Gums and Stabilisers For The Food Industry

t
for the Food Industry 11
Edited by
Peter A. Williams
North East Wales Institute, Wrexham, UK
Glyn 0.Phillips
Research Transfer Ltd. Cardifi UK
RSeC
ROYAL SOCloY OF CHEMISTRY
The proceedings of the Eleventh Gums and Stabilisers for the Food Industry Conference -
Crossing Boundaries held on 2-6 July 2001 at The North East Wales Institute, Wrexham,
UK.
Special Publication No. 278
ISBN 0-85404-83 6-7
A catalogue record for this book is available from the British Library
0 The Royal Society of Chemistry 2002
All rights reserved.
Apart from any fair dealingfor the purpose of research or private study, or criticism or
review as permitted under the terms of the UK Copyright, Designs and Patents Act, 1988,
this publication may not be reproduced, stored or transmitted, in any form or by any means,
without the prior permission in writing of The Royal Society of Chemistry, or in the case of
reprographic reproduction only in accordance with the terms of the licences issued by the
Copyright Licensing Agency in the UK,or in accordance with the terms of the licences
issued by the appropriate Reproduction Rights Organization outside the UK.Enquiries
concerning reproduction outside the t e r n stated here should be sent to The Royal Society of
Chemistry at the address printed on this page.
Published by The Royal Society of Chemistry,

Thomas Graham House, Science Park, Milton Road,
Cambridge CB4 OW, UK
Registered Charity No. 207890
For further information see our web site at www.rsc.org
Printed and bound by Athenaeum Press Ltd, Gateshead, Tyne & Wear
Preface
It is now a pleasure to introduce Volume 11 of this Series. These volumes are based
on the well-established Gums Conferences held in Wrexham every two years.
Looking back over the series, it is evident that as we enter the second decade, the
subject is stronger and more scientific than when we started 20 years ago. Then
hydrocolloids were commodity materials, with little interaction between researchers
and academic institutions, producers and users of additives and ingredients in food
products. The subject has accelerated meanwhile. The 10 previously published vol-
umes have contributed greatly in this respect. They are widely quoted and I have
found that they are well stocked on the bookshelves of industry and academia.
Alongside, the journal Food Hydrocolloids has also progressed, with its impact fac-
tor growing steadily over the years. This volume reflects this change in attitude, with
hydrocolloids now becoming technical products whose functional behaviour is
increasingly being recognised. This new book will again push forward the subject,
which is now at an interesting, if critical, juncture. There has been a rash of take-
overs and mergers in the companies involved in the hydrocolloid field, necessitating
cross-material use and an extension of specific functionalities. This book reflects this
trend.
A major emphasis is placed on current issues, both scientifically and in the mar-
ket. Denis Seisun, the originator of the series of business conferences entitled Food
Hydrocolloids, is the expert in Market Trends, and his predictions for the markets in
hydrocolloids, given here, are very positive but do also highlight differences in the
prospects between various hydrocolloids. Technically, all aspects of the subject are
covered in the book: microstructure of products, structure, interactions and charac-
terisation, new techniques and methods efc. The heartening feature is the emphasis
now on real food systems, and the increased participation of proteins within the
hydrocolloid classification.
Possibly the most significant trend is the extension of the functionality of tra3i-
tional materials in areas outside their conventional usage. This is particularly true of
the galactomannans. For a traditionalist, it is strange too that we should be now look-
ing to the pectins and for emulsion-stabilising properties.
New materials too are being devised using a biomimetic approach, tailor-making
hydrocolloids using nature as a model. Biomaterials such as konjac mannan, gum
arabic and other acacia gums, chitosan, dextran etc. along with a wide range of pro-
teins are now coming out of the shadows.
Increasingly too the hydrocolloids are leading the charge into healthier foods,
much demanded now by the public. There is a major section dealing with Hydrocol-
loids and Health. The subject, however, is a minefield, particularly with regard to
the health claims which can be made, and the positioning of the products between
healthy food supplements and pharmaceuticals. The legislative situation in Europe is
described, as are the broader aspects of the influence of hydrocolloids on health.
There is another new avenue opening up also and covered in the book, namely the
specific behaviour of hydrocolloids with regard to their influence on cells directly. If
these can direct cells to behave or misbehave, then we are entering an exciting new
era. Our long held belief in the goodness of these materials may well have a scien-
tific cellular basis.
iv Gums and Stabiliers for the Food Industry 11
I hope too that the social pleasure of the conferences is reflected in the book. We
now have become a family, and this series of volumes cements these relationships.
Glyn 0. Phillips
Chaixman,Gums Organising Committee and Food Hydrocolloids Trust
Contents
Market Overview
Overview of the Hydrocolloid Market 3
D. Seisun
Structure, Characterisationand Interactions
Capillary Electrophoresis as a Tool for the Characterisation of Pectin Fine

Structure 13
M.A.K. Williams, G.M.C. Buffet and T.J. Foster
Application of Two Dimensional (2D) NMR Spectroscopy in the Structural

Analysis of Selected Polysacchaxides 27
S.W. Cui
a and p Mechanical Relaxations in High Sugar Biopolymer Mixtures 39

S. Kasapis and I.M.A. Al-Marhoobi
Fluorescence Anisotropy Measurements of the Dynamic Behaviour of

Biopolymers 54
T.A. Smith, L.M. Bajada and D.E. Dunstan
Time Resolved Polarised Attenuated Total Reflection Spectroscopy of a

Protein Film at the SilicdSolution Interface 65
M.L. Jayme, M.L. Gee and D. Dunstan
Elucidation of Protein-Lipid Interactions 73

N.K. Howell
Rheological Aspects
Helix-Coil Transition in ThermoreversibleGels 85

K. Nishinari, Y. Nitta, E. Miyoshi, S, Ikeda and T. Takaya
Microstructural Origins of the Rheology of Fluid Gels 95

W.J. Frith, X. Garijo, T.J. Foster and I.T. Norton
Dynamic Mechanical Characteristics of Red Algal Polysaccharide Gels

Composed of Soluble Starch and Starch Granules 104
V.M.F. h i , H.-J. Liang and C.-y. Lii
V
vi Contents
Morphology Control in Disperse Biopolymer Mixtures with at Least One

Gelling Component 112
B. Wolf; W.J. Frith and I.T. Norton
Determining the Yield Stress of Xanthan Solutions via Droplet Rise

Experiments 120
H. C. Mielke and D.E. Dunstan
Predicting the Rheology of Water-in-water Emulsions 128

J.R. Stokes and W.J. Frith
Influence of Pentosans on the Rheological and Thermal Properties of

Starch Isolated from Two Different Wheat Varieties 137
D.M.J. dos Santos and J.A.L. da Silva
Gelatinisation Properties of Sago Starch in the Presence of Salts 145

F.B. Ahmad and P.A. Williams
Effect of K+ and Ca2+Cations on Gelation of K-Carrageenan 158

J. Doyle, P. Giannouli, K.Philp and E.R. Morris
Rheological Properties of K-Carrageenan Weak Gels 165

S. Ikeda, K.Nishinari and V.J. Morris
Mixed Systems of Fish Gelatin and Ic-Carrageenan 172

I.J. Haug, H.Devle, K.I. Draget and 0. Smidsr@d
Effect of Gelatidcarrageenan Interactions on the Structure and the

Rheological Properties of the Mixed System 177
C. Michon, K. Konate, G. Cuvelier and B. Launay
Hydrocolloids in Real Food Systems
Hydrocolloids in Real Food Systems 187

I. T. Norton and T.J. Foster
Interaction of Carrageenan with Other Ingredients in Dairy Dessert Gels 20 1

J. de Vries
Emulsions as Ingredients in Foods - Surface Compositions and Their

Modifications 21 1
D.G, Dalgleish
Contents Vii
Influence of Hydrocolloids on Flavour Release and Sensory-Instrumental

Correlations 217
R. Clark
The Yield Stress Phenomenon and Related Issues - An Industrial View 226
N.W.G. Young
Interfacial Behaviour and Gelation of Proteins
The Roles of Proteins and Peptides in Formation and Stabilisation of

Emulsions 237
P. Walstra
Effect of Various Factors on Emulsion Stabilising Properties of Chitosan

in a Model System Containing Whey Protein Isolate 245
S. Luplante, S.L. Turgeon and P. Paquin
Depletion Flocculation by Polysaccharides in Whey Protein Stabilised

Emulsions at High Whey Protein Concentrations 256
T.B.J. Blijdenstein, E. van der Linden, T. van Vliet and G.A. van Aken
Aggregation and Gelation of Globular Proteins with and without Addition

of Poly saccharides 263
D. Durand, C. le Bon, P. Croguennoc, T. Nicolai and A.H. Clark
Kinetic and Equilibrium Processes in the Formation of Weak Gelatin Gels 27 1

D. Oakenfull and A. Scott
New Materials
Using Nature to Tailor Hydrocolloid Systems 28 1

M.J. Gidley
Formation of Strong Gels by Enzymic Debranching of Guar Gum in the

Presence of Ordered Xanthan 289
C.E. Cronin, P. Giannouli, B.V. McCleary, M. Brooks and E.R. Morris
New Textures with High Acyl Gellan Gum 297

N.A.Morrison, G. Sworn,R.C. Clark, T. Talashek and Y.-L. Chen
Rheological Properties and Potential Industrial Application of Kongoli

(Maesopis eminii) Seed Gum 306
J. T. Baminas and I. C.Eromosele
Contents
I..
Vlll
Emulsifying Properties of Depolymerised Citrus Pectin: Role of the

Protein Fraction 311
M.Akhtar, E. Dickinson, J. Mazoyer and V. LangendorfS
Modified Hydrocolloids with Enhanced Emulsifying Properties 318

F.M. Ward
Hydrocolloids and Health
European Legislation and Health Claims 325

P. Berry Ottaway
Managing Healthy Levels of Blood Glucose and Cholesterol with Konjac

Flour 338
G. Crosby
The Role of Polysaccharides in the Gastric Processing of Foods 342

A. Fillery-Travis, L. Marciani, P.A. Gowland and R.C. Spiller
Cell Signalling Potential of Hydrocolloids: Possible New Applications? 349

A.O. Phillips
Alginates in Relation to Human Health; from Commodity to High Cost

Products? 356
K.I.Draget and G. SkjGk-BrRk
Subject Index 365

Acknowledgements
The Conference owed its success to the invaluable assistance of the Organising
Committee.
Members of the Organising Committee
Mr P Cowburn Danisco Ingredients Ltd

Professor E Dickinson University of Leeds
Dr T J Foster Unilever Research
Mr D Gregory D G Associates
Dr R Harrop (Treasurer)
Dr I Hodgson (Vice Chairman)
Professor D Howling Kelloggs
Mr H Hughes (Secretariat),North East Wales Institute
Mr D Lloyd Cerestar UK Ltd
Dr R G Morley Delphi Consultant Services Inc. USA
Professor J R Mitchell University of Nottingham
Professor E R Morris Cranfield University
Dr V J Moms Institute of Food Research
Dr J C F Murray Hercules Ltd
Ms L Patterson Du Pont (UK) Ltd
Professor G 0 Phillips (Chairman), Research Transfer Ltd
Dr G Sworn Kelco Biopolymers
Dr R White Optokem Ltd
Professor P A Williams (Secretariat),North East Wales Institute
ix
Market Overview
MARKET OVERVIEW
Dennis Seisun
IMR International
The Food Hydrocolloid industry represents a market of over US$3.0 BILLION ( $ 3 ~ 1 09).
The top two hydrocolloids, starch and gelatin, account for no less than 50% of the total
value. Starches comprise a wide range of diversified hydrocolloids. Native, modified, corn,
potato, wheat, tapioca and rice starches are all included. The degree of modification
allowed in starch production has given starch producers much leeway in supplying literally
hundreds of different starches to the food industry. Gelatin is probably the most
commoditized of the hydrocolloids surveyed. It has the distinction of having suffered the
most severe reversal in consumer image during the last 5-6 years. Gelatin, for most of its
'food hydrocolloid life' has been one of the most widely accepted texturizing agents. It's
declaration on a food label was no problem and consumers were familiar and comfortable
with gelatin on a food label. This cozy situation for gelatin took a sharp turn for the worst
with the advent of Bovine Spongiform Encephalopathy (BSE/Mad Cow Disease) and the
consumer food scares that ensued. Gelatin is now the only hydrocolloid with a negative
growth rate forecast during the first five years of the new millenium. Pectin is the other end
of the 'consumer image' spectrum and is most readily and widely accepted by consumers. It
is perceived as a traditional, healthful food ingredient. Grandma used pectin to produce her
homemade jams. Pectin comes from healthy, natural, fruits such as oranges and apples.
Readers of this article however, are reminded that hydrocolloid images can change rapidly
in the mind of consumers. Gelatin was, until recently, one of the most 'label friendly'
texturizing agents.
Realistic Market Potential
IMR developed the concept of realistic market potential to allow hydrocolloid producers a
more accurate estimation of their markets. The realistic market size for each hydrocolloid
consists of two parts. First, the volume of that particular hydrocolloid that is sold. Second,
a PORTION of the market for other hydrocolloids which can be realistically targeted by a
given hydrocolloid. For example, the market for Xanthan is $219 million which is the
value of xanthan estimated to be sold PLUS $153 million which represents the value of
xanthan that could realistically be sold in the marketplace when competing with other
hydrocolloids. The estimate for replacement depends on a number of factors including;
Copyright IMR International
Tel858-45 1 6080 email info@hydrocolloid.com
WWW .HYDROCOLLOID,COM
4 Gums and Stabilisers for the Food Industry I 1
price, functionality, availability, synergy and most important, consumer image. Gelatin for
example, has been difficult to substitute with other hydrocolloids. The recent concerns over
BSE however, have made it a much easier target for suppliers of other hydrocolloids. The
total potential market for xanthan is significantly larger than the volume of xanthan sales
alone. It is however much smaller than the total food hydrocolloid market. IMRs estimate
represents a realistic estimate between the two extremes. An estimate of the replacement
potential market for xanthan is presented below as an example of how this principle can be
used.
Source : IMR estimates from The Quarterly Review of Food Hydrocolloids
The realistic market potential for xanthan gum is $382 million. The potential is
significantly larger than the $219 million estimated sales of xanthan gum but much smaller
than the $3.0 BILLION total market for food hydrocolloids. IMR has conducted this
exercise for ALL food hydrocolloids surveyed. A computer program is available to analyse
various replacement scenarios. Please contact us for further information.
Commercial Aspects
A brief description of each hydrocolloid as it is seen and used in the food industry is given
below. These brief descriptors are in contrast to the highly scientific studies of
hydrocolloids that are normally presented at Wrexham conferences.
Agar Limited market, capital intensive, limited raw material supply. DifJicult
growth prospects in food applications. Increasing competition from other
gelling agents either used as single ingredients or combinations such as
carrageenadbg, xanthadbg or Curdlan. Use of agar in bacteriological
plates is resistant to competitive hydrocolloids and technologies.
Market Overview 5
Arabic Notoriously unstable in supply. Is expected to have a steady decline in

