the
technology is refmed, the procedures for carotenoid analysis will be modified for
accuracy and efficiency, advancing our understanding of chile carotenoids.
359
CHILE PUNGENCY ANALYSIS
The nature of the pungency constituents has been established as a mixture of
seven homologous branched-chained alkyl vanillylamides, named capsaicmoids (47).
Capsaicin is the most prevalent form, while dihydrocapsaicm is usually the second
most prevalent capsaicinoid. The other five compotmds, norcapsaicin, nordihydrocapsaicilL nomordihydrocapsaicin, homocapsaicin, homodihydrocapsaicin, are considered
minor capsaicmoids because of their relative low abundance in most natural products.
The production of capsaicmoids is restricted to the placenta of the fruit pod. No other
plant part produces capsaicmoids. Seeds do not contain capsaicinoids, but because of
their close proximity to the placenta, they can acquire some pungency.
Bucholtz (48) was the first to realize that the pungent constituents could be
extracted by macerating the pods with organic solvents, and in the following year
Braconnot (49) observed that the pungent principles could form salts with alkalis.
Braconnot named it capsicm (sic). The primary pungent principle was first isolated in
a crystalline state from the crude extract by Thresh (50) who named the compound
capsaicin. Experimemation by Micko (51) improved on the isolation technique of
Thresh, thereby proving capsaicin to be the pungent principle. Micko also
demonstrated that capsaicin possessed hydroxyl and methoxy groups, and he postulated
a structural relationship to vanillm. By 1920, Nelson (52) had been able to synthesize
capsaicm by reacting synthetic vanillylamme with decenoic acid extracted from natural
capsaicm. It was not until 1930 that Spath and Darling (53) were able to completely
synthesize capsaicm, without using natural products. In 1955, Crombie et al. (54)
demonstrated by an unambiguous synthesis that the configuration of the double bond
in the acid unit of natural capsaicm is tram. The structure of capsaicin has been
established as N-(4-hydroxy-3-methoxybenzyl)-8-methylnon-trans-6-enamide (55).
Capsaicmoids are biosynthesized from L-phenylalanme and L-valine or Lleucme through vanillylamme and C9 to C~1branched-chained fatty acids (56, 57). The
pathway proposed for the biosynthesis of the vanillylamme moiety of capsaicinoids
from L-phenylalanine is as follows: L-phenylalanine, trans-cinnamic acid, trans-pcoumaric acid, trans-caffeic acid, trans-ferulic acid, vanillin, and vanillylamine. The
enzymes which catalyze ferulic acid into vanillylamine are not known. Also,
capsaicmoid synthetase which catalyzes the condensation of vanillylamme and a C9 to
CI~ branched-chain fatty acid has not been purified. Chile powder is a complex
mixture of these closely related amides. The term "capsaicinoids" is used to represent
these homologues and analogues of capsaicin.
While the capsaicmoids may be devoid of"flavor" and odor, they are some of
the most pungem compounds known, producing a detectable heat in the mouth at
concentrations as low as 10 ppm. Beside food uses, medicinal applications of chile
pungency has brought renewed interest to the capsaicinoids. Chile pungency has long
been associated with medicinal properties. The Aztecs of Mexico have used chile
pungency to relieve pain, e.g. toothaches. Moreover, accurate determination of the
360
level of various capsaicmoids also is needed due to their increase use in the
pharmaceutical industry (58). The muscle liniments "Sloan's Liniment" and "HEET"
have capsaicin as an active ingredient. It was a pharmacist, Wilbur Scoville, at ParkDavis, the company producing "Heet", that developed the first heat measuring test for
chiles (59).
Medicinally, capsaicm is being used to alleviate pain. Its mode of action is
thought to be from nerve endings releasing a neurotransmitter called substance P.
Substance P informs the brain that something painful is occurring. Capsaicm causes an
increase in the amount of substance P released. Eventually, the substance P is depleted
and further releases from the nerve endings are reduced. Cream containing capsaicin
is used to reduce the pare associated with post-operative pare for mastectomy patients
and for amputees suffering from phantom limb pain. Prolonged use of the cream has
also been found to help reduce the itching of dialysis patients, the pare from shingles
(Herpes zoster), and cluster headaches. Further research has indicated that capsaicin
cream will reduce the pain associated with arthritis. The repeated use of the cream
apparently counters the production of substance P in the joint, hence less pain. A
decrease in substance P also helps to reduce long-term inflammation. Inflammation can
cause cartilage break down.
Accurate measurement of pungency has become important because of the
increased demand for pungent foods. Food industry researchers need reliable, safe, and
standard analytical procedures that are useful for comparing pungency among products,
and to produce a product with a known and consistent pungency level. The
capsaicmoid content varies among cultivars of the same species and among the fruit of
a single cultivar (60). The pungency of a given cultivar varies with growing location.
More specifically, any stress to the plant will increase the amount of capsaicinoids in
the pod. The pungency level of Capsicum has genetic and environmental components
(60). If the same cultivar was grown in both a hot semi-arid region and in a cool
coastal region, the fruit harvested from the hot semi-arid region would be higher in
capsaicmoids than the fruits harvested in the cool coastal climate. The capsaicinoid
content is affected by the genetic make-up of the cultivar, weather conditions, growing
conditions, and fruit age. Plant breeders can selectively develop cultivars with varying
degrees of pungency. Also, growers can somewhat control pungency by the amount of
stress to which they subject their plants. Capsicum is hottest after it has survived a
more stressful growing environment. This can be too little or too much water, high or
low temperatures, low soil fertility; any factor that is stressful to the plant. A few hot
days can increase the capsaicmoid content significantly. In New Mexico, it has been
observed that even after a ~ o w imgation, the heat level will increase in the pods.
The plant has sensed the flooding of its root zone as a stress, and has increased the
capsaicmoid level in its pods.
361
Measuring Chile Capsaicinoids
Kosuge et al. (61) showed that extracts of Capsicum and the crystalline
"capsaicin" from the fruit contained capsaicin and dihydrocapsaicin and proposed the
term capsaicinoids for the mixture of pungency stimuli in the fruit. Many papers on the
determination of the total pungency principles continue the use of the term capsaicin
or natural capsaicin. In this review, the term capsaicinoids is used for the total mixture,
and the individual components are given specific names. Analytical methods are
needed to measure pungency routinely, because chile pungency is variable from lot to
lot. The estimation of the capsaicmoids needs to be reproducible and accurate when
determining pungency. There are more that 200 papers published on the determination
and estimation of the capsaicinoids in Capsicum, the oleoresin, and products containing
their extracts. The methods could be grouped into:
1) organoleptic
2) colorimetric methods: chromogenic reagents reacted directly with the
phenolic hydroxyl of the vanillyl moiety on the extracts of the fruits
3) thin-layer chromatography (TLC) and paper chromatography
4) gas chromatography
5) high-performance liquid chromatography
There are many modifications to each group listed above. A representative sampling
of the methods will be discussed.
Org .anoleptic Method
The first reported reliable measurement of chile pungency is the Scoville
Organoleptic Test (59). The organoleptic method or taste test has been the standard
method for pungency analysis in the food industry. This test uses a taste panel of five
individuals that validate a chile sample and then record the pungency level. A sample
is then diluted until pungency can no longer be detected orally. The dilution is referred
to as the Scoville Heat Unit. Historically, this has been the most important and the only
sensory method for the assessment of heat in chile. While the Scoville Organoleptic
Test originally filled the need for a means of measuring and expressing heat in chile
products, it has been criticized for its lack of accuracy and precision (62-64). Specific
problems noted with the Scoville Organoleptic Test are build-up of pungency, rapid
taste fatigue and increased taste threshold as a result of the 5 samples required for
tasting, ethanol bite interfering with the pungency, poor precision, and more
importantly, the taste panel cannot determine the amount of the individual
capsaicinoids present in the sample.
Gillette et al. (62) developed a replacement method for the Scoville
Organoleptic Test for chile powder. They took 5 grams of ground chile powder,
steeped it for 20 minutes in 1995 ml of 90C water, filtered it, and then took 20 ml of
the filtrate and diluted it into 180 ml of 20C water. Trained testers compared this
362
concoction to a standard concentration of synthetic capsaicm. Because most testing
laboratories were using the HPLC method, there had to be a validation of the
msmmaental methods with the new sensory method.
To evaluate the correlation of sensory responses with HPLC capsaicmoid
quantitation, Gillette et al. (62) took samples from 60 lots of ground chile to represent
the normal range of Scoville Heat Units found in red chile powder. These 60 samples
were analyzed msmunentally for the 3 capsaicinoid analogs (nordihydrocapsaicin,
capsaicin, and dihydrocapsaicin) and sensorially by the Scoville Organoleptic Test.
The samples were also assessed using their new sensory method for pungency ratings.
All possible single and multiple regressions were performed in order to determine the
optimal instrumental altemative for the new sensory method, as well as to further
substantiate the precision of the new sensory method. Several very strong relationships
(r-0.90) were found between the instrumental and sensory measurements. They
concluded that the HPLC method and the new sensory method can provide accurate
measurements of pungency and can be converted to Scoville Heat Units that are
universally understood.
Colorimetric Methods
A number of colorimetric and ultraviolet spectrophotometric procedures have
been reported for the determination of capsaicin. A comprehensive review of these
techniques has been published in the first (65) and second (66) report of the Joint
Committee of the Pharmaceutical Society. It was found that these procedures do not
differentiate between capsaicin and its synthetic analogs and, therefore, have limited
utility.
Historically, an early colorimetric method was developed using vanadium oxytrichloride or ammonium vanadate and hydrochloric acid to react with the phenolic
hydroxyl group of the vanillylarnides. A resulting blue color was measured. The
variable natural color of the extracts proved to be a source of variability in the color
matching method. The different fruit colors; i.e. orange, light red, dark red, etc. gave
dissimilar readings for the same amount of capsaicinoids. Attempts were made to
compensate for this natural color by use of natural carotenoid, mixtures of synthetic
color, and mixtures of the inorganic salts (cupric nitrate and potassium chromate), all
adjusted to the color of the sample. This was tedious because the color of the different
fruit samples varied with cultivars, harvest maturity, drying conditions, and storage.
North (67) obtained the separation of capsaicinoids from pigments in the extract
by repeated partition between alkaline polar and nonpolar solvents. The capsaicinoids
with only a trace of color and fat were estimated by reacting with phenolic reagents,
e.g. phosphomolybdic and phosphotungstic acid. Pure vanillin, then more readily
available than capsaicinoids, was used for the standard curve, and the value obtained
from reference to the standard curve was multiplied by a factor of two, based on the
relation of the molecular weights to vanillin and capsaicin. This method was a major
363
advance in minimizing the interference from pigments and fat. However, the number
of steps in the clean-up made the method time consuming, and recovery and reproducibility were often reported to be poor.
Kosuge and Inagaki (61) determined capsaicmoids in ether-extracted
concentrates, taken in carbon tetrachloride, washed with acetic acid, and reacted with
Folm-Ciocaltew reagent. The blue color was measured at 750 nm and quantified using
pure vanillin as a standard and a conversion factor of 2.15 to give capsaicmoids. They
analyzed many samples by this method in their study of cultivars, effect of maturity,
cultivation practices, etc. but details of sensitivity, reproducibility, and repeatability are
not available. The vanadium oxytrichloride reagent had problems of stability both with
the reagent and the blue color formed.
Today, the colorimetric method is limited to plant breeding programs where a
direct, rapid method for detecting the presence or absence of capsaicinoids in cultivars
requiring no pungency, i.e. paprika, bell peppers, pimento, etc. is needed. The vesicles
along the placenta containing possible capsaicmoids are challenged with a 1 percent
solution of vanadium oxytrichloride in carbon tetrachloride. If there is a color reaction
(green) capsaicinoids are present (68).
Separation and Quantification of Capsaicinoids
With the rapid advancement of analytical instrumentation, many methods were
developed to overcome the traditional sensory method and the difficult colorimetric
methods. Newer separation methods emerged including paper chromatography, thinlayer chromatography, gas chromatography, GC-mass spectrometry, and high
performance liquid chromatography to objectively determine capsaicinoids. These
gave rapid and more efficient separations of complex mixtures of natural compounds,
and at the same time were simple and offered great operational flexibility. They were
quickly used in the determination of total, and later individual, capsaicmoids.
Thin-layer Chromato~aphy
One of these methods was thin-layer chromatography (TLC). The first report of
the separation of capsaicmoids from other substances in a capsicum extract was made
by Tiechert et al. (69). Dohmann (70) developed the separation into a quantitative
method that appears simple and clear for routine assay of total capsaicmoids.
Chloroform extracts of the powdered chile were applied on silica gel-G plates along
with standard capsaicmoids solutions equivalent to 100, 150, and 200 ug. The plate
was developed with chloroform-methanol-acetic acid to a distance of 14 cm. The bands
were marked as dark areas under ultraviolet light. The capsaicinoids near the solvent
front were clearly separated from the pigments near the starting line. The separated
bands of the capsaicinoids from the sample and standards were scraped off carefully
into individual tubes and reacted with Folm-Denis reagent. The blue color formed was
364
cleared by centrifugation and measured at 725 nm. The total capsaicinoids in the
sample was calculated by reference to the standard curve of absorption vs. micrograms
of crystalline capsaicinoids run on the same plate and the dilutions in making the
extracts.
Most of the methods used silica gel plates, but with many variations in the
developing solvents from single solvents, e.g. dimethyl ether, to mixed solvents of
varying polarity, with varying clarity of separation of the pigments and capsaicinoids.
However, one of the TLC methods where the different optimization steps have been
studied and found to be satisfactorily simple, rapid, and reproducible was detailed by
Jentzsch et al. (71). From a review of earlier methods, TLC on silica gel for separation
and Gibbs' dichoroqumonechlorimide reagent for greater specificity and sensitivity was
chosen, and the color reaction was optimized. Pankar and Magar (72) used the Gibb's
reagent (2,6-dichloro-p-benzoqumone chloramme) along with multi-band thin-layer
chromatography to estimate capsaicin quantity. They claim that their method required
no preliminary purification of chile extract; was less tedious, and more rapid and
accurate.
TLC as a quantitative method has not been found to be very satisfactory m
repeatability. It requires skill in making the plates, spotting of quantitatively microliter
amounts of samples, and collection of separated component areas quantitatively for
colorimetry. Permanently coated plates and automated densitometers are now available
to improve performance, but are costly.
Paper chromatography for the separation of capsaicinoids can also be
accomplished. However, the higher resolution capability of TLC techniques is more
useful in separation of individual capsaicinoids. Salzer (73) has stated that for routine
analysis the paper chromatography method is acceptable. The method was
recommended for routine use when the composition is natural and no synthetic
compound has been added. This simple, sensitive, and reproducible method is one of
the two recommended methods for capsaicinoids measurements by the Indian
Standards Institution for oleoresin capsicum.
Gas Chro _mato~raohv
v
_
_
Gas chromatography is a more sophisticated approach to analyzing
capsaicinoids. This separation method requires a higher level of insmunentation, skill
in operation, and preparative work on extracts. However, this method is rapid and
sensitive for the analysis of microgram quantifies. This method has been used to detect
adulteration of capsaicinoids with synthetic vanillylamides. The higher resolving power
and the discovery of the volatility of the silyl derivatives of poorly volatile and
nonvolatile compounds have subsequently been used in the direct quantitation of
individual capsaicmoids. The rapidity of analysis was attractive to developing methods
for total capsaicinoids and for laboratories that used this equipment routinely for other
analyses. Morrison (74) showed that a benzene solution of capsaicinoids injected into
365
a GC gave a clear, dominate peak, indicating the possibility of direct chromatography
of capsaicmoids. Hollo et al. (75) by similar direct GC of crystalline capsaicmoids on
a mixed glycol esters column on silianized glass powder at a lower temperature
obtained a major and a minor peak. These peaks were quantified by their peak areas
with reference to cholesterol as the internal standard. The capsaicmoids value by the
major peak was only 91 to 93% of the value obtained by TLC methods. The difference
was accounted for by the area of the minor peak, that was considered a minor analog.
Similar results were also obtained by direct analysis of paprika extracts. Tailing of
peaks and degeneration of columns and consequent problems of poor reproducibility
lead to common use of the GC separations of the more volatile and stable silyl
derivatives. The sensitivity and accuracy were improved by the use of an alkali flame
detector that had 25 times the response of the flame ionization detector.
Jurenitsch et al. (76) analyzed the capsaicmoids composition of fruits from
different Capsicum species and cultivars using a combination of TLC and GC. Using
extracts from lg of sample, they separated the capsaicinoids by rapid TLC, developing
with ether, free from color and fat. The capsaicinoids were extracted with methanolwater (95 + 5, v/v). Under carefully controlled GC separation conditions, the sequence
of elution with TLC-purified capsaicmoids mixture showed noncapsaicmoids, the
internal standard, an unknown noncapsaicinoid peak, octonoyl vanillylamide,
nordihydrocapsaicm, capsaicm, dihydrocapsaicin as partially resolved peaks, and
homodihydrocapsaicm.
Mueller-Stock et al. (77) purified the capsaicinoids from fruits by a 70% acetic
acid treatment. The components were identified by a combined mass spectrometer and
quantitated by an automatic area integrator of the peaks. There was clear separation of
a minor peak of nordihydrocapsaicin, a dominant capsaicm, and an overlapping but
distinct peak of dihydrocapsaicm. The higher homologs, homocapsaicm and
homodihydrocapsaicin, usually in minor amounts, were not separated. Todd et al. (63)
used programmed temperature GC conditions for distinct separation of five branchedchain homologs and three straight-chain analogs of capsaicin in a comprehensive study
of the different analytical steps. They described a defmitive method to quantify these
components that correlated highly with the organoleptic test. These GC methods
require considerable preparative work for isolation and derivatization and operative
skill in separation. Masada et al. (78) found that a 3% SE-30 column is optimal for the
gas chromatographic separation of the capsaicinoids.
Lee et al. (79) developed a mass fragmentographic method in concert with gasliquid chromatography for the quantitative microanalysis of capsaicm, dihydrocapsaicm, and nordihydrocapsaicin. The molecular ions at m/e 377, 379, and 365 in
mass spectra were used for monitoring the trimethylsilyl derivatives of capsaicm,
dihydrocapsaicm, and nordihydrocapsaicm, respectively. The ratios of the height of
each molecular ion to that of an internal standard (cholestane) were linear over the
range 5-60 nanograms. This was the first work to determine capsaicin, dihydrocapsaicin, and nordihydrocapsaicm at the nanogram level in chile powder.
366
Gas chromatographic methods usually require derivatization in order to convert
the analytes to more volatile compounds before analysis. Because capsaicmoids cannot
be determined using gas-liquid chromatography without derivatization, several HPLC
methods have been developed that are based on underivatized direct injections.
High Performance (Pressure) Liquid Chromatography Analysis
With the emergence of high performance liquid chromatography (HPLC)
techniques, the GC methods were replaced with HPLC analytical methods. Techniques
using HPLC, known earlier as high pressure liquid chromatography, have superior
separation capabilities. It provides accurate and efficient analysis of content and type
of capsaicmoids present in a chile sample (80). The separation is also generally
effected directly on the extracts of the natural materials without any preliminary cleanup. This high performance separation method found ready application in the separation
of closely related compounds, e.g., the homologs and analogs of capsaicm With the
availability of ready-made columns, automated high-pressure pmnps, optical detection
devices (ultraviolet and fluorescence), and recording and integrating accessories, HPLC
analysis has become the standard method for routine analysis by the processing
industry. The method, being rapid, can handle a larger number of samples and, with the
sensitivity of the detection systems, can be worked at submicrogram levels. With less
separation efficiency, the technique is also used on a preparative scale. Besides the
advantages of rapid direct analysis of natural extracts chile products, HPLC has
superior separation capabilities for closely related compounds and is operated at room
temperature. Combined with additional operational parameters, e.g., reversed-phase
systems, silver-ion complexing of olefmic compounds, and optical detectors, the
separation efficiency, sensitivity, and quantification at submicrogram levels of
capsaicmoids have been demonstrated in recent years. HPLC analysis accurately
determines the homologs and analogs of capsaicin and, combined with mass spectral
analysis, can identify the structural isomers of the minor components. Nanogram levels
of the individual capsaicmoids, as is required in biosynthetic and metabolic studies, can
be determined using HPLC.
Lee et al. (79) reported separation by HPLC of capsaicin and dihydrocapsaicm
from commercial capsaicmoids on a reversed-phase column by gradient elution,
increasing the methyl alcohol in water. The separation was effected in 20 min and
easily monitored by UV absorption at 254 nm. Woodbury (81) developed a HPLC
method that allowed analysis of as many as 50 samples a day that varied from less than
300 ppm to 13,000 ppm in capsaicmoids. This was possible by the combination of the
efficient separation and spectroflurometric measurement that was more sensitive and
selective even at low levels of capsaicmoids. Woodbury used HPLC for the estimation
of capsaicmoids and correlated the Scoville Heat Units (SHU) based on the ppm
numbers with the Scoville Heat Units determined by the organoleptic test. The common
practice, today, is to multiple ppm by 15 to convert to SHU.
367
Woodbury's HPLC method is the foundation for all subsequent modifications
to the HPLC method. His method entails extracting chile powder at 60C for 5 hours
using 95% ethanol saturated with sodium acetate. Separation was effected by injection
of the sample onto a LiChrosorb RP-18 (10#m) column (250 X 4.6 mm i.d.).
Capsaicmoids were eluted with water-acetonitrile-dioxane-2M perchloric acid as the
primary eluent and methanol-dioxane-acetonitrile as the secondary eluent used either
isocratically or with a gradient elution. Detection was by monitoring fluorescence
emission at 320 nm, with excitation at 288 nm.
Jurenitsch et al. (82) accomplished clear separation of the capsaicm homologs
and analogs directly from ground fruit extracts in about 70 mm by HPLC on a
reversed-phase system, eluted with dioxane-water. Four samples of Capsicum fruits
containing a range of 0.2 to 1% of total capsaicmoids were analyzed by this HPLC
method and also by the TMS-GC method. There was fairly close agreement between
the methods for total capsaicinoids. The HPLC method was clearly superior.
Chiang (83) developed a highly sensitive HPLC method with electrochemical
and UV detection to separate and quantify the capsaicinoids. Capsaicmoids are highly
electroactive compotmds. Their electroactivity was believed to be due to the easily
oxidized phenolic functional group. The electrochemical detection of capsaicinoids
provided a high degree of sensitivity and specificity. Non- and less electroactive
interfering compounds were eliminated. Thus it allowed sample analysis to be
shortened to 30 minutes, including sample preparation. She was able to detect to a
level of 0.06 ppm. When she combined it with UV detection, she was able to
simultaneously determine capsaicmoids and piperine ( the pungent compound in black
pepper, Piper nigrum) in a mixture.
Games et al. (84) used a combination of HPLC-mass spectrometry with a
moving-belt interface, field desorption mass spectrometry, and high-resolution accurate
mass electron impact mass spectrometry to identify the capsaicinoids in oleoresins.
Field desorption is a soft ionization technique that directly examines a complex mixture
without prior separation. They concluded that HPLC-mass spectrometry using a
moving-belt interface is well suited for the identification of compounds present in
oleoresms, especially when high-resolution accurate mass data are also available. They
stated that a clear advantage of the moving-belt interface is the ability to obtain electron
impact spectra that can be directly compared with existing electron impact libraries.
The field desorption-mass spectrometry technique complemented HPLC-mass
spectrometry, because it produced a total profile of the sample to be obtained.
Therefore, if components had degraded or were lost during HPLC- mass spectrometry
this would be evident.
A high performance liquid chromatography-chemilummescent nitrogen detector
(HPLC-CLND) was used to quantitatively analyze for capsaicin and dihydrocapsaicin
in chile (85). The principle is that non-nitrogenous compounds in the sample matrix are
transparent to the detector. Quantitative analysis of capsaicin and dihydrocapsaicin in
chile extract was demonstrated. Thus, it was reported that utilization of this technique
368
allows one to "see" through the non-nitrogenous components in the sample in order to
simplify the detection, identification, and quantiation of nitrogen containing analytes.
This method is so recent that there has not been an opporttmity for the capsicum
industry to test its usefulness.
Analysis for capsaicmoids by reversed-phase HPLC is recommended and has
become the standard method in the processing industry. Table 2 is a comparison of the
most common HPLC methods referenced in the literature. The method of Collins et al.
(80) has the shortest analysis time, increased sensitivity, and safety. All the HPLC
methods require a dry powder for extraction of the capsaicmoids.
Because most HPLC methods require a dry ground powder for analysis, a quick
assessment of fresh ripe chiles would be useful in determining pungency value before
the product is brought to the processing plant for handling. This is intended to replace
trained buyers tasting the raw fruit in the field and relying on their sensitivity and
pungency response memory to make purchase decisions.
A unique solvent rejection-extraction procedure that allows direct measurement
of relative capsaicmoids content of flesh jalapefio chiles has been described by Rymal
et al. (86). In the method proposed, the cross walls of the pod are ruptured without
puncamng the pod wall by rotating a hypodermic needle inserted from the stem end
from opposite sides. The pod is then filled with absolute methanol through one of the
holes made using the needle and syringe and allowed to stand for 30 mm. The extract
is flushed with three 10-mQportions of methanol and extract made up to a volume of
50 mQ, and absorbance is measured at the absorption maxima, 275 nm. This method
depends on extracting single whole pods, thus representative sampling could be a
problem, especially when the fruits in a plant are of varying maturities. At New
Mexico State University, another method is being tested where the flesh fruit is ground
in a blender and the capsaicmoids extracted with methanol. The solution is then filtered
and injected into the HPLC. The value difference between the "wet" product and the
standard dry ground product should be attributed to the water amount in the flesh
product. Preliminary tests indicate this to be true.
Methods for estimating capsaicmoid levels in blood, tissues and waste fluids
have been developed. One such use of the capsaicmoids is to monitor dietary intake
in controlled food studies or where designer foods are being feed. Saria et al. (87)
described a method for separation and determination of nanogram levels of capsaicin
and analogs in blood and deproteinized extracts of animal tissues, necessary in
pharmacological studies of capsaicin. They used a reversed-phase C~8 column and
elution with isocratic methanol-water (40 to 60 v/v) for separation. In the place of the
UV absorption at 280 nm, they used a more sensitive fluorometric detector set for
excitation at 270 nm and emission at 330 nm with the detection limits at 3 ng and
linearity range of 3 to 5000 ng. Johnson et al. (88) developed procedures for the
determination of trace amounts of capsaicmoids, as low as 500 ppb in feed and 10 ppb
in urine and waste water, required in toxicological studies.
Table 2. Comparison of HPLC methods for capsaicinoid detection.
Protocol
Extraction
bear)
method
Cleanup
ASTA (1985)
Sodium-acetate-saturated
Allow solids
95% ethanol, 3-h hot
to settle
plate or water bath
ASTA (1993)
Attuquayefio
and Buckle
( I 987)
Collins et al.
(1995)
Copper et al.
(1991)
Hoffman
et al. (1983)
Woodbury
(1980)
Flow rate
Mobile phase
Acetonitrile, dioxane,
perchloric acid, water
methanol
(mlmin-')
0.6-1.8
Fluorescence detector rnm)
Excitation
Emission
288
320
95% ethanol,
5-h refluxing
0.45-pm syringe
filter
Acetonitrile, water,
acetic acid, column
purge used
1.5
280
325
Acetonitrile
Sep-pak filtration
Methanol, water
3.5
NIA
NIA
Acetonitrile,
4-h water bath
0.45-pm syringe
filter
Methanol. water
1.o
280
338
Method 1: methanol and
centrifugation
Method 2: silica gel,
hexane, methanol,
water bath
0.45-pm filter
1.5
229
320
NIA
Methanol, water,
citric acid
Methanol, water,
citric acid
1.5
229
320
95% ethanol,
5-h heating
Allow solids
to settle
Acetonitrile, water
1.5
NIA
NIA
Sodium-acetate-saturated
95% ethanol.
Allow solids
to settle
Acetonitrile, water
dioxane. methanol.
1.o
228
320
5-h hot plate
perchloric acid .
W
Q,
\o
370
CONCLUSION
Chile is an internationally traded commodity. Because most chiles are used in
food preparations, the quality of the product is important. Color, together with
pungency, is used to assess the quality and value of chile and determines the price of
the product in the international market arena. Analytical methods allow standard
specifications to be set by the industry and provide a means to verify whether
purchased ingredients meet contract specifications. This removes uncertainty for the
chile processors, food manufacturers, and consumers regarding the carotenoid content
of chile products. The method of the American Spice Trade Association provides a
measurement of total pigment content, and is widely used by the spice industry for
color analysis. However, HPLC is the most accurate method for quantifying individual
carotenoids and for evaluating whether synthetic colorants have been added to a
product. As the HPLC technology is refined, the procedures for carotenoid analysis
will be modified for accuracy and efficiency, advancing routine analysis of individual
carotenoids by the food industry.
The pungency of chiles can be mimicked by the amides of vanillylamine and
fatty acids. These amides are made easily and cheaply, and unfortunately, offer
attractive adulterants for chile powders. Synthetic amides may have toxic properties,
and they do not have the same pungency as capsaicin, giving poor correlations between
chemical and organoleptic procedures. Thus, analytical methods allow for their
detection and assist in guaranteeing a safe and uniform product. When the analytical
methods are reviewed, it is apparent the benefits of high performance liquid
chromatography make it the first choice. Colorimetry, spectrometry, paper
chromatography, and thin-layer chromatography methods either lack specificity or are
not suitable for quantitation purposes. Gas chromatographic methods are inferior to
high performance liquid chromatography methods because derivatization prior to
analysis is necessary.
At the same time, it must be recognized that the method chosen for analyzing
chile products is dependent on the operational cost. Currently, HPLC is the best
method for analyzing carotenoids and capsaicinoids. However, the cost of the
equipment and of the trained personnel may warrant a less expensive method of
analysis. In the final judgement of a product, it is important to remember that the
analytical method chosen must correlate with the consumer's perception of the product,
whether visual, olfactory, or organoleptic.
REFERENCES
1
2
3
4
Matus Z, Deli J, Szabolcs J. J Agric Food Chem 1991; 39:1907-1914.
Poppel VG. Eur J Cancer 1993; 29:1335-1344.
Silver WL, Maruniake JA. Chem Senses 1982; 6:295-305.
Voss D. HortScience 1992; 27(12):1256-1260.
371
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Commission Intemationale de rEclairage (CIE). Supplement 2, CIE Publ 15 (El.3.1) Paris:Bureau Centrale CIE, 1978.
McGuire R. Hortscience 1992; 27(12):1254-1255.
Govindarajan VS. CRC Crit Rev Food Sci Nutr 1988; 25(3): 185-282.
Essential Oil Assoc (EOA). Specification for oleoresm paprika, EOA No. 239.
New York: EOA, 1975.
Assoc Off Anal Chem (AOAC). Official Methods Analysis. Virginia: AOAC,
1984.
Woodbury JE. J Assoc Off Anal Chem 1977; 60:1-4.
Amer Spice Trade Assoc (ASTA). Official Analytical Methods. New Jersey:
ASTA, 1985.
Baranyai M, Szabolcs J. Acta Alimentaria 1976; 5:87.
Bartley G, Scolnik P. Plant Cell 1995; 7:1027-1038.
Davies BH, K/3st H, In: K/3st H, Zweig G, Sherma J, eds. CRC Handbook of
Chromatography, Plant Pigments, vol 1. Florida: CRC Press, 1988; 3-8.
Goodwm TW. In: The Biochemistry of Carotenoids, Plants, vol 1. London:
Chapman and Hall, 1980.
Goodwin TW, Britton G. In: Goodwin TW, ed. Plant Pigments. London: Academic
Press, 1988; 62-132.
Dogbo O, Camara B. Biochem Biophys Acta 1987; 920:140-148.
Kuntz M, R0mer S, Suire C, Hugueney P, et al. Plant J 1992; 2:25-34.
Camara B. Pure Appl Chem 1985; 57:675-677.
Dogbo O, Laferriere A, d'Harlmgue A, Camara B. Proc Natl Acad Sci 1988;
85:7054-7058.
Hugueney P, R6mer S, Kuntz M, Camara B. Eur J Biochem 1992; 209:399-407.
Camara B, Dogbo O, d'Harlmgue A, Bardat F. Phytochem 1985; 24(11):2751-2752
Camara B, Moneger R. Dev Plant Biol 1980; 6:363-367.
Bouvier F, Hugueney P, d'Harlmgue A, Kuntz M, Camara B. Plant J 1994; 6:4554.
Zeichmeister L, Cholnoky L. Ann Chem 1927; 454:54.
Camara B, Moneger R. Phytochem 1978; 17:91-93.
Buckle K, Rahman F. J Chrom 1979; 171:385-391.
Rahman F, Buckle K. J Food Tech 1980; 15:241-249.
Candela M, Lopez M, Sabater F. Biol Plantarum 1984; 26(6):410-414.
Vinkler M, Richter K. Acta Aliment 1972; 1:41-58.
Baranyai M, Matus Z, Szabolcs J. Acta Aliment 1982; 11:309.
Almela L, Lopez-Roca J, Candela M, Alcazar M. J Chrom 1990; 502:95-106.
Almela L, Lopez-Roca J, Candela M, Alcazar M. J Agric Food Chem 1991;
39:1606-1609.
Biacs P, Daood H, Pavisa A, Hajdu F. J Agric Food Chem 1989; 37(2):350-353.
Biacs P, Daood H, Huszka T, Biacs P. J Agric Food Chem 1993; 41:1864-1867.
Fisher C, Kocis J. J Agric Food Chem 1987; 35:55-57.
372
37 Ittah Y, Kanner J, Granit R. J Agric Food Chem 1993; 41:899-901.
38 Levy A, Harel S, Palevitch D, Akiri B, et al. J Agric Food Chem 1995; 43:362366.
39 Mejia L, Hudson E, Gonzalez de Mejia E, Vazquez F. J Food Sci 1988;
53(5):1448-1451.