overall consumption lfood and industrial) in the medium to long ternfuture
unless supply and price instability can be resolved.
Alginates Mature product with no new major applications. Threat of Chinese alginate
supply remains. Chinese threat is tempered by a growing demand within
China itsel$ May benefitfiom positive image as a marine product less tainted
by agricultural tampering.
Cassia Gum A relatively new player in the world of food hydrocolloids. Similar to guar in
that it is a galactomannan and mostly originates in the Indian Subcontinent.
Mainly approved in pet food but human food approvals are being sought by
its major developer. A strong patent situation makes cassia virtually a single
source hydrocolloid.
Carrageenan Multi-functional hydrocolloid, good long term potential. The use of

carrageenan has been greatly expanded with its approval in meat and poultry
applications. Thefull approval of semi-refined carrageenan in the US and W.
Europe has changed the pricing and profitability equation for carrageenan.
The market is now far more competitive and less able to bear the added
value, which the higher prices of refined carrageenan allowed. The potential
for consumerconcerns over carrageenan is always present and has recently
been raised again by a US academic publication. carrageenans reactivity
with milk proteins makes it an ideal stabilizer in the continuously growing
market of dairy products.
CMC Low profitability hydrocolloid. Fairly mature in USfood applications with no

new applications in sight. Large capital investment required. Close to being a
commodity as a food additive. Can however, be highly direrentiated in
production.
MC/HPMC Major markets are in the industrial and pharmaceutical sector. Major
investment required to enter production. These hydrocolloids have a
chemical sounding name. If the t e r n carbohydrate gum and vegetable gum
become disallowed in the US it could be a problem. There seem to be many
under-promoted and under-used finctional properties for these
cellulosics. Dow, the worlds largest producer of MCBPMC has steadily
increased its finding and commitment to growing its food hydrocolloid
business.
MCC Micro-crystalline cellulose is a unique hydrocolloid with some niche markets.

It is unlikely to ever become anotherxanthun gum During the 1970sand
ll.
1980s FMC was virtually the only supplier of MCC and was able to obtain a
healthy return on sales. This allowed for major development eforts and
investments in MCC growth. Since that time, MCC markets have become
more competitive. Suppliers of colloidal MCCfiom Brazil and Taiwan have
steadily improved the quality and reliability of their products such that they
are taking an increasing share of a market that FMC once had to itself.
6 Gums and Stabilisers for the Food Industry 11
Gelatin Close to being a commodity hydrocolloid. Low profitability, produced by

some companies as a by-product from beef and pork packing operations.
Capital intensive production. Environmental issues in the production process.
Heightened awareness of gelatin following the Mad Cow Disease Scare
(Bovine Spongiform EncephalopathyBSE) and Foot & Mouth outbreak in the
UK. Inadequate for Halal, Kosher and Vegetarian applications. Gelatin on
the other hand, remains unique in some functional properties. No ideal
replacement has yet been found.
GelEan Gum Characterized commercially by being a Sole Source hydrocolloid. Gellan

gum is also characterized with the highest per lb. price of any food
hydrocolloid at $18-2OAb. Good future as a biopolymer. Potential for process
improvement and price reduction.
Guargum Uncertain and cyclic source of raw material. Most price competitive and
least profitable of hydrocolloids surveyed. Opportunity for physical and
enzymatic modification offers very large mark-ups. Inexpensive source of
dietary fiber. Good consumer recognition.
Karaya An obscure hydrocolloid with poor consumer and food processor

connotations. Many food processors view Karaya as a hydrocolloid for low
end foods. Also known as Indian Tragacanth because it has often been used
to extend tragacanth.
LBG An interesting hydrocolloid with excellent functional properties. LBG exhibits

synergy with xanthan gum and carrageenan. Notoriously cyclic in
availability and price. LBG prices skyrocketed from $3.OOAb to $15-2OAb in
the 1994B.5 season. LBG sufsersfrom poor long term investment b y its major
supp1iers.
Pectin One of the more interesting hydrocolloids surveyed. Pectin ofsers several
avenues for product diferentiation; HM, LM, physically modified (Slendid
TM) The excellent consumer connotation for pectin makes it one of the more
attractive to have on any food label and assures it of a long term future as a
texturizing agent.
Starch A major investment requirement. Poor profitability on the bulk of volume.

Ofen purchased as a commodity item. Requires any producer to be very
active in non-food applications of starches. Modified starches benefit from a
blanket approval for several grades and types of modijication. Labeling
issues cloud the future of modijied starches.
Tragacanth Similar to Karaya. Poor long term prospects. Has virtually been eliminated
from food applications by xanthan gum. Expensive and uncertain supply
mainly from Iran.
Market Overview 7
Xanthan Hydrocolloid of choice for long term future. Increasing competition has
maintained prices at reduced levels and attracted an increasing base of
users. Margins have declined, but the long term future of xanthan gum has
benefited from the strong competitive forces. Xanthan is perceived as having
the most diverse set of functional properties to offer the food processor.
Xanthan is a fermentation product. Biotechnology ofers excellent long term
opportunities, both for new products and for process improvements on the
production of existing hydrocolloids.
Hydrocolloid Classification
The scientific and academic classification of food hydrocolloids is divided along functional
properties or raw material origins. For example gelling agents vs. thickening agents or
seaweed extracts vs. seed gums vs. plant exudates vs. fermentation polymers. In the
commercial terms which IMR uses in reviewing these markets it is more useful to classify
the hydrocolloids in terms of price and availability cycles. The more stable hydrocolloids
include Starch, cellulosics, and gelatin. The typically cyclic hydrocolloids are guar, LBG
and gum arabic. These latter gums have suffered cycles of pricing that are literally 300-
400% in spread i.e. the high price reached in a complete cycle can be 300-400% higher
than the low price. For example, LBG prices have been as low as $2.75/1b and as high as
$12-15 in the last cycle. A number of other hydrocolloids fall in the questionable category
not so much because of price variation but because of availability problems. Xanthan,
pectin, and alginates have not had as dramatic cycles in price but they have suffered
periods of shortage and long delivery delays of 3-4 months.
Distribution Channels
Distribution channels are in constant state of flux, but overall the split between direct sales
vs. blenderhervice company has stayed the same in the last 15-20 years. Hydrocolloids
that are mainly sold directly from producer to end-user include starches, gelatin and agar.
Blends and services are mainly supplied to the dairy industry and use carrageenan, guar
and LBG. These three hydrocolloids are therefore distributed through third parties that buy
a number of hydrocolloids and supply custom blendshystems to their clients. The
remaining hydrocolloids are distributed through a variety of channels depending on the
end-use application and volumes required by end-users. Many large end-users have the
capability of developing and producing their own systems but prefer to delegate the work
and responsibility to a service company. The perennial problem is for service companies to
obtain fair compensation for the extensive work that is often required to develop a custom
system for specific process conditions and a particular client.
Corporate Changes
There have been and continue to be extensive strategic changes in this industry. Food
hydrocolloids are a specialty additive segment. Margins are much better than on
commodity chemicals although volumes are far lower. Large corporations seeking
diversification and bottom line improvements continue to look at food hydrocolloids as an
opportunity for expansion. The following is an indication of some of the major
hydrocolloid transactions that have taken place over the last decade.
Importer Sewices buys RP Gum Arabic Interests (1/92),

RP acquires Oatrim Rights from USDA (1/92),
A.E. Pellet & Sarcom merge (3/92),
Mare buys Cesalpinia & Italgumfiom Auschem (1/93),
Sandoz Acquires BeneJiber Technology (3/93),
HP Bulmer Divests-Citrus Colloids Formed (4/93),
Citrus colloids buys Yakhin Pectin Interests (I/94),
Nabisco buys Knox gelatin (1/94),
Obipektin buys Cesalpinia Pectin Plant (2/94),
SBI sale to SKW Trostberg Div. of Viag (4/94),
Merck Sells Kelco to Monsanto (1/95),
Hercules Tarragona LBG Facility Closed (1/95),
Freedom acquires Diamalt (1/95),
Kelco Acquires Cellulon From Weyerhaeuser (2/95),
Danisco Purchases Pectin Plant in Czech Republic (2/95),
Cerestar buys American Maize (3/95)
Valmar acquired (4/95),
Kelco Omega-3 Fatty Acids Venture (4/95),
Pfizer Food Ingredient sale to Cultor (1/9)
Shemberg/RP deal nearly done (2/96)
Metsa Serla Sale to Industrii Kapital(4/96),
Unilever Sells National Starch & Quest to ICI (2/97),
Hoechst Sale of Foreign Domestic Chemical (2/97),
Goodman Fielder Acquisition of Hormel Gelatin (12/97),
Citrico Formed (1st Qtr 98),
FMC Sells Cork Carrageenan plant, Dow Chemical acquires Courtaulds (4/98),
SKW Acquires Bunge FID (I0/98),
Monsanto Sells Algin Business to ISP I st Qtr 99,
CP Kelco founded @om Kelco Biogums and Hercules CarrageenadPectin (3rd
qtr 2000),
SKW announces intent to sell Gelatin (2ndQuarter 2000)
Obipektin acquired by Braes Group
Hercules Cellulosics up for sale
SKW acquires Ital SA in Mexico.
Goodman Fielder announces sale of Gelatin business (Feb 14, 01)
Goodman Fielder announces intent to sell Germantown in 2001.
BF Goodrich spins of Specialty Polymers toform Noveon Inc. (2nd qtr 2001),
Market Overview 9
Industrie Kapital sells Noviant to JM Huber (2nd quarter 2001)
There seems to be no indication that corporate changes are reaching a plateau. Continued
mergers, acquisitions, and divestments can be expected. The stabilizer industry is in need
of stabilization itself!! These corporate upheavals have for the most part not resulted in
improvements and streamlining in the supply of hydrocolloids and hydrocolloid related
services.
The future of hydrocolloids is assured in the food industry. Texture modification and
control are an integral and key arena in food processing. Functional issues are important,
but IMR believes that as important if not more important will be regulatory and social
issues. Regulatory issues are primordial. A food additive that does not have regulatory
food approval is not a food additive. Conversely changes in regulations can impact the use
levels and/or approved applications within the food industry. Social issues such as
vegetarianism, halal and kosher continue to grow in importance. Organic, GMO and
irradiation are three other social concerns that must be addressed by the food additive
industry.
Structure, Characterisation and Interactions
CAPILLARY ELECTROPHORESIS AS A TOOL FOR THE CHARACTERISATION
OF PECTIN FINE STRUCTURE
M.A.K. Williams*, G.M.C. Buffet and T.J. Foster
Unilever Research Colworth, Colworth House, Sharnbrook, MK44 1LQ, Bedford, UK
1. INTRODUCTION
1.1 Pectin Fine Structure
While pectin is essentially a linear co-polymer of galacturonic acid and its methylesterified
counterpart, it is arguably one of the most complex of the plant cell polysaccharides. This
complexity, manifest in a host of possible fine structure variants, gives pectin its utility of
function and, combined with its horde of accompanying enzymes in-vivo, its ability to
respond to a myriad of environmentally triggered stimuli. More generally, any process
capable of modifying pectin's molecular level polymeric characteristics, such as the molar
mass, the degree and distribution of methylesterification, amidation and acetylation, and
the content, length and distribution of neutral sugar side-chains, may yield a tangible
change in its manifest macroscopic properties. The determination of such polymeric
structure attributes forms an extensive and exciting field of research of which only a small
part can be covered in this article. In particular, the work reported here is focused on the
characterisation of homogalactan regions and, more specifically, on the elucidation of the
distribution of methylesterified residues.
1.2 Conventional Characterisation of Methylesterification
1.2.I Average Degree of Methylesterijkation. At present the classical titration method''2