40 Minguez-Mosquera M, Hornero-Mendez D. J Agric Food Chem 1993; 41:16161620.
41 Minguez-Mosquera M, Jaren-Galan M, Garrido-Fernandez J. J Agric Food Chem
1993; 41:2120-2124.
42 Minguez-Mosquera M, Hornero-Mendez D. J Agric Food Chem 1994; 42:38-44.
43 Minguez-Mosquera M, Jaren-Galan M, Garrido-Fernandez J. J Agric Food Chem
1994; 42:645-648.
44 Khachik F, Beecher G, Goli M, Lusby W. Meth Enzy 1992; 213:347-359.
45 Gregory G, Chen T, Philip T. J Food Sci 1987; 52(4):1071-1073.
46 Scott K. Food Chem 1992; 45:357-364.
47 Hoffman PG, Lego MC, Galetto WG. J Agric Food Chem 1983; 31:1326-1330.
48 Bucholtz PA. Almanach Taschenbuch Scheidekunstler Apotheker Weimar 1816;
37:1.
49 Braconnot, A. Ann Chim Phys 1817; 6:122.
50 Thresh JC. Pharm J 1846; 6:941.
51 Micko L. Z Unters Nahr Genussm Gebrauchsgegentaende 1898; 1:818.
52 Nelson EK. J Amer Chem Soc 1920; 42:597-599.
53 Spath E, Darling S. Chem Abstr 1930; 24:3504.
54 Crombie L, Dandegaonker SH, Simpson KB. J Chem Soc 1955; 1025-1027.
55 Nelson EK, Dawson LE. J Amer Chem Soc 1923; 45:2179-2181.
56 Bennett DJ, Kirby GW. J Chem Soc 1968; 442-446.
57 Leete E, Louden M. J Amer Chem Soc 1968; 90:6837-6841.
58 Carmichael JK. Amer Family Physician 1991; 44:203-210.
59 Scoville WL. J Amer Pharm Assn 1912; 1:453.
60 Lindsey K, Bosland PW. Capsicum & Eggplant Newsletter 1995; 14:36-38.
61 Kosuge S, Inagaki Y. Nippon Nagei Kagakui Kaishi 1959; 33:470.
62 Gillette MH, Appel CE, Lego MC. J Food Sci 1984; 49:1028-1033.
63 Todd PH, Bensinger MG, Biftu T. J Food Sci 1977; 42:660-665,680.
64 Suzuki JI, Tausig F, Morse RE. Food Tech 1957; 11:100-104.
65 Anonymous. Joint Committee Pharm Soc 1959; Analyst 84:603.
66 Anonymous. Joint Committee Pharm Soc 1964; Analyst 89:377-388.
67 North H. Anal Chem 1949; 21:934-936.
68 Ting SV, Barrons KC. Proc Amer Soc Hort Sci 1942; 40:504-508.
69 Tiechert K, Mutscher B, Rochelmeyer H. Z Anal Chem 1961; 181,325.
70 Dohman K. Lab Pract 1965; 14:808.
71 Jentzsch K, Kubelka W, Pock H. Sci Pharm 1969; 37:153-162.
72 Pankar D, Magar N. J Chrom 1977; 144, 149.
373
73
74
75
76
77
78
79
80
81
82
83
84
85
Salzer UJ. CRC Critical Rev Food Sci Nutr 1977; 345-373.
Morrison JI. Chem Ind 1967; 42:1785-1786.
Hollo J, Kurucz E, Dodor J. Lebensm Wiss Technol 1969; 2:19-21.
Jurenitsch J, Kubelka W, Jentzsch K. Sci Pharm 1978; 46:307.
Mueller-Stock A, Joshi RK, Buchi J. Pharm Acta Hev 1973; 48:504-516.
Masada Y, Hashimoto K, Inoue T, Suztd~ M. J Food Sci 1971; 36:858-860.
Lee KR, Suzuki T, Kobashi M, Hasegawa K, Iwai K. J Chrom 1976; 123:119-128.
Collins MD, Wasmund LM, Bosland PW. HortScience 1995; 30(1):137-139.
Woodbury JE. J Assoc OffAnal Chem 1980; 63:556-558.
Jurenitsch J, Bingler E, Becker H, Kubelka W. Plant Med 1979; 36:54.
Chiang GH. J Food Sci 1986; 51:499-503.
Games DE, Alcock NJ, Greef J, Nussen LM, et al. J Chrom 1984; 294:269-279.
Fujinari EM. In: Charalambous G, ed. Dev Food Sci: Spices, Herbs and Edible
Fungi, New York: Elsevier 1994; 367-379.
86 Rymal KS, Cosper RD, Smith DA. J Assoc Off Anal Chem 1984; 67:658-659.
87 Saria A, Lembeck F, Skofitsch G. J Chrom 1981; 208:41-43.
88 Johnson K, Thompson HC, Bowman MC. J Agric Food Chem 1982; 30:324.
This Page Intentionally Left Blank
375
Chemiluminescent Nitrogen Detectors (CLND) for GC,
SimDis, SFC, HPLC and SEC Applications
In memory of Dr. George Charalambous (Editor), to whom
this manuscript was originally promised. A large portion of
this chapter is related to Food/Flavor applications. The CLND
techniques however, reach far beyond these boundaries
in chromatographic detection.
With the help of friends and colleagues, we will show the use of
nitrogen-specific detection for chromatographing compounds
as small as acetonitrile (Mw - 41 dalton) by GC to a macromolecule as large as poly(acrylamide) (Mw = 6,000,000 dalton)
by SEC. The CLND is also a valuable tool for SimDis, SFC and
HPLC.
Eugene M. Fujinari, Ph.D.
Applied Research Chemist
376
D. Wetzel and G. Charalambous (Editors)
Instrumental Methods in Food and Beverage Analysis
9 1998 Elsevier Science B.V. All rights reserved
Chemiluminescent Nitrogen Detectors (CLND) for GC,
SimDis, SFC, HPLC and SEC Applications
Eugene M. Fujinari
Antek Instruments Inc., 300 Bammel Wesffield Road, Houston, Texas 77090
U.S.A.
Preface
In 1970, A. Fontijn and his colleagues reported a study on a gas-phase
chemiluminescence reaction of nitric oxide and ozone [1]. The elemental
total nitrogen analyzer using a pyro-chemiluminescent nitrogen detection
method was developed several years later [2]. The most recent Model 7000N
(Part 1) from Antek Instruments Inc. (Houston, Texas, U.S.A.) is available for
analyzing nitrogen containing solid, liquid, and gas compounds in an
automated analytical laboratory. Total nitrogen on-line process analyzers are
also available from Antek Industrial Inc. (Houston, Texas, U.S.A.). The
development of the chemiluminescent nitrogen detection for gas
chromatography (GC-CLND) was also accomplished by Parks et. al. [2]. The
Model 705D has been refined over the years, and additional improvements
were incorporated into the new Model 705E. Many thanks to everyone one
on Antek's present and past product development and manufacturing teams.
Analytical and research applications for GC-CLND has blossomed
significantly over the past several years and some examples are discussed in
Part 2. Nitrogen detection for simulated distillation (SimDis-CLND) is an
important emerging technique and is briefly described in Part 3. Recently, a
novel high performance liquid chromatography - chemiluminescent nitrogen
detector (HPLC-CLND) was described [3]. Applications using the Model
7000 HPLC-CLND are presented in Part 4 for ion chromatography (IC),
reversed-phase HPLC (RP-HPLC); and size exclusion chromatography (SEC)
of food grade protein hydrolysates, peptides and free amino acids.
Macromolecules such as poly(acrylamide) as large as Mw--6,000,000 dalton
have been chromatographed by S EC with dual DRI/CLND detectors (Part
5). In Part 6, professor L. T. Taylor et. al. from Virginia Tech (Chemistry
Dept., Blacksburg, VA, U.S.A.) most recently interfaced the CLND to
capillary supercritical fluid chromatography (SFC). Although SFC-CLND
technique is just budding, using packed columns, the newest technique to date
is expected to be in full bloom in the very near future.
In order to provide readers the true capability of the CLND in
chromatography, a wide range of applications for spice, food/flavor and
377
additive, beverage, biochemical, pesticide, polymer, and petrochemical
applications are presented in Parts 1 to 6 of this chapter. This manuscript is
intended to update readers to some of the latest CLND techniques in
chromatography, and to some extent, serve as a selected compendium of
CLND applications. The intent is to inspire and facilitate, new approaches
with the CLND to solve difficult research/plant production problems in the
food/flavor, biochemical, pharmaceutical, and petrochemical industries.
I would like to acknowledge the other contributors to this manuscript
and give a special note of thanks for taking time to share their expertise in
chromatography and the CLND.
REFERENCES
1.
2.
3.
A. Fontijn, A. J. Sabadell, and R. J. Ronco, Anal. Chem., 42 (1970)
575.
R.E. Parks and R. L. Marietta, U. S. Patent 4,018,562, 24 October,
1975.
E.M. Fujinari and L. O. Courthaudon, J. Chromatogr., 592 (1992) 209.
378
LIST OF CONTRIBUTIONS
AND CONTENTS
Part 1 Elemental Total Nitrogen Analyses by Pyro-Chemiluminescent
Nitrogen Detection
John. Crnko*, Bob C. Kibler, Eugene. M. Fujinari, Antek Instruments Inc.,
300 Bammel Westfield Road, Houston, Texas 77090, U.S.A.
Part 2 Gas Chromatography - Chemiluminescent Nitrogen Detection:
GC-CLND
Eugene M. Fujinari, Antek Instruments Inc., 300 Bammel Wesffield Road,
Houston, Texas 77090, U.S.A.
Part 3 Simulated Distillation-Chemiluminescent Nitrogen Detection:
SimDis-CLND
Richard J. Young*, Shell Canada Products, Ltd., Scofford Refinery, Fort
Saskatchewan, Alberta, Canada
Eugene. M. Fujinari, Antek Instruments Inc., 300 Bammel Westfield Road,
Houston, Texas 77090, U.S.A.
Part 4 High Performance Liquid Chromatography - Chemiluminescent
Nitrogen Detection: HPLC-CLND
Eugene. M. Fujinari, Antek Instruments Inc., 300 Bammel Westfield Road,
Houston, Texas 77090, U.S.A.
Part 5 The Determination of Compositional and Molecular Weight
Distributions of Cationic Polymers Using Chemiluminescent
Nitrogen Detection (CLND) in Aqueous Size Exclusion
Chromatography
Frank J. Kolpak*, James E. Brady, Hercules Inc., Research Center, 500
Hercules Road, Wilmington, DE 19808, U.S.A.
Eugene M. Fujinari, Antek Instruments Inc., 300 Bammel Westfield Road,
Houston, Texas 77090, U.S.A.
Part 6 Chemiluminescent Nitrogen Detection in Capillary SFC
Heng. Shi, J. Thompson. B. Strode III, Larry T. Taylor *, Department of
Chemistry, Virginia Polytechnic Institute and State University, Blacksburg,
VA 24060, U.S.A.
Eugene M. Fujinari, Antek Instruments Inc., 300 Bammel Westfield Road,
Houston, Texas 77090, U.S.A.
D. Wetzel and G. Charalambous (Editors)
Instrumental Methods in Food and Beverage Analysis
9 1998 Elsevier Science B.V. All rights reserved
379
1
Elemental Total Nitrogen Analyses by PyroChemiluminescent Nitrogen Detection
John Crnko*, Bob C. Kibler, and Eugene M. Fujinari
Antek Instruments Inc., 300 Bammel Westfield Road, Houston, Texas 77090
U.S.A.
1.1
INTRODUCTION
The chemiluminescent nitrogen detection mechanism is described
below (Equations 1.1 and 1.2). The total elemental nitrogen analyzers and
the various chromatographic CLND techniques collectively have the same
detection mechanism.
R
t
R-N
R
+ 02
9NO + 0 3
1000~
~
~
C02
NO2*
+ H20
+ -NO
(1.1)
--- NO2 + hv
(1.2)
Sample components are oxidized at high temperatures (1000 ~ - 1100 ~ C). The
nitrogen containing compounds are converted to nitric oxide (-NO), as shown
in Equation 1.1. Nitric oxide gas is mixed with ozone (03) in the reaction
chamber (Equation 1.2). The resulting nitrogen dioxide in the excited state
(NO2*) is formed by this chemical reaction. Light (hv) is emitted as NO2*
decays to the ground state NO2. Chemiluminescence is detected by a
photomultiplier tube (PMT). The signal is then amplified and recorded on a
computer for report generation. The linear detector response using nitric
oxide as the standard is shown in Figure 1.1. Linear regression analysis was
used for the nitric oxide calibration curve with correlation coefficient = 0.999.
Using the same total nitrogen detection scheme above, the various
chromatographic CLND instrumentation and techniques (Figure l.2) were
evolved. These will be described in the following passages of this manuscript
along with meaningful applications.
380
Z
40 J
J
O
r~
J
30J
2O-
10~x0
9
Z
J
0
1
1
!
!
1
!
i
5000
10000
15000
20000
25000
30000
35000
Peak Area Integration
Figure 1.1 CLND nitric oxide calibration curve.
Elemental Nitrogen Analyzer
GC-CLND
SFC-CLr
Total
HPLC-CLND
rnDis-CLND
Figure 1.2 Chemiluminescent nitrogen detection.
381
1.2
TOTAL NITROGEN ANALYSES
Boehm et. al. reported a collaborative study for the determination of
total nitrogen content in urine by pyro-chemiluminescence. Each of the
twelve participating laboratories analyzed 5 blind duplicate samples of human
urine. The nitrogen content ranged from 650 mg/L to 8800 mg/L. Results
showed repeatability standard deviation (RSDr) values from 1.49 to 3.91%
and reproducibility standard deviation (RSDR) ranged from 3.66 to 9.57%.
The total nitrogen determination by pyro-chemiluminescence method was
adopted first action by AOAC INTERNATIONAL [1.1]. Konstantinides
describes the chemiluminescence measurement for nitrogen balance studies in
the field of clinical nutrition [ 1.2]. Luli et. al. presented submicron assay for
proteins, peptides and amino acids [1.3]. An automated chemiluminescent
nitrogen analyzer for routine use in clinical nutrition was assessed by Grimble
et. al. [1.4]. Pyro-chemiluminescence as a real-time, cost-effective method for
total urinary nitrogen (TUN) in clinical nitrogen-balance studies was
presented by Konstantinides et. al. [1.5].
Jancar et. al. determined the protein content in beer and wort by pyrochemiluminescence [ 1.6]. Although, the brewing industry typically estimates
the total protein content of beer and wort by the Kjeldahl nitrogen method, the
total nitrogen content by Antek pyro-chemiluminescent measurement method
demonstrated detection linearity and less than 4% coefficient of variance in
the typically observed protein range. The proposed method is rapid (approx. 2
minutes per determination). They reported that the pyro-chemiluminescent
nitrogen detection data correlated very well with values by the Kjeldahl
method. Hazardous wet chemical reagents are also eliminated by this
alternate method.
Hernandez discussed earlier, general applications of total bound
nitrogen analysis by pyro-chemiluminescence [1.7]. Nitrogen estimation in
biological samples using chemiluminescence detection was reported by Ward
et. al. [1.8]. Drushel reported nitrogen content determination by pyrochemiluminescence in petroleum fractions [1.9]. Figure 1.3 shows a
multimatrix elemental total nitrogen analyzer Antek model 7000N.
Determination of total bound nitrogen has proven to be important to the
petroleum industry. Two ASTM test standards are currently used extensively.
D-4629 is a general purpose method for syringeable liquids, while D-5762 is
used for viscous materials. Total nitrogen in waste water has proven to be an
important control parameter. ASTM test standard D-5176 is used for various
aqueous waste streams. An on-line total nitrogen process analyzer model
6000 is also commercially available and shown in Figure 1.4
Figure 1.3 Model 7000 N total nitrogen analyzer equipped with the boat inlet option
is acceptable for use in ASTM test standards D-4623, D-5176, and D-5762.
383
384
1.
REFERENCES
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
K. A. Boehm and P. F. Ross, J. AOAC Inter., 78 (1995) 301.
F. N. Konstantinides, Nut. Clin. Prac., 7 (1992) 231.
G. W. Luli and S. L. Lee, Am. Biotech. Lab., Feb. (1989).20.
G. K. Grimble, M. F. West, A. B. Acuti, R. G. Rees, M. K. Hunjan, J.
D. Webster, P. G. Frost, and D. B. Silk, JPEN, 12 (1988) 100.
F. N. KonstantinidesK. A. Boehm, W. J. Radmer, M. C. Storm, J. T.
Adderly, S. A. Weisdorf, and F. B. Cerra, , Clin. Chem., 34 (1988)
2518.
J. C. Jancar, M. D. Constant, and W. C. Herwig, Am. Soc. Brewing
Chem. J.., 41 (1983) 158.
H. A. Hemandez, Am. Lab., June (1981) 72.
M. W. N. Ward, C. W. I. Owens, and M. J. Rennie, Clin. Chem., 26
(1980) 1336.
H. V. Drushel, Anal. Chem., 49 (1977) 932.
D. Wetzel and G. Charalambous (Editors)
Instrumental Methods in Food and Beverage Analysis
9 1998 Elsevier Science B.V. All rights reserved
385
2
Gas Chromatography-Chemiluminescent Nitrogen
Detection: GC-CLND
Eugene M. Fujinari
Antek Instruments Inc., 300 Bammel Westfield Road, Houston, Texas 77090
U.S.A.
2.1
INTRODUCTION
The chemiluminescent nitrogen detection (CLND) mechanism for
chromatography is the same as the total nitrogen detection (Equations 1.1 and
1.2). For GC-CLND, sample components are eluted from the column and
then oxidized at high temperatures (1000 ~ 1100 ~
T h e - N O and 0 3
chemiluminescence detected by the PMT (in Equation 1.2) is proportional to
the amount of each nitrogen containing compound eluting from the
chromatographic column. Figure 2.1 shows Antek's new GC-CLND Model
704E with the new Hewlett-Packard GC Model HP6890. Applications using
both new generation instruments will be reported in the near future.
2.2
SAMPLE INLET AND GC COLUMN
In order to obtain the maximum benifit of the CLND, the
chromatographer must quantitatively get the nitrogen containing analytes to
the detector. Therefore it is very important that 1) sample injection system
such as split/splitless, cool on-column, or valve (sample loop) injection and
transfer lines etc., 2) capillary column (also consider: stationary phase, film
thickness, column I,D., and length etc.), and 3) interface to the detector
(CLND pyro-furnace and transfer line) are clean and fully operational for the
type of sample to be analyzed. Although this is very basic, it is the criteria for
every successful chromatographic application using any detector and certainly
applies to the CLND. Yet more often than not, we sometimes come across an
analyte or the sample matrix hanging up (perhaps a portion) somewhere
between these three chromatographic entities and causing a problem. The
injection port of the GC should be cleaned as needed and the selection of a
386
387
good column is also of utmost importance. The flow diagram for the GCCLND is illustrated in Figure 2.2. At the detector, the detector inlet, pyrotube, transfer line to the reaction chamber, and the entire reaction chamber
must also be clean. This will insure optimum nitrogen response for analyte(s)
by the CLND. Next, nitrogen to carbon selectivity (N/C) is optimized by
providing sufficient pyro-oxygen to convert nitrogenous solute(s) to nitric
oxide and the sample to carbon dioxide, water, and other oxides. The flow
rate should be adjusted to maximize the nitrogen response to where N/C ratio
= 106 or better, increasing ozone will increase the nitrogen response, but an
excess will also decrease the nitrogen selectivity (N/C). Therefore, ozone
flow rate should also be set at optimum nitrogen selectivity. Under these
conditions, the CLND can operate in the nitrogen-specific mode.
Applications illustrating the utility of GC-CLND will follow.
2.3.1 GC STATIONARY PHASES
A chromatographic separation study of some nitrogen heterocycles such
as A=l-methylindole, B=indole, 1C=3-methylindole, 2C=2-methylindole,
D=acridine, 1E=9-methylcarbazole, and 2E=carbazole was pursued to better
understand the chromatography of these nitrogen containing compounds [2.1 ].
Figures 2.3 and 2.4 show the structures of GC stationary phases used in this
study and the nitrogenous analytes, respectively. GC-CLND was used as the
detection method along with 0.32 mm and 0.53 mm I.D. capillary collumns
coated with these selected stationary phases. Split (2 ~tL) and splitless (1 laL)
injection modes were also tested in this study.
2.3.2 EXPERIMENTAL
Apparatus
GC analyses were performed on a GC Model 3000 and Model 705D
nitrogen specific detector (GC-CLND) both from Antek Instruments
(Houston, TX, U.S.A.) with a Delta chromatography software from Digital
Solutions Pty. Ltd. (Margate, Australia) on an IBM 286 compatible computer.
The following widebore and/or megabore capillary columns were utilized in
this study (see Table 1): DB-1, DB-5, DB-5MS, DB-17 and DB-210 from
J&W Scientific (Folsom, CA, U.S.A.).
w
00
m
I
r --
I
I
I
--i
(NOT PAW OF 705D)
GC
INJEClU-
I
SEPTUM
I
PURGE
I
SPUTlER
cc
I
I
CARRIER IN
L--
I
I
1
1
I
I
MEMBRANE DRYER
0
SCRUBBER
+
PYRO N E E
I1
I I
I
I
FURNACE
+==+
----A
COLUMN
Figure 2.2 Antek chemiluminescent nitrogen detector GC-CLND flow diagram.
389
o_s,(•H3 -!
CH3 I n
Dimethylpolysiloxane
c~
~H2
~H2
-O
--
Si
-
I
CH3
Trifluoropropylmethylpolysiloxane
~H3
1
~
CH3 ~ ~O
S,I
n
C6H5
Diphenyldimethylpolysiloxane
Figure 2.3 Sructures of the GC stationary phases.
390
A
CH3
H
CH 3
1C=
2C= C H 3 ~ ~
H
H
D _.
1E=
2E =
CH3
H
Figure 2.4 Sructures of the nitrogen containing analytes:
A= 1-methylindole, B=indole, 1C-3-methylindole,
2C=2-methylindole, D-acridine,
1E=9-methylcarbazole, and 2E=carbazole.
391
Reagents and Standards
The nitrogen containing heterocycle standards were purchased from
Aldrich (Miliwaukee, WI, U.S.A.). Toluene reagent (99+%) was obtained
from Fisher Scientific (Fair Lawn, NJ, U.S.A.). All standards and reagents
were used without further purification. The 7 analyte standard mixture was
prepared in toluene and consisted of the following concentrations:
A= 1-methylindole (20.1 ppm), B=indole (14.9 ppm), 1C=3-methylindole
(16.0 ppm), 2C=2-methylindole (14.3 ppm), D=acridine (15.4 ppm),
1E=9-methylcarbazole (15.0 ppm), and 2E=carbazole (18.9 ppm) as w/v.
Chromatographic Conditions
Helium was used as carrier gas at flow rate of 2.1 mL/min for 0.32 mm
I.D. column and 6 mL/min for 0.53 mm I.D. column. Oven program:
temperature 100-250~ C at 3~ C/min. Detector and injector temperatures were
280~ and 275 ~ respectively. CLND conditions: pyrolysis temperature
1000 ~
PMT voltage 900, range x50 detector output 1 volt. Sample
injection techniques and sizes are presented in Figures 2.5 to 2.10. Split ratio
of 31/1 was used in the split injection mode.
2.3.3 RESULTS AND DISCUSSION
Since sample introduction and columns are an important part of each
and every chromatographic detection, it is worthwhile to spend some
discussion on this topic from the CLND point of view using nitrogen
containing analytes. In this study, both split and splitless injection modes
were tested together with several different stationary phases and results are
presented in Table 2.1. The difference in the functional group interaction of
each analyte with the stationary phase provides the means for the separation
processes. The measure of the amount of separation between two peaks is
resolution. Separation numbers (TZ) are valid for a homoligous series of
compounds varying by one methyl unit [2.21. Since resolution and separation
numbers provide useful information as to the solute to stationary phase
interaction, different stationary phases were tested in this study in search for
better chromatographic separations of several nitrogen heterocycles.
A mixture containing seven nitrogen heterocycles were analyzed by
GC-CLND. Figures 2.5a and 2.5b, show a 1 ~tL splitless injection on a 0.53
mm I.D. DB-5 column and 2 ~tL split injection on a 0.32 mm I.D. DB-5MS
column, respectively. The split injections in general provided sharper peaks
and the 0.32 mm I.D. column resulted in a better resolution for the carbazoles
(peaks D, 1E and 2E). No resolution between 3-methylindole (peak 1C) and
Table 2.1
Separation Number TZ and Resolution (Rs) as measure of stationary phase to
analyte functional group interactions
Nitrogen
Containing
Analyies
Phase:
DB-1
30mx0.32mm
df=1w
,.~.= l ~ . m e t ~ y l i r i a O .
le
............................................
..................
jEc=
.......
........ 10m8tl)......
......
........................................................
Phase:
DB-210
30mx0.32mm
dfd.5pm
Phase:
DB--17
30mx0.32mm
dfS.5pm
Phase:
DBdMS
30mx0 32mm
dfd.25pm
._
Phase:
DB-5
30mx0.53mm
df=l . 5 ~
..........
_
.......................................
..
........................
~~-.(46-5~-.-~
1 3 ~ l . .....................
6 ~ ~ lJ
~Al!.:.?3
....................................
:.....
................
~1-.~~7~,.03~
............... . . (71-.~4g~
~
................ 5 ~ ~ . 7 .
1-1~~
................................ ,......................... .7!2.@!.1.!?1
........... 4s..
....................... - ...............................................................................................................................
: 1.C._.3_methy!lndo!e.............".R#,,.
....................
...~3g..56
..............................................................................
NR .........................................
............................ ,......................................
..............
..........................
.............
..............................................................................
?.C.=.?:??!!T!dd!?
......................................................................................................................
........................
............................................
i
. ........
~ ..............................
..........._
i..........................
= acr'dine. ...........................................
^
..
~
......._ ..........
.....
1E = 9-methylcarbazole
........................................
l!?,PS)
$n.l.,sP,
@.5?1
..................
............
NF\
&!:.-:
................................
......
k??:.431........
_
......i?;(B,72)
._..................................................
il.2:J.2~.
.............................
I.!.s,sL................
4
I
:
......
~
.........
~
.....
-.
NR = No resolution observed.
._
......................
i:
_j
1
A
1
B
IIC
II I
a
+
2c
/-----
1c + 2c
1E
B
D
b
c
- I
I
0
I
5
1
10
I
I
15
20
time (min)
I
25
I
30
t
35
Figure 2.5 GC-CLND chromatogram of 7 nitrogen heterocycles on a DB-5 stationary phase:
a) 1 pL splitless injection, 30 m x 0.53 mm I.D. column;
b) 2 pL split (31/1) injection, 30 m x 0.32 mm I.D. column.
w
W
w
394
2-methylindole (peak 2C) was observed. This may be due to the phenyl-like
aromatic characteristic of the two methylindoles having similar interaction
with the phenyl (5%) groups on the DB-5 stationary phase.
The DB-1 (dimethylpolysiloxane) stationary phase typically used for
separating compounds by boiling point showed no resolution between 3methylindole (peak 1C) and 2-methylindole (peak 2C), similarly between 9methylcarbazole (peak 1E) and carbazole (peak 2E), see Figure 2.6.
DB-17 stationary phase contains 50% phenyl (Figure 2.7). When
comparing this chromatogram to the DB-5 in Figure 2.5b, the DB-17 phase
gave improved resolution (Table 2.1). However, like the DB-5 phase, no
resolution between 3-methylindole (peak 1C) and 2-methylindole (peak 2C)
was observed even with 10 times more phenyl in the stationary phase than
DB-5.
A trifluoropropylmethylpolysiloxane stationary phase (DB-210)
provided the best peak resolutions (Table 2.1) and all 7 nitrogen containing
compounds were resolved (Figure 2.8). The resolution (39.56) between 3methylindole and 2-methylindole was accomplished using this column. A
separation number between indole and 2-methylindole was measured to be 80
with a very good resolution (95.09). Data indicated that each of the 7 nitrogen
bearing compounds interacted to a different degree with the trifluoropropylfunctional group of the stationary phase structure, resulting in good
separation.
The elution order of the 7 standards on the DB-210 column was
confirmed by standard addition as shown by the GC-CLND chromatograms in
Figures 2.9 and 2.10.
2.3.4 CONCLUSION: (to sections 2.3.1 to 2.3.3)
GC-CLND technique was used in effort to search for better
chromatographic separation of the 7 nitrogen heterocycles than the typical
boiling point (DB-1) column. The trifluoropropylmethylpolysiloxane phase
was found to achieve optimum resolution, particularly for the separation of 3methylindole and 2-methylindole isomers. Both resolution and separation
numbers were obtained and provided useful information as to the degree of
stationary phase structure-to-analyte functional group interactions. Regarding
sample introduction for these 7 compounds, a 2 ~L split injection with a split
ratio of 31/1 and a 0.32 mm I.D. (widebore) column was preferred over the 1
I~L splitless injection with the 0.53 mm I.D. (megabore) column. In general,
better sensitivity and peak shapes were observed using the split injection for
these analytes onto a 0.32 mm I.D. column. Finally, the GC-CLND
chromatograms showed optimum nitrogen-specificity. No response to the
I _
IN
+
IN
§
0
-(~.~
In
-(~1
0
-(~1
_U~
-0
C:
E
m
E
0
o~O
x:E ~.
o~,,,i
,.. ,.o
0
I--, X
0
'-~
o'~'~
t=~
z~~
!
395
396
I
0
in
0
-(~1
E
am
E
. U') : ~
-0
0
i ~.
9
o~
~o
9
~.~
~
t~
E~
z~
!
i..)~
i~-
o~
o,..~
~:
......
r~
_
-
~~ ~
0
"('3
In
0
,1 ~
E
m
0
E
.U')~
0
-U')
-0
0
X
0
=i
~
=
~.~
~2
z~
!
397
398
/
lC
c
0
5
10
.
.
.
.
i
15
time
.
.
.
.
.
(rain)
w .
.20
.
.
.
.
i
. . . . . . .
25
Figure 2.9 Peak identification by standard addition.
GC-CLND with DB-210 column, 30 m x 0.32 m m I.D.
2 ~tL split (31 / 1) injection
A= 1-methylindole, B=indole, 1C=3-methylindole,
2C=2-methylindole.
399
1E
/
/
5
10
15
time (min)
20
2E
25
Figure 2.10 Peak identification by s t a n d a r d addition.
GC-CLND with DB-210 column, 30 m x 0.32 mm I.D.
2 ~L split (31/1 ) injection
D=acridine, 1E=9-methylcarbazole,
2E=carbazole.
400
toluene (hydrocarbon) solvent was observed.
This allowed the
chromatographer to easily identify and rapidly quantitate the nitrogen
containing analytes. Detector sensitivity of 12 pg N (1 uL x 100 ppb indole in
toluene) with a signal to noise (S/N) ratio of 2/1 was achieved. The next step
was to use the nitrogen-specific detector to analyze some real world samples.
2.4
GC-CLND OPTIMIZATION
A direct distillation gasoleum sample spiked with toluene and n-C7 to
n-C23 hydrocarbon standards was analyzed by GC-CLND. The purpose was
to demonstrate that only the nitrogen containing components in the distillation
fraction is observed by the CLND.
Chromatographic Conditions
Helium was used as carrier gas at flow rate of 2.0 mL/min. Oven
temperature program was 40-275~ at 4 ~ C/min, hold 2 min. Detector and
injector temperatures were 280 ~ C and 275 ~ C, respectively. Column: DB- 1,
30m x 0.32 mm I.D., 1 ~tm film thickness. FID conditions: 1 volt output,
range 10E-11, 249 mL/min air, 30 mL/min hydrogen, helium carrier 4.5
mL/min, helium make-up 20 mL/min. CLND conditions: pyrolysis
temperature 1015~ PMT voltage 750, range xl detector output 1 volt. A
ltttL splitless injection.
Results and Discussion
GC-CLND was optimized so that the non-nitrogen hydrocarbon
components in the direct gasoleum fraction was transparent to the detector and
only the nitrogen compounds were observed (Figure 2.11a). Total nitrogen
content in this sample was 150 ppm N. The corresponding FID
chromatogram shows toluene, n-C7 to n-C23 alkanes, and the hydrocarbons
of the gasoleum fraction (Figure 2.11 b). GC-CLND application on petroleum
light cycle oil (LCO) using a HP-1 (boiling point column) was previously
reported [2.31.
2.5
GC-CLND: Flavors and Essential Oils
Benn et. al. applied GC-CLND to selectively detect nitrogen containing
compounds in flavors and essential oils [2.5]. Using this technique, a simple
method was demonstrated to detect adulteration in food materials. Galbanum
401
C15
C17
C16
C18
Q)
C19
C14
C20
o
C21
[C22
3
C13
C7
0
0
-
. . . . . . . . . . .
2 0
2'0
,
b) GC-FID
.
. . . . .
Time (mini
Time (mini
40
|
40
"
'
60
6b
a) GC-CLND
Figure 2.11 Direct distillation gasoleum fraction.
DB-1 column, 30 m x 0.32 mm I.D.
1 pL splitless injection
a) GC-CLND showing only nitrogen compounds
b) GC-FID sample with toluene + n-C7 to n-C23
402
oil has a unique leafy odor and is used in low levels in some flavors.
Researchers have previously reported FID characterization of pyrazines in
galbanum oil [2.6-2.7]. Because of its cost, adulteration by addition of betapinene and other terpenes is not uncommon. Addition of 1-(2-pyridyl)-3chlorophenylpropane (Root Body) may make this type of adultration more
difficult to detect during organoleptic evaluation. Figure 2.12a shows FID
chromatogram of adulterated galbanum oil. The presence of Root Body in
this sample is not apparent by the FID and may not be detected by
organoleptic evaluation. Figures 2.12b, 2.12c, and 2.12d are GC-CLND
analysis of genuine galbanum oil, adulterated galbanum oil, and Root Body
standard, respectively. The chromatogram in Figure 2.12b shows the natural
pyrazines as viewed by the CLND. Peak assignments:
B= 2-methoxy-3-isopropylpyrazine,
C= 2-methoxy-3-isopropyl-5-methylpyrazine,
D=- 2-methoxy-3-sec-butylpyrazine,
E= 2-methoxy-3-isobutylpyrazine,
F= 2,6-dimethoxy-3-isopropyl-5-methylpyrazine
The CLND of the adulterated galbanum oil (Figure 2.12c) showed peaks
A=2,4,6-trimethylpyridine and the Root Body isomers (peaks G, H, and I) in
addition to the natural pyrazines. The presence of 1-(2-pyridyl)-3chlorophenylpropane in the galbanum oil clearly indicated adulteration of the
sample.
Figure 2.13a is a GC-FID profile of a natural peach flavor with 2isopropyl-4-methylthiazole. This additive is a synthetic chemical which,
when added to peach flavor at trace levels, provides a full, pleasant note. This
same sample was analyzed by GC-CLND (peak A=2-isopropyl-4methylthiazole) and presented in Figure 2.13c.
Note in the FID
chromatogram that this additive is co-eluting with a major component of the
peach flavor. The GC-CLND profile of only the peach flavor is provided in
Figure 2.13b.
Chromatographic Conditions
Helium carrier gas flow rate was 10.0 mL/min. Oven program:
temperature 50 ~ C, hold 2 min, ramp to 220~ at 7 ~ C/min, hold 5 min.
Detector and injector temperatures were 250 ~ C. Column: CP-Sil 5 CB, 25m
x 0.53 mm I.D., 1 tam film thickness (Chrompack Inc., Raritan, NJ, U.S.A.).
FID conditions: 1 volt output, 400 mL/min air, 30 mL/min hydrogen, helium
make-up 20 mL/min. CLND conditions: pyrolysis temperature 1000~ C, PMT
voltage 950, range x50 detector output 1 volt. Integrator conditions: 1 volt
input, attenuation 2 3 (FID) or 2 8 (CLND). A l laL splitless injection except
galbanum oil by FID which used 0.2 tttL splitless injection [2.5].
403
|
i
|-
0
10
20
-1
-
30
Time (rain)
Figure 2.12a GC-FID of adulterated galbanum oil.
Reprinted from S. M. Benn, K. Myung, and
E. M. Fujinari, G. Charalambous (Ed.), Food
Flavors, Ingredients and Composition, Elsevier
Science Publishers, Amsterdam, (1993) 65,
with permission.
404
b. genuine galbanum oil.
c. adulterated galbanum
oil.
d. Root Body std.
b)CLND
E
c
G
.c)CLND
D
.d)CLND
I
0
"
'
!
. . . . . .
10
I
20
'"
I
30
Time (min)
Figure 2.12b-d GC-CLND of galbanum oils.
Reprinted from S. M. Benn, K. Myung, and
E. M. Fujinari, G. Charalambous (Ed.), Food
Flavors, Ingredients and Composition, Elsevier
Science Pu blishers, Amsterdam, (1993) 65,
.... with perm!ssion.
.
405
a. flavor + 2-1sopropyl-4methylthlazole.
c. same sample as l a.
b. peach flavor only.
LJ
a) FID
c) CLND
-~I
0
.
.
-~
.
.
.
.
i
10
b) CLND
.
.
.
.
.
D
Z0
.
.
.
.
.
l-
30
T i m e (rain)
Figure 2.13 GC-FID and GC-CLND of peach flavor.
Reprinted from S. M. Benn, K. Myung, and
E. M. Fujinari, G. Charalambous (Ed.), Food
Flavors, Ingredients and Composition, Elsevier
Science Publishers, Amsterdam, ( 1993 ) 65,
with permission.
406
2.6
GC-CLND: Hydrocrackite Hydrocarbon Standard
GC-FID hydrocrackite hydrocarbon standard profile, containing
23.72% paraffins, 56.56% naphthenes, 16.94% aromatics, and 2.78%
unidentified olefins, is presented in Figure 2.14. This standard mixture
contains no nitrogenous compounds as shown by the corresponding GCCLND chromatogram after detector optimization, Young et. al. [2.8].
Chromatographic Conditions
Helium carrier gas flow rate was 2.0 mL/min. Oven program:
temperature 35-235~ at 4 ~ C/min. Detector and injector temperatures were
285~ and 280 ~ C, respectively. Column: DB-1, 30m x 0.32 mm I.D., 1 ~tm
film thickness. FID conditions: 1 volt output, range 10E-12, 262 mL/min air,
25 mL/min hydrogen. CLND conditions: pyrolysis temperature 1000 ~
PMT voltage 850, range x50 detector output 1 volt. A 0.2tttL splitless
injection (FID) and 0.4 laL splitless injection (CLND).
2.7
GC-CLND: Light Cycle Oil 0LCO)
The chromatographic optimization of the 7 nitrogen containing
standards is presented in the previous GC stationary phase study (section
2.3.1-2.3.4). The 7 nitrogen heterocycle standards and the LCO fraction, after
column and CLND optimization, are presented in Figures 2.15a. and 2.15b,
respectively. DB-210 phase does not elute solutes strictly by boiling point as
the dimethylpolysiloxane phase. The trifluoropropylmethylpolysiloxane GC
stationary (DB-210) support, however, provided excellent resolution of the
three methylindole isomers (peaks: 1= 1-methylindole, 3=3-methylindole, and
4=2-methylindole). For this reason, GC-CLND analysis of the LCO fraction
was pursued using the DB-210 column.
Chromatographic Conditions
Helium carrier gas flow rate was 2.0 mL/min. Oven program:
temperature 100-250~ at 3~ C/min. Detector and injector temperatures were
280~ and 275 ~ C, respectively. Column: DB-210, 30m x 0.32 mm I.D., 0.5
~m film thickness. CLND conditions: pyrolysis temperature 1025 ~ C, PMT
voltage 700, range x25 detector output 1 volt. A 1 laL splitless injection.
C6
$.
GC-FID
1,
c10
L
GC-CLND
0
5
10
15
Time (min)
Figure 2.14 GC-FID and GC-CLND chromatograms of a hydrocrackite hydrocarbon standard
3 0 m x 0.32 mm I.D., 1 pm film thickness, DB-1 column.
Note: peaks marked C5 to C 1 0 are n-alkanes.
408
b) GC-CLND
I
0
--
I
20
.
.
.
.
.
.
.
.
1
Time (rain)
40
2
0
.
.
.
.
.
.
.
.
.
.
.
.
.
of LCO
I
60
6
4
.
-
7
3
I
-
5
I
. . . . . . .
20
-
~-
"
Time (rain)
I
40
"
"
-
.
.
.
.
.
.
.
.
.
60
a) GC-CLND
of 7 stds
Figure 2.15 GC-CLNDof petroleum light cycle oil.
a) Nitrogen containing standards:
Peaks: l=l-methylindole, 2=indole,
3=3-methylindole, 4=2-methylindole,
5=acridine, 6=9-methylcarbazole, and
7=carbazole
b) Petroleum light cycle oil fraction.
409
2.8
GC-CLND: LCO Gasoleum and Distillation Fraction
GC-CLND of l ppm N indole standard in toluene, direct distillation
gasoleum fraction, and LCO gasoleum are presented in Figures 2.16a, 2.16b
and 2.16c, respectively. Note that the light cycle oils (early eluting peaks)
present in Figure 2.16c are no longer present after the distillation (Figure
2.16b). The FID profile of this distillation gasoleum fraction is provided in
Figure 2.11b. The CLND is only detecting the nitrogen components in the
two samples (Figures 2.16b and 2.16c) since the spiked hydrocarbon
standards n-C7 to n-C23 and toluene are not observed.
Chromatographic Conditions
Helium carrier gas flow rate was 2.0 mL/min. Oven program:
temperature 40-275~ at 4 ~
hold 2 min. Detector and injector
temperatures were 280~ and 275 ~ respectively. Column: DB-1, 30m x
0.32 mm I.D., 1.0~m film thickness. CLND conditions: pyrolysis temperature
1015~ PMT voltage 750, range xl, detector output 1 volt. A 1 taL splitless
injection.
2.9
GC-CLND: Photo-Active Compounds
GC-CLND of photo-active compounds in methanol are presented in
Figure 2.17. Peaks: A=Harmane (83.4 ng), B=Harmaline (78.0 ng), and
C=Harmine (114 ng). The stuctures are presented below.
H
Harmane
CH3
H
Harmine
C~3
H
C~ 3
Harmaline
Harmane is one of many light-activated naturally occuring biocide [2.9]. In
general, these types of compounds activate (biochemically) against
microorganisms, insects, nematodes, snails, and fish. Harmane alkaloids, type
I photosensitizers, have been found to undergo fugicidal activity via possible
410
i
r
. . . . . . . . . .
o
,.
0
J0
Time (min)
.
.
"
.
2()
Time (rain)
-~ ' - . . . . . . . . .
20
Time (min)
,
40
••
, '
40
"
,
c)
GC-CLND
LCO gasoleum
b)
GC-CLND
Direct distillation
gasoleum
a)
GC-CLND
lndolestd
1 ppm N
80
6()
,
60
Figure 2.16 GC-CLND of direct distillation fraction from
LCO gasoleum.
a) l ppm N indole standard in toluene
b) direct distillation gasoleum fraction
c) LCO gasoleum
411
A
I
'
C
B
I
0
I
4
. . . . . . . .
time (rain)
!
8
..................
9
12
Figure 2.17 GC-CLND of photo-active compounds.
Peaks: A=harmane, B=harmaline,
C=harmine.
15m x 0.32 mm I.D., 0.5 ~m film thickness, DB-17 column.
412
photo-binding mechanism to DNA.[2.10-2.11]. The analysis of these and
related compounds can be obtained by gas chromatography and detection
using CLND.
Chromatographic Conditions
Helium carrier gas flow rate was 1.5 mL/min. Oven program:
temperature 100-260~ C at 30 ~ C/min., 260-280 ~ C at 4 ~ C/min, hold 10 min.
Detector and injector temperatures were 280~
Column: DB-17, 15m x 0.32
mm I.D., 0.5~tm film thickness. CLND conditions: pyrolysis temperature
1000 ~ C, PMT voltage 950, range x25, detector output 1 volt. A 2 ~L split
(51/1) injection.
2.10 GC-CLND: 4-Formylmorpholine in Benzene
Determination of 4-formylmorpholine in benzene by GC-CLND can be
easily accomplished without hydrocarbon interference. The detector linearity
using a four point calibration was obtained from 0.1 ppm to 10 ppm. Linear
regression analysis was calculated (r = 99991, m = 0.00264, and b = -0.12215)
where r = correlation coefficient, m = slope, and b = y-intercept [2.12].
Nitrogen specific detection was demonstrated with a mixture of
benzene (50%) and other hydrocarbons such as toluene (10%), n-decane
(20%), and xylenes (20%) containing three nitrogen containing compounds:
A--4-formylmorpholine, B=indole, and C=3-methyl indole. The mixture was
analyzed by GC-CLND and FID as shown in Figure 2.18 [2.12]. FID showed
one large hydrocarbon peak, masking the 3 nitrogen containing compounds.
However, the hydrocarbons were transparent to the CLND, unveiling the 3
analytes.
Chromatographic Conditions
Helium carrier gas flow rate was 6.4 mL/min. Oven program:
temperature 70-220~ at 30 ~
hold 1 min. Detector and injector
temperatures were 220~
respectively. Column: SUPEROX (Carbowax
20M), 10m x 0.53 mm I.D., 1.2 tam film thickness. FID conditions: 1 volt
output, range 10E-12, 252 mL/min air, 25 mL/min hydrogen. CLND
conditions: pyrolysis temperature 1000 ~
PMT voltage 950, range x50,
detector output 1 volt. A 1 ~L splitless injection.
413
FID
A
CLND
i . . . . . . . .
0
l
3
"
time (min)
i
6
Figure 2.18 GC-CLNDand FID of a hydrocarbon mixture:
benzene (50%), toluene (10%), n-decane (20%),
xylenes (20%), and three nitrogenous compounds:
A=4-formylmorpholine, B=indole, and
C=3-methylindole
Reprinted from E. M. Fujinari, G. Charalambous
(Ed.), Food Flavors, Ingredients and Composition,
Elsevier Science Publishers, Amsterdam, (1993)
31, with permission.
414
2.11 GC-CLND: Liquified Petroleum Gases (LPG) of Refinery Streams
Liquified petroleum gases (LPG) can be analyzed by GC-CLND to
detect low levels of nitrogenous contaminants such as acetonitrile in refinery
streams. Flow diagram for the LPG/GC-CLND analysis is provided in Figure
2.19. The sample bomb is pressurized to 250 psi with argon (or another inert
gas) and is filtered through a 5 ~m (or 3 tam) particulate filter, on-line to a
sample (2~tL loop) valve. LPG sample is passed throught the quartz sight
glass and vented until no gas bubble is observed in the liquid sample. Valve
is switched and 2txL sample is swept into the GC split/splitless injector using
the helium carrier gas (note: study the flow diagram to your GC's
split/splitless injector for best sample introduction mode). Sample is
chromatographed on a capillary column and the nitrogen containing
compound(s) is detected by the CLND (see Figure 2.2, GC-CLND flow
diagram). For the analysis of acetonitrile standard, 2tttL of acetonitrile in
toluene may be injected into the GC split/splitless injector for calibration with
the CLND. The advantages of this method is that 1) the stability of standards
may be much better in liquid solution than in a gas balance, 2) preparation of
calibration standards and cold storage of the liquid standard solution(s) are
simple and convenient, and 3) ease of injecting liquid standard solutions for
generating calibration curves for LPG analysis.
2.12 LPG/GC-CLND: MTBE C4 Feed Stream and Depropanizer
Bottoms
LPG analysis using GC-CLND of a toluene blank, MTBE C4 feed
stream, and an acetonitrile (1.5 ppm) standard in toluene are presented in
Figures 2.20a , 2.20b and 2.20c, respectively. Peak A is the acetonitrile
analyte. Note that a minor (second) nitrogen component is also present in the
MTBE C4 feed stream. Similarly, LPG analysis of depropanizer bottoms
(main unit), depropanizer bottoms (auxiliary unit), and an acetonitrile (1.5
ppm) standard in toluene is shown in Figures 2 . 2 1 a , 2.21b and 2.21c,
respectively. See Table 2.2 for results and linear regression analysis of
acetonitrile calibration by the CLND. Figure 2.22 is the acetonitrile standard
curve using the CLND and 2tttL splitless injections.
I
I ACETONITRILE
STANDARD I
CAUBRAllON 1
L---
3
I
’I
CARRIER
I
II
I
T---’
I
3
(-
VENT TO HOOD
[
LOAD LPC
,--LOAD
LPC SAMPLE
SAMPLE
GC
INJECTOR
NECDLE V A L E
SIGHT GLASS
I
Y
SAMPLE VALVE
CC-CLND
Model 7050
m - J J I } i l
COLUMN
COMPUTER
PRINTER
Figure 2.19 Liquified petroleum gas (LPG) and GC-CLND application for refinery streams.
416
Z
IN
0
.,., .c_
E ~--
o - ~ .s
~
0
~_.n~
~9 o
'~-9
Otil~
~Z
0. o.
~E
Eo.
u
~
..1"
,xl::
z~"
r
-
v
~c5
.Q
to
(19
O
~
~
o
t-,!
N
N
~
o~MI
nO
z~
r--I
~c5
~tt~
9
C~
~
,..d
c,i
O
c",l
0
0
0
~--
~_ma
~00
z ~
EEE
9
~oo
~00
ONN
ooo
o00
~nO
oZ
z~
o
z~
CI~
~E
E n
~
Clio
T--V
~E
E~
~..,v
o
z~
~E
E
T--V
o
0
.Q
i
8
d~
-v.-q
~
~0
o~,~
417
418
\
"%,
\.
Q
"\.
"\.
"%~.
\
",~.
~
%..
N
\
N..
%..
"\
N.
""~
c2~
~
O
~
0
0
0
0
II
!
d
;d
O
O
O
O
O
9
d
II
E
.N
~.i.,,.I. ~
~N
-1,'-4
9
M
r
419
Table 2.2
LPG/GC-CLND analysis of MTBE C4 feed stream and depropanizer bottoms.
o,~,,,,~,.,,
gas/LPG
samples
I itlb'l
1 1 1 q~"II ,,7
Acetonitrile Sample
standard
size
ppm
~L
0.25
0.50
1.00
3.00
5.00
MTBE C4 feed stream
Depropanizer bottom
(main unit)
Depropanizer bottom
(auxiliary unit)
Integration
of
peak area
Acetonitrile
Acetonitrile
Lag N
ppm N
2
2
2
2
2
114.74
424.10
746.96
2103.64
3393.83
0.0005
0.0010
0.0020
0.0060
0.0100
2
2
263.03
360.16
0.00068
0.00096
0.34
0.48
2
364.49
0.00098
0.49
Samples were calculated bt linear regression analysis:
r = 0.999073
m = 0.0(0)003
b = -0.000098
Chromatographic Conditions
Helium carrier gas at flow rate of 10.4 mL/min. Oven program:
temperature initial hold 10 min, 40-180 ~ C at 20 ~ C/min., hold 15 min.
Detector and injector temperatures were 220~ and 200 ~
Column: DBWAX, 30m x 0.53 mm I.D., 1.0pxm film thickness. CLND conditions:
pyrolysis temperature 1050~ C, PMT voltage 900, range x50, detector output 1
volt. A 2 ~L splitless injection was used for the acetonitrile standard in
toluene solutions. Sample (2 [aL loop) valve injection was used for the
refinery LPG sample.
420
2.13 GC-CLND: Depentenizer Feed and Overhead
GC-CLND of an acetonitrile (10 ppm) standard in toluene, depentenizer
feed, and depentenizer overhead are presented in Figures 2.23a, 2.23b and
2.23c, respectively. Samples were introduced in a typical GC fashion using a
syringe with a l taL split injection. Peak A.is the acetonitrile analyte.
Chromatographic Conditions
Helium carrier gas at flow rate of 2.0 mL/min. Oven program:
temperature initial hold 10 min, 40-180~ at 20 ~
hold 15 min.
Detector and injector temperatures were 250 ~ C. Column: DB-1, 30m x 0.32
mm I.D., 1.0~m film thickness. CLND conditions: pyrolysis temperature
1000 ~ C, PMT voltage 800, range x50, detector output 1 volt. A 1 taL split
(30/1 ratio)injection.
2.14 GC-CLND: N'Nitrosonornicotine in Cigarette Smoke Extract
Fine et al. reported a TEA method of catalytic low-temperature
oxidation for determination of nitroso compounds [2.13-2.14]. However, the
nitrosamine detection mechanism by the CLND is depicted (equation 2.1)
I
!
R2N '--:N- O
I
500~
~--
"NO
+
R2N"
(2.1)
I
Bond
cleavage
by a thermal (450 ~ 650 ~ C) bond cleavage without a catalyst to generate
nitric oxide [2.3]
Determination by GC-CLND of N'Nitrosonornicotine (peak A - NNN)
in cigarette smoke extracted into methylene chloride is given in Figure 2.24
[2.12]. The structure of NNN is also provided in the figure.
9~
A
~
A
ell)
m
o
.c_
L
~
i
<
~C
m
<
~CE~
m
N
421
422
I
NO
A
i
i
5
10
J
15
Time (min
Figure 2.24 GC-CLND of cigarette smoke extracted in
methylene chloride.
Peak A = N'Nitrosonornicotine
Reprinted from E. M. Fujinari, G. Charalambous
(Ed.), Food Flavors, Ingredients and Composition,
Elsevier Science Publishers, Amsterdam, ( 1993 )
31, with permission.
423
Chromatographic Conditions
Helium carrier gas at flow rate of 1.3 mL/min. Oven program:
temperature 165 ~ C, hold 10 min, 165-220 ~ C at 2~
hold 10 min.
Detector and injector temperatures were 290~ and 220~
respectively.
Column: HP-1, 10m x 0.53 mm I.D., 0.88 lam film thickness. CLND
conditions: pyrolysis temperature 500 ~
PMT voltage 950, range x50,
detector output 1 volt. A 2 tttL splitless injection.
424
2. REFERENCES
2.1 E. M. Fujinari, presented at the 32nd Annual Eastern Analytical
Symposium, in Optimizing Gas Chromatography: Theory and Practice
Session, "Nitrogen-Specific Detection by Capillary GC-CLND",
Somerset, N.J., November 15-19, 1993.
2.2 W. Jennings, "Gas Chromatography with Glass Capillary Columns,"
Academic Press, N.Y, (1980) 110.
2.3 L.O. Courthaudon and E. M. Fujinari, LC-GC, 9 (1991) 732.
2.4 L.O. Courthaudon and E. M. Fujinari, Figure 1 correction, ibid., 10
(1992) 293.
2.5 S. M. Benn, K. Myung, and E. M. Fujinari, in "Food Flavors,
Ingredients and Composition", G. Charlambous (Ed.), Elsevier Science
Publishers, Amsterdam, (1993) 65.
2.6 A.F. Bramwell, J. W. K. Burrell, and G. Riezebos, Tetrahedron Lett.,
37 (1969) 3215.
2.7 J. W. Burrel, R. A. Lucas, D. M. Michalkiewicz, and G. Riezebos,
Chem. Ind. (London), (1970) 1409.
2.8 R.J. Young and E. M. Fujinari, unpublished work.
2.9 Y.Y. Marchant, "Light- Activated Pesticides", J. R. Heitz and K. R.
Downum (Eds.), ACS Symposium Series 339 (1987) 168.
2.10 G.H.N. Towers and D. E. Champagne, ibid. ACS Symposium Series
339 (1987) 231.
2.11 G.H.N. Towers and Z. J. Abramowski,.J. Nat. Prod., 46 (1983) 576.
2.12 E.M. Fujinari, in "Food Flavors, Ingredients and Composition", G.
Charlambous (Ed.), Elsevier Science Publishers, Amsterdam, (1993)
31.
2.13 D.H. Fine, F. Rifeh, D. Lieb, and D. P. Rounbehler, Anal. Chem., 47
(1975) 1188.
2.14 D.H. Fine, F. Rifeh, D. Lieb, and F. Rufeh, J. Chromatogr., 107
(1975)351.
D. Wetzel and G. Charalambous (Editors)
Instrumental Methods in Food and Beverage Analysis
9 1998 Elsevier Science B.V. All rights reserved
425
3
Simulated Distillation-Chemiluminescent Nitrogen
Detection: SimDis-CLND
Richard J. Younga *and Eugene M. Fujinari b
aShell Canada Products, Ltd., Scotford Refinery, Fort Saskatchewan, Alberta,
Canada
bAntek Instruments Inc., 300 Bammel Wesffield Road, Houston, Texas 77090
U.S.A.
3.1
INTRODUCTION
The nitrogen boiling point distribution of refinery streams can be
studied by simulated distillation with a chemiluminescent nitrogen detector
(SimDis-CLND) to improve the production of fuels and petrochemicals.
Some of the valued features of the CLND for SimDis are: high sensitivity and
selectivity of the detector for nitrogen containing compounds. Detailed
information is presented in the original paper [3.1]. Some petrochemical GC
applications using the CLND have been reported by Britten [3.2].
3.2
SimDis: Hydrocarbon Calibration with AED
Simulated distillation is often used for estimating the hydrocarbon
boiling range distribution of the petroleum fractions in order to control the
efficiency of the plant refining processes [3.3]. The application using a
multielement simulated distillation software for the atomic emission detector
(AED) was reported by Quimby and Dryden [3.4]. The AED was used to first
calibrate the hydrocarbon (standards) boiling point (-89 to 522~ distribution
from C2 to C40 using both carbon and hydrogen detection modes. (see
Figure 3.1).
426
(I)
E
.....
cO
.m
c
(I)
L.
>
co
f-
0
Z
_J
a2 o
c~
0
Z
o9
(D
E
~
c
0
. ......
.4--.,
(D
(I)
I.=
to
>
m
c
o
_Q
-I-
c)
oI._.
-0
i~
tu
(
'~
~o
(0t~) \
(~0~1
(~c~)
(z(~) \
(,~)
~ E O \
(6~c)
(~cc)
(z~;) 0~a
(99G)
~
~
:~
~
oo o O oo
C~) ~
\
(~)
~a \
(~t[) 0[o \
( ~ ) 6o \
(~)
(to~) ~
19111010
~
o
o
(Oo) lU!Od 13U!l!OEl
C
E
E
c"
0
E
l-
q)
~
~
C
0
E
~
~
o
r,l
~
<
<
-~
0
"~'
c
~
~
~
Rd ~~
~ = ~ 0
9
427
3.3
SimDis: Nitrogen Calibration with CLND
A simulated distillation method with the CLND was then developed
using matched capillary columns (as AED) in the same GC oven.
Simultaneous automated sample injections (one injection to CLND and the
other to AED) were accomplished using two separate HP7673A systems
mounted on a single HP5890 GC. The CLND was used to calibrate the
nitrogen boiling point distribution from 171 to 349~ Figure 3.1. Table 3.1
shows a 4% RSD of the area counts per nitrogen using the CLND for the
boiling point distribution range of 171 to 349 ~ C.
Table 3.1
Nitrogen compounds used for SimDis nitrogen boiling point distribution.
N
(ppm)
B.P.
(o C)
5.07
5.00
5.04
5.00
5.01
5.02
2.02
5.01
171
184
210
232
254
273
334
349
Compound
2,4, 6-Trimethylpyridine
Aniline
Nitrobenzene
4-Chloroaniline
Indole
4-Nitroanisole
1,7-Phenanthroline
Phenanthradine
Area count
per N
301,142
299,713
313,801
308,969
299,890
313,899
111,281
301,147
4% RSD
Reprinted from R. J. Young and E. M. Fujinari, Am. Lab.,
October (1994) 38, with permission.
Chromatographic Conditions
Helium carrier gas flow rates were 3.5 mL/min for both AED and
CLND. Oven program: temperature initial hold 1 min, 35-350~ at 10~
C/min., hold 7.5 min. Detector bases were both 350 ~ C. Columns: both HP-1,
25m x 0.32 mm I.D., 1.051ttm film thickness. CLND conditions: pyrolysis
temperature 1090 ~ PMT voltage 750, range x l0, detector output 1 volt.
428
<
~PIJInaIpT~qnfl- ~
I~
LM
<
~ ~ ~ ~ -~ ~ ~ ~
I~
ILl
i~
<
~ueaapeaqaX
0
~
9
~aueaapoo
~qaaOT~lai~ I
u e o a o n i j -~
auanT ~
Z
...I
o
.
i
_
_~
-
_
-~
,_
.
: _~
~
~
, - o ,
.,_
i!2
--
~ ~ ~ -~ ~ ~ ~ .
kJ
o~.~
0
o~
0
<::
~ ~
~
~
~
c
~.a,
~ .~
.-
~h~
..~~
F
.o
~
if:
~~~
eq
e~
u.
o~
W
o"1
El
W
121
W
D
Z
-J
(NI
c-
0
, ,..,,
c
@
9
9
"a
,,I....I
.i..-I
o=
"~= ~~
,.,....,
o~ .~
9
o
o,.4~
r,n
t~
ov,,,,l
429
430
Simultaneous 0.1 tttL cool on-column injection at 35~ using two robotic
autosamplers (HP7673A) for the same C~ oven (HP5890 GC).
3.4
SimDis-CLND: Simulated Distillation of Refinery Streams and
Products
In a refinery process, certain types of catalysts may be poisoned by
nitrogen bearing compounds. Detection and removal of the nitrogen
compounds can result in prolonged catalyst life. The purpose here is to
demonstrate the nitrogen detection technique for simulated distillation. In this
application, gasoline was fortified with components containing 90-100 ppm N
which elute within the nitrogen boiling point calibration. A SimDis AED
checkout for the C, H, and S modes is shown in Figure 3.2. The
corresponding N mode by the CLND shows only nitrobenzene. Total
nitrogen found by the CLND was 90 ppm N in the spiked gasoline (Figure
3.3). The N distribution is clearly different from the AED's C, H, and S
detection modes. Because of the high sensitivity and nitrogen-specificity of
SimDis-CLND, the technique can be useful to study refinery streams and final
products.
3. REFERENCES
3.1 R. J. Young and E. M. Fujinari, Am. Lab., October (1994) 38.
3.2 A. J. Britten, R&D, 31 (1989) 76.
3.3 ASTM standards: D2887-93. Standard test method for boiling range
distribution of petroleum fractions by gas chromatography.
Philadelphia, PA, 1993.
3.4 B. D. Quimby and P. C. Dryden, Multielement simulated distillation
with HP 5921A atomic emission detector. Application note 228-205.
Avondale: Hewlett-Packard, 1992.
D. Wetzel and G. Charalambous (Editors)
Instrumental Methods in Food and Beverage Analysis
9 1998 Elsevier Science B.V. All rights reserved
431
4
High Performance Liquid ChromatographyChemiluminescent Nitrogen Detection: HPLC-CLND
Eugene M. Fujinari
Antek Instruments Inc., 300 Bammel Westfield Road, Houston, Texas 77090
U.S.A.
4.1
INTRODUCTION
The chemiluminescent nitrogen detection (CLND) mechanism for high
performance liquid chromatography (HPLC) is the same as in Equations 1.1
and 1.2. Although the detection mechanism is the same for elemental total
nitrogen analyzer (section 1.1) and GC-CLND (section 2.1), HPLC-CLND
was developed for handling HPLC mobile phases and subsequent nitrogen
detection. Schematic flow diagram for HPLC-CLND and photograph of the
detector are shown in Figures 4.1a and 4.1b, respectively. Configurations of
the dual HPLC-CLND and UV detection with a 4.6 mm I.D. analytical
column are shown in Figures 4.2 and 4.3. In the former configuration, the
HPLC-CLND and fraction collection are accomplished post-UV detection. In
this case, a high pressure UV cell is recommended. More common set-up is
shown in Figure 4.3, where CLND and UV detection are accomplished with a
post-column split. Three types of splitters have been used: 1) Y-splitter
(Valco), T-splitter (Keystone Scientific), and capillary GC splitter (SGE). All
three splitters worked well. Most significant reason for using HPLC-CLND is
that amines (primary, secondary, and tertiary, as well as quaternary
ammonium) attached to compound which do not contain UV chromophore(s)
are difficult to detect by conventional UV detectors. On the other hand,
CLND can readily detect amines and other nitrogen containing compounds.
An example is given in Figure 4.4a, piperazine, nitrogenous compound, does
not have UV chromophore within its molecular structure, and consequently
not observed by UV (254 nm) detection. Piperazine (peak A) without
derivatization was easily detected by CLND as shown in Figure 4.4b [4.1].
Analysis of ammonium nitrogen in waste water was described earlier by
HPLC-CLND [4.2]. The reported method utilized an ion chromatography
(IC) column for cation separation.
432
D~
Zl
I:
!
'
i
!
!
!
I
I
I
!
I
1
I
w~
i
I
I
I
i
I
!
!
I
!
!
i
i
!
!
i
!
!
!
,
~
!
!
i
i
!
i
i
!
!
i
L
0
"r"
(D:
.._Jl
i
I
i
-l-:
a.i
1
1
I
I
I
I
I
I
I
I
I
I
I
'
'
I
t
I
I
:
1
I
I
L_ . . . . . . . . .
i_
I
(/i
d
qJ
d.,.a
:::s
<
c~
c~
"o
9
Z
I
c~
qa
~T~
~162
433
434
a
0
o
r~
4-1
0
L.
..I
~
()
--.
1
3
)
!
)
_L_
m
I1. ca
"r
m
L.
u- 0
0
=-
,<
~
;>
z
I
,.d
r
o~,.~
('
IS
t:::
~E
IX, In
a.
i
|
<
9
>
z
!
435
,.d
r,.)
,..d
436
N
B
~
Et
Et
A
2,3-Diethylpyrazine
a) UV (254 nm)
|
. . . . .
0
!
I
5
Time (mln)
10
N
A
B
Piperazine
i
i ......
0
~__L_
i
"-
5
Time (rain)
b) CLND
)
10
Figure 4.4 RP-HPLC of a two s t a n d a r d mixture.
Peak A = piperazine which does not have a UV
Chromophore, 13 = 2,3-diethylpyrazine.
Reprinted from E. M. Fujinari a n d J. D. Manes,
J. Chromatogr. A, 676 (1994) 113, with permission.
437
Anions such as nitrite and nitrate ions have also been separated by IC and
detected by CLND (Figure 4.5).
1.00
NO2
.75
V
o
I
t
s
NO 3
.50-
.25-
time (min)
Figure 4.5 HPLC-CLND trace of nitrite and nitrate ions.
Reprinted from E. M. Fujinari and L. O.
Courthaudon, J. Chromatogr., 592 (1992) 209,
with permission.
In the following sections 4.2 and 4.3, high performance liquid
chromatography - chemiluminescent nitrogen detector (HPLC-CLND) is
demonstrated, as a tool which can provide the means to facilitate an analytical
method development process. Reversed phase HPLC-CLND technique is
presented where ethylene thiourea (ETU) standard is fortified in
H
N
~ N
H
Ethylene thiourea
438
apple juice and recoveries analyzed after sample clean-up. Since sample
preparation is also a very important part of most analytical processes, a
relatively new solid phase extraction method using the SPEC-Microcolumn
technology was demonstrated. Microcolumn is a solid phase extraction disc
having more surface area than particles which are used in most solid phase
extractions (SPE). A preliminary apple juice ETU recovery study using
Microcolumn clean-up followed by HPLC-CLND analysis is discussed.
CLND results were also compared to data obtained by UV detection.[4.3].