is still the official method used for the determination of the average degree of
methylesterification (DE) of a pectin ample.^^ Traditional alternatives involve the
liberation of methanol by de-esterifying enzymes or by acid or alkali treatments, and its
subsequent quantification by ~hromatography,6.~ while more recently methods have been
described using infra-red spectroscopy.*
I . 2.2 Intermolecular Distribution of Methylester$cation. Previously reported
chromatographic methods for obtaining the intermolecular DE distribution (the variation in
DE between different molecules in the same sample) involve binding pectin onto an ion
exchange column under conditions of low ionic strength and subsequent elution of the
pectin with a buffer, the ionic strength of which is gradually increased?*" Fractions are
collected and classically the pectin mass in the column eluates has been determined by a
colorimetric assay with meta-hydroxy diphenyl (MHDP) while the DE has been inferred
14 Gums and Stabilisers for the Food Industry I I
from the elution time. However, it should be noted that the ion-exchange chromatography
(IEC) method is not without its problems. Pectins with blockwise intramolecular DE
distributions are found to elute from the column in an irregular manner, which is
potentially a serious problem, as the intramolecular charge distribution will be polydisperse
to some degree even in a sample in which all chains possess the same average charge per
residue. Furthermore, the binding of pectin to the column will be a function of its degree of
polymerisation since the equilibrium constant for the binding of a polyanionic molecule to
a polycationic IEC stationary phase depends on electrostatic interactions summed over all
groups.
In an attempt to correct for such problems a method has been described in which eluates
from IEC have been analysed by size exclusion chromatography (SEC) using a dual
detection system monitoring refractive index and conductivity. In this method the DE of
the eluates is actually measured, using a calibration linking the refractive index /
conductivity ratio to DE, developed using pectin standards, rather than inferred from the
IEC elution time. However, the eluates from IEC may still be polydisperse with respect to
DE and it should be noted therefore that the DE distribution derived using these average
values is still an approximation to some degree.
I .2.3 Intramolecular Distribution of Methylesterification. In general there are two
approaches to the determination of intramolecular DE distribution (the spatial distribution
of methylesterified residues along the pectic backbone). These are i) to attempt to measure
the distribution directly by recording a property of the residues that is sensitive to the local
residue type environment, and ii) to fragment the chain according to rules that depend upon
the residue type environment, and analyse the fragments generated. Broadly speaking
NMR methods have been applied that follow the first meth~dology,~~ while many
e n ~ y m a t i c ~and
~ ~chemical methods2072have been described that utilise the second
approach. Recent advances in this area include the separation of methylesterified enzyme
di est fragments using high performance anion exchange chromatography (HPAEC) at pH
5!-16 and the use of tandem mass spectrometry, which is able to locate the position of
methylesterified residues within the surviving enzyme-resistant
The question of how to exploit the rich vein of experimental data that is now being
obtained, in order to derive fine structure descriptors that efficiently encode the maximum
structural information and offer useful substrate discrimination, dnves a further interesting
area of the work. Recently the degree of blockiness of a pectin substrate has been
defined using the results from endo-polygalacturonase (endo-PG) digests,I4 based on the
liberation of unesterified mono-, di-, tri- and tetra-mers of galacturonic acid. Furthermore,
using the proportion of unesterified mono-, di- and tri-galacturonic acid released in
combination with the percentage of the total substrate galacturonic acid that is liberated in
an unesterified form, and the ratio of the sums of peak areas obtained for methylesterified
and unesterified fragments, sequence similarities have been represented by distance trees
using com utational techniques commonly exploited in the analysis of DNA and
These approaches have been used to compare pectins with various average
DE values and differing intramolecular distributions (generated by random or blockwise
de-esterification) and clearly demonstrate a useful discriminatory ability.
Complementary approaches focus on minimising the set of pectin fine structures that
are consistent with available digest information, in order to gain the most realistic
representation of the polymeric backbone. This philosophy is essentially one of chain
reconstruction. Practically, this type of work involves the computer simulation of test
substrates and their digestion, with subsequent comparison of digest characteristics with
those measured experimentally. To date this has been attempted simply with the aim of the
reconstruction of the measured molecular weight di~tribution.~~ However, recent attempts
Structure, Characterisation and Interactions 15
have been made at building more specific models of enzyme action25 that could, in
principle, be used in such a fashion. An obvious problem here is the practical issue of
computational time. To construct, and subsequently digest according to specific enzyme
rules, all (statistically) possible fine structures that exist for a pectin molecule of some 500
residues in length is simply impractical. However, as more is understood about pectin
biosynthesis it is possible that by limiting the set of trial pectin substrates to those that are
biosynthetically likely such an approach may become more realistic.
1.3 Capillary Electrophoresis
While capillary electrophoresis (CE)26-28has been routinely used to perform efficient

separations of proteins and nucleic acids for a number of years,2930applications to the
analysis of carbohydrates have taken by comparison longer to develop owing in part to a
lack of charge or a chromophore for many members of this class of biomolecule. A number
of techniques that circumvent this difficulty have now been successfully applied including
complexation, for example of sugars with anionic borate3 and lant starches with iodine /
iodide:* derivatization with U V absorbing or fluorescent and the use of indirect
U V detection at high pH.35For large carbohydrate polymers derivatization methods are of
limited scope as tagging can only be carried out at reducing sites, typically the end group,
imparting only one label per molecule. However, unlike many other natural
polysaccharides, pectins possess both charge and a UV chromophore, the carboxylate
group, making CE an attractive analytical tool for the study of these carbohydrates.
In fact, one of us has previously reported that CE can be successfully used in order to
measure the degree of methylesterification of a pectin sample, since there is a linear
relationship between the electrophoretic mobility and the average charge per residue.36
While many other methods perform this feat equally well, an advantage of the
electrophoretic method is its inherent separation quality. A symmetrical scaling of charge
and hydrodynamic friction coefficient with degree of polymerisation (DP) means that each
polymer chain, regardless of its DP, will elute according to its average charge density.
Each migration time, therefore, marks species with a unique DE and peak shapes thus
reflect the intermolecular DE distribution of the sample. This aspect of the CE
methodology has also been investigated previously and while some differences were found
between the detailed shapes of intermolecular DE distributions obtained by CE when
compared with the results of a more conventional IEC-SEC methodology, average DE
values, and moreover the spans of the extracted distributions, showed excellent agreement
between technique^.^^
The aims of this article are two-fold. Firstly, further evidence that the CE peak shape is
a useful measurement of the intermolecular DE distribution is presented, and secondly, it is
demonstrated that the exact same CE methodology previously reported as a useful tool in
the measurement of average DE values and intermolecular distributions, also offers an
attractive alternative for the measurement of endo-PG digest patterns.
2. MATERIALS AND METHODS
2.1 Sample Preparation and Enzyme Digests
2.1.I Substrates. All pectin samples were lemon peel pectins, supplied by Copenhagen
Pectin (www.cpkelco.com). Initial characterisation was carried out by the manufacturer
16 Gums and Stabilisersfor the Food Industry 1 I
using the titration method. The pectin samples were supplied as powder and solutions were
prepared by heating in de-ionised water at 60 "C for 30 minutes. Polygalacturonic acid was
purchased from Sigma (www.sigma-aldrich.com) and prepared similarly.
2.1.2 Enzyme Digests. A commercial sample of endo-polygalacturonase from
Aspergillus Niger was obtained from Megazyme (www.megazyme.com). Typically 1.7 U
was added to 2.5 mls of 0.25 % pectin solution and digests were carried out for 48 hours at
40 "C unless otherwise stated.
2.1.3 Standards. Mono-, di- and tri-galacturonic acid were purchased from Sigma and
solutions were prepared by mixing known weights of sample and de-ionised water in order
to obtain the desired w / w concentration.
2.1.4 Buffers. Phosphate buffer at pH 7.0 was used as a CE background electrolyte
(BGE) and was prepared by titrating aqueous 50 mM NaH2P04with 1 M NaOH. Sodium
acetate at pH 5.0 was used as a HPAEC eluent and was prepared by titrating 1 M NaOAc
with 1M HOAc. All buffers were filtered through 0.2 pm filters (Whatman).
2.2 Anion Exchange Chromatography and Fraction Collection
2.2.I HPAEC. High performance anion exchange chromatography (HPAEC)was

carried out using a Dionex-300 equipped with a Carbopac PA1 column and a PA100
guard. The separation was carried out at pH 5 and post-column addition of NaOH allowed
for the use of PED detection, as reported previ~usly.'~-'~ Samples were eluted with a 65
minute linear gradient of 0.05 to 0.7 M sodium acetate using a flow rate of 0.5 ml min-I,
which was typically made up to 0.8 ml min-' with the NaOH.
2.2.2 Fraction Collection. Fractions were collected from HPAEC for subsequent use in
capillary electrophoresis. These were diluted, post collection, in order to achieve a sample
ionic strength of 50 mM prior to injection. In contrast to higher ionic strengths that caused
significant peak broadening, this was found to give negligible perturbation to the CE
electropherograms, and dilution was favoured over complex desalting procedures. In order
to collect sufficient quantities of the relevant methylesterified oligomers so that they could
be clearly observed, even after dilution, a partial digest of a 5 % pectin sample was carried
out and injected from a 500 pL injection loop. While this lead to significant peak
broadening compared with the experiment carried out at lower column loadings the
chromatogram was still clearly recognisable and individual peaks of interest sufficiently
resolved to allow fraction collection. 100 fractions of 0.25 ml were collected.
2.3 Capillary Electrophoresis
Experiments were conducted using an automated CE system (HP 3D), equipped with a
diode array detector. Electrophoresis was carried out in a fused silica capillary of internal
diameter 50 pm and a total length of 64.5 cm (56 cm from inlet to detector). The capillary
incorporated an extended light-path detection window (150 pm) and was thermostatted at
25 "C. All new capillaries were conditioned by rinsing for 1 hour with 1 M NaOH, 1 hour
with a 0.1 M NaOH solution, 1 hour with water and 2 hours with BGE. Between runs the
capillary was washed for 10 minutes with BGE. Detection was carried out using UV
absorbance at 191 nm with a bandwidth of 2 nm. Samples were loaded hydrodynamically
(various injection times at 5000 Pa, typically giving injection volumes of the order of 10
nL), and electrophoresed across a potential difference of 30 kV. All experiments were
carried out at normal polarity (inlet anodic) unless otherwise stated. Electrophoretic
Structure, Churacterisation and Interactions 17
mobilities, p, are related to the migration times of the injected samples relative to a neutral
marker, t and to respectively, by the equation:
where L is the total length of the capillary, 1 is the distance from the inlet to detector, V is
the applied voltage, Fobs is the observed mobility and pea is the mobility of the
electroosmotic flow (EOF). 26
3 . RESULTS AND DISCUSSION
Figure 1 shows typical electropherograms obtained from 20 s injections of 0.5 % wlw

solutions of pectin samples of (a) 77.8 %, (b) 55.8 YOand (c) 31.1 YODE. The absorbance
dip at around 4 minutes indicates the neutral marker position. It can clearly be seen that the
migration time increases with decreasing DE (increasing charge).
3 4 5 6 7 8 9 10
t / min
Figure 1 Typical electropherograms obtained from pectin samples of varying DE.
At pH 7.0 the galacturonic acid groups, pKa = 3-4, are fully charged, and although pectin is
susceptible to base-catalysed p-elimination above pH 4.5, no problems were encountered
during run times of less than 2 0 minutes, at 25 "C, in the CE capillary. All these anionic
polysaccharides migrate after the neutral marker. The observed mobility pobs is the vector
sum of pe0and p (see equation 1 ) and since p is negative and smaller in magnitude than
/Je, the anions having the most negative mobility have the smallest pobs and thus the longest
migration times. Also shown is the result obtained following a consecutive 7 s injection of
all three samples. It can clearly be seen that the individual peak shapes are maintained in
the mixed sample and, as reported previously, the peak widths are invariant to injection
length, demonstrating that peak variance arising from the length of the injection plug is
minimal. Owing to a linear relationship between electrophoretic mobility and average
charge density, data such as that shown in figure 1 can be used to construct a simple
transform that maps the migration time axis onto degree of methylesterification. In order to
do this the electrophoretic mobilities of the peak maxima are taken to correspond to species
with the sample average degree of methylesterification as determined by titration. It has
previously been shown that following such a calibration, based on the simultaneous
injection of three resolvable pectin standards, this approach is capable of extracting the
average DE value of further pectin Sam les to within 2% (which is comparable with the
uncertainties in the titration method)! Furthermore, peak shapes transformed in this
manner have been found to give intermolecular DE distributions whose spans agree well
with those obtained by more conventional IEC-SEC methods.37
Figures 2(a) and 2(b) show comparisons of the intermolecular DE distributions,
obtained as described above, for high and low calcium sensitive pectins of a similar DE. In
each case the average DE values obtained by CE are found to be in good agreement with
those obtained by titration.
, , . , . I .
low Ca Sensitive (a) -Low Ca Sensitive