4.2
EXPERIMENTAL
Apparatus
High performance liquid chromatographic separations were achieved
on a microbore HPLC system: pump Micromeritics Model 760 from Alcott
Chromatography (Norcross, GA, U.S.A.). Sample injections were achieved
with a 20 ~L loop on a Model 9125 injection valve from Rheodyne (Cotati,
CA, U.S.A.). BDS-Hypersil-C18 microbore HPLC column was purchased
from Keystone Scientific Inc. (Bellefonte, PA, U.S.A.). Sample clean-up was
achieved by soild phase extraction on (SPEC-47-C18AR and SPEC VC MP3)
SPEC-Microcolumns from SPEC, a Division of ANSYS, Inc. (Irvine, CA,
U.S.A.). The detection and quantitation of ETU analyte was accomplished
with the nitrogen specific detector, model 7000 HPLC-CLND, from Antek
Instruments Inc. (Houston, TX, U.S.A.) and Delta chromatography software
from Digital Solutions (Margate, Australia) run on an IBM 286 compatible
computer. The variable wavelength UV spectrophotometric detector Model
770 from Spectra-Physics (Santa Clara, CA, U.S.A.) was also used in this
study.
Reagents and Standards
The ethylene thiourea (ETU) analytical standard was obtained from
Aldrich (Miliwaukee, WI, U.S.A.). HPLC grade methanol (99+%) were
obtained from Fisher Scientific (Fair Lawn, NJ, U.S.A.). Sodium free
distilled water was obtained from Ozarka Drinking Water Co. (Houston, TX,
U.S.A.). All standards and reagents were used without further purification.
HPLC mobile phase was filtered through a Millipore (Bedford, MA, U.S.A.)
HV filter with a 0.45 lam pore size.
Standards and Analytical methods
Aqueous analytical standard solutions (0.5, 1.0, 2.0, 10.0, and 20.0 ppm,
w/v) ETU were prepared and analyzed by HPLC-CLND with a 6 laL partial
filled injections (using a 20 [aL sample loop) to a BDS-Hypersil-C18
439
microbore column: 150mm x 2mm ID, 5 lttm particle size. An isocratic
mobile phase (MP) methanol/water (5:95 v/v) mixture was utilized with a
flow rate of 200 ~tL/min. The CLND conditions were" 1050~ pyrolysis
temperature, PMT voltage 750, range x50, and detector output 1 volt. SPE
conditions: first, a 47 mm disc C18AR SPEC-Microcolumn was used. The
solid phase was activated with 2 x 5 mL methanol followed by 2 x 5 mL
water. A 100 mL apple juice containing 10 ppm ETU was extracted using a
vacuum (1 inch Hg) manifold at a flow rate of approx. 1 drop/sec. The eluent
contained ETU + water soluble juice components, leaving behind the
methanol soluble components. Next, VC MP3 SPEC-Microcolumn (with 15
mg capacity) was used for additional clean-up. The MP3 phase was activated
with 3 x 1 mL water. Then 1 mL of the recovery (10 ppm ETU) apple juice
extract from C18AR SPE was added to the MP3 solid phase and eluted
slowly, 0.5 mL of a 3% acetic acid methanol/water (50:50 v/v) solution was
added. Both eluent (total of 1.2 mL) was collected and analyzed by HPLCCLND and UV (240 nm) detection. Control apple juice (25 g) was similarly
extracted by microcolumn SPE and analyzed by this method.
4.3
RESULTS AND DISCUSSION
Solid phase extraction portion of this study was facilitated by using the
SPE method development flow chart Figure 4.6. Stationary phases are
selected based on solubility of analyte in various solvents. ETU (10 mg) was
soluble in 1 mL water. In similar solubility tests, ETU dissolved in 1 mL
methanol, but not in methylene chloride (1 mL). Figure 4.6 indicated that
group 3A: SCX, SAX, CN, and NH 2 solid phases should be tested to obtain
the most efficient extraction for the ETU analyte. Clean-up procedure with
microcolumn solid phase extractions of ETU in apple juice, followed by
HPLC-CLND, are outlined in Figure 4.7. Two microcolumns were used in
series. First, 100 mL apple juice containing 10 ppm ETU was extracted on a
47 mm (diameter) C18AR microcolumn. Next, 1 mL aliquot of the aqueous
eluent was further extracted on a VC MP3 (15 mg) microcolumn. SPEC VC
MP3 is a moderately polar SCX solid phase. The addition of 0.5 mL of a 3%
acetic acid in methanol/water (50:50 v/v) solution was used to help improve
ETU recovery. The combined eluent, aqueous juice fraction, and the 3%
acetic acid solution were collected (total of 1.2 mL) then analyzed by HPLCCLND and UV detection. Aqueous ethylene thiourea standard solutions were
used for calibrating the CLND response. Linearity of HPLC-CLND is shown
in Figure 4.8. Linear regression analysis (r=0.99990; m=0.01983; b=0.59071) was calculated where r = correlation coefficient, m = slope, and b =
y-intercept. The minimum detectable limit (MDL) of the CLND is 3 ng of
440
ETU C o m p o u n d (10 mg + 1 mL)
I
soluble in water?
I
No
i
I
Yes
I
I
I "
soluble in
methanol?
"
_ I
=" "
I
I
Yes i
No
I
"1"
I
I
3-A
"J
soluble in
methylene - - - chloride?
I
I
I
I
I
w
'
I
I
'
Yes
No
'l
(3_A)
I
2-B
1-F
I
I
L . . . .
I
I
soluble in
- ' - - " hexane? "--'--
I
No
I
3-A
2-C
1-H
Yes
I
3-A
1 -I
Figure 4.6 Method development flow chart for solid phase
extraction where condition selection is based on
compound solubility (3-A: SCX, SAX, CN, or NH 2.
Reprinted from The Supelco Guide to Solid Phase
Extraction, with permission.
441
Apple Juice
(fortified with 10 ppm E-IU)
SPEC-47-C18AR
MICROCOLUMN
(47 mm disc)
Aqueous Juice Fraction
SPEC VC MP3
MICROCOLUMN
(PP, 15 rag)
I.
VACUUM
-~
PHASE
C8
SAMPLE
~F.se~
C18
C18AR
SCX
L._~ ~'~
L~r
SAX
~.SEFW~R~ CN
NH2 "
~
sxn=JmTmN
1~o~c
Si
or .ocor~ MP1
PSA
MP3
HPLC-CLND analysis
Figure 4.7 SPEC Microcolumn solid phase extraction (SPEC
Division ANSYS, Inc.) and HPLC-CLND (Antek
Instruments, Inc).
442
.....j
R
e 8.6
S
/
P
o 0,4
/
//"
S
e 8.2
j
/v
.
/
,
/
.!..
...:
4
18 0.8
8.8
i...
,r
!
i
,,
[
188.8
58.8
Concentration
Figure 4.8 L~FU s t a n d a r d c u r v e b y HPLC-CLND.
~omI~1 ' 1
Depth ' I
8.11~
V
ZTU
0
i
tB.l~
S
B.M
8.1111
5.~
18.~
15.88
Figure 4.9 MDL o f ETU b y HPLC-CLND: 6pL x 0 . 5 n g / p L ETU.
443
a.
b.
c.
d.
6~L x 1 p p m ETU standard.
6~L x 2 p p m ETU standard.
6gL x 10 p p m ETU standard.
6~L x 20 p p m ETU standard.
a)
I
0
5
Time (min)
~-"
I--
0
i. . . . . . .
10
~----~-~
. . . . . .
b)
:
l
i
0
5
Time (rain)
i
5
Time (min)
c)
. . . . .
i
r --- C
I
i
10
0
.......
10
~
I
d)
'
5
Time (rain)
10
Figure 4.10 HPLC-CLND c h r o m a t o g r a m s of ETU s t a n d a r d s
in water.
444
ETU on-column (61aL x 0.5 ppm ETU or 0.816 ng N, with S/N ratio of 2/1)
and presented in Figure 4.9. HPLC-CLND chromatograms of the ETU
standards at 1, 2, 10, and 20 ppm concentrations are shown in Figure 4.10.
Figure 4.11 is the HPLC-CLND chromatogram of the 100 mL recovery
apple juice sample containing 10 ppm of ETU after extraction with the 47 mm
(diameter) disc with a C18AR solid phase. ETU peak is on the shoulder of a
large nitrogen containing component. It is clear that additional cleanup of the
juice matrix is necessary, therefore 1 mL of this aqueous juice fraction was
placed on a MP3 microcolumn (second SPE clean-up step) and eluted with 0.5
mL of 3% acetic acid methanol/water (50:50 v/v) solution. Analysis of the 10
ppm ETU recovery sample shows a significant cleanup (Figure 4.12c) as
compared to the first cleanup step (Figure 4.11). The apple juice control
sample is shown in Figure 4.12b. A water solvent blank showing the baseline
after the MP3 solid phase extraction is presented in Figure 4.12a. HPLC with
UV detection at 240 nm of the control (no ETU) apple juice through only the
first clean-up step is shown in Figure 4.13a. Recovery sample with 10 ppm
ETU after the first and second SPE clean-up is provided in Figures 4.13b and
4.13c, respectively. A side by side comparison of the apple juice ETU
recovery sample analyzed by HPLC-CLND and UV detection is shown in
Figure 4.14. Apple juice ETU recovery sample #2 analyzed by HPLC-CLND
calculated to 74.8% and UV detection (72%) agreed within 3.7% as depicted
by Table 4.1.
y
_0
O
b"
<
!
!
I--..
.=
~3
9
,.d
~.)
,.d
6~
~
445
446
I::I
b
r
~:ro
<~~.~
_o
_0
L -O
_0
-11r
-0
[--,
ow,,,,,~
I=I
r-,-I
0o
I::I o
r
o.,!
o~
Z
r,..)
!
rj~
,.~ u-I
,,,,,,,,,i
":~
-
t.r
t~
\
..0
-0
0
..0
_0
" tr
-0
o~,,,,q
o~,,,q
~
[--
oo
<
!
o1,,~
~
ol-.,q
E.-
o
o
0
>
f12
v
8
0
447
~o
2.
s
rj
>
OJO
ol--,q
448
~4
5o
-Ir
-f,,~
[-..
9- -
~
t~
o
0
"tJ ,-:,
>~:
~-~
C~,.
z~
O~
449
Table 4.1
HPLC-CLND analysis of ETU recoveries in apple juice after two different
microcolumn solid phase extractions performed in series.
ErU
standard
ppm
Sample
size
laL
Integration
of
peak area
0.5
1.0
2.0
10.0
20.0
6
6
6
6
6
146
362
617
3102
6059
Recovery #1, CLND
Recovery #2, CLND
6
6
2160
2294
42.24
44.90
70.4
74.8
Recovery #2, UV
6
1,796
43.20
72.0
Apple
juice
ETU
ETU
ng
% recovery
3
6
12
60
120
The average recovery of the two samples analyzed by CLND calculated to be
72.6%. The utility of HPLC-CLND and the SPEC Microcolumn solid phase
extraction can provide a quick scouting approach at the beginning of a method
development process. A volume of 100 mL apple juice sample was easily
extracted on the 47 mm SPEC Microcolumn disc. The HPLC-CLND
sensitivity of 3 ng ETU on-column was observed in this study.
4.4
HPLC-CLND: African type Capsicum Oleoresin
The hot flavors in many foods are often due to the presence of
capsaicinoids, a class of nitrogen containing compounds. The means to obtain
a standardized method to measure the heat levels in foods is essential. A
clean-up method reported by Cooper was used [4.4]. The solid phase
extraction was scaled down to 100 mg of silica prior to HPLC and the new
nitrogen detection technique described herein. HPLC-CLND application of
capsicum oleoresin is presented in Figure 4.15. The elution order of the
capsaicinoids are as follows: A=nordihydrocapsaicin, B=capsaicin, and
450
I
0
'
'l
i
....
20
40
Time (min)
I
60
Figure 4.15 HPLC-CLND of African type capsicum oleoresin.
Peaks: A=nordihydrocapsaicin, B=capsaicin, and
C=dihydrocapsaicin.
451
C=dihydrocapsaicin. Capsaicin and dihydrocapsaicin have been quantitated
in red hot peppers by HPLC-CLND [4.5].
Chromatographic Conditions
Capsicum oleoresin (20 mg) was extracted from the Supelclean LC-Si 1
mL (100 mg) SPE tube using 2 x 1 mL methanol elution and analyzed by
HPLC-CLND. LC-Si bed was activated with 1 mL hexane. The oleoresin
was directly weighed into the tube, solid phase was washed with 2 x 1 mL
hexane before extracting with methanol. A 20 9L loop injection was made to
a SUPELCOSIL LC-18S column: 250mm x 4.6mm ID, 5 lam particle size,
100A pore size. An isocratic mobile phase (MP) methanol/water (60/40 v/v)
with 0.1% citric acid (pH 3.0) mixture was utilized with a flow rate of 650
taL/min. A post-column split was used with a flow rate of 200 ~tL/min to the
CLND and 450 9L/min to waste. The CLND conditions: 1050~ C pyrolysis
temperature, PMT voltage 760, range x 10, and detector output 1 volt.
4.5
HPLC-CLND: Aspartame and Nitrogen Compounds in Beverages
Aspartame, scientifically known as L-aspartyl-L-phenylalanine methyl
ester, can be readily detected and quantitated by HPLC-CLND. A
simultaneous HPLC-CLND and UV (214 nm) detection of aspartame and
other nitrogenous compounds in diet soft drinks have been reported [4.6].
Dual detection was performed with a post-column split as shown in Figure
4.3. Diet cola samples #3, #4 (caffeine free), and #5 will be reviewed in
Figures 4.16, 4.17, and 4.18, respectively. Table 4.2 provides the quantitative
results by CLND of aspartame levels found for these three diet cola and citrus
beverages. Each CLND profile (Figures 4.16, 4.17, and 4.18) is different. In
particular, diet cola #5 has a large unidentified peak A (a nitrogen containing
component) and a low aspartame level (90 ppm) as compared to diet colas #3
and #4. Sodium benzoate does not contain nitrogen and is not observed by
the CLND. Because sodium benzoate contains an aromatic chromophore, it is
easily observed by UV detection. Sample profiles obtained by simultaneous
detection can save time and provide useful sample information. DKP, a
decomposition product of aspartame, was also observed using this detection
technique [4.6]. HPLC-CLND analysis of caffeine in coffee and other
beverages has been reported [4.7].
452
A=saccharin
B=caffeine
C=aspartame
D=sodium benzoate
(no nitrogen in D)
I
\
HPLC-CLND
B
UV
!
0
!
8
16
Time (min)
Figure 4.16 Simultaneous HPLC-CLND and UV (214 nm)
detection of diet cola sample #3.
Reprinted from E. M. Fujinari, G. Charalambous
(Ed.), Shelf Life Studies of Foods and Beverages,
Elsevier Science Publishers, Amsterdam, (1993)
1033, with permission.
453
A=saccharin
(caffeine free)
C=aspartame
D=sodium benzoate
(no nitrogen in D)
n' I
HPLC-CLND
A
C
i_
|--
0
9
UV
',
8
16
Time (min)
Figure 4.17 Simultaneous HPLC-CLND and UV (214 nm)
detection of diet cola sample #4 (caffeine free).
Reprinted from E. M. Fujinari, G. Charalambous
(Ed.), Shelf Life Studies of Foods and Beverages,
Elsevier Science Publishers, Amsterdam, (1993)
1033, with permission.
454
c
A=N-compound
B=saccharin
C-caffeine
D=aspartame
E=sodium benzoate
(no nitrogen in D)
0
HPLC-CLND
UV
|
0
!
8
16
Time (min)
Figure 4.18 Simultaneous HPLC-CLND and UV (214 nm)
detection of diet cola sample #5.
Reprinted from E. M. Fujinari, G. Charalambous
(Ed.), Shelf Life Studies of Foods and Beverages,
Elsevier Science Publishers, Amsterdam, (1993)
103 3, with permission.
455
Table 4.2
HPLC-CLND analysis of aspartame in diet soft drink beverages.
Diet
beverage
or brand
Citrus # 1
Citrus #2
Cola #3
Cola #4
Cola #5
Aspartame
standard
ppm
Sample
size
~L
Integration
of
peak area
Aspartame
Aspartame
lag
ppm
50
100
200
400
5
5
5
5
5
5
5
5
5
4808
10027
20772
42182
43428
44051
50918
46399
9044
0.250
0.500
1.000
2.000
2.059
2.088
2.409
2.198
0.451
412
418
482
440
90
Reprinted from E. M. Fujinari, "Shelf Life Studies of Foods a n d
Beverages", G. C h a r a l a m b o u s Ecl., Elsevier Science Publishers,
Amsterdam, (1993) 1033, with permission.
Chromatographic Conditions
A 50 lxL loop injection was made to a Deltabond ODS column: 150mm
x 4.6mm ID, 5 ~m particle size, 300A pore size. An isocratic mobile phase
(82/18 v/v) mixture: aqueous solution containing 0.2% H3PO 4 at pH 2.2 (with
0.05M KH2PO4/methanol was utilized with a flow rate of 1 mL/min. A postcolumn split was used with a flow rate of 0.450 mL/min to the CLND
and0.550 mL/min to the UV (214 nm) detector. The CLND conditions:
1050~ pyrolysis temperature, PMT voltage 700, range x25, and detector
output 1 volt.
4.6
HPLC-CLND: Important Biochemicals and Food Components
A preliminary study for a direct detection of biochemicals such as
amino acids, peptides, and proteins by HPLC-CLND without pre- or postcolumn derivatization has been reported [4.8].
Since nature's own
biochemical processes have provided a nitrogen label on each of these classes
456
of compounds, the use of a nitrogen detector such as the CLND is inherently
suitable. A reversed phase HPLC peptide mapping with UV (214 nm)
detection and a peptide mapping based on 'nitrogen content' detection using
HPLC-CLND of peptides isolated from casein hydrolysate were reported
earlier [4.1 ]. The advantage of simultaneous peptide mapping technique using
CLND/UV detection is presented in Figure 4.19. Peak A is a peptide with no
aromatic UV chromophore and appears as a minor component by UV
detection. Peak B is a peptide which contains aromatic UV chromophores (4
Tyrosines) and consequently showed a strong UV response. It was
determined by the CLND that peptide A contained a greater nitrogen content
than peptide B. Results of the amino acid analysis for peptides A and B are
discussed in detail [4.1].
Nucleotides, nucleosides, and their corresponding bases (pyrimidines
and purines) are hydrolysis products of nucleic acids, such as DNA and RNA.
Figure 4.20 shows HPLC-CLND chromatogram of an aqueous standard
mixture consisting of 5 nucleotides, 6 nucleosides, and orotic acid. The
aqueous standard mixture of their corresponding bases, purines and
pyrimidines, have also been detected by HPLC-CLND as depicted in Figure
4.21 [4.9]. These important biomolecules play a paramount role in foods and
flavors, in biochemical processes, and in pharmaceutical and related
industries.
Monosodium glutamate (MSG) is the monosodium derivative of an
H
HO zCCHzCHzCOOzNa~
NHz
MSG
amino acid, L-glutamic acid. MSG, like L-glutamic acid, does not contain
aromatic UV chromophore and may have difficulties with the sample matrix
or baseline interference(s) when using the UV detector. However, a quick and
easy determination of MSG in various types of soup bases have been recently
accomplished using a nitrogen-specific detector, HPLC-CLND [4.10].
Results of the MSG quantitation in various soup bases are presented in Table
4.3. The HPLC-CLND chromatograms of MSG in some different soup bases
are shown in Figure 4.22.
457
UV
(k = 214nm)
B
.
10
.
.
.
.
.
20
.
.
.
.
i
.
-
time (rain)
30
4~0
CLND
A
B
7---
6
.to
'
20
I .
.
.
.
3~
Time (min)
Figure 4.19 RP-HPLC chromatograms of casein hydrolysate.
Peaks: A=peptide with no aromatic UV
chromophore, B=peptide with aromatic UV
chromophore.
Reprinted from E. M. Fujinari and J. D. Manes,
J. Chromatogr. A, 676 (1994) 113, with permission.
458
A = Adenosine
C = Cytidine
Oro~ic
acid
~/
/h~P+Ur
I dine
IMP
CHP
~/
GHP~
"
9
|
'
0
'
'
Inosine
Guan
'
'
)
'
Thymidine
~
~
'
10
'
'
l
20
Time (min)
Figure 4.20 HPLC-CLND Chromatogram of a standard mixture
consisting of 5 nucleotides (5'-substituents),
6 nucleosides, and orotic acid in aqueous solution.
Reprinted from E. M. Fujinari and J. D. Manes, G.
Charalambous (Ed.), Food Flavors: Generation,
Analysis and Process Influence, Elsevier Science
Publishers, Amsterdam, (1995) 379,
with permission.
459
1
2
3
4
5
6
7
=
=
=
=
=
=
=
Cytosine
Uracil
Guanine
S-Methyl c y t o s i n e
Xanthine
Thymine
Adenine
4
1
2
'
'
3 5
1
6
'
10
7
'
'
'
I
20
Time (min)
Figure 4.21 HPLC-CLND C h r o m a t o g r a m of a s t a n d a r d m i x t u r e
consisting of t h e c o r r e s p o n d i n g p u r i n e a n d
p y r i m i d i n e bases of the n u c l e o t i d e s a n d
n u c l e o s i d e s in a q u e o u s solution.
R e p r i n t e d f r o m E. M. Fujinari a n d J. D. Manes, G.
C h a r a l a m b o u s (Ecl.), Food Flavors: G e n e r a t i o n ,
Analysis a n d Process Influence, Elsevier Science
Publishers, A m s t e r d a m , ( 1 9 9 5 ) 379,
with p e r m i s s i o n .
460
0
>
9
LL
0
>
It_
Q..
E
I,...
u?
(_9
O9
"x
_0
-~
Io..,~
E
.m
-,~'E
-04
%
-0
Q
r.
, , N
v
-~E
-0
-Cxl
___J
L
L.
0
>
u_
,.---.
. ,,,,..
0
E
0
0
c,
~9
1-0
~
..C
0
S
_ O
-~
v
rE
r-
~
-~tE
-CXl
- 0
0
v
E
-~E
-~t
~-o
..~
o
o
.w.,,!
o
E
t~
0
E
o
Z
0
-1
fxl
o...i
461
Table 4.3
HPLC-CLND analysis of MSG in flavored soup bases.
Soup
flavor
MSG std
ppm
500
1,000
2,500
5,000
Chicken
Chicken & Mushroom
Shrimp
Beef
Oriental
Sample Integration of
MSG
size (taL) peak area (xl03) ng (xl03)
2
2
2
2
2
2
2
2
0.815
1.668
4.880
8.997
4.358
2
5.548
4.632
6.020
1
2
5
10
4.791
6.738
6.085
5.089
6.599
MSG
ppm
2,3%
7.379
3,043
2,545
3,300
Chromatographic Conditions
A 2 ~L partial filled (5 laL loop) injection was made to a Hamilton
PRP-X100 column (Keystone Scientific Inc.): 50mm x 3mm ID, 10 ~tm
particle size. An isocratic mobile phase (MP) methanol/water (80/20 v/v)
with 0.1M citric acid (pH 2.6) mixture was utilized with a flow rate of 500
laL/min. A post-column split was used with a flow rate of 100 laL/min to the
CLND and 400 laL/min to waste. The HPLC-CLND (Antek Instruments Inc.)
conditions: 1050~ C pyrolysis temperature, PMT voltage 760, range x50, and
detector output 1 volt.
4.7
SEC-CLND: Size Exclusion Chromatography (SEC) of Peptides
and Food Grade Protein Hydrolysates
A chemiluminescent nitrogen detector (CLND) with size exclusion
chromatography (SEC) was developed to estimate average molecular weight
distribution of peptides and food grade protein hydrolysates, as well as protein
hydrolysate-based foods [4.11 ]. SEC separations of the 7 component standard
mixture was achieved from 30000 to 75 dalton range using a silica based
TSK-G2000SWXL column (Figure 4.23). The SEC calibration for this
462
Mw
2 9 0 0 0 = carbonic a n h y d r a s e
14400 = a-lactalbumin
3550 = insulin
1620 = b o m b e s i n
777 = n e u r o t e n s i n (1-?)
425 =/~-casomorphin (1-3)
204 = t r y p t o p h a n
O
O
O
~
I~"
I~
t--
I.~
~1
~t"
s
oa
c,4
CLND
c,4
o
2
o
uV
o
I
0
I
7.5
. . . . . .
I
15
.
.
.
.
I
18.5
I
30
Time (min)
Figure 4.23 Size exclusion c h r o m a t o g r a p h y with TSKG2000SWXL c o l u m n of 7 c o m p o n e n t s t a n d a r d
mixture, s i m u l t a n e o u s CLND a n d UV
(214 nm) detection.
Reprinted from E. M. Fujinari a n d J. D.
Manes, G. C h a r a l a m b o u s (Ed.), Food Flavors:
Generation, Analysis a n d Process Influence,
Elsevier Science Publishers, A m s t e r d a m , (1995)
929, with permission
463
column (the molecular weight distribution of the seven standards over the
separation time) is presented in Figure 4.24. Response factors ( f = l ) relative
to bombesin were calculated for ct-lactalbumin, a milk protein (Mw - 14400
dalton), bombesin, a bioactive peptide (Mw = 1620 dalton), and the smallest
peptide, glycyl-glycine (Mw - 132 dalton), indicating equimolar nitrogen
responses for the CLND. One example to highlight the advantage of using a
dual detection system for SEC will be reviewed. A simultaneous CLND and
UV (214 nm) detection was used to analyze an extensively hydrolyzed casein
(22.0 ~tg) by SEC (Figure 4.25). The manufacturer of this product reported
approximately 50% degree of hydrolysis with 50% free amino acid content.
The results obtained from the dual detection technique provided an inherently
different Mw profile. The UV profile indicated a minor component (Mw=680
daltons) as a shoulder to the major peptide with average Mw=334 D. The
CLND however, revealed this peptide (average Mw=680 D) as the third major
component in the casein hydrolysate. This major component (average
Mw=680 D) in the sample did not contain a significant UV chromophore as
depicted in the UV chromatogram. The advantage of using the CLND is that
(on-line) nitrogen content measurement can be obtained for each sample
component, as they elute from the chromatographic column. The CLND also
showed the two major components (as UV detection), peptide average
Mw=275 D, and the free amino acids Mw-108 D. We demonstrated a very
powerful technique, i.e. dual CLND/UV detection for S EC and a better means
to characterize food grade protein hydrolysates, peptides, and amino acids
than by stand-alone UV detection.
464
MW
2 9 0 0 0 = carbonic a n h y d r a s e
14400 = a-lactalbumin
3550
= insulin
1620 = b o m b e s i n
777 = n e u r o t e n s i n (1-7)
425 = ~-casomorphin (1-3)
204 = t r y p t o p h a n
4.8
0
4.4
1
4.2
'r
L3
~U
3.s L_
3.4
I
7
13
14
15
16
17
18
~
[]
EXPT D A T A
19
29
21
22
23
24
25
1]~ MINLrFE5
-
I
~
VALUF~
Figure 4.24 Molecular weight calibration by SEC with TSKG2000SWXL c o l u m n of 7 c o m p o n e n t s t a n d a r d
mixture, Detector:. CLND.
Reprinted from E. M. Fujinari a n d J. D.
Manes, G. C h a r a l a m b o u s (Ed.), Food Flavors:
Generation, Analysis a n d Process Influence,
Elsevier Science Publishers, A m s t e r d a m , (1995)
929, with permission
465
O ~ oo
oo I'~ O
~D c,.l ,-~
CLND
~D
oo
0
uv
,q
o
-it --'~' _
[. . . .
o
6
o
. ---. . . . Ai
1'2
1
18
. . . . . . . . . .
i
24
Time (min)
Figure 4.25 Size exclusion chromatography with TSKG2000SWXL column of extensively hydrolyzed
casein with simultaneous CLND and UV (214 rim)
detection. Reprinted from E. M. Fujinari and J. D.
Manes, G. Charalambous (Ed.), Food Flavors:
Generation, Analysis and Process Influence,
Elsevier Science Publishers, Amsterdam, (1995)
929, with permission.
466
4.
REFERENCES
4.1 E.M. Fujinari and J. D. Manes, J. Chromatogr. A, 676 (1994) 113.
4.2 E.M. Fujinari and L. O. Courthaudon, J. Chromatogr., 592 (1992) 209.
4.3 E.M. Fujinari, presented at the 206th ACS National Meeting, Residue
Analytical Methods and Emerging Technologies II Session, Chicago,
IL, August 22-27, unpublished.
4.4 T.H. Cooper, J. A. Guzinski, and C. Fisher, J. Agric. Food Chem., 39
(1991) 2253.
4.5 E.M. Fujinari, "Spices, Herbs and Edible Fungi", G. Charlambous Ed.,
Elsevier Science Publishers, Amsterdam, (1994) 367.
4.6 E. M. Fujinari, "Shelf Life Studies of Foods and Beverages", G.
Charlambous Ed., Elsevier Science Publishers, Amsterdam, (1993)
1033.
4.7 E. M. Fujinari, in "Food Flavors, Ingredients and Composition", G.
Charlambous (Ed.), Elsevier Science Publishers, Amsterdam, (1993)
55.
4.8 E. M. Fujinari, E. Ribble, and M. V. Piserchio in "Food Flavors,
Ingredients and Composition", G. Charlambous (Ed.), Elsevier Science
Publishers, Amsterdam, (1993) 75.
4.9 E.M. Fujinari and J. D. Manes, "Food Flavors: Generation, Analysis
and Process Influence", G. Charlambous Ed., Elsevier Science
Publishers, Amsterdam, (1995) 379.
4.10 E. M. Fujinari and E. Ribble-Garlick, presented at the 1995 Pittsburgh
Conference, Liquid Chromatography - Food Analysis Session, New
Orleans, LA, March 5-10, 1995.
4.11 E. M. Fujinari and J. D. Manes, "Food Flavors: Generation, Analysis
and Process Influence", G. Charlambous Ed., Elsevier Science
Publishers, Amsterdam, (1995) 929.
D. Wetzel and G. Charalambous (Editors)
Instrumental Methods in Food and Beverage Analysis
9 1998 Elsevier Science B.V. All rights reserved
467
5
The Determination of Compositional and Molecular
Weight Distributions of Cationic Polymers Using
Chemiluminescent Nitrogen Detection (CLND) in
Aqueous Size Exclusion Chromatography
Frank J. Kolpak a *, James E. Brady a, and Eugene M. Fujinari b
aHercules Inc., Research Center, 500 Hercules Road, Wilmington, DE 19808
U.S.A.
bAntek Instruments Inc., 300 Bammel Wesffield Road, Houston, Texas 77090
U.S.A.
5.1
INTRODUCTION
Differential refractive index (DRI) is the most common means of mass
detection for the molecular weight distribution analysis of polymers by size
exclusion chromatography (SEC). For aqueous SEC mobile phases
comprised of simple salts or buffers, this detection strategy is usually
adequate except for the presence of a system or buffer peak at moderately
long retention times, which may interfere with detection of the low moleular
weight end of the distribution. For polymers containing a chromophore, UV
detection is an improvement. The direct determination of nitrogen in liquid
chromatographic eluents by a chemiluminescent nitrogen detector (CLND),
now affords a potentially more sensitive means of characterizing nitrogen
based cationic polymers by SEC. This approach also allows for the use of a
greater selection of mobile phase compositions than the simple buffer or salt
solutions (i.e. aqueous/organic mobile phases) because the system peak is
transparent to the element specific nature of CLND. Chemiluminescent
nitrogen detection is shown to be a viable, and potentially powerful, means of
characterizing polymeric materials.
468
5.2
EXPERIMENTAL
Apparatus
Aqueous size exclusion chromatographic (SEC) separations were
achieved using a Waters 510 solvent delivery system and a Waters WISP.717
autoinjector (Milford, MA). A stainless steel T-splitter from Valco
Instruments Co. Inc. (Houston, TX) was used to achieve dual DRI/CLND
detection. Synchrom CATSEC columns which contain cationically modified
silica porous packing material, were purchased from Keystone Scientific Inc.
(Bellefonte, PA). The primary (mass) detection of cationic polymers was
accomplished with a Hewlett-Packard (Wilmington, DE) 1047A differential
refractometer.
Nitrogen specific detection of the SEC eluent was
accomplished with a chemiluminescent nitrogen detector model 7000 HPLCCLND from Antek Instruments Inc. (Houston, TX). Data collection was
accomplished with the Waters ExpertEase data acquisition and analysis
software run on a VAXstation 3100.
Reagents and Standards
The 20K, 250K, 350K, and 6,000K poly(acrylamide) standards were
purchased from American Polymer Standards (Mentor, OH). House distilled
water was used to prepare the mobile phase. Glacial acetic acid, lithium
chloride and ethylene glycol reagents were purchased from Aldrich Chemical
(Milwaukee, WI). All standards and reagents were used without further
purification. The SEC mobile phase was 0.20M Li acetate plus 2% ethylene
glycol (pH 4.5). It was filtered through a Rainin Instruments Inc. (Woburn,
MA) Nylon-66 0.22 lam pore size filter. The internal flow standard was acetic
acid.
Standards and Analytical methods
The poly(acrylamide) standards were dissolved at a concentration of 2
mg/mL in the mobile phase and injected without further preparation. The
cationic resin was dissolved at the same concentration in mobile phase and
filtered through a 0.45 ~tm PVDF membrane (Millipore Millex-HV Marborough, MA) prior to injection. Sample and standard injections were
50tttL and mobile phase flow rate was 0.25 mL/min. The columns were
CATSEC 4000A + 1000A + 300A + 100A in series. The temperature of the
columns and the DRI detector were thermostatted to 40~ The flow was split
50:50 to the DRI and CLND detectors after exiting the columns. The CLND
conditions were: 1050~ pyrolysis temperature, PMT voltage 650, range x50,
and detector output 1 volt. A schematic flow diagram for the DRI/CLND dual
detection SEC system is shown in Figure 5.1.
ao
__10
Zo
rE
r
oE
0
ll,,i,
a
9
9
o J0
Z
~
~
<
t~
~
469
470
5.3
RESULTS AND DISCUSSION
An example of the DRI/CLND dual detection SEC analysis that can be
performed is shown in Figure 5.2. In this plot, the DRI and CLND signals are
not adjusted for registration differences. However, it is clear that the
chromatographic envelopes for the two signals are essentially the same, which
would be expected for a homopolymer. The significant system peaks in the
DRI chromatogram (31.3 and 32.6 min) are virtually absent in the CLND
chromatogram, indicating that they are not due to the standard. Figure 5.3
shows the CLND chromatograms of three poly(acrylamide) standards with
order of magnitude increments in molecular weight. This plot demonstrates
the ability of the CLND to handle very high molecular weight polymers as
readily as low molecular weight ones.
In Figure 5.4, the dual SEC analysis of a chain extended polyamine is
shown. The polymer exhibits a bimodal molecular weight distribution due to
the chain extension step, which is accomplished with a nitrogen free reagent.
For a sample which is not a linear homopolymer, an analysis such as this can
provide chemical composition data as a function of the molecular weight
distribution. This approach can be particularly valuable in analyzing off-spec
material. The DRI/CLND S EC analysis can detect differences between two
lots of polymer which have similar molecular weight distributions, but
dissimilar chemical composition distributions. Such subtle, but previously
undetectable, discrepancies could account for noticeable performance
variation between the resins.
5. REFERENCES
5.1
F. J. Kolpak, J. E. Brady, and E. M. Fujinari, presented at the 18th
International Symposium On Column Liquid Chromatography,
HPLC'94, Polymer Analysis and Characterization Session,
Minneapolis, MN, May 8-13, 1994.
o
0
~
~
m
4)
~-
<(
Z
"
r
_~~I j 1
6
U.I
U)
j
~J
!
_,q.
o
9 o'J
-0 2 ,
E
E
_~1"-
9
E
~
E
o,...q
9
9
9
E
.o
x
s
o,.~
t~
471
d
472
u~
c~
I
0
I
I
I
O
I
U3
I
c5
t~
,-:
O
,-:
od
U3
04
o
o
o
~G
c5
O
<
(AUJ) esuodseH JoloeloCI ClN-IO
0
0
0
u,i
d~
0
E
V-
v
O ~
0
0
0
-~iii
-
-
0!
~0
~ oo
0
0
o,...4 ~
e~
t~
0
~g
0
.~ -~.
X
N
m~
m
0
~
I:::
0
I:::
I:::
ltl
~ .,..,,
J::
.I:::
.,.,,,,
I
e,i
|
r'1
Z
_J
(9
'r-
~--
(AUJ) osuodsoEl
!
0
-
rr
JOlOO]O(]
O
0
0
~
!
0
0
~t
0
co
E
:~
,...,
c
0
l-
~
v
O ~
c~
0
cJ
0
C~
cq
0
t-
~
9
9
.-I
C~
Z
C~
~
9
o
N
.o
~
t~
.,,.-i
473
This Page Intentionally Left Blank
D. Wetzel and G. Charalambous (Editors)
Instrumental Methods in Food and Beverage Analysis
9 1998 Elsevier Science B.V. All rights reserved
475
6
Chemiluminescent Nitrogen Detection in Capillary SFC
Heng. Shi a, J. Thompson. B. Strode IIIa, Larry T. Taylor a *
and Eugene M. Fujinari b
aDepartment of Chemistry, Virginia Polytechnic Institute and State
University, Blacksburg, VA 24060
U.S.A.
bAntek Instruments Inc., 300 Bammel Westfield Road, Houston, Texas 77090
U.S.A.
6.1
INTRODUCTION
Chemiluminescence detectors have become increasingly important in
the analytical world because of their inherent advantages in chromatographic
detection, i.e. extremely high element-selectivity and sensitivity. Applications
with the nitric oxide/ozone based chemiluminescent nitrogen detector
(CLND) for GC have been reported earlier [6.1-6.2]. Benn et al. showed
detailed examples for (GC-CLND) detection of nitrogen containing
components in flavors and essential oils [6.3]. CLND for HPLC was first
described for the detection of ammonium nitrogen in waste water [6.4]. Many
examples for nitrogen-specific (CLND) detection of various nitrogencontaining compounds in complex sample matrices have been shown in the
earlier sections (Parts 1 -5).
The CLND was sucessfully interfaced for the first time to capillary S FC
and studied extensively in terms of detector optimization under supercritical
fluid conditions 1) without a column for flow injection analysis and 2) for
capillary chromatography. More recently, new pharmaceutical applications
are reported where polar nitrogen containing compounds are eluted from
packed column SFC with methanol modified CO2 and detected using the
CLND [6.5]. S FC-CLND has opened a new dimension in analytical
chemistry. Its intrinsic use in the nitrogen-specific detection mode is
presented for analysis of horseradish oil and other nitrogen containing
compounds. Applications with the novel simultaneous CLND/FID for
capillary S FC are also discussed.
476
6.2
EXPERIMENTAL
Apparatus
A model 705D CLND nitrogen specific detector from Antek
Instruments (Houston, TX, U.S.A.) was interfaced to either a Hewlett-Packard
(Wilmington, DE, U.S.A.) 1250A supercritical fluid chromatograph (for flow
injection analysis) or a Dionex Lee Scientific (Salt Lake City, UT, U.S.A.)
series 600 SFC (for chromatographic separation). SB-cyanopropyl capillary
column (20 m x 100 lam I.D., 0.25 ~xm film thickness) from Dionex and a DB1701 (10 m x 100 ~m I.D., 0.4 ~tm film thickness) column from J & W
Scientific (Folsom, CA, U.S.A.) were utilized in this study. A 25 lam I.D.
Integral restrictor was used with the DB-1701 capillary column and a 50 ~m
I.D.Frit restrictor was used with the SB-cyanopropyl capillary column. Time
split injector from Valco (Houston, TX, U.S.A.) was used with a 500 nL rotor.
A Hewlett-Packard 3310 integrator was used for data acquisition.
Reagents and Standards
Allyl cyanide, allylisothiocyanate, 2,3-dimethylindole, 2,6dinitrotoluene, diphenylamine, indole, 4-nitrotoluene, 2-phenylindole, phenyl
ethyl isothiocyanate, and pyridine were purchased from Aldrich (Miliwaukee,
WI, U.S.A.). p-Nitroaniline was purchased from Fisher Scientific (Pittsburgh,
PA, U.S.A.). Caffeine was purchased from Sigma Chemical (St. Louis, MO,
U.S.A.), 2,5-1utidine from Chem Service Inc. (West Chester, PA, U.S.A), 2butyl isothiocyanate from Lancaster Synthesis Inc. (Windham, NH, U.S.A.),
and dimethoate from Accu Standard Inc. (New Haven, CT, U.S.A.).
Alkyldimethylamine mixture was obtained from Albemarle Corporation
(Baton Rouge, LA, U.S.A.). All chemicals were used without further
purification. Horseradish oil standard, Wasabi and the hot yellow mustard
were received from commercial sources and extracted into methanol. HPLC
grade solvents from EM Science (Gibbstown, NJ, U.S.A.) were used for
preparing standard solutions. Grade 4.3 oxygen from Airco (Murry Hill, NJ,
U.S.A.) was used as both pyrolysis and ozone-generator gas. SFC-grade CO2
was obtained from Air Products and Chemical Inc. (Allentown, PA, U.S.A.).
Chromatographic Conditions
All analyses were performed with pressure programming. In all cases,
the oven temperature was held constant throughout the run. Integrator
conditions: 1 volt input, attenuation 7. CLND conditions: pyrolysis
temperature 1050 ~ PMT voltage 750, range x50, detector output 1 volt.
FID conditions: 360 mL/min air, 65 mL/min hydrogen, 34 mL/min nitrogen
477
(make-up gas), 1 volt output, and detector temperature was 3 5 0 ~
Additional chromatographic conditions are cited in the Figure legends.
6.3
RESULTS AND DISCUSSION
The first successful interface between the chemi|uminescent nitrogen
detector (CLND) and capillary SFC systems was accomplished. Frit and
Integral restrictors were used and positioned at the bottom of the cylindrical
burner of the CLND as shown in Figure 6.1. In optimizing the detector,
several effects such as restrictor position, and pyroreactor oxygen flow rate
have been investigated by flow injection analysis (FIA) using the HewlettPackard SFC system. It was found that the restrictor position (of 15 cm from
the base of the nut to the restrictor tip) was critical to the detector
performance. The response factor relative to indole for several nitrogen
containing compounds has been measured by the CLND under S F conditions
without a column and are listed in Table 6.1. Results indicated an equimolar
Table 6.1
Response Factors ( f ) relative to indole with CLND
and 100% CO 2 supercritical fluid conditions
Compound
caffeine
3,3-dimethylindole
2,6-di nitrotol uene
diphenylamine
indole
4-nitrotoluene
2-phenylindole
pyridine
(fx)/N
1.07
1.04
1.05
1.06
1
1.01
1.01
1.01
nitrogen response by the detector. The limit of detection of 60 pg of nitrogen
was determined in 100% CO2 mobile phase. Detector linearity was found to
478
to reaction chamber
pyrolysistube
oxygen
,,
frit restrictor
! ..... d
! ---= ;:--.:.11 i
I
- - -~ . . . . .
/
15 cm
l column effluent
Figure 6.1
Schematic illustrating the interface between the
supercritical fluid c h r o m a t o g r a p h y (SFC) system
and chemiluminescent nitrogen detector (CLND).
A frit or integral restrictor (from the exit end of
the capillary column) is placed at the base of the
CLND pyro-furnace to achieve best oxidation and
pyrolysis of the sample eluent. Chemiluminescence
takes place in the reaction c h a m b e r of the detector.
Reprinted from H. Shi, J. T. B. Strode III, L. T.
Taylor*, and E. M. Fujinari, J. Chromatogr. A, 734
(1996) 303 with permission.
479
be at least three orders of magnitude at a single range setting. A selectivity
(N/C) of 107 was obtained.
The CLND was then coupled to the Series 600 Lee Scientific SFC
system using S B-cyanopropyl and DB-1701 capillary columns for
chromatographic separation and subsequent nitrogen-specific detection.
Figure 6.2 shows structures of the 5 nitrogen containing components found in
horseradish oil standard. SFC-CLND chromatogram of the horseradish oil is
profiled in Figure 6.3. Levels of horseradish oil in foods for flavor
assessments are typically quantitated by measuring the allyisothiocyanate
component (peak C). In addition to the unidentified peak C, the CLND
chromatogram of the hot yellow mustard (Figure 6.4) showed two compounds
(peaks:A=allyisothiocyanate and B=2-butylisothiocyanate) which are also
present in the horseradish oil. Figure 6.5 shows the SFC separation and
CLND detection of 7 nitrogen containing compounds, including caffeine
(food and beverage ingredient) and dimethoate (insecticide), peaks 4 and 6,
respectively.
Simultaneous UV and sulfur chemiluminescence detector (SCD,
Sievers Instruments Inc., Boulder, CO, U.S.A.) has been previously interfaced
to SFC by Howard et al. [6.6]. Bornhop et al. reported the analysis of a
polysufide by capillary SFC and a dual SCD/FID system I6.7]. Chang et al.
demonstrated SFC analysis of some sulfur compounds in garlic oil using dual
SCD/FID system [6.81.
A simultaneous detection using CLND and FID for capillary SFC
systems was achieved in this study and a schematic flow diagram is illustrated
in Figure 6.6. The dual CLND/FID chromatogram for SFC of a mixture of 6
alkyldimethylamines in toluene is shown in Figure 6.7. A good peak for peak
retention time correlation between the two detectors was observed. The
advantage of the CLND is shown where the toluene (hydrocarbon solvent)
peak is transparent to the CLND. Much stronger CLND response than the
FID was observed for these dimethylamines (nitrogen compounds) with n-C8,
n-C10, n-C12, n-C14, n-C16, and n-C18 alkyl functional group, respectively.
Figure 6.8 is the capillary SFC and dual CLND/FID chromatogram of a
Japanese horseradish Wasabi extract in methanol. Notice the sensitivity
improvement by the CLND over that of the FID. The methanol solvent blank
by the CLND is shown by only the baseline without the methanol peak. The
CLND profile for the Wasabi very clearly shows a much different
composition from the CLND profile of the yellow hot mustard (Figure 6.4)
and the horseradish oil (Figure 6.3). Wasabi consisted of two identified peaks
A and B, allylisothiocyanate and phenylethylisothiocyanate, respectively.
Both the hot mustard and Wasabi samples consisted of compounds found in
the horseradish oil, but each having a different composition of hotness. When
used in conjunction with organoleptic evaluations, the compositional
480
HN
H/
H
C"
-C ~ C
f
H
H\
H/
I
i
----- C ~
N
A = but-3-enonitrile
H
H
C ~
H\
H/C ~
I
C -----C ----- S
I
H
I
H
-C~N
B = allylthiocyanate
H
I
~ ----C---- N ~C
I
H
H
C = allylisothiocyanate
N=C~S
D = 2-butylisothiocyanate
N~C~S
E = phenylethylisothiocyanate
Figure 6.2 Structures of the nitrogen containing components
in horseradish oil standard.
O-
,,
~o
tlJ~
0----~
,i
--
0
E
b.
0
~
c~
oE~
~
u
~
0
c~'~
~o..~
~9 ~ ~
9
t~
E ~-~8~
,-~
~.r
9
0
481
482
<
~
w
~,~d
-o
o
0
~
o
"~
.'~
o
0~
o f~
~ 0
o~
0
=
.~
o
r~
o
14
0
E
o
rj
~
~t-..
~f~ ~ .Z.," o
o
od
o_:
•
:IL I:I
~ c'q i~
~
~
~.~
o~,
"~
o~
I.., I-,
.~
r.~ 0
1.2,5-lutidine
2. Cnitrotoluene
3.2,S-dinitrotoluene.
4. Caffeine
5. lndole
6. Dimethoate
7. p-nitroaniline
3
14
L
7
0
I
20
Time (min)
Figure 6.5 Supercritical fluid chromatography and chemiluminescent nitrogen detection
of a mixture of several organic nitrogen containing compounds.
Conditions: pressure program 100 atm (hold 4 min), ramp to 250 atm at
15 atm/min, then ramp to 340 atm at 30 atm/min (hold 3 min); Cyano
column ( 2 0 m x 100 pm I.D., 0.25 pm film thickness); time split injection
0.2 sec; sample solvent: methanol; decompressed CO, 8.5 mL/min.
H. Shi, J. T. B. Strode 111, L. T. Taylor*, and E. M. Fujinari, J. Chromatogr. A,
7 3 4 (1996) 303 with permission.
P
00
W
484
o
c
N
0
m
c
~ m,,,~
i
E
I11
5
D
m
It..
\
D
Z
_J
0
m
c
E
0
0
U_
.=.d z_o~
,.4
em
o~,,~
_
~9
~
0.~..
9
,
E.E_
~, ~, ~ ~ ~
"
.
E
o~
~
o
~
.-E
---_._~
ooo~~
o
m
1.1.
t~
Z
~
.
.
.
.~,
I1.
.
.
m
~.~
~.~~
: ~ ~~-~
9
.
.
.aO~o
~
.,i .
.
0
E
~
~
9
a
9
~ •
9
~'~,
~-
~ ~
~.~
..
.
~~ !~ ~ ~
~ . ~ ~o_
.-~~~~
~
485
486
0
te-
a
m
n_
(1)
e0
0
~ 0
m
'~
~
oN,
<-)
a
z
..i
0
m
1
o~=~ ~
a9 b -g' :
A
v
13=
z~
,-,1'5
,,,,,,,,,t
rr
,,---
v
E~
Z
rj
o
.~
~
"t3
s
~
rj 3
00
~
,.,.
0
I:3,,o
~176
I=
.o
E~
=L
x
o
0
o
_=
o
487
differences of these nitrogen containing compounds found by S FC-CLND in
samples can be readily correlated to their flavor enhancing properties in a
variety of foods. Additional information about S FC-CLND are reported in
[6.5, 6.9, and 6.10].
Acknowledgements
The authors thank Antek Instruments Inc., for the loan of 705D
nitrogen specific detector. We would also like to acknowledge HewlettPackard Company and Lee Scientific Division of Dionex Corporation for the
loan of the S FC systems. In addition, many thanks to Air Products and
Chemicals, Inc. for the donation of carbon dioxide.
6. REFERENCES
6.1
6.2
6.3
A. J. Britten, R & D, 31 (1989) 76.
L. O. Courthaudon and E. M. Fujinari, LC-GC, 9 (1991) 732.
S. M. Benn, K. Myung, and E. M. Fujinari, in "Food Fl,avors,
Ingredients and Composition", G. Charalambous (Ed.), Elsevier
Science Publishers, Amsterdam, (1993) 65.
6.4 E. M. Fujinari and L. O. Courthaudon, J. Chromatogr., 592 (1992)
209.
6.5 H. Shi, L. T. Taylor, and E. M. Fujinari, J. Chromatogr.A, 757 (1997)
183.
6.6 A.L. Howard and L. T. Taylor, Anal. Chem., 65 (1993) 724.
6.7 D.J. Bornhop and B. J. Murphy, Anal. Chem., 61 (1989) 797.
6.8 H.C. Karen Chang and L. T. Taylor, J. Chromatogr., 517 (1990) 491.
6.9 H. Shi, J. T. B. Strode III, L. T. Taylor, and E. M. Fujinari,
J. Chromatogr.A, 734 (1996) 303.
6.10 H. Shi, L. T. Taylor, and E. M. Fujinari, J. High Resolu. Chromatogr.,
19 (1996) 213.
488
"There is a growing number of researchers who are lucky enough
to be working on the cutting edge of science. Some are fortunate
to hang glide and soar beyond this edge to establish new boundaries.
These are the hang gliders of science."
Eugene M. Fujinari
Facilitator of Applied
Chromatography
and Detection
May 27, 1997
D. Wetzei and G. Charalambous (Editors)
Instrumental Methods in Food and Beverage Analysis
9 1998 Elsevier Science B.V. All rights reserved
489
The SPECMA 2000 data bank applied to flavor and fragrance
materials*
F. Colon and G. Vernin**
Laboratoire de Chimie des Ar6mes-Oenologie (CNRS, URA 1411)
Facult~ des Sciences et Techniques de St-J~r6me, Case 561,
Avenue Escadrille Normandie-Ni~men F 13397 Marseille C~dex 20
ABSTRACT:
After briefly reviewing our previous works since 1982 on the
SPECMA bank, its originality, its specificity and its limitations due to the data
processing of the time, we describe the modifications, improvements and the main
functions of the new SPECMA 2000 bank. This, like previous versions, is based on
the comparison between an unknown mass spectrum and a computerized library
of reference spectra. Its contains a certain number of data such as the Kovats
indices used as filters, the Registry Number, the descriptors and the olfactory
thresholds when kown, the name, origin and bibliographical reference of the
spectrum and a whole series of useful programs. It was conceived with a Pentium
(Intel processor) whose designed RAM has been brought up to 16 Mo. It functions
with ACCESS software via WINDOWS 95. A program of molecule design
CHEMWINDOWS DB has been incorporated into the previous system. Thus, it has
maintened all its interactivity and its originality while being more time saving and
including several options. We give some examples of numerous spectra (among
more than 4500) taken from our main topics of research: essential oils, spices and
flavorings, Maillard and related model systems, fruits, wine and alcoholic
beverages.
1.
INTRODUCTION
Among the different methods of identifying by mass spectrometry of an
unknown compound: theoretical, pattern recognition and comparative methods,
the latter have proven to be the most efficient. That is why they have developed
over the past 20 years. Initially, concerned with organic molecules in general, they
became more and more specialized, in the form either of spectrum libraries, or of
computerized banks, namely in the field of fragrances and flavors which interests
the perfume and food industries. They are reviewed in another chapter of this
book.
* The bank is not presentlycommercialyavailable but informationcan be obtainedon demand.
** Author to whom correspondenceshould be addressed.
490
Aware of the necessity for this specialization as early as 1982 we conceived a
first version of a bank called SPECMA, devoted to the heterocycles of the Maillard
reaction and of food flavors in general.
As a result of the progress made in data processing and software, this bank
which was initially on CP/M was translated into the TURBO PASCAL (version 6) in
the years 1985 to 1988 and extended to all the volatile compounds of essential
oils, natural aromas and Maillard reaction as well as to similar related model
systems (1-3).
Furthermore, we have introduced Kovats indices as filters in the search in
order to distinguish between two products with similar spectra but different eluting
orders on polar and non polar columns, when the mass spectrometry is coupled
with the gas phase chromatography. This is the analytic technique which has been
in general used since 1967 in this type of study.
With the incessant increase in the performance of data processing and
sotfware, we have again created a new version of this bank called SPECMA 2000
with the same basic principles but using ACCESS via WINDOWS 95 software and
a Pentium equipped with the Intel processor. Moreover a molecule design
program (CHEMWlNDOWS DB) has been incorporated into the previous system.
This is the version which we present with its applications in this chapter.
2.
RECALL OF OUR PREVIOUS WORKS (1 -8)
The identification of a given organic molecule by mass spectrometry is based
on three groups of methods
Theoretical methods or "Artificial intelligence" consist either in drawing a
logical inference from structural information by examining the different spectra or in
making suppositions concerning the structures, deducing the fragmentation
patterns, then matching the theoretical spectrum with the unknown compound. The
very insufficient theoretical knowledge of mass spectrometry appears to a non
specialist to be very difficult to overcome.
Statistical methods called "Pattern Recognition" computerized under the
name of "Learning machine", They consist in utilizing spectrum structures,
empirical correlations enabling a rapid classification of the unknown compound
into a given chemical family. This method seems to be of interest from only one
point of view: the classification of a compound absent from the reference library by
using some structural features.
Comparative methods or "Library Search" consist in comparing the
unknown spectrum with every spectrum present in a MS data bank which must be
as complete as possible. These last methods have appeared to many analysts to
be by far the best ones since the development of the GC/MS technique in 1967.
491
Owing to the drawbacks of the early MS data bank (lack of specificity,
insufficient selection criteria, no discrimination between the spectra in the literature
for a similar product, Kovats indices not used as filters etc.), in 1982, we decided
to remedy this situation and to design our own data bank called "SPECMA".
The construction of a computerized data bank using comparative methods
requires the solution of a certain number of problems which have been described
elsewhere (5, 7). They are summarized in Scheme 1.
a) The first problem is the mass spectrum specificity based on the double
implication:
Identical compounds
--~
Identical spectra
The first implication depends on the spectral reproducibility. This condition is
connected with experimental conditions:
i) the ionization voltage (usually 70 eV)
ii) the type of apparatus (magnetic, quadrupole and more recently ion trap
developed by Adams (9). Mass spectra obtained by this type of instrument are, in
most cases, quite similar to those obtained with a quadrupole. Differences are
greater between magnetic and quadrupole apparatus. Low intensity of fragments
(m/z: 43, 41,39, 29, 27) are, as a general rule, higher with a quadrupole than with
a magnetic.
iii) the scanning position on the chromatographic peak. At the top a saturation
phenomenon can be observed (several base peaks at 100%). If the mass
spectrum is recorded very close to the edge, the presence of impurities can modify
the spectrum.
iv) with a too small amount of product, the spectrum is altered by a background
noise. The second implication cannot always be borne out since various
compounds (homologous, cis and trans, (Z) and (E) isomers have similar
spectra. This is particurlarly the case with monoterpene hydrocarbons (with the
exception of limonene) certain sesquiterpenic alcohols, and many other products.
b) The second problem is the data acquisition and validation. Data acquisition can
be made either from literature data or from our own GC/MS analyses.
i) Literature data. They have been compiled in data bases (Wiley, TNO etc.)in
specialized libraries, works or Atlases, in original papers, reviews, in the reports of
congresses and theses (for references see the previous chapter) and (10 -13).
Some of these sources are not always usable. Mass spectra are too small, without
numerical values for intensities and uninterpretable; others give main fragments or
a limited number of fragments classified in decreasing order of intensity causing a
loss of information. On the other hand, experimental conditions are inaccurate or
non-existent. Furthermore, some published spectra are very different for the same
product due to an erroneous interpretation.
To solve these problems it is necessary to compare several spectra of the
same product when available and to take into account authors' specializations in
the field. Errors can also be detected by a good knowledge of the fragmentation
process.
492
IDENTIFICATION OF VOLATILE COMPOUNDS
BY GC/MS
LIBRARY SEARCH
WITH LOW RESOLUTION MS DATA
1
MASS SPECTRUM SPECIFICITY
IDENTIFICATION CRITERIA I
AND FILTERING (KI)
I OATAAOOU,'s't'~ I
,,,
1
CONSTRUCTION OF PROGRAMS[
l
1
! co~sT~OCT,O. OF ~'~Sl
1 MS DATA BANK I
Scheme 1. The flowchart of the MS-KI bank design (1)
493
ii) Our own data. Laboratories possessing a combined GC-MS system set up their
own bank. It is the most efficient solution but also the most costly and timeconsuming. For the identification of each mass spectrum of a GC-MS listing it is
necessary to first constitute a bank or a library of mass spectra arising from the
literature. Answers given by the data bank (EPA/NIH, for example) coupled with
the GC-MS system afford useful but limited information.
Encoding of mass spectra
At the early stages of GC-MS analyses, microcomputers were not powerful
enough to store all MS data, i.e. fragments and corresponding intensities. For this
reason, only a limited number of fragments with their intensities (5<N<10) were
taken into account. The various methods described in the literature have been
summarized elsewhere (5).
The best encoding method for a spectrum seems to consist in (7):
- Selecting N highest fragments with N high enough to take all important ones.
- Taking N variables for every mass spectrum, because some of them are very
simple and others contain a large number of fragments
- Fragments of low intensity (< 10%) but characteristic of a given molecule: primary
alcohols (m/z : 31), secondary alcohols (m/z: 45), alkyl esters (m/z: M + - OR and
M + - COOR), alkyl derivatives (m/z" M + -CH3 and m/z" M + - 43) sulfur derivatives
(m/z: M + - SH and M + - SCH3) etc.. must be taken into account. The peaks at m/z'
28 and 32 a.m.u, must be discarded to avoid air peaks as well as the fragment at
m/z: 207 (artifact)(with some exceptions). Whatever, there is always a loss of
information between measurement and publication of a spectrum.
Choice of a comparison function
If $1 are the intensities of an unknown spectrum and $2 those of a
reference spectrum, the comparison criterion can be reduced to a function F ($1,
..=,,,,1==~
$2) such that :
$1 = $2
~
m
F(S1,
~
~2)
=0
The most simple bijection arises from attributing the ratio m/z = k, the kth
coordinate of S (provided that m/z values are integer (5).
In practice, the function F is always positive and the closer to zero F becomes,
the more similar the spectra are.
A threshold value FS is defined such that :
F ~< FS
F > FS
Identicalcompounds
Different compounds
Different functions have been described in the literature and previously
summarized (5). An original approach was applied by Petitjean (1) for the
SPECMA data bank. The function used in this case allows one to work with
variable tolerance with respect to the gaps in the spectra according to their origin.
494
In this function"
F
Sr, X, R) = F(
Dr)
oT,-
X and R are the lists of parameters for the unknown and reference compounds.
These parameters take into account variations in experimental conditions and
spectral quality.
C h o i c e of K o v a t s indices as filters
For many years we have recommanded the use of Kovats indices on polar and
non polar columns to differenciate various compounds giving the same mass
spectra (3, 5). They possess a number of properties which can profitably be used
as a route of identification.
i) By definition, the Kovats index of a linear alkane CnH2n+2 is equal to 100 n;
ii) They are little influenced by temperature changes and may be used with
programmed temperature;
iii) For the higher numbers of a homologous series, the Kovats indices may be
expressed as:
Kin= KI + 100n
iv) For unsymmetrical substituted compounds ( R - X - R') Kovats indices may be
calculated from symmetrical compound indices by"
KI(R-X-R')=
[KI(R-X-R)+
KI(R'-X-R'
],/~
v) On the basis of the Kovats' index of a parent molecule (PH), it is possible to
calculate the indices of its derivatives by using the additivity of the substituent
increments"
i
KI = P- R = KI (P- H) + F_,(A)Ri + f (Ri, Rj)
where f(Ri, Rj) is a negative function taking into account the interaction of various
groups Ri and Rj in the molecule. Its value depends on the polarity of groups.
vi) The difference of Kovats indices (KID) on a polar phase (KIP) and on a non
polar phase (KIA) is characteristic of a functional group:
KID = KIP - KIA
By definition, for a linear alkane, it is equal to zero. The more polar the column, the
greater will be the KID of a polar solute and its family.
Other filters such as the GC-FTIR technique are also used but they are very
expensive and more or less efficient.
495
The computerized data
They have been initially divided into two sets called SPECMA.DAT and
NOMREF.DAT files, respectively. The SPECMA.DAT contains the following
information: empirical formula (C, H, N, O, S, Z), molecular weight, mass spectra
(up to 25 peaks with their El intensities), spectra in PCI and NCI, Kovats indices
(KIA, KIP, KID) lower mass limit. The NOMREF.DAT file contains all other
information which is not specifically useful for mass spectra searches, i.e, the
name, odor and flavor descriptors with threshold values, registry number and
reference. In the two files extensions have been anticipated.
This data bank was first realized on a North Star Horizon microcomputer of 64 K
bytes, equipped with a dual 8" floppy disk unit. It was run by CP/M. The source
program (PL1 80) was parameterized so as to optimize adjustments to various
types of diskettes.
In 1985, the bank was implanted on an IBM microcomputer using TURBO
PASCAL, version 3 and then version 6 as software. This version was saturated
with 4.000 spectra. On the other hand, products being classified by increasing
molecular weight, input of a new product of low molecular weight was very timeconsuming.
For all these reasons and taking into account the increased performance of
microcomputers and software, a new revision of the bank was undertaken using
ACCESS via WINDOWS 95 and a Pentium (RAM16 Mo), Intel processor (14).
3.
SPECMA 2000 DATA BANK: REALIZATION
3.1. New modifications relative to the previous versions
-
Choice of ACCESS as data base.
The first versions of the bank were developed by using proprietory languages
(PL1/CPM, TURBO PASCAL) whose major disadvantage was the lack of versatility
and evolutions before a new developing standard, such as WINDOWS. By
choosing ACCESS developed by MICROSOFT, these disadvantages could be
mastered and it was hard to imagine that the inventor of WINDOWS, MICROSOFT,
would not continue to develop its only data base. The use of a language such as
MICROSOFT'S VISUAL BASIC would have equally been a good choice. But
ACCESS gives the user greater liberty, allowing personal modules to be added
without being enclosed in a compiled program. After a few "bugs" in the first
versions of ACCESS a rather radical modification of the data access programming
methods was to be found in version 2 which lead us partly to reprogram once
those.
The current version of the bank functions on ACCESS version 7. The
programming methods have again been modified to come close to VISUAL
BASIC. Finally, MICROSOFT has created an efficient product , completely
integrated in the 32 bit environment of WINDOWS 95. It is pleasant and easy to
use so that a non-expert may question the data base or create his own display
screens.
496
- The different tables used
Because the new SPECMA bank is not limited by the number of compounds,
the use of a data base was called for, in order to overcome the inherent difficulties
in programming file management and numerous recordings. In ACCESS the latter
are stored in tables. The main tables are defined as follows:
* The TBANQUE table contains the list of all the compounds and each recording is
numbered singly, unlike the previous versions.
* The TPICS table contains all the fragments of all the compounds. A field
containing the number of the corresponding compound enables all the fragments
of a given compound to be found.
These two tables are essential for the operation of the bank. Other tables allow
the application to be managed. These tables are bonded to each other so as to
maintain a certain logic. For example, it is important for each peak of the TPICS
table to be matched by an associated compound in the TBANQUE table. This is
checked by ACCESS thanks to the referential integrity mechanism.
- Referential integrity
This is an essential concept when relations between different tables are to be
dealt with. Its guarantees the integrity of the base. When using the bank, the
consequences are numerous for the constraints it imposes are effective in the
options: up-dating, deleting, additions. In our case, this integrity is confirmed
between the TBANQUE and TPICS tables. One single compound corresponds to
each peak recorded in TPICS. When a compound is deleted, the associated peaks
must be deleted in the TPICS table.
- Numbering the compounds
Each compound possesses a single, permanent number coded by ACCESS.
This number is chosen interactively but without reusing the attributed numbers,
even if the number belonged to a deleted compound. But after repeated creations
and deletions, this numbering is not at all continuous. The solution of renumbering
all the compounds is too complex but is conceivable.
3.2.Running
of t h e d a t a
bank:
various
options
The SPECMA 2000 data bank has been conceived in such a way to simplify
the operations which were necessary in the previous versions written in TURBO
PASCAL (versions 3 and 6). Numerous options have been added rendering the
management of the compounds and the search program easier.
The "MAIN MENU" called "COMPONENTS MANAGING" allows the various options
of the bank to be accessed (See Figures 1 and 2).
-" FAST LIST " allows all compounds to be shown. The latter are classified
according to alphabetical order. In the present case we choose the ~ letter from
the lower left-hand part of the screen (A, B, C...E .... Z) and check the 2-ethoxy-3ethylpyrazine as an example (N ~ 1465) (darker on the screen). The right-hand part
contains the seven options of the bank which are summarized below (Figure 2).
497
Figure 1. SPECMA 2000 Data bank: Presentation and
"MAIN MENU" called "COMPONENTS MANAGING"
498
t~9
"0
t-O
O.
E
0
t--
0
O9
~z
0
. 0i
..d
0")
F-O9
LL
c~
.~
499
-"CHECKING OF A COMPOUND". The "check" switch allows a compound to be
shown (N~ 1465, see Figure 2) and subsequently, printed or modified using the
corresponding options.
-",,s
allows the mass spectrum of the checked compound to be
displayed directly on the screen. A switch "ZOOM" enables the spectrum to be
enlarged and a choice between the different ionization techniques (El, PCI, NCI
modes) to be made (See Figure 3).