o High Ca Sensitive o High Ca Sensitive
40 50 60 70 40 50 60 70 80
DE 1 % DE/%
Figure 2 Intermolecular DE distributions (calculated from the CE peak shape) for high
and low calcium sensitive pectin fractions that have titration determined average DE
values of (a) 51.7 and 54.2 % and (b) 62.0 and 64.4 %, respectively.
It is particularly interesting to note that there is a common feature in the distributions

obtained for both fractions with high calcium sensitivity, manifest as a low DE tail. This is
clearly what would be expected for a calcium sensitive daughter fraction obtained by
separation on the grounds of calcium affinity and, as such, the result adds weight to the
interpretation of the CE peak shape as an intermolecular DE distribution, and further
demonstrates its usefulness.
Figure 3 shows the results obtained, under identical electrophoretic conditions to those
used above, following the injection of endo-PG digests carried out as described in the
experimental section, of (a) polygalacturonic acid, and (b) and (c), two pectin substrates of
31.1 and 55.8 % DE respectively. Previous work on polygalacturonic acid digestion using
the same enzyme source as this work found that mono-, di- and tri-galacturonic acid were
the main digestion products, implicating these compounds as responsible for the CE peaks
observed at around 7 , 9 and 10.5 minutes. Indeed, figure 4 shows the endo-PG digest of the
3 1.1 % DE pectin substrate run under the same conditions but spiked with commercial
samples of (a) mono-, (b) di- and (c) tri-galacturonic acid, and serves to unequivocally
confirm the identity of these peaks. (The notation nx is used throughout to indicate an n-
mer with x residues methylesterified).
Structure. Characterisation and Interactions 19
4 6 8 10 12
t I min
Figure 3 Electropherograms obtained following I 0 s injections of endo-PG digests of (a)

polygalacturonic acid, (b) pectin of DE 31.1 %, and (c) pectin of DE 55.8 % .
I I I I
I
I
lo 2O 3O
4 6 8 10 12
t / min
Figure 4 Electropherograms obtained following 5 s injections of the 31.1 % DE pectin

digest spiked with 5 s injections of 0.1% w/w (a) mono-, (b) di- and (c) tri-galacturonic
acid solutions.
It can be seen then that the scaling symmetry discussed in the introduction and
exemplified by the measurement of the intermolecular DE distribution has been broken.
While galacturonic acid oligomers of 10, 20 and 30 residues in length would co-elute
20 Gums and StabiEisers for the Food Industry 1I
according to their identical charge density, those of 1 , 2 and 3 residues are clearly resolved
despite their common average charge per residue. In fact, this symmetry breaking at low
DP values has been predicted theoreticall$* and has previously been detailed
experimentally for single stranded DNA.39In simplistic terms, the DP at which the scaling
no longer holds depends on how the average molecular conformation changes with the
length of the molecule and, as such, contains information about the flexibility of the
analyte.
For the use of CE in pectin fine structure elucidation, far from this being a
disadvantage, such a symmetry breaking enables fragments obtained from endo-PG digests
of pectic substrates to be separated. This is of particular interest because, as described in
the previous section, digest patterns of this type can be used to infer information regarding
the intramolecular DE distribution of the pre-digested substrate. Having clearly identified
mono-, di-, and tri-galacturonic acid in the electropherograms, attention is turned to
quantitation. Figure 5 shows a plot of absorbance peak areas (normalised for differences in
migration times4') obtained from CE runs of mono-, di-, and tri-galacturonic acid as a
function of the number of moles of galacturonic acid injected.
Figure 5 Observed normalised peak areas as a function of the concentration of

galacturonic acid residues injected in picomoles.
The latter is calculated from the concentration of standard solutions and the injection
volume, calculated using Poiseuille's law. Reasonable linear calibration plots are found. A
HPAEC method using W detection has also recently reported a linear relationship
between absorbance and concentration for these analyte~.~'It is worth noting, however,
that we find that the absorbance per galacturonic acid residue, while statistically
indistinguishable for the dimer and trimer (and presumably higher oligomers), is
significantly different for the monomer. Using this calibration information it can already be
seen that CE can successfully measure the degree of blockiness of a pectic substrate as
defined in the introduction, simply by quantifying the amount of unesterified mono-, di-
and tn-galacturonic acid liberated in an endo-PG digest. Indeed, as expected, figure 3
shows that significantly less of these unesterified species are liberated from a more
esterified starting material (55.8 % c.f 3 1.1 %).
In order to examine further the usefblness of CE as an additional technique for the
analysis of pectic digests, figure 6 shows a comparison of the results of (a) a conventional
Structure, Characterisationand Interactions 21
HPAEC experiment and (b) a CE experiment, carried out on the same endo-PG digest of a
+
3 1.1 % DE pectin substrate.
0.2
lo 2O 3O
y
.0 1
-
me
.-
; 0.0
w
a
-0. I
10
I
20
. ,
30
. ,
40
.
(a)
,
50
1
II
4
.
6
I . I
8
. I
10
. 1
12
t I min iI min
Figure 6 A comparison of the results of (a) an HPAEC experiment and (b) a CE

experiment carried out on an endo-PG digest of a 31.1 % DE pectin sample. ( n e major
HPAEC peaks have been assignedfrom existing literature' 8,.
On the whole CE compares favourably, especially when it is noted that some 10000 times
less sample was analysed in this particular experiment when compared with the HPAEC
methodology. However, it should be noted that there appears to be one more major peak in
the HPAEC chromatogram as compared to the CE electropherogram. In order to rationalise
this observation and to extract the maximum information from digest electropherograms all
the CE peaks have been identified. In order to achieve this fractions were collected from an
HPAEC separation of a concentrated endo-PG digest of a 31.1 % pectic substrate, as
described in the experimental section, and run in CE.
I ' I I I I I 1
I 5 2 6241 5' I
Ad)
L
I I I I
4 6 8 10 12
t / min
Figure 7 Electropherograms obtained following 5 s injections of a 31.1 % DE pectin

digest spiked with 20 s injections of diluted HPAEC fractions rich in the methylesterged
oligomers (a) 5', (6) tj2, (c) 4' and (d) 5'.
22 Gums and Stabilisersfor the Food Industry 1 1
From current literature on similar digest^'^^'^ (which had used mass spectrometry for peak
identification) the HPAEC chromatogram could be identified so that fractions of known
methylesterified oligomers could be collected. Figure 7 shows the endo-PG digest of a 3 1.1
% DE pectin substrate run under the same conditions as shown reviously, but spiked with
P
samples of the methylesterified oligomers (a) 52, (b) 62, (c) 4 and (d) 5', and serves to
unequivocally confirm the identity of the peaks in the CE experiment. It is apparent that
under the present conditions the methylesterified oligomer 5' co-elutes with digalacturonic
acid, which clearly explains the differing numbers of major peaks seen in figures 6(a) and
(b).
In order to investigate the quantitation of these methylesterified peaks the digest was
de-esterified using a base saponification. (It should be noted that, prior to this, the sample
is heated in order to denature the present endo-PG in order to prevent hrther fragment
digestion upon the alkali-mediated stripping of the methylester defenses.) Figure 8 (a) and
(b) shows the result of this modification on the CE electropherogram.
I'
I I
l o g 2 624l(2'0+4 3' d0506; I
4 6 8 10 12
t / min
Figure 8 Electropherograms obtained following I0 s injections of (a) an endo-PG digest

of a 31.1 % DE pectin, (b) the same digest post de-esterijkation and (c) the same digest
again after thefurther addition offresh endo-PG.
The result of de-esterification is the disappearance of the peaks previously found in

between those of the monomer and dimer of galacturonic acid and the corresponding
appearance of three hrther peaks eluting at longer times. This is entirely consistent with
the previous assignment of these peaks as 4', 52, and 62, which after de-esterification
appear as the tetramer, pentamer and hexamer of galacturonic acid. It can also be seen that
while the 1' and 3' peaks remain unchanged by the de-esterification procedure, the (2'+5')
peak reduces in size as the 5' component is shifted to contribute to the 5' peak. (It should
also be noted that there are indications of the presence of 7' and 8', showing that some
methylesterified septamers and octomers were present in the original digest, but at low
concentrations). The most important observation here is that the normalised peak areas of
the 4' and 62 peaks are the same, within experimental uncertainties, as the normalised peak
areas of the 4' and 6' peaks that they subsequently generate after de-esterification. This is
highly significant for the quantitation of such digest patterns as it suggests that, at least for
oligomeric digest products of interest here, the differences between the absorbance of
esterified and unesterified groups is minimal. This fact, coupled with the previous assertion
that for molecules larger than the dimer the absorbance per galacturonic acid residue is
constant, means that the quantitation of all digest peaks is trivial and can be carried out
using the calibration data shown in figure5.
The results of such an analysis for an endo-PG digest and its de-esterified derivative, as
shown in figure 8(a) and (b) are shown in table 1 . The total number of moles of
galacturonic acid residues contained in a 10 s injection (13.68 nL) of each of these samples
can be seen, within experimental uncertainties, to be comparable, thus supporting the
correct quantitation of all peaks. Furthermore, it can be calculated that an equivalent
volume of the pre-digested substrate solution, accounting for the dilution that occurs on
enzyme addition, would contain around 1 10 pmoles of galacturonic acid. The excellent
agreement between this value and those determined for the digests further supports the
quantitation strategy. As a final check the de-esterified digest has been disassembled
hrther by the addition of more enzyme. This procedure reduced the digest content to
purely mono-, di- and tri-galacturonic acid as expected, as shown in figure 8(c). It is now
possible, from this digest pattern, to calculate the total amount of galacturonic acid
residues in the sample based purely on the calibration obtained with commercial standards,
without making any assumptions about absorbance with regard to its modification by
methylesterification or DP.
Table 1 Quantitation of the endo-PG digest of a 31. I % DE pectin substrate using CE.
Normalisedpeak areas, the number of galacturonic acid residues in a I0 s injection (13.68
nL), and the molar concentration of oligomers in the digest are given.
Again it can be seen that, within experimental uncertainties, the calculated amount agrees
well with that calculated for the initial digest, showing that the quantitation of the
methylesterified oligomers is sound. Thus, it has been demonstrated that not only the
detection, but also the quantitation, of endo-PG digest patterns can be performed, simply
and efficiently, by the CE method.
While the primary goal of this article has been to demonstrate the potential of the
capillary electrophoretic method as an additional technique for use in pectin fine structure
elucidation, there are many topics of further work, worthy of note. These include the
detailed analysis of the way in which the mobility of the galacturonic acid oligomers scale
with DP. In addition, the quantitative characterisation of digests of different substrate types
will be key to the demonstration of the practical usehlness of the method. Ideally such
work should be carried out using pure enzymatic isoforms on which the most detailed
biochemical information exists so that the experiments can, in addition to yielding
information on substrate discrimination, help with the model of enzyme action. Following
the time-course of enzyme digestion by the CE method is also a real possibility.
Preliminary work in this area shows that, unsurprisingly, additional species to those
described here are observed in the electropherograms after shorter digestion times.
4. CONCLUSIONS
It has previously been reported that CE can be successhlly used in order to measure the
DE of a pectin sample. While many other methods perform this feat equally well, an
advantage of the electrophoretic method is its inherent separation quality. A symmetrical
scaling of charge and hydrodynamic friction coefficient with degree of polymerisation
means that each polymer chain, regardless of its DP, will elute according to its average
charge density. Each migration time, therefore, marks species with a unique DE and thus
peak shapes reflect the intermolecular DE distribution of the sample.
It has also been shown, however, that this exploited scaling symmetry breaks down at
small DP values. Far from being a disadvantage, this symmetry breaking at low DP enables
the separation of fragments obtained from endo-PG digests of pectic substrates, regardless
of their same average charge density. The quantification of such digest patterns has been
demonstrated and can be used to infer information regarding the intramolecular DE
distribution.
In conclusion, by carrying out the same CE experiment on a pectin sample and its
subsequent endo-PG digest, information can be obtained not simply on the average DE, but
also on the inter- and intramolecular DE distribution.
Acknowledgements
The authors gratefully acknowledge excellent technical assistance from Chris Tier and
Alison Russell and helpful discussions with Mike Gidley, Ian Norton and Allan Clark.
References
1. C.L. Hinton, Special Report N0.48, Dept. Scientific and Industrial Research Food
Investigation, London, 1939.
2. T.H. Schultz, in Methods in Carbohydrate Chemistry, ed. R.L.Whistler, Academic
Press, New York, 1965.
3. EEC, Council Directive 781663, Off: J. EEC, 14/08/1978.
4. Food Chemicals Codex, FCC I11 Monographs, National Academy Press, Washington
D.C., 1981,215-216.
5. FAO, Food and Nutrition Paper 52, Addendum 1, Rome, 1992,87-88.
6. R.F. McFeeters and S.A. Armstrong, Anal. Biochem., 1984,139,212-217.
7. A.G.J. Voragen, H.A. Schols and W. Pilnik, Food Hydrocoll., 1986, 1, 65-74.
8. R. Gnanasambandam and A. Proctor, Food Chem., 2000,68,327-332.
9. A. Ploger, J. Food. Sci., 1992,57, 1185-1186.
10. P.-E. Glahn and C. Rolin, in Gums and Stabilisers for the Food Industly 8, ed.
P.A.Williams and G.O. Philips, Oxford University Press, London, 1995, pp 393-402.
1 1. N. Blumenkrantz and G. Asboe-Hansen, Anal. Biochem., 1972,54,484-489.
12. H. Grasdalen, A.K. Andersen and B. Larsen, Carbohydr.Res., 1996,289, 105-1 14.
13. T.G. Neiss, H.N. Cheng, P.J.H. Daas and H.A. Schols, Macromol.Symp., 1999, 140,
165-178.
14. P.J.H. Daas, K. Meyer-Hansen, H.A. Schols, G.A. DeRuiter and A.G.J. Voragen,
Carbohydr. Res., 1999,318, 135-145.
15. P.J.H. Daas, A.G.J. Voragen and H.A. Schols, Carbohydr. Res., 2000,326, 120-129.
16. P.J.H. Daas, G.J.W.M. vanAlebeek, A.G.J. Voragen and H.A. Schols, in Gums and
Stabilisers for theFood Industry 10, ed. P.A.Williams and G.O. Philips, The Royal Society
of Chemistry, Cambridge, 2000, pp 3- 19.
17. G. Limberg, R. Korner, H.C. Buchholt, T.M.I.E. Chritensen, P. Roepstorff and J.D.
Mikkelsen, Carbohydr. Res., 2000,327,293-307.
18. G. Limberg, R. Korner, H.C. Buchholt, T.M.I.E. Chntensen, P. Roepstorff and J.D.
Mikkelsen, Carbohydr. Res., 2000,327,321-332.
19. T.P. Kravtchenko, M. Penci, A.G.J. Voragen and W. Pilnik, Carbohydr. PoZym., 1993,
20, 195-205.
20. E.M.W. Chen and A.J. Mort, Carbohydr. Polyrn., 1996,29, 129-136.
21. A.J. Mort, F. Qui and N.O. Maness, Carbohydr.Res., 1993,247,21-35.
22. R. Komer, G. Limberg, T.M.I.E. Christensen, J.D. Mikkelson and P. Roepstorff,
AnaLChem., 1999,71, 1421-1427.
23. P.J.H. Daas, B. Boxma, A.M.C.P. Hopman, A.G.J. Voragen and H.A. Schols,
Biopolymers, 2001, 58, 1-8.
24. P.J.H. Daas, A.G.J. Voragen and H.A. Schols, Biopolymers, 200 1,58, 195-2003.
25. M.A.K. Williams, G.M.C. Buffet, T.J. Foster and I.T. Norton, Carbohydr. Res., In
Press.
26. P.D. Grossman and J.C.Colburn, Capillary electrophoresis-theory and practice,
Academic Press, San Diego, 1992.
27. R. Weinberger, Practical capillary electrophoresis, Academic Press, San Diego, 1993.
28. C.A. Monnig and R.T. Kennedy, AnaLChem., 1994,66,286.
29. B.L. Karger, Y.H. Chu and F. Foret,. Ann.Rev.Biophys and Biomol.Structure, 1995, 24,
579.
30. A.G. Barron and H.W. Blanch, Separation and Purijcation Methods, 1995,24, 1 .
3 1 . Z. ElRassi, Adv. Chromatogr.,1994,34, 177.
32. J.D. Brewster and M.L. Fishman, J.Chromatogr.A, 1995,693,382.
33. J. Sudor and M. Novotny, Proc. Natl.Acad.Sci. U.S.A., 1993, 90, 945 1 .
34. M. Stefansson and M. Novotny, Carbohydr.Res., 1994,258, 1 .
35. M.D. Richmond and E.S.Yeung, AnaLBiochem., 1993,210,245.
36. H.-J. Zhong, M.A.K. Williams, R.D. Keenan, D.M. Goodall and C. Rollin, Carbohydr.
Polym., 1997,32,27-32.
37. H.-J. Zhong, M.A.K. Williams, D.M. Goodall and M.E. Hansen, Carbohydr.Res.,
1998,308, 1-8.
38. A.R. Volkel and J. Noolandi, J.Chem.Phys., 1995, 102(13), 5506-551 1 .
39. A.R. Volkel and J. Noolandi, Macromol., 1995,28,8182-8189.
40. D.M. Goodall, S.J. Williams and D.K. Lloyd, Trends AnaLChem., 1991, 10,272.
41. T.J-M. Deconinck, A. Ciza, G.M. Sinnaeve, J.T. Laloux, P. Thonart, Curbohydr. Res.,
2000,329,907-911.
APPLICATION OF TWO DIMENSIONAL (2D) NMR SPECTROSCOPY IN THE
STRUCTURAL ANALYSIS OF SELECTED POLYSACCHAFUDES
Steve W. Cui
Food Research Program, Agriculture and Agri-Food Canada, Guelph, Ontario,

Canada
ABSTRACT
The basic principles of 1D and 2D NMR spectroscopy and the most recent
application of 2D NMR in the structural analysis and sequencing of some native
polysaccharides are reviewed. Native polysaccharides discussed include wheat p-D
glucan, a rhamnogalacturonan from yellow mustard gum and a galactomannan from
fenugreek. Because of the rapid development in 2D NMR technology, an integrated
approach using 2D N M R correlations is becoming a powerful tool for elucidating the
complete structure and sequence of some simple native polysaccharides without the use
of chemical modification or methylation analysis.
1. Introduction
Elucidation of the structure of a polysaccharide requires the knowledge of
monosaccharide composition, configuration, linkage patterns and sequence of the
individual sugars. Molecular weight and molecular weight distribution are also important
properties of polysaccharides, which are beyond the scope of this review, therefore, will
not be discussed herein. In order to acquire information necessary to solve the structure
of a polysaccharide, traditionally we use chemical methods, such as monosaccharide
analysis, methylation analysis, peroxidatiodSmith degradation, partial hydrolysis and
specific enzyme hydrolysis. During the last two decades, modern analytical techniques,
e.g. double Fourier transformation and two-dimensional (2D) NMR spectroscopy (Aue et
al., 1976), have demonstrated enormous power in the determination of the structure of
polysaccharides.
2D NMR methods offer dramatic improvements not only in resolving structural
information in NMR spectra that was lost in one dimensional studies due to spectral
overlap, but also providing detailed linkage and sequencing information of sugar units
through different experimental techniques. In addition, complete structural information
including monosaccharide composition, configuration, linkage patterns and sequences
can be obtained fi-om the 2D NMR spectroscopy without breaking down of the
polysaccharides. However, because of the complexity and severe overlapping of the
NMR signals, 2D NMR spectroscopy has only been most successfully used in
determining the structures of monosaccharides, oligosaccharides and their conjugates,
such as glycolipids and glycopeptides, but not for native polysaccharides. During the past
few years, 2D NMR methods have been directly applied to the elucidation of the
structures of naturally occurred polysaccharides, either standing alone or in combination
with other analytical methods (Abeygunawardana and Bush, 1990; Cui, 2001; Cui et al.,
2000; Ramesh et al., 2001). In this paper, the author intends to review the principles, and
methodologies 2D NMR spectroscopy and their applications in elucidating the structure
of selected native polysaccharides.
2. NMR Principles and Methodology

2.1. NMR Principles
NMR spectroscopy is based on the magnetic property of nuclei in atoms, called
s in. All nuclei have a spin, however, only those nuclei with unpaired spins, such as 'H,
lgF, 13C, 31P and "N, are used in NMR studies,. In the structure analysis of
polysaccharides, 'H and 13Care most fiequently used.
Since atomic nuclei are associated with charge, a spinning nucleus generates a
small electric current and has a finite magnetic field associated with it. When a spinning
nucleus is immersed in a strong magnetic field, the nuclear magnet experiences a torque
which intend to align it with the external magnetic fields. There are two possible
orientations of the nucleus: parallel to the applied magnetic field (low energy) and against
the applied magnetic field (high energy). Since the parallel orientation is lower in energy,
this state is more populated than the anti-parallel state. At this stage it is possible to
transfer energy into the spin system by introducing radio frequency pulses that can cause
the nuclei to jump from the low energy level to the high energy level, as shown in Figure
1.
*
Radio Relaxation
+-- Frequency
+-&$&
a
f#-
b
* C
Bo
Figure 1. Illustration of energy status of nuclear spin in a magnetic field: (a),

equilibrium; (b), excitation by radio frequency and (c) relaxation to equilibrium and
release of signals.
After the excitation, the nuclei relax back to the state of equilibrium, sending a
weak signal that can be recorded. Because every nuclear spin in a molecule is affected by
the small magnetic fields of its nearest neighbors, it is possible to separate the signals
coming from different atomic surroundings. Based on this principle, the structure of the
molecule can be determined by interpreting these individual signals.
Structure, Churacterisationand Interactions 29
2.2.Chemical Shifts and Spin-Spin Coupling

2.2.1. Chemical Shift
As described in the previous section, nuclei in a molecule have small magnetic
fields that tend to oppose the applied field. As a result, the magnetic field at the nucleus
(effective field) is generally less than the applied field by a fraction of (T:
Because the electron density around each nucleus in a molecule varies according to
the types of nuclei, bonds in the molecule and the opposing field, the effective field at
each nucleus varies as well. This is called the chemical shift phenomenon. The resulting
shift in the NMR signal for a given nucleus is referred to as chemicaZ shift, and in
general, protons or carbons adjacent to electronegative atoms will be deshiezded and will
move to a hgher chemical shift (undergo transition at a lower applied field). In practice,
chemical shift of a nucleus is defined as the difference between the resonance frequency
of the nucleus and that of a standard, relative to the frequency of the standard; and this
standard is often tetramethylsilane or TMS (Si(CH&). The quantity is reported in ppm
and the chemical shift is given the symbol delta (6):
where v is the frequency of the sample, and V ~ isFthe frequency of the standard. The
factor of lo6is introduced into the equation to give a whole number scale for
convenience. The magnitude of the screening depends on the atom. For 'H NMR, the
scale generally extends from 0-12 ppm; the scale for 13Cnuclei, however, is much lar er
and covers the range fkom 0-220 ppm. This explains why there is less overlap in the C 'B
NMR spectrum than in the 'H spectrum.
2.2.2. Spin-Spincoupling
Nuclei experiencing the same chemical environment or chemical shift are called
equivalent. Those nuclei experiencing different environment or having different chemical
shifts are nonequivalent. Nuclei that are close to one another exert an influence on each
other's effective magnetic field. This effect shows up in the NMR spectrum when the
nuclei are nonequivalent. If the distance between non-equivalent nuclei is less than or
equal to three bond lengths, this effect is observable and it is called spin-spin coupling or
J coupling. In proton NMR spectrum, the spin-spin coupling arises from the interactions
of the proton magnetic field with bonding electrons. Since each proton can have two
possible spin orientations in the applied field, the magnetic field received by adjacent
protons can have two possible values. As a result, n protons will split adjacent protons
into n+l peaks. The intensity of these peaks is a result of the possible orientations. The
separation between these peaks is called the coupling constant (J, in Hz), and sets of
coupled protons will display exactly the same coupling constant. More details and
applications of spin-spin coupling can be find in related NMR text books.
The coupling constant in '3C NMR spectra is not typically observed due to the
low natural abundance of 13Cisotope (1.l%), however the attached protons will couple
with the 13Cnucleus to give spin-spin coupling. As a result, the signal from the carbon
30 Gums and Stabilisers for the Food Industry I1
will be split into n+l peaks, where n is the number of attached protons. In practice, all of
the split I3C peaks are reduced to sharp singlets by a proton noise-decoupling process, in
which the sample is irradiated at a second frequency to promote all the protons in the
molecules to the higher spin states to disallow the spin-spin coupling process. The
couplings between proton-proton (scalar coupling), proton-13C and I3C-l3Care the basic
foundation of 2D NMR spectroscopy to solve the structural problems of molecules.
2.3.Introduction of 2D N M R Spectroscopy
2.3.1. Principles of 2D NMR spectroscopy
2D NMR spectroscopy was first suggested conceptually by Jeener in 1971.
Unfortunately his work was not published. It was not until 1975 that the 2D NMR
concept was confirmed by experiments (Aue et al., 1976).
The difference between 2D NMR and traditional NMR is that the response intensity
is a function of two frequencies rather than a single frequency in the latter.
Experimentally, data is collected as a function of two independent time domains to give
time domain data set, which is subjected to a double Fourier transformation into two
frequency domains as shown in equation (3)
One-dimensional NMR experiments are limited to the portrayal of response intensity

as a function of the observation frequency. However, 2D NMR spectra have a second
frequency domain, which adding vast new information to the spectrum. The introduction
of the second domain provides the means of establishing correlations and hence
additional connectivity information that are very usehl in determining molecular
structure, particularly the structure of polysaccharides. In section 3, a number of
frequently used 2D methods for polysaccharides will be discussed.
2nd Pulse
Pulse andor time
Preparation Evolution Mixing Detection
Figure 2. A typical pulse sequence in 2D NMR experiments.