Figure 3. SPECTRUM option. As an example: 2-Ethoxy-3-ethyipyrazine (N~
-"MODIFY" The modification switch allows all the data concerning the checked
compound to be modified (See Figure 4)i.e. Name, Kovats indices, Positive lists
(15), Molecular formula, Registry Number (CAS), Reference to Chemical Abstracts,
Descriptors and Threshold values, Family and Sub-Family, Occurrence, the Mass
spectrum and the Drawing of the molecule in the lower right-hand part of the
screen (or to be corrected, if necessary).
The "CANCEL" switch allows exit from the page without any change.
The "SAVE" switch, on the contrary allows the modifications to be stored.
500
e-0
o
9
Z
I.-r
tl
9
z~
1.1..
501
- "PRINT" This switch allows access to a dialog box which suggests the standard
printing options (See Figure 5).
Figure 5. "PRINT" option
- Printing of a checked compound
Printing of a set of checked compounds
- Search parameters used when the operator wants to print the result of the
unknown compound search (See Figures 7 and 8).
-"DELETE" This option deletes each compound after entering its corresponding
number. This is not used again. The program asks YES or NO.
- "NEW" This very important option allows the input of a new component into the
bank (See Figure 6). The data acquisition screen for a new compound is divided in
two distinct parts: the left side contains the field of data acquisition of the
compound description and the right side the mass spectral data. Pressing this
switch induces the appearence of a blank data acquisition screen. These data
include:
- Usual name, synonyms and chemical name
Origin and reference of the mass spectrum
- Molecular formula" Cu, Hv, Ow, Nx, Sy, Zz, and molecular weight which is
immediately calculated.
- Kovats indices on non-polar (IKa) and polar (IKp) chromatographic columns at
-
-
programmed temperature, as well as DIK which is characteristic of a given family.
It is also used as a filter in the search option.
502
0
O_
0
9
D
Z
9
13.
o
U.I
~
z
.
503
Numbers on positive lists (FEMA, CoE, IOFI )(15)
- Registry number (CAS)
- Reference to Chemical Abstracts of the paper giving the mass spectrum.
- Descriptors including odor, flavor, aroma taste and the corresponding threshold
values when they are available (16-24). Computerized structure-odor relationships
are of interest to classify an aroma compound according to its structure. Numerous
methods have been developed and works are in progress (25, 26).
- Families to which the compound belongs. They have been classified as follows:
Alicycliques
- Irregular monoterpenes
Aliphatics
- Monoterpenes
Aromatics
Nor-isoprenoids (C13)
Diterpenes
- Sesquiterpenes
- Fused heterocyclic compounds
- Triterpenes
- Heterocycles
- Sub-Family. In each family compounds have been classified in various
categories. Sub-families for heterocycles are given in Figure 7.
- Occurrence. This option concerns the flavor and fragrance field in which the
compound has been found. The following topics have been chosen:
- Alcoholic beverages
- Fruit, wines, alcoholic beverages
Essential oils
- Grapes
Essential oils and Maillard
- Maillard/Model systems
Fruits and essential oils
- Spices
- Fruits and wines
- Synthesis
- Vegetables
- Tea, coffee, cocoa
- Wines, alcoholic beverages and Maillard
This list is not exhaustive.
When several compounds are added to the bank, they are numbered by
increasing order. This number is automatically attributed by ACCESS to the file
during the data acquisition and cannot be modified.
- Mass spectrum. After the occurrence data, the runner set on the mass spectrum:
mass
intensity
Mode (El, PCI or NCI)
-
-
-
-
-
-
-
-
-
The number of fragments and their intensities are unrestricted. After "SAVE" the
compound is recorded under is own number. The "MODIFY" option then allows
the corresponding molecule to be drawn using the CHEMWINDOW DB program
and to be inserted into the mass spectrum.
- " S E A R C H " option. With ~
it is one of the two options most in use (See
Figure 8). It allows a compound to be searched using all the described data for
the I.'.NEW" option. By pressing the "SEARCH" (or SEARCH COMPOUNDS) switch,
the screen appears showing all the various possible data to be entered.
- If we want to search a compound using "USUAL NAME", it is enough to type a
part of the name. However, the latter must be quite precise in order to avoid too
great a number of answers.Kovats indices are given at +/- 2% and DIK at +/- 4% in
order to take into account the difference in experimental conditions for polar and
non-polar columns, respectively. A range for molecular weight is scheduled in the
case where it is unknown.
504
Figure 7. "FAMILY" sub-option.
~
o
9
Q.
0
a
z
r~
9
o
T
0
r~
iii
ii
505
506
In opposition to the previous versions of the bank, it is not necessary to enter by
the keyboard all fragments and their intensities for the mass spectrum. Usually, the
two, three or four most important fragments are sufficient. The corresponding
intensities are entered with a range of error from 0 to 100%, chosen by the user. If
the result is not satisfactory the range can be changed with the switch "CLEAR
ALL". The input of molecular weight and Kovats indices greatly improves the
search. It is possible to correct an error at any time.
As an example, the search for 2-ethoxy-3-ethylpyrazine is given in Figures 9 and
10. We choose various intensities and error ranges (20, 40, 40, 50%) for the four
highest peaks at m/z: 124, 123, 95 and 152, respectively. In both cases, the
program gives the right result, showing the interest of this approach relative to the
use of a function.
When we want to execute a new search, the "CLEAR LAST SEARCH" switch
allows all the parameters of the previous search to be deleted. The fields of the
screen are automatically deleted at the time of starting up the bank or when
"COMPONENTS MANAGINg" appears.
Second example of search: Gleenol case (Figure 11 ).
Recently, in his book Adams published the mass spectrum of gleenol (9). This
product has been found in our recent study on the Helichrysum Stoechas
(overlasting) essential oil harvested in (Alpes de Haute Provence). The two mass
spectra (El) have been input in the SPECMA bank (See Figure 11). In order to
study the influence of the range of error concerning intensities and to find these
spectra again by the SEAR(~H option, this latter has been varied. The three most
abundant fragments have been taken into consideration and they are reported
below with their corresponding intensities.
ADAMS
m/z
Int.
GC/ITMS a)
Error
range
121
81
41
100
73
70
30-50%
40-50%
80-100%
a) Ion trap mass spectrometer(9)
VERNIN
GC/QMS b)
m/z
81
121
41
Int.
100
83
44
Error
range
30-40%
20-30%
60-80%
b) Quadrupole mass spectrometer
Only the two mass spectra have been found with the above reported values for the
range error. Below these values only a mass spectrum was found.
These preliminary results emphasized that for an unknown compound fragments
of high intensity (60 to 100%) near the base peak, a range of error from 30 to 50%
must be taken into account.
507
FIRST PROPOSAL
ISearch parameters:-" I
Usmd mmae:
Orlgta &
M
ref:
~
CHONS
fonaula:
IKA:
1080
IKP:
FEMA:
1430
I~N (CAS):
(152)
DIK:
COE
IOFI:
lOO
124
-El
80
CA:
Descriptor odor/Flavor.
350
.=
123
95
,=
152
,
....
, ....
20
40
, ....
, ....
60
80
, ....
L
,
,,-
120
100
140
, ....
160
~ .... i
,, ,' ~w;~, ,~,,,-i,~-,r,
180
220
200
240
260
•
280
300
320
El: 124(100) 123(70)95(50) 152(43)
100
-EI
124
123
80-
60-
152
41
411
81
67 ++
20
~
| |
0
+
w r l w
20
t
~
|
|
, I
.
40
.
.
.
i
I'TI
60
i
II+
_
+
i
,
80
•
,
100
120
-,,,,,
1~
,~ . . . .
1~
l~
~",,',.-:1;,,-~
2~
~O
....
2~
2~
. .......
2~
El: 124(100) 123(94) 96(71) 152(60) 41(60) 81(42) 57(32) 68(30) 107(27) 108(25) 56(25)
Figure 9. Search for 2-ethoxy-3-ethylpyrazine using different range errors.
300
320
508
FIRST PROPOSAL
[Search parameters: I
Usual name:
Orlrl. & rd:
M o l e c u l a r formula:
IKA:
l l 15
FEMA:
100
C H O N S (152)
IKP:
1460
COE
--
ILN (CAb'):
DIK:
345
IOFI:
'
-El
CA:
D e s c r i p t o r odorfFlavor:.
'
123
80-
124
152
60-
95
40-
200
....
0
i ....
20
i ....
, ....
t, ,,
40
60
80
, ....
!
100
....
,,,,
120
,,,,,,
140
160
....
180
; ....
i,,r,;
200
220
....
240
I ....
, ....
260
280
;
....
34}0
320
El: 123(1(30) 124(80)95(60) 152(70)
100
1:24
:123
-El
8O
95
60
40
81
20
, ....
0
0
20
. . . .
40
iI'1"!
!
'
60
'
,
'
i,
80
,
,
I,"l
,
100
,,I
120
,,,t,,1
140
,t
....
160
I ....
180
t ....
200
I ....
220
I ....
240
I ....
260
t ....
280
t ....
3O0 32e
El: 124000) 123(94) 95(71) 152(60) 41(60)81(42) 57(32) 68(30) 107(27) 108(25) 56(25)
Figure lO. Search for 2-ethoxy-3-ethylpyrazine using different range errors.
509
lOO
80
-
121
_
60-
HCY'""
40-
o
16s69
41,
}1
,.,,,,
....
[~9 ~71~[~"~
~
-
i
-,,
93
1
,I,
108
09 122 IJ11'345
,,
........
l l6l
i ....
i'"',~,
,,~nA
....
222
, ....
, ....
, ....
i ....
, ....
0
20
40
60
SO 1O0 120 140 160 180 200 220 240 260 ZSO 300
El: 81(100) 121(83) 93(45) 108(45) 41(44) 69(29) 39(10) 43(18) 53(7) 55(20) 57(10) 67(9) 68(8) 91(13) 107(22) 109(11)
122(10) 135(5) 136(6) 161(6) 179(3) 191(2) 204(4) 222(15)
Figure11.
Mass
spectra
of
gleenol.
t20
510
~
o
o
9
i--..<
....i
Z3
r
._d
r
r
1.13
r
IZI
Z
O9
I--<
9
,,e'
z~
'T"--
e,i
~
!1
511
For the fragment at m/z 41 whose intensity greatly changes according to
experimental conditions and apparatus 80 -100% error range is necessary (or it
must even be squarely eliminated). The choice of this range of error is left to the
skilled user!
- Misr
There are other options which have not been described" "SHOW ALL
COMPOUNDS" and "LAST ENTRIES (See Figure 2). The first switch allows all
the components present in the bank to be filed off using the runners set on the right
side. To select a compound more quickly, it is necessary to use the switches (i.e.
alphabetical letters) set below the list. Compounds beginning with the selected
letter can be visualized. The switch "LAST ENTRIES" allows the compounds to be
sorted according to the recording date. This option is particularly suitable for
finding the compounds last recorded. These compounds appear first in the list
allowing them to be found easily again.
The "MAIN MENU" (See Figure 1) besides "COMPONENTS MANAGING"
contains the "OTHER FUN(~TIQN$" option. The latter concerns K.OVATS INDICES
CALCULATIO..N which allows from scans (X) and known Kovats indices (Y) of
compounds identified in the GC/MS listing on the left-hand part of the screen to be
calculated (See Figure 12).
In the middle part, the program gives the linear equation 9
Y= ax=b
(Y=KI and X=scans)
as well as the correlation coefficient (r). It ranges usually between 0.98 - 0.99 for a
limited number of Kovats indices unities range (~200). The equation is not linear
along the scan listing. The right-hand part allows all Kovats indices from the
corresponding scans to be calculated. This version is an enlargement of our
previous MBASIC SCAN1 program (See the previous chapter).
Applications
The earlier versions of the SPECMA bank have been tested with more than
several hundreds of volatile components of essential oils (27-39), spices, herbs
and flavorings (40-53) Maillard reactions (54-60) and related model systems,
fruits (61-65), wines and alcoholic beverages (66, 67). The reader is referred to
these references for more details. Further on, several examples using the
SPECMA 2000 data bank and belonging to the previously cited works are
reported. They will be completed subsequently.
Acknowledgements-
The authors are indebted to the Central Analytical Service of the CNRS,
Vernaison and the Mass Spectrometry Service, Marseilles (Mrs. C. Chariot) for
recording GC/MS analyses. Thanks are also due to Mrs. G.M.F. Vernin and
Mrs.R.M. Zamkotsian for their collaboration.
512
100
i
-EI
93
80
_
60
79
40
gO
20
l43
0
....
0
El: 9 3 ( 1 0 0 )
~ ....
20
79(60)
i
40
121(60)
121
)1
3
,
105
107
,,
60
77(41)
,,,
, I,
80
100
80(41)
,, t ....
'13o,,
120
140
160
t ....
91(41) 41(34)
67(32)
105(23)
I ....
180
I ....
200
I ....
220
53(20) 63(20) 55(18)
I ....
240
43(16)
I ....
I ....
I ....
260
280
300
107(16)
136(2)
320
513
Chemical name:
1,3,3-TRIME TRICYCLO-2,2,1- HEPTANE
Origin & ref:
ARTEMISIA HERBA ALBA,(ALGERIA),VERNIN et aI.,"FOOD FLAVORS",ELSEVIER,37A, 147-205,1995.
Molecular formal C10 H16 (136)
IKA:
925
FEMA: 0
IKP: 1038
COE 0
N ° 723
R.N (CAS): 508 32 7
DIK 113
IOFI 2
CA: 0:0
Family: Monoterpenes Hydrocarbons (C10H16; MW 136)
Descriptor odor/Flavor
100
'
-EI
40
121
....
i ....
20
El:
93(100)
136(27)
~ I~_, 1~7l
, ....
, ....
40
60
121(26)
92(23)
9_4 ]11°¢~7_
136
i,,
,~ . . . .
,,,,
80
100
120
140
I ....
160
I ....
91(22)
41(20)
79(15)
39(14)
77(13)
I ....
I ....
180
200
105(10)
94(9)
I ....
I ....
220
55(6)
240
67(6)
Chemical name:
BICYCLO[3.1.0] HEX-2-ENE,2-METHYL-5-(I-METHYLETHYL)
CELERY SEEDS E.O.;VERNIN et aI.,"SPICES,HERBS...",ELSEVIER,34,329-345,1994.
Molecular formal C10 H16 (136)
IKP: 0
COE 0
R.N (CAS): 2867 05 2
DIK 105
IOFI 2
,,I
....
280
I ....
300
320
107(6)
Origin & ref:
IKA:
930
FEMA: 0
I,
260
N ° 4871
CA:
Family: Monoterpenes Hydrocarbons (C10H16; MW 136)
Descriptor odor/Flavor
1oo
80
60
40
77
20
27 _ 41
]79
,
,, i 165
,,,
,i
20
40
60
80
100
91(48)
92(33)
77(38)
. . . . . . . . . . . . . .
El:
93(100)
136
i
41(17)
27(15)
l ln~
....
,
i .'-01 , I . . . .
120
136(12)
140
121(3)
I ....
160
105(5)
I ....
180
79(20)
I ....
200
65(6)
I,,,,I
220
....
240
I ....
260
I ....
280
I ....
300
320
514
1~)
-EI
/
80
173
60
0
411
20
!
o
20
40
60
80
1oo
El: 44(100) 73(~) 42(5O) 3O(3O)2O(25) ~(10)
12o
14o
16o
18o
200
220
240
260
280
300
320
515
100
43
80
60
0
44
40
,~
,~~
2I
20
40
~6 1
114
I'o
60
. . . .
80
100
120
140
',t ....
160
I ....
180
'1 . . . .
I ....
! ....
I ....
t ....
!
200
220
240
260
280
300
El: 4 3 ( 1 0 0 ) 7 3 ( 9 3 ) 7 2 ( 7 2 ) 4 4 ( 4 8 ) 5 5 ( 1 9 ) 4 2 ( 1 8 ) 4 1 ( 2 5 ) 6 0 ( 2 6 ) 3 9 ( 1 6 ) 8 6 ( 2 3 ) 1 1 4 ( 1 4 ) 129(5) 7 0 ( 1 0 )
320
516
100
80
60
43
40
2O
163
o
0
zo
40
6o
so
El: 104(100) 43(54) 91(28) 72(27) 163(20)
100
120
140
~60
180
zoo
zzo
z40
260
zso
300
320
517
100
44
-El
0
29
80
27
:1
56
60
43
40
7
20
57
14:5I[~
0
20
40
60
i72
80
82
100
120
140
160
180
200
220
240
260
El: 44(100) 27(82) 29(90) 39(31) 41(75) 43(52) 45(17) 53(5) 55(16) 56(73) 57(42) 67(10) 71(5) 72(18) 82(13)
280
300
320
518
i
100
43
80
0
C8H17~,,~
144
H
60
6
40
82
I12
20
lO
.... i ............
20
40
i,,,l'i
60
80
,,I,
100
, I , , ! 2 ~ , ' ? , ~
120
140
_
. . . .
160
,
. . . .
180
,
. . . .
200
.
. . . .
220
.
. . . . .
240
, , , , ~ ,
260
280
,
. . . .
300
El: 57(100) 43(88) 44(70) 70(59) 71(56) 68(56) 55(53) 82(49) 56(45) 112(34) 67(33)81(32) 41(31) 83(29) 95(26) 45(24)
96(22) 42(21 ) 110(15) 128(6) 138(5)
320
519
100
nl
-El
108
4O
95
2O
o
........
0
, 1~9 ,
20
40
"'17:!7 ll"['
. . . . . .
60
|,
109
.
.
.
.
I . . . . . . .
80
100
i1~9
120
I ....
140
I ....
160
t ....
180
I ....
200
t ....
220
t ....
t ....
I ....
t ....
240
260
280
300
El: 1 0 8 ( 1 0 0 ) 6 7 ( 3 7 ) 4 1 ( 3 5 ) 9 5 ( 2 9 ) 1 0 9 ( 1 9 ) 8 2 ( 1 7 ) 9 3 ( 1 7 ) 3 9 ( 1 6 ) 4 3 ( 1 6 ) 5 5 ( 1 5 ) 8 1 ( 1 4 ) 5 7 ( 1 1 ) 2 9 ( 8 ) 1 1 9 ( 6 )
320
520
I0080
IEl_
60_
40-
-
4
20-
51
132
77
1781]~021103
~19
11104
,t
0
El:
20
40
H
60
80
100
120
....
140
I ....
160
I ....
I ....
I ....
I ....
t ....
I ....
I,~,',
180
200
220
240
260
280
300
131(100)132(63)51(46) 77(56) 78(38) 103(39) 104(15) 102(8)39(12)43(6)
320
521
11111
i
i||1
-El
80-
40-
0
0
20
40
60
80
El: 83(100) 55(34) 41(11) 39(10) 69(4)
100
120
140
160
180
200
220
240
260
280
300
320
522
loo
0
80
60
43
164
121
0
2o
4o
60
so
]oo
]2o
]40
]60
El: 107(100) 164(45) 43(45) 77(18) 121(15) 94(12) 91(10) 149(6) 65(6)
]~
2oo
22o
240
260
z~
3oo
32o
523
lOO
-EI
0
60
40
[43
103
45
1
27
20
1
40
121
92
'11 t
79
60
80
100
120
140
160
180
200
220
240
260
280
300
El: 91(100) 103(45) 121(44) 27(13) 39(18) 41(9) 43(40) 45(28) 51(15) 63(9) 65(23) 77(18) 79(9) 92(28) 120(17) 146(24)
320
524
100 -EI
0
8O
6O
69
_
123
40
20
0
192
8 l
II
I
0
''- .... i , , j , , ,
20 40 60
~.51107~ 21 f
7I
109.1
.
.
.
'
.
.
.
80
"!
,
,
, !
1149
.
.
.
.
.
i
_
.
.
.
.
100 120 140 160
I,,,',1
180
I
....
I ....
I ....
I ....
I ....
200
220
240
26,0
280
I ....
300 320
177(100) 41(33) 43(33) 51(4) 53(8) 55(17) 67(10) 69(65) 77(8) 79(15) 81(42) 91(15) 93(17) 95(21) 107(27) 109(10)
121(25) 123(56) 135(21) 149(38) 192(40)
El:
525
lib
58
0
\ (CH2~'~
60
40
20
0
0
20
40
60
80
100
120
14o
16o
18o
El: 43(100) 58(88) 59(50) 41(28) 55(18) 71(17) 57(13) 29(12) 85(5) 96(4) 226(2)
200
220
240
260
280
300
320
526
i
100
108
OH
80
60
109
43
152
0
20
40
60
80
100
120
140
160
180
El: 108(100) 41(15) 43(20) 55(14) 69(7) 77(6) 91(7) 152(7) 207(2) 109(22)
200
220
240
260
280
300
320
527
100
-El
41
_
80"
~
O
H
_
60_
40 -
67
0
El:
....
i ....
, ....
0
20
40
41(100)
55(35)
67(42)
, ....
60
82(22)
i ....
80
69(15)
I' l u t t
100
31(19)
, I , , ,', t T", , , I . . . .
120
29(15)
140
27(17)
160
42(22)
I ....
180
43(10)
! ....
200
39(20)
100(1)
t ....
220
t ....
240
i ....
I ....
I ....
260
280
300
320
528
100
80
85
l
139
0
20
40
60
80
100
120
140
160
180
200
220
240
El: 43(100) 59(45) 85(27) 41(24) 81(20) 139(12) 67(6) 55(15) 96(12) 31(9) 95(11)97(8) 69(7) 121(6)
260
280
300
320
529
100
-El'
_
67
4O
20
55
!
43
1
96
80
100
20
40
,i~,~
60
120
.11t9
140
. .
160
180
200
220
240
El: 95(100) 67(49) 41(29) 55(26) 81(15) 53(14) 43(12) 79(12) 96(12) 77(8) 139(5) 121(3) 111(1) 154(1)
260
280
300
320
530
-EI
100
61
80
~
S
~
O
H
60
31 4 ~4~
9
40
20
0
I ....
0
El:
106(100)
....
i,,
20
4O
60
I ....
80
I, ,,,I
100
....
120
I ....
140
I ....
160
I ....
180
I ....
200
61(82) 58(70) 57(69) 41(50) 49(50) 47(47) 31(45) 48(38) 45(36) 55(25)
I ....
220
I ....
240
I ....
I ....
I ....
260
280
300
320
531
100
-EI
4
73
80
~ ~ ' , , . ~ S ~ s / S
ll3
~
60
40
15
20
I ........
....
0
20
El: 7 3 ( 1 0 0 ) 4 1 ( 9 8 )
40
I ....
i ....
60
80
113(80)45(40)39(47)
l'~
i ....
100
178(10)
I ....
120
I ....
140
180(2)103(15)
i ....
160
.~$6.!
180
....
200
I ....
220
I ....
240
I ....
! ....
I ....
260
280
300
320
532
100
80"
60-
40-
20108
150
o - I.
0
20
40
60
80
1O0
120
140
160
180
200
220
El: 43(100) 27(22) 39(15) 41(31) 44(6) 59(3) 45(3)66(3) 69(3) 72(6) 108(17) 109(3) 150(11)
240
260
280
300
320
533
100
111
71
40
"~
..
27
56
[83
20
....
0
0
i ........
i,
20
60
40
,I
80
.....
100
, I, , 1 2 6 1
120
....
140
I ....
160
I ....
180
I ....
200
I
....
220
El: 111(100) 43(86) 55(80) 71(53) 27(45) 67(45) 56(36) 41(29) 99(30) 83(24) 126(13) 39(25)
I ....
240
I ....
i ....
i ....
260
280
300
320
534
100
-Ez
85
'
60
08HI
0
0
40
.
0
20
.
.
40
.
60
.
.
80
.
.
100
~
120
140
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
160
180
El: ~(100) 27(12) =~39) 41(18) ~(14) ~(~4) ~(~5) 57(~6) ~(12) ~(6) ~
200
220
240
260
~(4) ~(6) 128(9) 136(3)
280
300
320
535
100
-EI
~o~
6o.~
40
20
142155 I~
-
IIII41 I,,;
....
0
El
99(100)
~ ....
20
~ ....
40
114
i ....
,,rg4
60
80
....
100
i ....
120
| I, L'-a-. , i . . . .
140
160
| ....
180
42(44) 55(44) 71(44) 70(44) 41(28) 27(31) 29(27) 68(11) 69(11)
i ....
200
114(14)
i ....
220
! ....
! ....
I ....
I ....
240
260
280
300
84(6) 142(3)
320
536
100
-El
107
_
8O
611,
2
40 "~
108
167
"
141'
20 -
| .....
-
0
El:
9
20
....
40
91
53
165
I ....
60
109
I
;
100
120
, . . . . . . . . . . . .
80
1
i
....
140
1151
i
....
160
1182
I ....
180
I ....
200
I
....
220
I ....
240
I
....
260
I
....
280
107(100) 108(44) 167(39) 41(35) 91(23) 108(22) 39(20) 53(17) 135(15) 43(13) 77(11) 79(11) 151(8) 182(7) 65(6)
I
....
300
320
537
IO0
131
80"
o~
60-
40-
20-
176
0
0
20
40
zu..t
60
80
100
z..to
120
140
160
180
El: 131(100)43(7)51(18) 77(15) 103(28) 147(10) 132(8) ~76(~2) ~ ( 6 ) ~04(6)
200
220
240
260
280
300
320
538
100
81
123
80-
H~
......
60-
41
40-
/
55 67
I07 22
20-
i ....
0
0
El: 8 1 ( 1 0 0 )
2O
123(75)
....
40
, ........
60
i ....
80
100
i ....
120
107(30)41(45)55(35)59(25)67(38)69(25)
lli"M~
, ....
140
|''
160
79(26)
1182
l~r/| ....
i ....
I ....
t ....
180
200
220
240
122(28)
139(5)
140(4)
152(19)
I ....
i ....
I ....
260
280
300
153(15)
167(1)
182(7)
320
539
i
100
iin
80
0
I
40
20 -
55
_
92
00
20
40
60
80
100
120
140
160
El: 91(100) 83(33) 82(3) 55(22) 65(7) 77(3) 79(3) 92(10) 190(2) 172(3)
180
200
220
240
260
280
300
320
540
100
80
C6H5CH:CHCH200C
(Z)
~
H
60
40
20
. 143
....
i .... I ....
20
40
_
I ....
60
i ......
. . . 197
I',,
80
100
' 1~ ' ,~ , ,',:?r3
120
140
.
.
.
I.
.
160
El: 83(100) 82(3) 55(70) 43(12) 97(6) 115(25) 117(21) 133(3) 216(5)
.
.
.
t
.
.
180
.
.
.
I
.
.
200
.
.
.
. .
I~1~
220
,
. . . .
240
,
. . . .
260
,
. . . .
280
-
.
.
300
.
.
,320
541
100
'55
183
gl
80
41
0
'
'82
123
I
i
40
138
11o
20
,
.
0
20
40
60
80
109
1
.
.
100
.
.
.
.
.
120
....
140
I ....
160
I,,
180
I.~I.
....
200
I ....
220
I'.
240
,,,
....
t ....
I ....
I
260
280
300
83(100) 55(97) 81(89) 95(78) 82(74) 41(73) 69(64) 123(62) 67(49) 138(43) 68(33) 109(29) 96(18) 43(14) 56(13) 53(12)
57(12) 101(12) 80(8) 94(7) 191(2)
El:
320
542
100
~
-EI
80
60
40
I ....
0
El:
oc 89
91
151
0
0
105(100)
i ....
20
212(27)
i ....
40
51(10)
65
i ....
1194
~ ....
i ....
i ....
i ....
60
80
100
120
140
65(10)
77(24)
91(45)
194(7)
165(2)
i,',,,5~i
160
....
180
'; . . . .
200
i ....
220
;,,,:;
240
....
260
i,,,
- .....
280
300
320
543
100
-EI
~(CH2)14-'~0~
0
80
6O
li
40
101
20
....
0
, ....
20
, . . . . . . . . . . . .
40
60
80
= ....
100
; ....
120
~,A,J,
140
~'z
160
; ....
180
I ....
200
i
....
220
, ....
240
260
,,
280
,z~4;
....
300
El: 88(100) 101(50) 43(51) 41(46) 55(33) 57(23) 60(13) 61(10) 69(14) 70(12) 73(13) 89(12) 143(5) 157(5) 239(4) 284(1)
320
544
I00
-El"
164
OH
CHa
6O
4O
lll,,, 19,1
104
,o.
....
0
El:
164(100)
i ....
20
149(40)
- ....
40
103(35)
, ....
60
77(35)
I ....
80
I ....
100
133(35)91(30)
12
I ....
120
137
i ....
140
I,
160
131(30)55(27)41(25)
,,,I
....
180
I ....
200
104(20)
I ....
220
121(20)
I ....
240
137(20)
I ....
I",',
260
280
, , I ....
300
320
545
100
109
OH
I
124
81
40
20
....
i .....
20
El:
....
40
163 I ~9
i ........
60
80
I ....
100
....
120
]25
I ....
140
t ....
160
i ....
I ....
I ....
I ....
I ....
~ ....
180
200
220
240
260
280
109(100) 124(77) 81(60) 27(24) 39(26) 53(19) 63(7) 77(12) 79(10) 107(22) 108(21) 125(7)
|
....
300
~20
546
100
135
150 ....
_
OH
80"
_
6091
40
77
20
51
79
-
[ 171
[
....
0
; ....
20
. ....
40
El: 1 , 5 0 ( 1 0 0 ) 1 3 5 ( 9 9 ) 9 1 ( , 5 6 )
!11
; ....
, ....
60
80
; ....
100
, ....
120
77(40) 107(32) 115(29)39(28)
,33
; ....
140
-
....
160
~, , , , ~ . . . .
180
200
79(23)51(21)41(20)65(18)
; ,...-,. , , 9 . . . .
220
240
; ....
; , ,,,,,,, , ; . . . .
260
280
105(17) 117(13) 121(13)
300
320
13,3(12)63(11)
547
100
m
...
-El
80"
=,
.
17g
Of,,
91
107
03
6O
1351147 163
I
0
. . . .
I
. . . .
20
I
,''r
40
,
,
,
. . . .
60
,
. . . .
80
,
. . . .
100
,
. . . .
120
,
....
140
,'
160
180
200
El: 178(100)91(81) 107(72) 103(64) 147(40) 163(40) 72(40) 79(35) 105(28) 135(24)
220
240
260
280
300
320
548
100
119
55
40
105
1
20
....
El:
I
l
~ ....
20
i ....
40
, ....
60
,,,
80
,I
204
. . . . . . .
100
120
~I
....
140
I ....
160
I..I.I,2~
180
" ....
200
I ....
220
I ....
240
I ....
260
i ....
280
119(100)93(65) 105(48) 133(10) 55(48)67(8) 69(30) 77(26) 79(30) 91(36) 134(14) 161(10) 189(3) 204(30)
. ....
300
320
549
II
11111
H"---
-El
H
40
OH
:
,2,,
121
20
0
....
0
I ....
20
i
40
,~,
~ ....
, ....
60
80
,
~,,
100
120
137
,', I
140
11164
, , ,~ . . . .
160
I,,~,,6,:5',
180
200
,zv],;222,
220
"
240
i
~ ....
-
260
280
300
320
El: 4 3 ( 1 0 0 ) 9 5 ( 5 3 ) 1 2 1 ( 3 0 ) 4 1 ( 2 0 ) 5 5 ( 1 7 ) 7 1 ( 1 5 ) 7 9 ( 1 7 ) 8 1 ( 1 6 ) 1 0 5 ( 1 7 ) 1 0 9 ( 1 8 ) 137(7) 1 6 1 ( 1 5 ) 1 6 4 ( 1 0 ) 189(2) 2044115) 2 2 2 ( 3 )
205(2)
550
IO0 -El
8O
H
97 /107
4O
I3119
20
....
i ....
20
95 lO 1191351
67
1169
....
40
161
11211133
~ ....
~ ....
60
80
i ....
100
~ ....
120
i ....
140
1 204
162
; ....
160
189
i ....
180
; ....
200
i !222, ; . . . .
i ....
; ....
i ....
220
260
280
300
240
El: 59(90) 41(100) 31(25) 39(27) 43(85) 53(12) 55(35) 67(21) 69(10) 77(22) 79(30) 81(30) 91(41) 93(50) 95(31) 105(35)
107(47) 109(17) 119(25) 133(13) 135(26) 121(12) 161(43) 162(14) 189(15) 204(20) 222(3)
320
551
iii
100
-El
I1
H
40
20
5
]31
Iti9
[!~~7
]81~s
!149
119 Itl~ , 1 1 6 1 1 2 0 4 1 8 9
t'
0
20
40
60
80
1O0 120 140 160 180 200 220 240 260 280 300 320
El: 59(100) 43(67) 41(50) 31(17) 55(26) 67(11) 69(19) 81(16) 91(20) 93(20) 95(17) 105(15) 107(20) 109(18) 119(12) 135(12)
133(9) 149(30) 161(20) 189(14) 204(18) 222(4)
552
100
$9
-EI
8O
6O
4O
2O
,,,[3,1 ,, , I! 5,
............
20
40
60
80
I tt, .4i. . .
100
120
140
w , ....r l
160
I,
180
, l l-~,---~. Inn
. . . J I I,~ -'-'-'-' ~ ....
200
220
240
~ ....
260
280
300
59(100) 41(47) 43(50) 53(10) 31(15) 55(20) 67(15) 69(10) 77(11) 79(21) 81(22) 82(16) 91(20) 93(21) 95(18) 105(18)
107(20) 108(23) 109(23) 119(11) 121(14) 123(11) 135(5) 149(14) 161(3) 164(14) 189(3) 204(3) 222(2)
El:
320
553
100
-EI
204(
80
161
60
43
40
59
5
20
7
0
i
i
20
40
1
[
81 91
l~l
5
,
60
i
133
80
1~ l
135
,,,|
100
163
205
,',,
120
140
|
160
180
"
|
200
~ j~
220
.
240
.
.
260
.
280
300
320
El: 189(100) 204(78) 161(75) 41(73) 59(52) 133(51) 43(54) 81(48) 55(45) 91(45) 123(40) 79(39) 105(37) 93(35) 69(30) 67(30)
107(31) 147(23) 149(21) 77(20) 53(15) 57(15) 135(18) 95(15) 222(5) 205(10) 153(20)
554
100
143
80
40
0
....