2.3.2 Procedures of 2 D NMR experiments

All two-dimensional NMR experiments contain four steps: preparation, evolution,
mixing period and detection.
Preparation serves the purpose of initially bringing the nuclear spins to a thermal
equilibrium or return the spins to equilibrium following an acquisition. The nuclear spins
need to be prepared to the same state to start a 2D experiment. The preparation period
is ended by the application of a pulse that perturbs the equilibrated spins to initiate the
NMR experiment.
Evolution is the period after the disruption of the equilibrium state created during
the preparation period. Nuclear spins will be allowed to precess in the x/y-plane of the
rotating frame. This precessional event is referred to as evolution, which is responsible
for providing the information encoded into the second frequency domain.
Mixing is introduced after the completion of the evolution period. At the end of
the evolution, there is an option to redistribute the nuclear magnetization among the
spins, and this distribution may involve the use of pulses and/or time period. To use or
not to use the mixing period will determine the way we interpret the interactions of the
nuclei.
Detection period in each 2D NMR experiment is identical; however, the
information detected will be different as a function of the changes introduced by the
evolution period.
3. 2 D NMR Spectroscopy and Polysaccharide Structural Analysis

3.1. Homonuclear autocorrelation: COSY and proton-proton connectivity
COSY stands for Correlation SpectroscopY. It is the first 2D NMR experiment and
the proton spectrum is correlated with itself via the scalar (J) couplings between
resonances. Contour plots are frequently used to present a COSY spectrum, as shown in
Fig. 3. The responses of the diagonal correspond to the resonances of a normal spectrum.
The usehl information is contained in the off-diagonal peaks, also called cross-peaks.
The spectrum is symmetrical about the diagonal, and the 2 triangular halves are mirror
images (Fig. 3). The off-diagonal responses actually show the correlation of resonances
that are associated with one another by scalar coupling. For example, in a partially
overlapped ABX spin system, the x spin (furthest downfield) is coupled to the B spin but
not to the A spin. The B spin is coupled, in turn, to the A spin (furthest upfield).
AB
Fig. 3. COSY contour plant schematic

showing an AE3X spin system. Diagonal
responses are the normal spectrum while
the off-diagonal shows the correlations
between two spin systems.
/I X B A
This autocorrelations can be used to determine the proton-proton connectivity in a

sugar ring, as shown in Fig. 4.
Fig. 4. Schematic shows the proton-proton connectivity that can be obtained

through COSY experiment. Based on the principles described in Fig. 3., H-1 correlates
with H-2; H-2, in turn, correlates to H-3. Using the same principle, the proton-proton
connectivity from H-1 to H-6 can be established.
The strategy for assigning the COSY spectrum is to find one unmistakably
characteristic signal from which a spin system or network can be traced. An anomeric
proton signal, marking one of the ends of a spin system, is a good example of such
system. The main advantage of COSY over traditional J-resolved experiment is the
disentanglement of the overlapping multiplets in NMR spectra, especially for
carbohydrates, the sugar proton signals are mostly crowded in the range of 3-5.5 ppm.
Fig. 5 is the COSY spectrum of (1 +3)( 1 +4) mix-linked p-D-glucan from wheat.
Following the above strategy for assigning the COSY spectrum, the unmistakable
characteristic resonances are the 3 anomeric protons at chemical shift 4.38 to 4.47 ppm,
which can be obtained from known knowledge and literature data. Based on the ABX
spin system (Fig. 3), we can assign the chemical shifts of the three H-Zs from the coupling
with their respective anomeric protons, and the three H-3s, which were derived from their
corresponding H-2s. Similarly, the H-6 AI3 systems and respective H-5s can be assigned,
but a clear assignment of the H-4s was unobtainable due to a high degree of overlap
between the H-3, 4 and 5 protons. Other experiments are needed in order to assign all the
resonances completely and unambiguously.
H-IH TOCSY (Total Correlated SpectroscopY) is useful for determining whch
signals arise from protons within one spin system, especially when the multiplets overlap
or there is extensive second order coupling. TOCSY correlates protons that are in the
same spin system (protons are never more than 2 or 3 bonds apart from another proton).
It yields long range as well as short range correlations and the cross peak intensity is not
an indicator of distance. TOCSY serves the purpose of verifying the assignments from
the COSY spectrum, and also establishes the coupling network within a sugar residue,
and frequently, it also provides further information to assign the ambiguous resonances
from the COSY spectrum. The TOCSY spectrum of the (1+3)( 1+4) mix-linked p-D-
glucan from wheat is shown in Fig. 6.
4.4 4.2 4.0 3.8 3.6 3.4 3.2 tmm
Fig. 5 . COSY spectrum of a (1+3)(1+4) mix-linked P-D-glucan from wheat (adapted

from Cui et al., 2000).
2
c
% Z
8
I s
A 4.4 4.2 4.0 3.8 3.6 3.4 3.2 ppm
Fig.6. TOCSY spectrum of a (1+3)( 1+4) mix-linked P-D-glucan from wheat. A,

B and C represent the glucosyl residues in Fig. 7.
The strategy of assigning a TOCSY spectrum is to draw a horizontal or vertical

line starting from the anomeric resonances, as shown in Fig. 6. All the resonances falling
on the same line are derived from the same sugar residue. Due to the similar values of the
3 J ~for, ~all protons of P-D-glucose (7 to 9 Hz around the hexose ring), polarization is
transferred completely from H-1 to H-6 in each residue, and the resolution of the anomeric
protons then permits the assignment of all protons within each of the three residues. From
the three lines in Fig. 6, it is found that the middle row is distinct from the others, whch was
assigned to the 0-3 substituted residue (B in Scheme l), whereas the other two were
attributed to the flanking 0-4 substituted residues (A and C).
A B C A B C D A B C
+ 4G1+ [5G1+4G1+ 4G1+ ]J5Gl+ 4G1+ 4G1+ 4G1+ ]n)G1+ 4G1+
Fig.7. A model structure of a (1+3)(1+4) mix-linked P-D-glucan from wheat
(Adapted from Cui et al., 2000).
3.2
a-4
a
B-3 3-6
9.B
I
1.0
4.a
Fig. 8. 'H/I3C correlation of a (1+3)( 1+4) mix-linked P-D-glucan from wheat (Adapted
3.2. Heteronuclear chemical shift corrolation- H/C connectivity

As described earlier, the identification of the relative position of protons to one
another in a molecular structure helps to define the structure, the same can be said for the
position of a given proton relative to its corresponding carbon. Heteronuclear chemical
shift corrolation ('W13C)spectroscopy allows a one to one match of the protons to the
corresponding carbons in a molecule. Fig. 8 is an example of a 'W13C correlation
spectrum of of a (1+3)(1+4) mix-linked P-D-glucan from wheat was achieved. The 1D
spectra of this polysaccharide (projected on the x,y axis) appeared crowded. Much clearer
signals can be identified from the 2D spectrum, enabling a one to one match of each of
the protons and their corresponding carbon signals. The method of assigning a 'H/13C
heteronuclear correlated spectrum is to draw a horizontal and a vertical line from any
resonance of interest, to the x, y axis. The knowledge of one assignment ('H or 13C)will
easily lead to the assignment of the corresponding other signals. A complete assignment
of the 'H and I3C resonances of a (1+3)(1+4) mix-linked P-D-glucan from wheat was
achieved (Cui et al., 2000). The signals labeled on the spectrum (Fig. 8) only shows the
assignments of the signals of Residue B in Fig. 7.
3.3.Relayed coherence transfer and related 2D-NMR experiments-residue

connectivity and sequence
3.3.1. NOESY
Another class of experiments, know collectively as nuclear Overhauser effect (NOE),
provide information on through-space rather than through-bond couplings. Such
experiments are very important in the use of NMR to establish tertiary and quaternary
structures. NOESY is one of the most useful techniques as it allows to correlate nuclei
through space via cross-relaxation effects between protons, observed only for pairs of
protons that are separated by less than about 5A,regardless of covalent structure. This
information is complementary to that obtained from spin-spin coupling, and can
distinguish between stereo-isomers. In a NOESY contour plot, a normal 'H s ectrum is
shown along the edges. The resonances along the diagonal match the normal H P
spectrum. A cross-peak "connects" 2 resonances, indicating that an NOE exists between
them.
There are numerous NOESY peaks which are uninteresting, because they arise
from the covalent structure and provide no information about stereochemistry, isomers or
conformation. The first step in analyzing a NOESY spectrum is to eliminate these
uninteresting resonances; the uninteresting resonsnces can be identified by comparing the
NOESY against COSY spectrum. Fig. 9. is an example of NOSEY of a
rhamnogalacturonan from yellow mustard gum. Signals lebaled with arrows were
identified as glycosidic linkage signals. For more detailed assignment of the spectrum,
please refer to reference (Cui et al., 1996).
3.3.2. Heteronuclear Multiple Bond Connectivity (HMBC)

The HMBC experiment detects long range coupling between proton and carbon
(two or three bonds away) with great sensitivity. This technique is very valuable to detect
indirectly quaternary carbons coupled to protons - specially useful if direct Carbon-13 is
impossible to obtain due to low amount of material available. It is also very useful as a
sequence analysis tool that provides unique information concerning connectivity across
PCQ-
1 5 0
. -I,r,
4 5
,
4 3
. . . ,
3 5
-
Fig. 9. NOESY spectrum of a rhamnogalacturonan from yellow mustard gum (Adpted
L i
~
pm
. -
5.7
~ - -
5
- w - *
c 3
--r-.,
,
4 4
, .
4 8
. 4 2 40 jn
.
IE
Y .
Fig. 10. H-I3C HMBC spectrum of a fenugreek galactomannan (Adapted from Ramesh
et al., 2001).
Structure, Churacterisation and Interactions 37
glycosidic linkages. For example, Fig. 10 shows the HMBC spectrum used for
sequencing a galactomannan from fenugreek gum (Ramesh et al., 2001). The relevant
long range C-H connectivities of the constituent galactosyl and mannosyl residues are
labeled on the graph: the H-1 (gal) at 6 5.02 is coupled with the I3C resonance at 6 68.00
(C-6 of @-man);the H-4 at 6 3.87 is coupled with the I3C resonance at 6101.8 (C-1 of @-
man). This glyconsidic linkage is also supported by another cross peak at 6 3.90 (H-6 of
@-man)which is coupled with the 13Cresonance at 6 100.2 (C-1 of a-gal). By
summarizing these results, the inter-residue linkages, @ - m a - (1+ 4)-@-manand a-gal-
(1+ 6)-@-manare established.
60
70
00
90
A1 ('H) to B3 (I3C)
lm PPM
100
Fig. 11. . 'H-I3C HMBC spectrum of a ( l + 3)(1+ 4) mix-linked P-D-glucan from wheat.
A, B and C are glucosyl residues in Fig. 7. (Adapted from Cui et al., 2000).
Information to resolve ambiguities in 13Cassignments and to establish the sequence

of the three P-D-glucosylresidues of wheat @-D-glucanwas obtained from analysis of the
long-range heteronuclear correlation spectrum (HMBC), as shown in Fig 11. The important
correlations arise from 3J~,c,cand from 3 J ~ , ~Correlations
,c. of each of the anomeric
resonances at 4.47,4.42and 4.38 ppm to the carbon atoms across the glycosidic linkage (C-
3 at 86.4 and the two C-4s at 79.5 and 79.3 ppm, respectively) define which anomeric
proton can be assigned to structures A and C (Fig. 7). The anomeric proton at 4.47 is
assigned to H-1 of the 0-4 substituted P-D-glucosyl residue (A) which is glycosidically
linked to the 3-position of residue B and attached via 0-4 to another 0-4 substituted residue.
Similarly, the anomeric proton at 4.38 pprn is assigned to H-1 of the 0-4 substituted P-D-
glucosyl residue(C) which is linked through 0-4 to C-1 of the 0-3 substituted P-D-glucosyl
residue (B) (Fig. 7). The concomitant carbon to proton correlations in the reverse direction
(i.e. 103.3 ppm (C-1 of A) to 3.46 ppm (H-3 of B) and C-1 of B and C to H-4 of C and A),
are also observed, but are not as usehl due to overlap of the H-4s from A and C. This, along
with the additional intraresidue correlations observed between the resolved H-2s of each
residue and their respective C- 1 and C-3 resonances, serves to confirm the assignments
within each glucosyl residue, and resolves all ambiguities between the C-4 assignments for
A and C.
Summary
2D NMR spectroscopy is a rapid developing technology and has become a

powerfwl tool in the structural analysis and sequencing of polysaccharides. For some
simple polysaccharide structures, 2D NMR technique provides all the necessary
information to elucidate the structure and the sequence of the glycosidic linkages without
the use of traditional chemical modification, e.g. methylation analysis. Homonuclear
correlations (COSY and TOCSY) help establish the scalar connectivity of a sugar residue
while long range correlations (NOESY and HMBC) are very useful in establishing the
sequence of the individual sugar residues on the polysaccharide chain. There are other 2D
NMR methods not mentioned in the present paper but can be very useful in analyzing the
structure of polysaccharides. Nevertheless, 2D NMR spectroscopy has its limitations: it
can only be applied to simple polysaccharides at the moment. For larger molecular
weight polysaccharides, solubility of the sample in a proper solvent is frequently
encountered problem. To solve the structure of more complex polysaccharides,
combination of partial hydrolysis, methylation analysis and integrated approaches using
2D NMR spectroscopy could be very useful.
References:
Abeygunawardana, C. and Bush, C.A. 1990. Biochem. 29:234-248.