0
El:
- ....
20
, ....
40
I
93
5
; . . . . . . . . . . . . . . .
60
80
43(100)81(84)41(60)55(21)69(20)93(17)80(16)
119(8)
123
19
100
120
1161
I ....
140
123(16)
| ....
160
71(14)
I ....
180
105(14)
I ....
200
I ....
220
79(13)95(11)107(11)
, ....
240
, ....
, ....
-
260
280
300
109(11)53(10)
161(8)
320
555
100
-El
H
,,,~
,,
8O
41
f43
40
119
li1
'~
20
....
i ....
20
i
40
,,
I . . . .
60
80
I,
100
,,,
. . . . . .
120
140
t ....
160
I ....
180
i ....
200
t ....
220
I ....
240
t ....
t ....
I,,',,
260
280
300
320
El: 6 9 ( 1 0 0 ) 4 1 ( 5 4 ) 4 3 ( 4 5 ) 1 1 9 ( 3 7 ) 1 0 9 ( 2 9 ) 5 5 ( 2 0 ) 6 7 ( 1 6 ) 7 9 ( 1 6 ) 1 2 1 ( 1 6 ) 8 1 ( 1 2 ) 1 0 7 ( 1 2 ) 1 2 3 ( 1 2 ) 1 3 7 ( 1 2 ) 1 3 9 ( 1 2 ) 2 0 4 ( 1 2 ) 1 6 1 ( 8 )
556
100
-El
91
80
6198, ,05
60
55
40
zo
79
I
.-
7
9
II2
I3
I
'7
7
147
133
i
20
. . . .
. . . .
40
i
60
.
.
.
.
.
.
80
.
.
.
.
.
.
100
!
. . . .
120
i
.
140
272
161 175 187
~
. . . .
257
229
.
.
.
59
.
.
.
160
.
3
!
. . . .
180
-,,,~
i ....
i
200
i
,
.
220
240
.
.
.
.
258
.
260
.
i
.
.
.
|
.
.
280
.
.
.
.
300
320
El: 39(14) 41(100) 43(26) 53(17) 55(52) 57(17) 69(65) 77(28) 67(38) 79(56) 80(16) 81(65) 82(24) 83(21) 91(83) 92(31)
93(50) 94(24)95(57)96(10)97(14)105(65)106(32) 107(32) 108(17) 109(33) 117(13) 119(31) 120(12)121(15)123(34) 125(43)
131(14) 133(20) 134(11) 137(12) 145(12) 147(34) 148(13) 159(12)161(18) 173(11) 175(18) 187(20)213(26)229(34)257(41)25~
272(38)
557
100
il
109
80
60
0
53
40
20
'
0
20
'
I
.
.
40
El: 110(100) 109(87) ~ ( ~ )
.
.
.
60
.
.
.
.
.
80
I,
100
,
120
39(13) 51(11) 41(10) 81(8)
140
160
180
200
220
240
260
280
300
320
558
1oo
-EI
8O
6O
0
39
40
=
I10
1143
2O
0
20
40
60
80
100
120
140
160
El: ~ ( 1 ~ ) 110(,~) 39(5O)43(4O)67(3O) 68(2O) ~(15) 1 ~1(10) 82(4)
180
200
220
240
260
280
300
320
559
100
-El
126
60
0
29
40 -
20
....
0
El:
97(100)
20
126(61)
40
69(30)
, ....
,[ Q1
, . . . . . . . .
60
80
29(48)
39(65)
100
41(87)
| ....
120
51(17)
! ....
140
53(17)
| ....
160
81(4)
i ....
180
109(9)
I ....
200
I ....
220
I ....
240
! ....
i ....
; ....
260
280
300
i
320
560
100
-El
HO
80
60
0
128
40
85
20
0
0
20
40
60
80
100
El: 43(100) 57(69) 128(61) 85(39) 55(16) 39(10)
120
140
160
180
200
220
240
260
280
300
320
561
,=
11111
,,
-El
III
43
60"
55
411"
7O
20"-
[lO
I12
0
20
40
60
80
100
120
140
160
180
200
220
El: 111(100) ~(96) 42(63) 41(19) ~ ( ~ ) ~(41) 69(11) 7O(26) 82(~ 1) 98(7) ~0(19) 112~
240
260
280
300
320
562
100
J-.s
N
80-
60-
40-
200
o
20
40
6o
so
100
El: 127(100) 71(77) 41(14) 86(73) 85(27) 59(27)
120
140
160
180
200
220
240
260
:zso
300
320
563
100
-El
"
2
72
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
71
144
I
i
40
45
TM
"/9
0
20
40
60
80
10
100
I
120
140
160
180
200
El: 72(100) 71(73) 144(52)45(45) 111(35)39(25) 73(12)97(12) 103(11)69(8) 79(8)
220
240
260
280
300
320
564
100
55
i
8060-
I 60
59
40-
180
200
0
20
40
60
80
100
120
140
160
180
200
220
El: 55(100) 43(78) 45(62) 60(60) 59(47) 180(46) 88(25) 115(25) 64(24) 92(23) 116(20) 182(5)
2~
2~
2~
300
320
565
100
I-EI
44
60
_
59
40-
-
....
0
El:
17
3
0
44(100)
i ....
9 i
i . . . . . . . . . . . .
20
40
60
80
59(48)
60(56)
70(30)
71(26)
I
1
100
69(11)
,,,|
!
....
120
103(15)
! ....
140
163(26)
li,,,I
....
160
180
42(22)
43(11)
t ....
200
t ....
220
t ....
240
i ....
,
-,,,,
260
280
300
320
566
S---S
S----S
60
40
184
124
t
200
65
0
20
40
60
92
80
100
186
120
140
160
El: 59(100) 124(30) 184(37) 119(11) 45(15) 60(44) 65(7) 92(7) 186(7)
180
200
220
Z40
260
ZSO
300
320
567
m
-El
104}
i
]24
i
80"
I
60-
40-
20-
151
0
o
2e
40
6o
so
100
120
140
160
180
200
220
El: 124(100) 151(20) 166(7) 95(10) 94(16) 125(8) 165(4) 81(11) 53(12) 123(10) 125(8) 54(10)
240
26o
z80
300
32e
568
II
100
I!
80
60
40
20
149
1177
0
20
40
60
80
100
120
140
160
El: 136(100) 121(19) 149(15) 177(10) 137(10)41(6)53(6) 135(5) 192(1)
180
200
220
240
260
280
300
320
569
Chemiegl name:
2-(5'-HYDROXYMETHYL-2'-FORMYLPYRROL-I'-YL)-3-PHENYLPROPIONIC ACID LACTONE
Origin & ref:
GLUC.-PHENYLALANINE MODEL SYSTEM;VERNIN et aI.,INSTR.ANAL. OF FOODS,ELSEVIER,1982,97.
Molecular formul C15 HI3 03 N1 (255)
IKA:
0
FEMA: 0
IKP: 0
COE 0
ILN (CAS): 60026 28 0
DIK 0
IOFI 2
N ~ 4829
CA:
Family: Heterocycles Pyrroles
Descriptor odor/Flavor TOBACCO,CHOCOLATE-LIKE / IDEM
100
-EI
91
80
t~~CHO
4O
0
39
20
0
~ . . . . . . . .
0
20
40
1
~ ....
60
I ....
80
100
| ....
120
1255
t ....
140
,,I
160
t ....
180
I,'~
200
,,,
....
220
~ ....
240
r ....
~ ....
260
280
El: 91(100)255(22)39(9)51(4)65(13)77(4)78(4)108(13)120(9)131(9)136(4)147(4)148(4)164(4)210(4)211(4)
256(4)
300
320
570
4.
REFERENCES
1.
Petitjean M, Vernin G, MetzgerJ. In: Charalambous G, Inglett G, eds.
Instrumental Analysis of Food: Recent Progress, Academic Press, New York:
1983: Vol. 1: 97-123.
Petitjean M, Mass spectra and Kovats indices data bank of flavoring
heterocyclic compounds (in French), Thesis Sciences, University of AixMarseilles III ,1982.
Vernin G, Petitjean M. In: Vernin G, ed, The Chemistry of Heterocyclic
Flavouring and Aroma Compounds., Ellis Horwood Publ., Chichester,
England: 1982; 305-342.
Vernin G, Petitjean M, Metzger J. Parf. Cosm. Ar6mes 1983: 51: 43-51.
Vernin G, Petitjean M, Poite JC, Metzger J, Fraisse D, Suon KN. In:
Vernin G, Chanon M, eds. Computer Aids to Chemistry, Ellis Horwood
Publ., Chichester, England: 1986; 294-333.
Vernin G. Le point sur la banque de donn~es SPECMA, flaveurs et
fragrances, Parf. Cosm. ArSmes 1985; 66" 27.
Vernin G, Petitjean M,, Poite JC, Metzger J. In: Sandra P, Bicchi C, eds.
Capillary Gas Chromatography in Essential Oil Analysis, Huethig Verlag Publ.
Heidelberg: 1987; 287-328,
Vernin G, Boniface C, Metzger J. Analusis 1987:15 : 564-568.
Adams R. In: Identification of Essential Oil Components by GC/ITMS,
Academic Press, New York: 1992; Allured Publ. Co, 1995.
McLafferty FW, Stauffer DB. Mass Spectrometry Library Search System
Bench Top PBM version 3.0, Palisade 10, Newfield, New York: USA, 1993;
idem ; The Wiley/NBS.
Hennberg D, Wilmann B, Joppek W. MPI Library of Mass Spectral Data,
Max-Planck Institut fur Kohlenforschung, M01heim: Ruhr, Germany: 1994.
Ten Noever de Brauw MC, Baumann J, Gransberg LM, La Vos MGF. In:
Compilation of Mass Spectra of Volatile Compounds in Food, 1988-1996;
Vol. 1-16; TNO Nutrition and Food Research, P.O. Box 300, 3700 A.J. Zeist:
The Netherlands.
National Institute of Standards and Technology, PC Version of the
NIST/EPA/NIH Mass Spectral Data Base, Version 4.5, US Dept. of
Commerce,Gaithersburg, USA: 1994.
Colon F. SPECMA 2000 data bank, Thesis Sciences, University of AixMarseilles III (to be presented) 1997.
Allured's Flavor and Fragrance Materials. Worldwide reference list of Materials
used in compounding flavors and fragrances, Allured Publ. Co.362 Schmale
Road, Carol Stream, IL 601 188-2787 USA: 1996.
Arctander S. Perfumer and Flavor Materials, Montclair, NJ: 1969.
Furia TE, Bellanca N. eds. Fenaroli's Handbook of Flavor Ingredients, 2nd ed.
CRC Press, Cleveland, Ohio, USA: 1975.
Van Gemert CJ, Nettenbreijer AH. eds. National Institute for Water Supply,
Central Institute for Nutrition and Food Research, TNO, A.J. Zeist, The
Netherlands: 1977.
.
,
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
571
19. S. Fors, In: Waller GR, Feather MS, eds. The Maillard Reaction in Foods and
Nutrition, ACS Symposium Series, 1983: 215; 185-286.
20. Bauer K, Garbe D. Common Fragrance and Flavor Materials, VCH-VerlagsGesellschaft, Weinheim, Germany: 1985.
21. Takahashi A, Akiyama H. In: Charalambous G, ed. The Shelf-life Studies of
Foods and Beverages, Elsevier, Amsterdam: 1993: 33: 1003-1032.
22. Calkin RR, Jellinek JSt. eds. Perfumery, Practice and Principles, Wiley J and
Sons, New York, Chichester: 1994.
23. IFF International Flavors and Fragrances Inc. Perfumes compendium 4th ed.
World Headquarters, 521 West 57th Street, New York 10019: 1995.
24. Aldrich, Flavors and Fragrances, International Edition, 1101 West St-Paul
Avenue, Milwaukee, Wisconsin, 53233, USA: 1996.
25. Schnabel KO, Beliz HD, Von Ranson C. Z. Lebensm. Unters. Forsch 1988:
187:215-223.
26. Chastrette M, Cretin D, El A'(di Chafei. J. Chem. Inf. Comput. Sci 1996: 36:
108-113 and references cited therein.
27. Vernin G, Metzger J, Fraisse D, Scharff C. Parf. Cosm. Ar6mes 1983: 52"
51-61.
28. Fraisse D, Scharff C, Vernin G, Metzger J. IXth Int. Congres Essential Oils,
Singapore: 1983. Essential Oils Technical Paper, 1983: Book 3: 100-120.
29. Vernin G, Chakib S. Geranium essential oils from Morocco (Unpublished
results ).
30. Vernin G, Metzger J, Fraisse D, Suon KN, Scharff C. Perfumer Flavorist 1984:
9; 71-86.
31. Vernin G, Boniface C, Metzger J, Ghiglione C, Hammoud A, Suon KN, Fraisse
D, Parkanyi C. Phytochemistry 1988: 27: 1061-1064.
32. Vernin G, Metzger J, Suon KN, Fraisse D, Ghiglione C, Hammoud A, Parkanyi
C. Lebensm. Wiss u. Technol. 1990: 23: 25-33.
33. Vernin G, Faure R, Pieribattesti J.C.J. Essent. Oil Research 1990: 2(4): 211214.
34. Vernin G, Metzger J, Mondon JP, Pieribattesti JC. J. Essent. Oil Research
1991 : 3:197-207.
35. Vernin G. J. Essent. Oil Research 1991:3: 49-53.
36. Vernin G, Merad O, Vernin GMF, Zamkotsian RM, Parkanyi C. In:
Charalambous G. ed. Food Flavors: Generation, Analysis and Process
Influence, Elsevier, Amsterdam: 1995: 37A: 147-205.
37. Vernin G, Merad O. J. Essent. Oils Research 1994; 6: 437-448.
38. Vernin G, Colon F. Roman Camomile (Anthemis Nobilis) essential oil (to be
published).
39. Vernin G. Poite JC. Everlasting (Hefichrysum Stoechas) essential oil (to be
published).
40. Vernin G, Metzger J. Perfumer and Flavorist 1986; 11 : 79-84.
41. Vernin G, Metzger J, Fraisse D, Scharff C. Planta Medica 1986; 96-101.
42. Randriamiharisoa R, Gaydou EM, Bianchini JP, Vernin G. Sciences des
Aliments 1986: 6:211-231.
572
43. Vernin E, Pujol L, Vernin G, Vernin GMF, Metzger J. Parf. Cosm. Ar6mes 1989:
89:81-94.
44. Vernin C, Vernin G, Vernin GMF, Metzger J, Pujol L. Parf. Cosm. Ar6mes 1990:
93: 85-90.
45. Vernin G, Metzger J. In: Linskens HF, Jackson JF. eds, Modern Methods of
Plant Analysis, Essential Oils and Waxes, Springer Verlag, Berlin, Heidelberg:
1991: Vol 6: 6; 99-130.
46. Vernin G, Metzger J, Azzario P, Barone R, Arbelot M, Chanon M. In:
Charalambous G. ed. Recent Developments in Food Science and Human
Nutrition Elsevier SC. Publ. 1992; 75-97.
47. Vernin G, Petitjean M, Metzger J. Tarragon essential oil (See Ref. 7).
48. Vernin G, Parkanyi C. In: Charalambous ed. Spices, Herbs and Edible Fungi,
Elsevier, Amsterdam: 1994; 34: 329-346.
49. Vernin G, Vernin C, Metzger J, Pujol L, Parkanyi C. In: Charalambous ed.
Spices, Herbs and Edible Fungi, Elsevier, Amsterdam: 1994: 34:411-426.
50. Vernin G, Metzger J, Parkanyi C. In: Charalambous ed. Spices, Herbs and
Edible Fungi, Elsevier, Amsterdam: 1994: 34: 457-468.
51. Vernin G, Gighlione C, Parkanyi, C. In: Charalambous ed. Spices, Herbs and
Edible Fungi, Elsevier, Amsterdam: 1994: 34: 483-500.
52. Vernin G., Vernin E, Metzger J, Pujol L, Parkanyi C. In: Spices, Herbs and
Edible Fungi, Elsevier, Amsterdam: 1994: 34: 501-578.
53. Vernin G, Parkanyi C. In: Spices, Herbs and Edible Fungi, Elsevier,
Amsterdam: 1994; 34: 579-594.
54. Vernin G. ed. The Chemistry of Heterocyclic Flavouring and Aroma
Compounds, Ellis Horwood Publ. Chichester, England: 1982.
55. Vernin G, Boniface C, Metzger J, Obretenov T. Kantasubrata J, Siouffi A,
Larice JL, Fraisse D. Bull. Soc. Chim. France 1987: 4: 681-694.
56. Vernin G, Metzger J, Obretenov T, Suon KN and Fraisse D. In: Lawrence BM,
Mookherjee BD, Willis BJ. eds. Flavors and Fragrances: a World Perspective,
Elsevier, Amsterdam: 1988; 999-1028.
57. Debrauwer L, Vernin G, Metzger J, Siouffi AM, Larice JL. Bull. Soc. Chim.
France 1991 : 128: 244-254.
58. Vernin G., Metzger J, Boniface C, Murello MH, Siouffi AM, Larice JL, Parkanyi
C. Carbohydrate Research 1992: 230: 15-29.
59. Vernin G, Debrauwer L, Vernin GMF, Zamkotsian RM. In: Charalambous G. ed.
Off Flavours in Foods and Beverages, Elsevier Sci. publ. 1992: 28: 567-623.
60. Vernin G. Metzger J, Sultan AM, EI-Shaffei A.K, Parkanyi C. In: ACS
Symposium Series, Molecular Approaches to the Study of Food Quality, Amer.
Chem Soc. Washington DC: 1993; 36-55.
61. Vernin G. Metzger J, Suon KN, Fraisse D. Parf. Cosm. Ar6mes 1985: 62: 6974.
62. Vernin G, Bouin D, Metzger J, Fraisse G, Scharff C. In: Charalambous G. ed.
The Shelf-life of Foods and Beverages, Proceedings of the 4th Int. Flavor
Conference, Elsevier, Amsterdam: 1986: 12: 255-284.
573
63. Vernin G, Vernin E, Vernin C, Metzger J, Soliman A. Flavour and Fragrance J.
1991: 6: 143-148.
64. Vernin G, Vernin GMF, Metzger J, Roque C, Pieribattesti JC. J. Essent. Oil
Research 1991 : 3: 49-53.
65. Vernin G, Vernin C, Roque C, Pieribattesti JC. GC/MS analysis of volatile
components of Psidium Cattleianum Sabine fruit from Reunion island,
submitted for publication.
66. Vernin G., Boniface C, Metzger J, Fraisse D, Doang D, Alamercery S. In:
Charalambous ed. Frontiers of Flavors, Elsevier, Amsterdam: 1988: 17: 655685.
67. Vernin G, PascaI-Mousselard H, Metzger J, Parkanyi C. In : Charalambous ed.
Shelf-life Studies of Foods and Beverages, Elsevier, Amsterdam: 1993: 33:
975-990.
This Page Intentionally Left Blank
D. Wetzel and G. Charalambous (Editors)
Instrumental Methods in Food and Beverage Analysis
9 1998 Elsevier Science B.V. All rights reserved
575
CAPILLARY ELECTROPHORESIS FOR FOOD ANALYSIS
Custy F. Femandes and George J. Flick, Jr.
Department of Food Science and Technology, Virginia Tech, Blacksburg, VA 24061-0418
1. INTRODUCTION
1.1 Basic Elements
The capillary electrophoresis methodology combines column chromatographic and gel electrophoretic
techniques. Fig. la, represents the basic elements of a capillary electrophoreUc system in its simplest form.
The separation is performed in a capillary filled with a carrier electrolyte and loaded with analyte. Following
application of an electrical energy at both ends of the capillary, the analyte migrates and is detected by the
detector (fig. lb).
Fig. la. Basic elements of a capillary electrophoretic system.
576
1.1.1 Capillary
The industrial production of a narrow internal diameter (d) (ca. 10-100 pm), polyimide coated (ca. 15
pm thick t') fused silica capillary resulted in rapid developments and extensive applications of the capillary
electrophoretic technology. The fine diameter capillary is essential to dissipate Joule heating. The polyimide
coating reduces the fragility of the fused silica capillary. The capillary is characterized by the effective (L,) and
total ~ lengths (fig. la). The I_, is the distance between the injector and detector, while l-t is the entire length
of the capillary across which the electrical voltage is applied.
For on-capillary detection (e.g., optical
spectroscopy), the L, is generally less than It. For off-column detection (e.g., mass spectrometry) the I-t and
L~ are equal.
1.1.2 Electrolytic Vessels
Submerged in each electrolytic vessel (reservoir) is the capillary filled with a carder electrolyte. The
vessels are characterized as anodic (injector-end) and cathodic (detector-end) electrolytic vessels. The
Fig. lb. Operating capillary electrophoretic system during spearation with injector end anodic vessel
(left) and detector end cathodic vessel (right).
577
electrolytic vessels and the capillary contain an identical electrolyte, it is essential to fill the electrolytic vessels
to same level. Thus reducing the imbalance due to hydrostatic flow.
1.1.3 Electrical Source
An electrode immerses in each electrolytic vessel. The electrical energy for separation is supplied
through the electrodes. A direct current power regulator with an output voltage (ca. 5-50 kV) and current (ca.
0.5-1.0 rnA) ranges is generally used. The applied electrical field strength (E in V/m) is the ratio of the applied
voltage (V in V) to total length of capillary. The electrical field strength facilitates comparison between
capillaries of different length.
E = V/I.~
[V/m]
(1)
1.1.4 Sample Application
A sample is loaded by exchanging the anodic (injector-end) electrolytic vessel with the sample vessel.
Generally, minuscule quantity (ca. 5-10 qL) of the sample is loaded at the anode (injector-end) either,
electrokinetically or hydrodynamically. The electrolytic vessel replaces the sample vessel and separaUon is
initiated within the capillary following application of electrical energy.
1.1.5 Detector
The detector detects the analyte as it migrates to the cathode. An analyte can be detected at
femtomole concentration (one fM is 10"15M). An optical detector is frequently used.
1.2 Basic Electrophoretic Principle
Suppose a sample (e.g., two cationic analytes) is subjected to capillary electrophoretic separation. The
two cationic analytes are loaded at the anode and an electrical voltage (V) is applied across the fused silica
capillary of total length (I-t). Both analytes will migrate from the anode to the cathode, through the detector.
The analyte migration time (t=) is the interval between the electrophoretic initiation and detection. The analyte
migration distance is the effective length (L,) for on-capillary detection. The electrophoretic velocity (V,~) is
expressed as:
578
V,, =l_Jt,
[m/s]
(2)
The electrophoretic mobility (p.,) rather than the electrophoretic velocity is preferred, because it
facilitates comparison between electropherograms. The relaUonship between electrophoreUc mobility and
velocity is given by:
p,~ = Vo,/E = L,/Et,
[m2Ns]
(3)
The electrical (F,) and frictional (F=)forces are given by the eq. (4) and (5). respectively. The observed
mobility is determined by equating the electrical force on the ionic analyte with the frictional drag through the
medium:
F. = q=E
Ff = 3nq=,d=V,,
[CV/m]
(4)
[Kgm/sZJ
(5)
[m2Ns]
(6)
from eqs. 4 and 5
q,E = 3nq=,d,V,~,
or
p,~ = q,/3nq=,d=V,~
Where rl=, is viscosity of carrier electrolyte
d= is stokes diameter of ionic analyte
q= is analyte charge
[Poise] or [kg/ms]
[m]
[Coulomb] or [C]
The observed mobility (Pob=) of the analyte is constant in a defined electrolytic environment and
facilitates comparison between electropherograms. The observed mobility is the summaUon of electrophoretic
(p~ and electroosmotic (P,of) mobilitJes.
579
(7)
or
(8)
Pob, -- I_JEt, + q,/3nn.d,V.p
1.2.1 Electroosmotic Flow
The wall of the fused silica capillary is caUonic at most pH conditions. There is a built-up of anionic
countedons in the solu'don adjacent to the capillary wall. Under the influence of the applied electrical field, the
anions are drawn toward the cathode, resul'dng in the bulk flow of electrolyte toward the cathode. The bulk flow
of carrier electrolyte under the influence of an electric field is termed "electroosmoUc flow" (EOF). The
magnitude of the EOF depends on the consb'u~on of the capillary and the electrolytic environment. A capillary
constructed with a nonionic material (e.g. Teflon) exhibits slow EOF (ca. pH 3-5), whereas an ionic material
(e.g., fused silica) shows a fast EOF (ca. pH 6-8) (see Fig. 2). The direction of the EOF can be modulated by
altering the internal charge (e.g., ionic or nonionic) of the capillary wall.
During coelectroosmotic
electrophoresis (Po~ = P~ + P~), the electroosmoUc and electrophoretJc mobil~es migrate the analyte in the
same direction. While, in counterelectroosmotic electrophoresis (Pob, = I~p - ~ ), the electroosmotic and
electrophoretic mobilitJes migrate the analyte in the opposite direction.
1.2.1.1 Zeta potential
The zeta potential (Z) results due to the electrical double layer. The electrical double layer develops
between the internal charge on the capillary wall and the charged carder electrolyte. The internal surface of
a fused silica capillary is anionic due to the ionization and/or adsorption of cations. The ionization (major
contdbuUon) of the silanol groups (SiO) and adsorption (minor contribution) of cations is responsible for the
anionic surface. The ca~ons in the carder electrolyte buildup and equilibrate the anionic surface. The electrical
double layer generates a potential difference near the capillary wall and is termed zeta potential.
Under the influence of an electrical field, the cations contribu~ng to the double layer migrate to the
cathode. The solvated cations drag the electrolytic fluid in the fused silica capillary toward the cathode. The
relationship between electroosmotic velocity and zeta potential is expressed by the Smoluchowske equation
580
v ~ = E(~_/nJ
(9)
or
p,~ = ~Z/q=,
where
(10)
~ is dielectric constant of cartier electrolyte
The zeta potential is influenced by capillary characteristics (surface charge) and carder electrolyte
composition (ionic concentration and strength) but independent of applied electrical field. The zeta potential
is primarily due to the charge on the internal wall of the capillary. The pH, of the carrier electrolyte, modulates
the charge on the surface as well as the charge and mobility of the analyte. For fused silica capillary the eof
increases with pH. This is because the silanol groups ( -SiO "1) and analytes are deprotonated at high pH and
vice-versa at low pH (fig. 2) (Lukacs and Jorgenson, 1985) [K. D. Lukacs and J. W. Jorgenson, 1985. J. High
Res. Chrom pg. 407-411]. The zeta potential is an inverse function of the ionic strength of the cartier
electrolyte. Increasing ionic strength reduces the electrical double-layer and lowers zeta potential.
1.3.1 Flow Profile in Capillary
Fluids flow due to the pressure difference between the ends of a capillary tube and a parabolic or
laminar (NR~Reynolds number < 2000) fluid flow profile develops due to the shear forces at the capillary wall.
However, a fiat flow profile is observed in a narrow bore fused silica capillary during electrophoretic separation
(fig. 3), a uniquely desirable feature. Additionally, there is no pressure drop (or hydrostatic imbalance) across
the capillary. Therefore, the driving force for the carrier electrolytic fluid is the EOF. The flat flow is nearly
uniform throughout the capillary and limits analyte zone dispersions. However, a thin quiescent electrolytic
layer extends a short distance into the bulk of the electrolytic fluid. This arises due to frictional forces against
the bulk fluid flow at the wall. The fluid velocity reduces rapidly at the wail. The frictional forces in the thin
quiescent electrolytic layer are small compared to the other dispersive forces that contribute to the overall
separation process. Additionally, the fluid velocity and flow profile are generally independent for a narrow bore
capillary (d= < 10-100 pm). In a wide bore fused silica capillary ICd 9200-300 pm), the fluid flow profile is
affected.
581
f
\
PEoFXl0~(cm2Ns)
5-
Pyrex
Silica
Teflon
2
t
3
t
4
t
5
pH
'
I
6
,
t
7
I
8
Fig. 2. Electroosmotic flow (EOF) as a function of pH for different capillaries.
Fig. 3. Illustration of flat flow profile during electrophoretic separation with narrow bore fused silica
capillary.
582
1.3.2 Flow direction in Capillary
Generally, under the influence of EOF all analytes migrate in the same direction and independent of
their charge. However, some exceptions are encountered. For fused silica capillary, the eof is from the anode
to the cathode. An anionic analyte migrates toward the cathode, as its electrophoretic mobility is smaller than
the electroosmolJc mobility. A mixture of three analytes (anionic, cationic and neutral) can be electrophoresed
in a single run. All analytes will migrate toward the cathode (fig. 4). The anionic analyte migrates the slowest
(Po=, = P,~ - IJ.p), because the electrophoretic mobility directs it to the anode but the electroosmotic mobility
carries it to the cathode. The neutral analyte migrates at EOF (Po=,= P.~). The cationic analyte migrates the
fastest (IJo=,= P,of + Pop), because the electrophoretic and electroosmotic mobilities are in the same direction.
If the magnitude of the electrophoretic mobility exceeds the electroosmotic mobility, then the anionic and
cationic analytes, migrate to their respectwe electrodes. Further, alteration of capillary wall charge can reduce
or eliminate EOF without influencing the electrophoretic mobility of the analyte. In such situations, cationic and
anionic analytes migrate to the cathode and anode, respectively. This has been observed for small ions (e.g.,
Na § K *1, C1-1, F-l).
1.4 Factors influencing EOF
In a fused silica capillary, the EOF increases with pH (fig. 5). At rapid EOF (e.g., high pH), analytes are
eluted without resolution. On the other hand, at slow EOF (e.g., low pH) the anionic fused silica capillary wall
will adsorb calJonic analytes through coulombic interactions. Under such conditions, separation or desorption
of analytes can be obtained through alteration of capillary characteristics (e.g., surface), carrier electrolyte
composition (e.g., viscosity) and/or operational parameters (e.g., electrical field strength). The EOF is directly
related to electrical field strength. Therefore EOF can be reduced by lowering the electrical field strength.
However, low field strength reduces efficiency as well as resolution and increases analyte retention time.
Pragmatically, the best approach is to reconstitute the carrier electrolyte pH. The zeta potential is altered by
carder electrolyte compos~on (e.g., ionic strength and concentration). Adjusting the pH can affect the charge
and mobility of the analyte. Low pH buffers protonate the capillary surface and the analyte, while the contrary
is true of high pH buffers. The pH of the carrier electrolyte can be selected based on the isoelectric point of
analyte (e.g., protein). Higher buffer concentration and ionic strength have been used to affect EOF. At high
583
Fig. 5. Increased resolution of analytes with low pH gradient.
584
buffer concentration, the coulombic interactions between the analyte and the capillary wall decrease the net
surface charge. A high ionic strength generates high current and additional Joule heating, thus limiting the
modulation of the EOF rate. While, low ionic strength results in adsorption of the analyte. The addition of
organic modifiers (eg. methanol, acetonitdle, trifluroacetic acid) to carder electrolyte alters the zeta potential
and viscosity. Temperature (an operational parameter) affects the viscosity of the carder electrolyte. Joules
heating lowers the viscosity of the carrier electrolyte.
1.5 Dispersion
The analyte zone spreading or broadening is termed as dispersion. Although, many factors influence
dispersion, it is principally due to the longitudinal diffusion along the capillary axis and little or no effect due to
radial diffusion. Other factors contributing to dispersion include the adsorption of analyte on the internal wall
of capillary, sample injection volume, Joule heating, anticonvective properties of capillary, and differences in
analyte mobilities within that zone. Dispersion should be controlled because it increases zone length as well
as the difference necessary for resolution. For a Gaussian peak the baseline peak width (w=) is:
wb = 4 0
[m]
where o standard deviation of peak
(11)
[m]
The efficiency is expressed as the number of theorelJcal plates (N). For on-capillary detection it is given
by:
N = (l_Jo)2
(12)
During capillary electrophoresis, there is molecular diffusion (D=), which leads to peak dispersion and
it is expressed as:
o == 2O=t== 2O,(L,l_~p,,)
where
D= is analyte diffusion coefficient
(13)
[m2/s]
585
The dispersion (o~, is affected inversely by electric field strength and directly by diffusion coefficient.
Generally, the diffusion coefficients are inversely related to molecular weights. Thus, macromolecules (e.g.,
polypepUdes, polynucleoUdes, polysaccharides) will disperse at a lesser extent than micromolecules (e.g.,
amino acids, nucleotides, monosaccharides). Therefore, macromolecules will generate the highest number
of theoretical plates. When an analyte is subjected to a high electric field, it will migrate rapidly through the
fused silica capillary leaving little or no time for dispersion.
1.5.1 Electrodispersion
Electrodispersion is due to difference in sample and carrier electrolyte conductivities. Electrodispersion
results in, skewed peak shapes, analyte focusing (low conductivity), defocusing (high conductivity), and
temporary isotachophoresis.
When the conductivities are equivalent electrodispersion does not occur.
However, when the conductivities are mismatched electrodispersion occurs. If the analyte zone has a higher
mobility than the running buffer, the leading edge of the analyte zone will be diffused and the trailing edge
sharp. Conversely, when the analyte zone has a lower mobility than the running buffer, the leading edge will
be sharp and the trailing edge diffused.
When the analyte zone has a higher mobility (e.g., higher conductivity, lower resistance) than the carder
electrolyte, the front edge of the analyte zone encounters a higher voltage drop as it enters the buffer zone.
This causes the diffusing analyte to accelerate away from the analyte zone and results in zone fronting. As the
analyte at the trailing edge diffuse into the carder electrolyte they also encounter an increase in voltage drop,
but in the same direction of migration, and accelerate back into the analyte zone, keeping the trailing edge
sharp. Electrodispersion has no effect on neutral analytes.
Although electrodispersion always occur, it is smaller than other dispersive effects (e.g., diffusion).
Dispersion is particularly evident in a sample containing analytes with a wide ranges in mobilities.
Electrodispersion can be overcomed by matching the conductivities of the sample and the carder electrolyte.
1.6 Joule heating
Joule heating (capillary temperature elevation) is a consequence of the resistance of the carrier
electrolyte to the current flow. Traditionally, the progress in applications of capillary electrophoresis was
586
impeded due to use of wide bore diameter capillary. The development of narrow-bore diameter capillary
accelerated the development of capillary electrophoretic technique.
The extent of Joule heating developed is the difference between electrical energy input and thermal
energy dissipated. If all the electrical energy is dissipated then there is no Joule heating effect (no rise in
temperature). This is possible at very low electrical field strength. Normally, high electrical field strength is
applied across the capillary. Under such conditions the electrical energy (ca. 0.5-5.0 W/m) is converted to
thermal energy (ca. 10 K temperature rise). The quantity of heat dissipated is determined by capillary
dimensions and buffer conductivity.
The rate of heat generation in a capillary can be approximated as:
dQ/dt = IV/Al..t = K=o (V/I~)2 = K. (E)2
[w/m3s]
(14)
since I = V/R. & R=. = I _ ~ , A
where
R=, is the resistance of the carrier electrolyte
A is cross sectional area of capillary
[ohms]
[rn ~]
Two major problems associated with Joule heating are a radial temperature gradient and temperature
changes with Ume due to ineffective heat dissipation. Other associated problems include viscosity changes,
analyte mobility, analyte zone dispersion, analyte migrational time, and disproportionate increase in EOF.
The Joule heating is problemaUc because it gives rise to radial temperature gradient. The temperature
along the capillary axis is higher than the capillary wall temperature. The temperature gradient results in
viscosity gradient. Since temperature change (ca. I~
affects the viscosity (ca. 2-3%) and mobility (ca. 2-3%),
temperature control is critical.
Joule heating can be influenced by capillary design. The temperature difference depends on the
thermal properties (K=p,,,~, K~y,,~,,
hc=pa,ary) and dimensions (d= and do as well as polyimide coating thickness
1') of polyimide coated fused silica capillary. Although the polyimide coating is ca. 15 pm thick, its low thermal
587
conductivity significantly limits heat transfer. A capillary with a narrow d~and large do dissipates heat readily.
The high ratio of inner surface area-to-volume facilitates Joule heat dissipation. A large do is beneficial due to
a reduction in the insulalJng properties of the polyimide and improves the surrounding heat transfer. A dramatic
decrease in temperature difference can also be realized by reducing the capillary d,. A wide range of capillary
dj are used. Problems encountered with small capillaries (ca. 10-25 IJm (;I) include clogging, detection and
sample loading. Medium size capillary (ca. 25-50 pm d~ are practical. Clogging problems are significantly
reduced by filtering carrier electrolyte through small pore size filters (ca. 0.2 um).
Theore~cally, Joule heating can be controlled by adjusting the operational parameter (e.g., electrical
field strength) and carder electrolyte formulations (e.g., ionic strength). Temperature gradient is directly
proportional to applied electrical energy. Therefore, a reduction in temperature gradient can be obtained by
performing capillary electrophoresis at lower field and ionic strength. These measures are useful but limit
capillary efficiency and performance.
Most of the capillary electrophoretic separations are performed at high electrical field strength and
Joules heating is often encountered. The amount of Joule heating that can be dissipated with design and
optimization of operational parameters and electrolyte concentration is limited. Hence, Joule heating is
dissipated by additional instrumentation. Regulating the capillary temperature enables heat dissipation and
holding the capillary temperature constant.
1.7 Sample LengthNolume
A small sample length/volume is desirable. An increase in sample plug length increases diffusion and
reduces resolution. The sample volume/length depends on the diffusion coefficient (i.e., analyte molecular
weight) and migrational time. The diffusion coefficient of macromolecules (e.g., proteins) is smaller than
micromolecules (e.g., amino acids).
The capillary dimension is another important factor in sample
length/volume. Generally, sample injection length is ca. 2% of the L~. For a narrow bore fused silica capillary
ca. 1.0 m Lt and 40 pm ql this corresponds to 20 mm sample length and 5 pL sample volume. Although
modem sample delivery systems can reproducibly deliver such small quantities, analyte detection limits often
necessitate longer injection lengths.
1.8 Analyte Adsorption
588
One would expect that the capillary electrophoretic separation of macroanalytes (e.g., proteins) would
yield high efficiencies due to their low diffusion coefficients. However, analyte adsorption on the fused silica
capillary internal wall decreases the performance. A decrease in performance results in peak tailing or total
analyte adsorption. The analyte adsorption on the fused silica capillary walls is primarily due to strong ionic
forces (e.g., cationic analyte and anionic capillary wall), and weak hydrophobic forces. Generally, protein
adsorption on the capillary wall is due to coulombic (ionic) and van der Waal (hydrophobic) forces. While a
large surface area-to-volume ratio of the fused silica capillary is preferred to dissipate Joule heating, a small
ratio is required to minimize adsorption.
Analyte adsorption on the capillary wall surface can be reduced by altering the carrier electrolyte
composition. This measure reduces the coulombic forces. A high ionic strength of carrier electrolyte reduces
the surface charge and analyte adsorption. The reduction in surface charge lowers the zeta potential and the
EOF, resulting in an increase in analyte migration time. The adverse effect of high ionic strength, is Joule
healing (due to increased current). An alternative strategy is to use zwitter ionic carrier electrolytes (e.g., TRIS,
MES, CHAPSO). Electrophoretic separation at pH extremes effectively reduces coulombic forces. At low pH
(ca. < 3) the electrical double layer is reduced as the protonated silanol groups (of fused silica capillary) are
non-ionic. This reduces the EOF and the cationic proteins will migrate slowly towards the cathode. An adverse
effect of this approach is protein denaturation due to low pH as well as precipitation of plant and milk proteins.
At high pH (ca. > 9), both the analyte and capillary wall are anionic (due to deprotonation) and the analyte
adsorption is limited by coulombic repulsion.
1.9 Capillary Wall Modification
The intemal wall of the capillary is altered to reduce adsorption of analyte. Two basic approaches are:
a) permanent modification by covalently bonded or physically adhered phases; and b) dynamic deactivaUon
through incorporation of carder electrolyte additives (e.g., hydrophilic polymers or detergents). Both methods
eliminate or reverse the surface ionic charge, vary hydrophobic forces and limit nonspecific adsorption. The
two methods are beneficial, but neither method has a disUnct advantage.
1.9.1 Permanent coating
589
Permanent coating (e.g., covalent bonded or physically adhered phases) can be accomplished with
various functional groups. Silylation followed by permanent coating with a suitable funclJonai group is the most
widely used approach. The EOF is eliminated or reversed by permanent coating. Neutral functional groups
(e.g., polyacn/lamide or polyethylene glycol) eliminates EOF. This results from both decreased surface charge
and increased viscosity at the wall. Cationic functional groups reverses the eof. Amphotedc functional groups
(e.g., proteins, amino acids) yield reversible EOF depending on the pl of the coating and carder electrolyte pH.
The permanent coating of choice is the one that requires little or no maintenance and is stable during
regeneration and hydrodynamic flow. Unfortunately, the stability of most permanent coatings is limited. The
siloxane bond (Si-O-Si) has limited stability (ca. pH 4-7) and hydrolysis limits long term stability.
1.9.2 Dynamic coating
Stability problems encountered in permanent coating are overcome with dynamic coating. Dynamic
coating has been achieved with the addition of modifiers to the carder electrolyte. An advantage of dynamic
coating is stability. A dynamic modifier repeatedly regenerates the coating and dynamic stability is attained.
The dynamic modifier coats the wall and alters coulombic and hydrophobic forces. Dynamic modifiers are
easily incorporated by dissolution in the carder electrolyte and their concentration can be readily optimized.
Dynamic coatings with cationic surfactants (e.g., CTAB) reverse the EOF. There are some drawbacks due to
dynamic coating. These include pH extremes, solutes as well as capillary surface are affected (e.g., SDS
denatures proteins), long regeneration time needed to obtain a reproducible coating and constant EOF.
1.10 Resolution
The resolution is of paramount importance in separational science. The maximum efficiency in
capillary electrophoresis is achieved during coelectroosmosis. Only then is the mobility contribution to the
separation efficiency maximized. However, the optimization of capillary electrophoretic separations is made
more complicated by a divergence of optimal conditions for separation efficiency (N) and resolution (R). The
resolution efficiency of two analyte zones is:
R = 0.25 (N)'~(~v/v)
(15)
590
where
Av/v is the relative migration velocity of the two analyte zones (denoted by subscripts I and 2)
~v/v = (po., - po.9/po=.v...
= (p.,, - p . , ~ ) / ( p , ~ . . . , .
+ p.~)
(~6)
Po==l= P~I + P~
(17)
Po~ = P..= + IJ.of
(18)
From eq. (16) p~ and P.o, inversely affect the magnitude of relative velocity for any given peak pair.
A decrease in relative velocity reduces the resolution (R). Fortunately, this is only valid when the inherent
electrophoretic and electroosmotic mobilities are the only contributions to the po..
N=(l/o) 89
(19)
o==2D=t==2D=LL~/p.,V
(20)
t== L/Vo.=L/p.,E= LI.t/p.pV
Eqs. 19 and 20 indicate that increasing the voltage is a limited means to improving resolution. To
improve the resolution by two folds, the voltage must be quadrupled. However, Joule heating is oRen the
limitation to increasing voltage. The resoluion can be increased by increasing ~p,~,.
2. MODES OF CAPILLARY ELECTROPHORESIS
Capillary electrophoresis comprises a group of techniques that have different operative and separative
characteristics.
The most widely used capillary electrophoretic techniques include capillary zone
electrophoresis, capillary gel electrophoresis, capillary isoelectric focusing, capillary isotachophoresis and
micellar electrokinetic capillary chromatography.
2.1 Capillary Zone Electrophoresis
The Capillary zone (also referred as free-solution or open tube) electrophoresis (CZE) is the most
extensively used among the capillary electrophoretic separational techniques. In CZE separation occurs
591
because analytes migrate in discrete zones and at migrational velocities based on differences in the charge-tomass ratio. It is essential to have carder electrolyte homogeneity and constant field strength throughout the
length of the capillary. Following sample (either electrokinetic or hydrodynamic) injection and application of
voltage across the capillary, the analytes in the sample separate into discrete zones (fig. 6). Both, macro and
micro analytes with ionic and nonionic characteristics can be separated. Cations had the shortest migration
time and are eluted first, followed by nonions and anions. Nonionic analytes are not electrophoresed but are
coeluted with the EOF. For zwitter ionic analytes (e.g., amino acids, peptides), the net charge is influenced by
the pH of the carrier electrolyte. They exhibit charge reversal at their pl as well as change in the direction of
the electrophoretic mobility.
2.1.1 Selectivity and the use of Additives
Various additives have been added to the carrier electrolyte to change the selectivity of separation. The
additives modulate the electrophoretic mobility. The carrier electrolyte (ionic strength and concentration)
composition, is varied to affect selectivity. Electrolytic additives include surfactants and chiral selectors.
Organic modifiers (e.g., methanol, acetonitrile, trifluroacetic acid) are excellent solvents for hydrophobic
proteins as they modify the electroosmotic mobility. Inorganic salts (e.g., CsCI, LiCi) are responsible for
conformationai changes in proteins.
2.1.2.1 Selection of Carrier Electrolyte
Many carrier electrolytes have been used in CZE. The electrophoretic mobility is affected by pH
changes. Therefore, carrier electrolytes must have good buffering capacity (@ 50-100 mM), low mobility (i.e.,
low charge-to-mass ratio) to reduce Joule heating and unabsorb at the detection wavelength. Effective carrier
electrolytes have a range of approximately two pH units centered on the pK= value. Certain carder electrolytes
(e.g., citrate, phosphate, succinate) have more than one useful pK= and can be used in more than one pH
range (Table 1). Zwitter ionic carrier electrolytes (e.g., bicine, tricine, CAPS, MES) are used routinely for
separating proteins and peptides. The major merit of deploying zwitter ionic carder electrolyte is their low
conductivity (ca. pH =pl), thus reduces Joule heating. The presence of salts in carrier electrolyte increase the
conductivity and reduce the electroosmotic mobility through annihilation of the charged double layer.
2.1.2.2 Hydrogen Ion Concentration
592
Fig. 6. Separation of analytes into discrete zones after application of voltage across the capillary.
593
Alterations in pH are particularly useful for zwitter ionic analytes (e.g., peptides and proteins). Carder
electrolyte pH above and below the pl value will change the net charge of the zwitter ionic analyte and cause
the analyte to migrate either before or after the EOF. Below its pl the analyte is cationic and migrates toward
the cathode, ahead of the EOF. Above the pl the opposite phenomenon occurs. Due to the high chemical
stability ofthe fused silica capillary, the accessible pH range is 2-12, but it is usually limited by the pH stability
of the analyte (e.g., denaturation of milk and plant proteins at its pl).
In addition to effecting the analyte charge, changing the pH will also cause a concomitant change in
EOF. This may require reoptimization of resolving conditions. For instance, adequate resolution may be
obtained at a low pH, but when increased to alter analyte charge, the EOF may be too high so that analyte
elutes before resolution is achieved. Under such conditions either increase the effective length (L,) of the
capillary or reduce the electrophoretic mobility.
2.2. Capillary Gel Electrophoresis
Capillary gel electrophoresis (CGE) (or molecular sieve) is a hybrid of traditional slab-gel and freesolution capillary electrophoretic techniques. It is used to separate charged macroanalytes (e.g., proteins,
nucleic acids) based on molecular size, through a suitable hydrophilic polymer that acts as a molecular sieve.
In CGE separation is mainly due to electrophoretic mobility and without any contribution from electroosmotic
mobility.
The capillary dimensions and gel concentration affect separation efficiency and migration time. Small
(ca. 25 pm), medium (ca. 50-100 IJm) and large (ca. 200 pm)internal diameter capillaries have been used.
The medium size capillary inner diameter balances between minimizing thermal gradients across the capillary
and maximizing the detection path length. CGE has been performed in short (@ 0.07 m), medium (ca. 0.15-0.4
m) and long (ca. 1.0 m) effective length capillaries. Separation efficiency is improved by increasing capillary
length, although migration time increases and higher field strength is required to achieve a given separation.
As the gel concentration increases, the separation efficiency is improved.
Ionic macro- and micro- analytes will migrate through the gel filled capillary when an electric field is
applied. Macroanalytes are subjected to more impediments or frictional drag during their migration through
the polymeric network, resulting in slower migration velocity than faster migrational velocity of microanalytes
594
that migrate relatively unhindered.
Covalently linked polymers (e.g., acrylamide/ N,N'-methylene-bis-
acrylamide filled capillaries) are usually employed. Hydrogen-bond linked polymers (e.g., agarose) are
unstable to Joule heating produced at high voltages used in CGE.
Macromolecules (e.g., DNA, oligonucleotide, SDS-denaturated protein, peptide) are separated by capillary gel
electrophoresis, since their charge-to-mass ratio is unaffected by molecular size. In case of DNA, each
additional nucleotide adds an equivalent unit of mass and charge and does influence the mobility in free
solution. Proteins adsorb a fixed quantity of SDS (ca. 1.4 g/g of protein). SDS is an anionic surfactant and
therefore all proteins become negatively charged and migrate toward the anode.
A major merit of the CGE is efficient dissipation of Joule heating. This allows use of high electrical field
strength without the detrimental consequences of Joules heating as resolution is unaffected.
Maintaining
separation range is particularly a severe problem as all the molecules must migrate the same effective/total
length of the capillary to reach the detector.
The gel is a cross-linked network, swollen with a fluid component or sol. The cross-links must be
uniformly dispersed as well as similar size, and fluid content generally dictates the size and size distribution of
the pores in the gel. The porosity of the gel influences the size separation of the analytes. Generally, the gel
concentration is inversely proportional to the size of the analyte being separated. During CGE the analytes
separate into bands with the smallest analyte migrating fastest (fig. 7).
Two classes of gels employed are physical and chemical gels. The porosity of the physical gel is due
to the entanglement of polymers and hydrogen-bonding (e.g., starch, agarose). They are quite rugged to
environmental changes. The porosity of chemical gel is due to cross-links produced by covalent, ionic, and van
der Waals interactions (e.g., acrylamide/N,N'-methylene-bis-acrylamide). These gels are less rugged, and
it is difficult to change the carrier electrolyte once the gels are formed.
Cross-linked polyacrylamide, a widely used matrix, is usually polymerized in situ and not removed from
the capillary. PreparalJon of these gels requires extreme care. Rapid polymerization, use of solutions without
degassing, or impure chemicals often lead to bubble formation or unstable gels. A potential disadvantage of
gel filled polyacnllamide is its rigid nature. This results in gel extrusion during hydrodynamic sample injections.
Therefore, gel filled capillaries are loaded electrokinetically. Several separations can be done in gel filled
595
Fig. 7. Differences in electrophoretic separation based on size of the analyte.
596
capillaries if they are handled properly. Soiled samples, clogged capillary ends, or bubble formation during use
reduces the performance of gel filled capillaries.
Linear polymers are an option to the crosslinked
polymers. They are essentially polymer soluUons and have increased flexibility. The linear polymer solutions
are polymerized in situ, but it is not necessary. Pre-polymerized polymer can be dissolved in carrier electrolyte
and hydrodynamically loaded into the capillary.
Selectivity can be modulated by the addition of chiral selectors, ion-paring reagents (e.g., TFA), or other
complexing agents (e.g., ethidium bromide for DNA, SDS for denatured proteins). These species can be
covalently bonded to the gel or added into the carder electrolyte.
2.3 Capillary Isoelectric Focusing
Capillary isoelectric focusing (CLEF) separates polyamphyolytes or zwitter ionic analytes based on their
pl in a stable pH gradienL An ampholyte is a zwitter ionic molecule that contains both positively and negatively
charged groups such as amino and carboxylic (e.g., peptJdes, proteins). At the pl, the net charge on the
ampholyte is zero and migration ceases under the influence of an electric field. This process is termed
isoelectdc focusing. ClEF is run in a pH gradient where the anodic and cathodic reservoirs are at low and high
pH, respectively (fig. 8). The carder electrolytes generate the pH gradient. A stable pH gradient (ca. 3-9 pH)
is developed within the capillary using a wide variety of amphyolytes. A pH gradient is established by filling the
capillary with a mixture of amphyolytes dissolved in the carder electrolyte. Before the application of an electrical
field, the pH throughout the capillary is constant and is an average from all the amphyolytes in the carder
electrolyte solution. A high electrical field strength is applied (ca. 5-50 V/m). When the electric field is applied
across the capillary, the amphyolytes migrate, their buffeting capacity enables establishment of a pH gradient.
The resulting cathodic and anodic electrolyte reservoirs are alkaline and acidic, respectively. For isoelectric
focusing, the cathodic reservoir is filled with alkali (e.g., NaOH) and the anodic reservoir is filled with an acid
(e.g., H3PO4). The pH of the anodic electrolyte reservoir must be less than the pl of the most anionic ampholyte
to prevent migration into the analyte. Similarly, the cathodic electrolyte must have a higher pH than the pl of
the most cationic ampholyte.
Polyamphyolytic sample (e.g., peptides, proteins) is loaded along with amphyloytes for separation.
They migrate to the positions corresponding to their pl. When the peptides and proteins in the capillary are
597
focused (i.e. steady-state) the current ceases to flow. The peptides and proteins are separated in very sharp
zones. As the focusing progresses, the current drops (ca. 1.0 pA). Overfocusing results in precipitation due
to protein aggregation at high localized concentrations. Any band broadening caused by thermal diffusion is
quickly reduced by the existing pH gradient. If a molecule diffuses (i.e., defocusing or band broadening) into
an adjoining pH zone it acquires a net charge and realigns (i.e., focuses or band sharpening) into its pl zones.
Preparative quanti'des of analytes can be separated using the CLEF, as the sample is loaded into the capillary
during filling with an amphoteric solution. Certain proteins (e.g., plant and milk) are precipitate at their pl, thus
limiting their separation by CLEF.
In CLEF, the internal surface of the capillary wall should be treated to eliminate EOF. The EOF flushes
the amphyolytes and the analytes from the capillary before the ClEF is complete. Hence, the EOF should be
lowered or eliminated. Reduction of EOF is achieved through dynamic or covalent coating of the capillary. The
capillary coating limits the adsorption of proteins on its internal wall. An important practical consideration in
ClEF is the elimination of EOF, which, if present, prevents stable focused zones from developing. Therefore,
coated capillaries are preferred for CLEF. Methylcellulose and polyacrylamide coated capillaries are used. A
potential problem with polyacrylamide coated capillaries is its attachment to the silica via a siloxane bond which
makes it unstable at high pH.
Nonionic surfactants (e.g., Triton •
Brij-35), or organic modifiers (e.g., glycerol, ethylene glycol)
have been added to carrier electrolyte to minimize protein aggregation. Non-denaturing protein modifiers are
used. Urea denatures proteins and consequently avoided.
ElectrophoretJc mobilization can be accomplished in either anodic or cathodic reservoir. Mobilization
is achieved through addition of a salt solution (e.g., NaCI) to an electrolytic vessel. For anodic mobilization, a
salt solution is added to the anodic reservoir. For cathodic mobilization, a salt solution (eg. NaOH/NaCI) is
added to the cathodic reservoir. The addition of a salt alters the pH of the carrier electrolyte in the capillary.
Both, carrier ampholytes and analytes are mobilized in the direction of the reservoir with added salt. As
mobilization proceeds, the current increases as the ions migrate into the capillary.
2.4 Micellar Electrokinetic Capillary Chromatography
598
Micellar electrokinetJc capillary chromatography (MEKC) is a dynamic mode of capillary electrophoresis.
It is a hybrid of electrophorelJc and chromatographic techniques.
It is used particularly for nonionic and
ionic analytes with a wide range of hydrophobic and hydrophilic characteristics (e.g., microanalytes, pepUdes,
oligonucleotides). Surfactants are added to the carder electrolyte that can form micelles. The most important
feature is controlled electrophoretic mobility.
Surfactants are amphophilic moieties with hydrophobic and hydrophilic characteristics. They can be
ionic, zwitter ionic or nonionic. Surfactants have a long hydrophobic tail and a hydrophilic head. As the
concentration of the surfactant increases, they aggregate and form colloidal-sized assemblies called micelles.
The concentration at which a surfactant forms a micelle is termed critical micelle concentration (cmc) (table
2). Normally, micelles are spherical with the hydrophobic tails of the surfactant molecule oriented toward the
center to avoid interaction with the hydrophilic carrier electrolyte. The charged head is oriented toward the
carder electrolyte.
During migration, the micelle and analyte complexes through both hydrophobic and
electrostatic interactions. The separation results from interaction between the micelle and the analyte.
The surfactants are broadly divided into two broad categories, synthetic (eg. anionic, cationic, zwitter
ionic, nonionic) and natural (eg. bile salt) (table 3). A variety of surfactants have been used in MEKC.
Many surfactants are adsorbed on the capillary wall. Adsorption of surfactants modifies the EOF and
limits potential analyte adsorption. The surfactant charge affects the magnitude and direction of EOF. The
direction of EOF is reversed by adding a cationic surfactant (e.g., CTAB, DTAB) to the carder electrolyte. The
CTAB monomers adhere to the wall through coulombic attraction.
The positive charge results from
hydrophobic interaction between the free CTAB monomer with those bound to the internal wall of the capillary.
The ionic micelles migrate either with or against the EOF (depending on the charge). Anionic surfactants (e.g.,
SDS, SOS) migrate toward the anode, in the opposite direction to the EOF. Since the EOF is generally faster
than the migralJon velocity of the micelles at neutral or basic pH, the net migration is in the direction of the EOF.
The anionic, nonionic and cationic analytes can be separated using SDS. Generally, the anionic
analyte will be separated first followed by nonionic and calJonic analytes. Anions spend more time in the carder
electrolyte due to coulombic repulsion from the anionic micelle (e.g., SDS). The greater the anionic charge
599
the more rapid the elution. Nonionic analytes are separated exclusively on hydrophobicity. CaUonic analytes
elute last due to strong coulombic attraction (e.g., ion pairing with the micelle). Neutral (e.g., hydrophobic,
hydrophilic) analytes are partitioned in and out of the micelle. Partitioning affects the separation and migration
time. The hydrophobic analytes (e.g., Sudan III) interact more strongly with the micelle than hydrophilic
analytes. The prolonged analyte-micelle interacUon results in increased migration Ume, since the micelle
carries the analyte against the EOF. When the analyte is dissociated from the analyte-micelle, the analyte is
carried with the electroosmotic flow.
The properties of the micelle are affected by operational parameters and electrolytic additives
incorporated in the carrier electrolyte. Operational parameter (e.g. temperature) dramatically effects the
micelle properties. A potential problem with the use of ionic surfactants, especially at high concentrations, is
an increase in current. As a result of which the capillary temperature rises considerably. Elevated or reduced
temperatures affect viscosity, EOF and migration lime. The addition of electrolytes to aqueous micelle systems
results in an increase in aggregation number and a reduction in cmc. The cmc for SDS in water and
phosphate/borate carrier electrolytes are 8 mM and 5 mM, respectively (Sepaniak and Cole, 1987). Organic
modifiers (e.g., methanol, 2-propanol, acetonitrile) have been used to affect analyte-micelle interaction. The
addition of organic modifiers to the carrier electrolyte reduces hydrophobic interaction between the
analyte-micelle complex. The organic modifiers affect the aggregation and the micellar ionization numbers.
Methanol enhances micelle formation at low concentrations but inhibits it at high concentrations. Acetonitrile
can hydrogen bond with water and at moderate concentration increases the cmc (Hinze, 1987).
2.4.1 Chiral Selectors
Inclusion complexes are molecular compounds with characteristic structural arrangement, in which
a compound (e.g., chiral selector) spatially encloses another (e.g., analyte) or at least part of it. The inclusion
phenomena are widely used in separationai sciences. Chiral recognition is dependent on the formation of
diastereoisomers either through covalent or electrostatic interactions. The most commonly used chiral
selectors for the formation of inclusion complexes in capillary electrophoresis are bile salts and cyclodextrins
that form steroselective interaction with the analyte. Chiral resolution by capillary electrophoresis comprises
600
the addilion of a chiral selector (e.g., bile salts, cyclodextrins) into the carrier electrolyte. The migrational time
of the analyte-micelle chiral complex is increased because it moves slowly in the carrier electrolyte.
The optically active chiral additives (e.g., bile salts, cyclodextrins) permit chiral separaUon by
steroselective interaction with the analyte. In case of bile salts (e.g., sodium taurocholate, sodium glycocholate)
the chiral interaction occurs at the surface of the micelles. Cyclodextrines are another class of widely used
chiral selector. They are nonionic cyclic oligosacchaddes. They are characterized as a, 13and T cyclodextrins
that contain six, seven and eight units of a-D-glucopyranoside, respectively. The stable L-complexes exhibit
increased migration time. Cyclodextrins are a hollow truncated cone with a cavity diameter determined by the
number of glucose units. The cyclodextrine cavity is relatively hydrophobic while the external surface is
hydrophilic. The circumference contains chiral secondary hydroxyl groups. Chiral resolution results from
inclusion of a hydrophobic portion of the analyte in the cavity and hydrogen bonding to the chiral hydroxyl
groups of the cyclodextrins. Modified ionic cyclodextrins (e.g., carboxyl, succinyl), have been used to alter
selectivity and improve detection properties. Selectivity can be modulated by using appropriate chirai selector
(chemical properties and quantity). Additionally, organic modifiers such as alcohols, surfactants, urea, and
metal ions also alter chiral selectivity as well as improve detection.
2.5 Capillary Isotachophoresis
Capillary isotachophoresis (CITP) is a moving boundary electrophoretic technique and uses a
discontinuous carder electrolytic system. The analytes condense between the leading and the terminating
electrolytes, producing a steady-state migration pattern composed of consecutive analyte zones. Either anionic
or cationic analytes are resolved during separation.
Generally, in capillary electrophoresis the analyte concentration is determined from the peak area in
the electropherogram.
However, in CITP the isotachopherogram contains a series of steps, each step
representing an analyte zone. The analyte concentration is evaluated from the zone length.
CITP relies on zero electoosmotic flow, and the electrophoretic system is heterogeneous. Generally,
the separations require modified capillaries with suppressed (e.g., hydrooxypropylmethylcellulose [ca. 0.25%])
or annihilated EOF. Recently, CITP with EOF has been demonstrated (ref). The capillary is filled with a leading
electrolyte. The analytes are loaded in the capillary (ca. 30-50%) without sacrificing separation quality. The
601
analytes are loaded (either electrokinetically or hydrodynamically) and the anodic reservoir is filled with
terminating electrolyte. For anionic analytes, the leading and the terminating electrolytes must contain anions
with an effective higher and lower mobility, respectively, than the anionic analytes being resolved. Under the
influence of an electric field the anions migrate toward the anode. Separation occurs between the leading and
terminaUng electrolytes based on analyte mobility (fig. 9). Since the leading anion has the highest mobility it
moves the fastest, followed by the higher and high mobility anions. Stable zone boundaries develop among
the anionic analytes, resulting in highly efficient separations. At the start of the separation, the current may be
quite high since the highest mobile leading electrolyte completely fills the capillary. As the separation
progresses, the current declines because the least mobile terminating electrolyte enters the capillary.
CIPT has two characteristics, the combination of which is unique to electrophoretic methods. The
anionic analytes migrate in discrete zones or bands. Additionally, all analyte zones move at the same velocity.
The migration velocity is influenced by the characteristics of the leading anion. The higher mobile zone has
higher conductivity (lower resistance) and a lower voltage drop across the zone. The mobility is the product
of the voltage drop and the conductivity. The voltage drop is an inverse function of conductivity. Thus, the
electrical field strength varies in each zone and normalizes to maintain constant velocity (velocity = mobility X
field), with the lowest field across the zone having the highest mobility. If an anion diffuses into adjoining zones
(i.e., defocusing or broadening), it is accelerated or decelerated based on the electrical field strength
encountered and realigns 0.e., focuses) into the corresponding zone. This phenomenon results in very sharp
analyte zones.
A demerit of CITP is the adjoining analyte zones are in contact with each other. Spacer molecules
have to be employed to overcome this drawback. A spacer is a nonabsorbing molecule with a mobility value
between the mobilities of two analytes being separated.
Various compounds have been added to the carder electrolyte in order to form complexes. These
complexes permit the effective mobilities of the analyte anions to be controlled and thus optimize separation.
The
13-cyciodextrins have
been
added to the leading
electrolyte.
Also, small
amounts of
hydrooxypropylmethylcellulose are added to the leading electrolyte. It dynamically coats the internal wall of
the capillary, reduces adsorption of proteins and annihilated the electrical double layer.
602
Selection of an appropriate leading and terminating electrolytic systems within the pH that contain both
leading and trailing anions/cations is a difficult task. The leading cathodic electrolyte might contain a strong acid
(more mobile e.g., H3PO4[ca. 5 mM]), while the terminating anodic electrolyte might contain a weak acid (less
mobile e.g., propionic acid or valine). For cationic system, consisting of acetate as the leading electrolyte (ca.
10 mM pH 4.75) and acetic acid as the terminating electrolyte (ca. 10 raM). On the other hand, for the anionic
system, formic acid titrated with ammediol (ca. 10 raM, pH 9.1) is deployed as the leading electrolyte and
alanine-ammediol (ca. 10 mM, pH 9.5) is deployed as the terminating electrolyte.
3. INSTRUMENTATION
InstrumentalJon development for capillary electrophoresis has progressed tremendously. The typical
capillary electrophoretJc system consists of: 1) Sample applicator; 2) Separating capillary; 3) Detector; 4) Liquid
handling system and 5) Electrical energy source.
3.1 Sample Application
The small capillary dimension limits the quantity of the sample that can be loaded. Generally, the
injection plug length should not exceed 1-2% of the I..tof the capillary. This corresponds to a sample injection
volume of 1 to 50 nL, depending on capillary dimensions. Generally, samples are loaded appropriately.
Severely underloading the sample may avoid analyte detection. An advantage of underloading is high
efficiency. On the other hand, sample overloading is generally routine and is detrimental to resolution. Injection
plug lengths longer than the diffusion controlled zone width will proportionally broaden peak width. Additionally,
it intensifies field inhomogeneities and distorted peak shapes due to mismatched conductivity between the
carrier electrolyte and the analyte zone.
Two methods commonly used in loading samples are electrokinetic and hydrodynamic injections. In
both methods the injection volume is indirectly calculated from defined operational parameters. The defined
operational parameters are voltage/'dme for electrokinelJc injection, or pressure/time for hydrodynamic injection.
3.1.1 Electrokinetic Injection
603
An electrokinetJc (i.e., electromigration) injection is performed by exchanging the injection-end reservoir
with the sample vial and replacing the sample vial with the reservoir buffer. Following this step the voltage is
applied across the capillary. Normally, the sample loading voltage is 3-5 magnitudes lower than sample
separation voltage. The sample enters the capillary by both electrophoretic and electroosmotic effects. Since
the sample loading is influenced by electrophoretic mobility, the high mobile ionic species are loaded at a
greater concentration than the low mobile counterparts. Despite the quan~ative and reproducible drawbacks,
electrokinetic injections are easy to perform, do not demand supplementary device, and suitable for viscous
media or gels utilized in the capillary electrophoresis.
3.1.2 Hydrodynamic Injection
A hydrodynamic injection is the most broadly used sample loading method. It is achieved by: 1)
applying pressure at the capillary injector-end 2) vacuum suction at the capillary detector-side, and 3) siphoning
action or hydrodynamic imbalance by lifting the injector-end reservoirs compared with the detector-side
reservoir. W'rth hydrodynamic injection, the quantity of the sample loaded is nearly independent of the sample
matrix.
The sample volume (V,) for a pressure/vacuum injection can be calculated by the Hagen-Poiseuille
eq. (21). The volume (V,) is dependent on capillary dimensions, the carder electrolyte viscosity, the applie