Aue, W. P., Bartholid, E. and Emst, R. R. 1976. J. Chem Phys., 64,2229
Cui, W. 2001. Polysaccharide gums from agncultural products: processing, structures and
functionality. Technomic Publishing Co. Inc. Lancaster, PA, USA.
Cui, W., Eskin, N.A.M., Biliaderis, C.G. and Marat, K. 1996. Carbohydr. Res., 292:173-
183.
Cui, W. Wood, P.J., Blackwell, B. and Nikiforuk, J. 2000. Carbohydr. Polym., 41:249-
258.
Ramesh, H.P., Yamaki, K., Ono, H and Tsushida, T. 2001. Carbohydr. Polym., 4559-77.
a AND p MECHANICAL RELAXATIONS IN HIGH SUGAR BIOPOLYMER
MIXTURES
Stefan Kasapis and Insaf M. A. Al-Marhoobi
Department of Food Science & Nutrition, College of Agriculture, Sultan Qaboos

University, PO Box 34, Al-Khod 123, Sultanate of Oman
1 ABSTRACT
In the present work a case is made for correlating mechanical dispersion phenomena to
identifiable segments of the polymer in the presence of a high sugar environment. The
argument was, in the main, based on the finding that at least two distinct dispersions could
be found in polymers. One dispersion reflects a process corresponding to the glass
transition or a relaxation at which the main polymer chain is thought to acquire
considerable mobility. In this temperature region, the viscoelastic nature of the material
gives rise to a predominantly viscous behaviour. Recently, we observed the second process
or p transition occurring at a somewhat lower temperature than the a transition. This is
attributed to the mobility of side chains with possible localised interactions with
immediately adjacent parts of the main chain. These characteristic features of
viscoelasticity were modeled using the synthetic polymer approach as follows: Time-
temperature superpositions of mechanical spectra yielded master curves whose fitting with
the Williams Landel and Ferry equation predicted the glass transition temperature (Tg). It is
proposed that the rheological T is the point between the glass transition region and the
glassy state. It acquires physical significance by identifying the transformation from free-
volume phenomena of the polymeric backbone in the glass transition region to an energetic
barrier to motions in the glassy state involving stretching and bending of chemical bonds.
Based on this definition, small additions of polysaccharide accelerate the vitrification
properties of sugar with the mixture undergoing vitrification as a whole.
a Transitions were characterised by i) following them as a function of time alone and a
function of temperature alone, ii) determining the monomeric friction coefficient in
relation to the molecular weight distribution of the polymer and iii) engineering a state of
iso-free-volume that could predict the development of viscoelasticity regardless of the
physicochemical characteristics of polymers. Secondary dispersions were observed at
temperatures just below the main softening region of a glassy material made of high acyl
gellan and glucose syrup. They are associated with rearrangements of the glycerate,
positioned at O(2) and acetate at O(6) of the 3-linked glucose molecule of the
polysaccharide. Approximate heats of activation were calculated according to the Andrade
equation on the assumption that this dispersion is a rate process. The process is absent in
the deacylated gellan where the expectation is that mechanical properties in the glass
transition are determined by the polymeric backbone. The mechanical response of the p
transition remains predominantly elastic and appears between the glass transition region
and the glassy state, a result which calls for modification of the master curve of
viscoelasticity.
2 INTRODUCTION
Previously, we reported on the application of the WLFIfree volume theory to high solids
systems (60 to 87% wlw) containing macromolecular ingredients at normal levels of
industrial use (e.g. up to 1% polysaccharide)'. These preparations are recently enjoying
increasing attention from researchers and product developers alike in the food and
pharmaceutical industries and there is a need to develop definitive theoretical and empirical
relationships. The rheological glass transition temperature was pinpointed by means of
measurements of shear moduli within the linear viscoelastic region. The technique of
dynamic oscillation was used capable of resolving the structural properties of materials into
a solid and a liquid-like response (GI and G", respectively)2. From these basic parameters,
tan 6, the loss tangent is defined to be the ratio of G" to G . Composite curves were
constructed by empirical shifts of data using the method of reduced variables or the time-
temperature superposition principle (TTS)3. In doing so, mechanical spectra were obtained
over a wide temperature range of up to a hundred degrees centigrade, which reflects the
glass transition region. Good matching of the shape of adjacent curves without
irregularities was observed, which allowed use of the same extent of shifting in the
superposition of both G' and G" data4. The approach is based on the assumption that during
a change in state all relaxation times of a molecular process depend identically on
temperature. Thus in the absence of a first order thermodynamic transition (e.g.
development of crystallinity), shear moduli measured at frequency o and temperature TI
are equivalent to those measured at frequency and temperature To. The shift factor, a T ,
is estimated by the following expression5:
where po is the density of the sample at the reference temperature, To. Equation (1) makes
the shift factor a fundamental descriptor of the temperature dependence of viscoelastic
functions. Then, we examined the applicability of the WLF equation to this pattern of
viscoelastic behaviou? :
In terms of the free volume theory, the parameters Cp and Cf correspond to B12.303fo
andLlaf; respectively, and B for simplicity is taken equal to unity. Equation (2) allows
calculation of the fractional free volume Cr,) and the thermal expansion coefficient, ah at
Tg.In doing so, the set of shift factors for each preparation was plotted using the linearised
form of equation (2) thus calculating Cp and Cf from the gradient and the intercept of the
linear fit, respectively. Gratifyingly, the WLF equation represents a good fit of the factor aT
as a function of temperature in the glass transition region, which allows prediction of the
value of rheological Tg for high sugarhiopolymer mixtures. These predictions are
congruent with the passage from the glass transition region to the glassy state in the
composite curve of the mechanical spectra and as recorded in the overlapping
coolingheating traces of G'and G"for the same samples. However, the equation is unable
to follow progress in mechanical properties at the lower temperatures of the glassy state.
These are better described by the mathematical expression of Andrade7:
Ea 1 1
log aT = (- - -) (3)
2.303R T T o
This yields the concept of activation energy (Ea) for an elementary flow process in the
glassy state, which is independent of temperature. In contrast, equation (2) predicts
increasing values of E, with decreasing temperature. It seems, therefore, that the glass
transition temperature is a true turning point where large configurational vibrations
requiring free volume cease to be of overriding importance. Instead, the emerging constant
energy of activation reflects the need to overcome an energetic barrier for the occurrence of
local rearrangements from one state to the other. The phenomenon appears to be universal
thus encouraging us to introduce the concept of the rheological glass transition temperature
as an index of physical significance standing at the threshold of two distinct molecular
mechanisms, namely: the transformation from free-volume derived effects in the glass
transition region to the process of an energetic barrier to rotation in the glassy state819.
3 DISTRIBUTION FUNCTIONS AND THE FRICTION COEFFICIENT
The composite curves of time or frequency provide a preliminary examination of the

pattern of mechanical relaxation processes in high sugarhiopolymer mixtures. This
relationship, however, is most conveniently expressed with the distribution function of
relaxation times, @, because it can be derived from both dynamic and transient
measurements". Furthermore, the fbnction connects together the values of G and G"
although they are independent at a single frequency of measurement. Storage and loss
modulus data is used in the first approximation calculation of @ in synthetic materials but,
in general, the treatment fails to harmonize the derived curves, especially at both ends of
the experimental window of observation. Similarly, inconsistencies were observed in the
'.
estimation of @ for our samples' The need to minimize the discrepancy between the two
predictions of the shear moduli led to the development of second a proximation methods.
An elegant account of the approach is given by Williams and Ferry' .P
Briefly, the refinement is based on the observation that the @curves rise at short times
with slopes being a simple power function of T. Thus the analytical form @ = kz-m is
assumed and substituted in the 'Maxwellian equations' of shear modulus to give
expressions which include gamma functions of m. The latter are used in the equations of
the first approximation to obtain:
d log G'
@T=l/o=AG(l- I -- 1 I>
d logo
d log G"
= l/o
QjT = BG"( 1 - I- I)
d logo
The correction factors A and B are introduced in order to eradicate the error emanating
fi-om the application of the first approximation expressions to the experimental data thus
allowing equations (4a) and (4b) to comply with the experimentally verified form @ = kT-m.
The adjustment is as follows:
Dynamic data can also be used to calculate the mechanical retardation distribution, i.e. the
fbnction L d lnz which is defined as the contribution to shear compliance associated with
retardation times whose logarithms lie between In z and In z + d In z. In doing so, the real
and imaginary parts of the complex compliance J' and J" were calculated. Although J* =
UG*, their individual components are not reciprocally related, but are connected by the
following equation~'~:
In a manner analogous to the derivation of the second approximation calculations for the
relaxation function, equations can be constructed for the shear compliancet2:
d log J'
L , = I /=AJ'(l
~ 1+----I--- I)
d logw
d log J"
L,=t/,=BJ"(l - I ~ 1)
d logo
The factors A and B remain the same as those defined in equations (5a) and (5b), but m is
the positive slope of the curve of log L , obtained from a first approximation calculation, vs.
log z.
Figure 1 depicts the outcome of such calculations for the gelatin/co-solute system. In
doing so, the slope m was taken at various points of the curves of the first approximation
calculation and used to calculate the gamma function. This, in turn, led to the estimation of
the numerical factors A and B which were employed to shift the @ curves to the second
approximation through equations (4) and (7). Within the glass transition region, the
predictions of @ from G' and G" and L from J' and J" are brought into a very good
agreement. At extremely short relaxation times the values of @ descend rapidly since the
change in modulus or in the rheological terminology 'the contribution to instantaneous
rigidity' between In z and In T + d In z is negligible. The approach is less successful to
normalise modulus readings at the onset of the rubbery region where the slope of the 0
curve deviates from being a simple power hnction of m. According to the constitutive
equation of G', its contribution to relaxation distributions is substantial at long times of
measurement making this calculation preferable in the rubbery regionI3.
Estimation of the glass transition temperature and the timekemperature function paves
the way for correlating the measurable viscoelastic parameters with molecular
characteristics. This has been a major goal in the physics of synthetic melts and glasses,
and is based on the observation that viscoelastic quantities are governed by the dimensions
of a macromolecule and the density dependent friction f a c t ~ r ' ~For
. meaningful
9.5 -5.5
8.5 -6.5
7.5 -7.5
2 -
CrJ
E:
.CI
p1
8 6.5 -8.5 2
M
4
M
3
5.5
s
-9.5
4.5 -10.5
3.5 - 1 1.5
Figure 1 Mechanical relaxation (0) and retardation ( L ) distributions as a function of

timescale of observation for a sample comprising 25% PS4 with 40% sucrose and 15%
glucose syrup (reference temperature: - 13C).
b
E
$ 8
l
d
a
.-c
8
M
4 7
6
-10 -8 -6 -4 -2 0 2 4 6 8
Figure 2 Shapes of relaxation spectra for the four gelatin fractions (25% of PS1 to PS4) in
mixture with co-solute (40% sucrose plus 15% glucose syrup) at the reference temperature
of -13C. Dashed line gives the slope predicted by the Rouse theory (The Mnof gelatin
samples is: PS1- 68 kD; PS2 - 55.8 kD; PS3 - 39.8 kD; PS4 - 29.2 kD).
comparisons between different molecular fractions, the frictional resistance per monomer
unit encountered by a chain segment is considered which constitutes the monomeric
c0.
friction coefficient, The resistance to segment relaxation is enhanced by local and long-
range forces in the form of intrachain and interchain physical bonds, and topological
entanglements. This increases the magnitude of co,
which is calculated by the relaxation
spectrum as fo~lows'~:
log c0= 2 log @+ log z + log (6/kT)+ 2 log (27cM0/apN0)

where k is the Boltzmann's constant, Mo the molecular weight of the monomer, p the
density, No the Avogadro number and a the root-mean-square end-to-end length per square
root of the number of monomer units16.
Figure 2 illustrates the application of this approach to the relaxation functions of four
gelatin fractions of different molecular weight. Reduction in the macromolecular
dimensions of the protein shifts the position of the transition zone to short time scale,
which reflects the absolute magnitude of the relaxation times and therefore that of the
monomeric friction coefficient. According to the Rouse theory, the fairly long-range
cooperative motions of the chain backbone are responsible for free-volume phenomena,
and their relaxation spectrum is linear on a logarithmic plot with a gradient of -%". As in
the case of synthetic polymers, experimental curves of gelatin approach the theoretical
slope of 4near the bottom where 0 is close to lo6 in Pa and the cooperative motions are
sufficiently long-range to conform with the conditions of the theory. The friction
coefficient is then calculated by substituting @ and z at this point into equation (8).
In doing so, the molecular weight of a peptide residue in gelatin was averaged to -
105, and a was taken equal to 3.33 A considering that the persistence length of the protein
is around 20 A which represents approximately two tripeptide units'*. Comparison is
meaningless without reduction to some sort of corresponding state and for this purpose the
relaxation spectra were normalised at - 13"C, a temperature which is close to the Tgof PS 1
but lies increasingly closer to the rubbery region of the gelatin materials with reduced
molecular weight (Table I). The density (p) of the gelatidco-solute mixtures was estimated
at the reference temperature and ranged from 1.68 to 1.54 g/ml for the PS1 and PS4
fractions, respectively. Table I reproduces the outcome of such calculations for the
monomeric fiction coefficient at - 13C. Evidently, the magnitude of tends to increase
dramatically with M,, which, qualitatively, can be interpreted by considering that
translation of a chain segment is hindered with increasing magnitude of intermolecular
forces and entanglements of long chains.
The effect of molecular weight on relaxation times can be uantified by considering
that the specific volume of a polymer is a linear function of l/Mnl'. It is then suggested that
the free volume should follow the same relation. However, an allowance is made for
additional free volume with decreasing molecular weight owing to imperfect packing
around the end of the macromolecule. Thus at a constant temperature, e.g. - 13C in our
study, the following relation is applicable:
wheref,, refers to the fractional free volume at infinite molecular weight. As shown in
Figure 3, a good linear relationship is obtained between the fractional free volume and the
reciprocal molecular weight of the gelatin fractions at - 13C. For gelatin samples with
non-uniform molecular weight, M, is the proper average because its reciprocal is
Table I
Parameters characterizing the temperature dependence of shift factors, and the monomeric
friction coefficients o f the four gelatin fractions.
To (OC) CP C ; (deg.) Tg (OC) Cf Cq (deg.) fg af log TO*

Samples
(deg-' x (dyne-sec/cm)
PS1 -2 9.68 62.3 -14.5 12.06 50 0.036 7.2 6.91
PS2 -15 11.29 62.0 -27.0 14.0 50 0.03 1 6.2 3.93
PS3 -22 10.61 62.0 -34.0 13.16 50 0.033 7.2 1.66
PS4 -25 9.77 63.5 -39.0 12.41 50 0.035 7.0 -0.02
*co was estimated at -13C for the four gelatin fractions

0.06
0.05
0.04
0.03
0.02
0 10 20 30 40
lmn,x lo6
Figure 3 Fractional free volume plotted against 1/M, at the reference temperature of
-1 3C for the four gelatin fractions (PS 1 to PS4) in mixture with co-solute.
sugar A
-
A 0
A 0
A gellan+Ca2+
K-carrag. A
-40 -20 0 20 40 60 80
Temperature ("C)
Figure 4 Temperature variation of shift factors normalised at 0C for polysaccharidei

co-solute, gelatidco-solute and single co-solute systems shown in Table 11.
proportional to the number of chain ends. Alp is the additional molar free volume
associated with a mole of pairs of gelatin molecular ends. The ratio of NpMn becomes
negligible at infinite molecular weight and for the gelatin fractionsf,, acquires the value of
0.025. Therefore the free volume attributed to the bulk of the molecule in close vicinity to
'T is 2.5% of the total volume, a result that is consistent with estimates for the synthetic
systems2'.
4 ISO-FREE-VOLUME STATES IN HIGH SUGARBIOPOLYMER SYSTEMS
In parallel with the gelatin work, composite curves were constructed by empirical shifts of
data using the method of reduced variables for several high sugar/polysaccharide mixtures
and single sugar preparations. The shift factors used in this reduction are plotted against
temperature in Figure 4. The composite curves produced at various reference temperatures
were nonnalised at 0C (Trio, in Table I1 which reproduces the systems analysed). T,,,
represents different states of free volume for each preparation as shown by the values of
horn in Table 11. Thus, curves of log aT against T for the samples cross at the point Trio, =
O'C, log aT = 0 with different slopes.
Table I1
Reference, glass transition, normalised and iso-free-volume temperatures, and their

fractional free volumes for the polysaccharide/co-solute, gelatinko-solute and single
co-solute systems.
Samples
Gellan+Ca2+ 90 0.093 -26 0.029 0 0.043 -10.9 0.0365
Agarose -22 0.042 -4 1 0.030 0 0.0545 -30.2 0.0365
K-Carrageenan 60 0.075 -7 0.032 0 0.0365 0 0.0365
Pectin DE 22 -21 0.039 -34 0.031 0 0.052 -25.1 0.0365
PectinDE 92 0 0.042 -12 0.034 0 0.042 -8.3 0.0365
Gelatin (PS1) -2 0.045 -14.5 0.036 0 0.0465 -13.6 0.0365
Gelatin (PS2) -15 0.0385 -27 0.031 0 0.048 -18.2 0.0365
Gelatin (PS3) -22 0.041 -34 0.033 0 0.0555 -28.7 0.0365
Gelatin (PS4) -25 0.0445 -39 0.035 0 0.062 -36.3 0.0365
sugar -23 0.038 -36 0.030 0 0.05 15 -25.2 0.0365

48 Gums and Stabilisers for the Food Industry 1 I
The universal application of the WLF equation to amorphous synthetic

macromolecules2prompted us to surmise that the disparate dependence of shift factors on
temperature for the samples in Figure 4 should be due to individual patterns of vitrification
with cooling which reflect the state of free volume in the system. To evaluate this
hypothesis, we chose the most efficient vitrifier, K-carrageenan, as the reference withf,,,
being equal to 0.365 at Tnom.For the remaining samples, a temperature was identified to
correspond to the same value of fractional free volume. The approach yields a new set of
reference temperatures, which represent iso-free-volume states and are given as Tsofvin
Table 11. Then, shift factors were calculated via the WLF equation for the iso-free-volume
temperatures. When log aT referred to each individual zsOfvis plotted against T - Ts0h,
where T is the experimental temperature of the mechanical spectra, a single composite
curve is obtained for all the samples shown in Figure 5.
It is gratifying to find that in biological preparations with distinct chemical and
conformational properties, molecular weight and ability to behave as gelling or thickening
agents22,the temperature dependence of viscoelastic rate processes can be described by a
single parameter TsOfv which in turn is related to free volume. The solid curve in Figure 5
corresponds to the WLF equation in the form:
in which the basic reference temperature for the K-carrageendco-solute sample is 0C.
For each of the samples,Jso) = 0.0365 at its own Tiso). The fractional free volumes of the
samples can all be compared at 0C by calculating as follows:
where af is the relevant thermal expansion coefficient, and these are given as fnorm in Table
11. In fitting these data, the three high-temperature points in Figure 5, which correspond to
states far above the vitrification points of the materials, do not follow the temperature
progression of shift factors predicted by the WLF fit. However, it is perhaps the most
convincing success of the free-volume interpretation of out data that the factor UT does not
turn out to be of the correct magnitude for the three points, since at the rubbery plateau
viscoelastic behaviour reveals distinct contributions depending on specific chemical
features.
5 p MECHANICAL DISPERSION PHENOMENA

Besides the glass transitions (amechanism), specific mechanical dispersions known as p
transitions have been documented for the rapid, uncorrelated, rotational motions of
polymeric segments of amorphous synthetic polymers, like the methacrylate As
far as we are aware, these have not been detected in high sugarhiopolymer mixtures and
the present section is in pursuit of the elusive p transition. This is demonstrated in Figure 6
for high acyl gellan in mixture with glucose syrup. Following the synthetic polymer
approach, a series of mechanical spectra were taken at constant temperature intervals of
three degrees covering the accessible frequency range of 0.1 to 100 rads. For data
reduction to a standard temperature, we chose -1C which lies well within the glass
transition region. Remaining mechanical spectra were shifted left and right of this
reference point along the log frequency axis.
Stmcture, Characterisation and Znteructions 49
-30 -10 10 30 50 70 90 110
T 'Tisoh
Figure 5 Shift factors of dynamic oscillatory mechanical spectra for the samples of Figure
4 reduced to their own zsofv
given in Table I1 (the line reflects the WLF fit).
2.0
10
9 1.5
Lo
1.o 4
E-
M
s
6
0.5
4 0.0
-5 -2 1 4 7 10
Log O U T
Figure 6 Real and imaginary parts of the complex modulus and tan 6 for 1% high acyl
geIlan plus 79% glucose syrup reduced to -1C and plotted logarithmically against reduced
frequency. Part of the rubbery plateau (11) is exhibited, the a dispersion (111), the glassy
state (IV) and in the between the p dispersion.
As shown in Figure 6, this superposition is closely fulfilled by both G' and G" within
the temperature range of 8 to - 40C. Thermal scans extended to 70C but the method of
reduced variables in its basic form as given above proved to be inapplicable at the upper
range of temperature. Master curves could be drawn by subjecting the data to an additional
vertical shift but the validity of this undertaking is rather tenuous. This failure was
attributed to the partial melting of the junction zones of the polysaccharide network at high
temperature^^^'^^.
In part I1 of Figure 6 , the plateau zone is unveiled where the gel-like character is
intensified with decreasing frequency hence showing increasing separation of the G' and
G" traces and reduced variation. This should be due to the minimum contribution of
configurational rearrangements between junction zones (short) and beyond junction zones
(long) of the high acyl gellan network to relaxation processes. The entirety of the glass
transition is captured in part I11 where the viscoelastic behaviour is dominated by
configurational rearrangements of regions of the polysaccharide backbone that are shorter
than the distance between junction zones. The temperature/frequency dependence of
relaxation processes associated with the chain backbone motions is known as the primary
mechanism and it is symbolised by d7.The characteristic feature of the a transition is
dissipation of energy seen in the values of tan 6 exceeding one in Figure 6.
Elastic rather than a viscous character is developed at the upper range of reduced
frequency (part IV in Figure 6 ) . This is the glassy state with whatever slight motions being
accomplished involving bending and stretching of chemical bonds28.Meanwhile the values
of loss tangent drop to well below one. A specific mechanical mechanism is observed at
frequencies just above the main softening region of our glassy material. This results in a
discontinuity in the downward slope of the tan 6 trace as it runs from part 111 to part IV in
Figure 6 . The motions of this region do not couple with motions of the main chain
backbone of the high acyl gellan and should reflect a secondary dispersion involving side
chain motions. The p transition in our system could be associated with rearrangements of
the glycerate, positioned at O(2) and acetate at O(6) of the 3-linked glucose molecule of the
polysaccharide. Certainly, the secondary mechanism has not been observed in the
mechanical properties of linear biopolymers, including the deacylated counterpart, which
follow the transformation from rubber-like to glass-like con~istency~'~~'~'.
As shown in Figure 6 , the glass transition region was clearly discernible thus allowing
logarithmic plotting of LZT versus T. The empirical a T values followed closely the
predictions of the WLF equation at the reference temperature (To) of -1C (Figure 7).
Fitting of the shift factors to the WLF/free volume framework can be achieved by plotting
l/log aT against 1/(T- To)and obtaining the two parameters Cp and C; from the slope and
intercept of the linear fit, respectively. These were 9.7 and 75 deg, respectively, withf,
being equal to 0.045. Experimental evidence and WLF modelling argue that the Tg of the
chain backbone motions of the gellan polysaccharide is about -26C. The values of Clp
and CZg appropriate to Tg as the reference temperature in equation (2) were found to be
14.5 and 50 deg, respectively. Thus the fractional free volume (fg = 0.030) and the thermal
expansion coefficient (a1= 6.0 x 10-4deg-') at Tg can also be calculated. Below Tg (-26"C),
long-range configurational changes of the gellan backbone take place exceedingly slowly.
Rapid viscoelastic responses dominate which develop an elastic rather than a viscous
character and cover a wide spectrum of relaxation times seen for the glassy state in Figure
6. Empirical shifting of data to construct the composite spectrum of the glassy state was
described using equation (3). As illustrated in Figure 8, this yields the concept of activation
energy (En& for an elementary flow process, which is independent of temperature.
1.5
1.1
0.7
+ 0.3
CI
4p
c, -0.1
-0.5
-0.9
-1.3
-13 -9 -5 -1 3 7 11
Temperature (C)
Figure 7 Reduction factors CZTcalculated from the horizontal superposition of mechanical

spectra within the glass transition region for 1% high acyl gellan plus 79% glucose syrup.
The solid curve represents the predictions of the Williams - Landel - Ferry equation.
Reference temperature is -1C.
3.5
3.0
2.5
2.0
1.5
1.o
0.5
0.0
0 0.5 1 1.5 2 2.5
[(1/r)-(1/r0)]x104
Figure 8 Reduction factors aT calculated fiom the horizontal superposition of mechanical
spectra within the /3 transition region and the glassy state for 1% high acyl gellan plus 79%
glucose syrup. The solid lines are the Andrade fits. Reference temperatures are -13C and
-28C for the p transition region and the glassy state, respectively.
52 Gums and Stabilisersfor the Food Industry 11
The zero value of log aT refers to the arbitrarily chosen reference temperature (-28C)
for data reduction within the glassy state of the high acyl gelldco-solute mixture. The
factor aT is an exponential hnction of the reciprocal absolute temperature, so the
logarithmic form with a constant energy of activation can be used for calculating a
numerical value. This was found to be in the order of 274 kJ/mole.The marked anomaly
in the master curve between the glass transition region and the glassy state also produced
reduced variables of a constant energy of activation; p transition of Eap = 194 kJ/mole in
Figure 8. This leads confidence to the postulate that the p transition reflects the need to
overcome an energetic barrier to rotation from one state to the other of the acyl substituents
of the gellan chain. However, the exact nature of the side chain motions, which cause the
viscoelastic response of the p mechanism is still somewhat obscure in synthetic polymer
Based on our finding that the p dispersion is also a rate process, the increased energy
of activation in the glassy state ( E a g > Eap) should be due to the difficulty for local
rearrangements to occur in a medium of extremely high rigidity (the limiting high
frequency modulus, Gm, was found to be Pa in Figure 6). The apparent activation
energies of the p mechanism and glassy state are two-to-three times higher than the
corresponding values of the methacrylate series2, a result which underlines the increased
steric hindrance of glycosidic-bond and side-chain motions of the gellan polysaccharide.
6 REFERENCES
1. S. Kasapis and I.M.A. Al-Marhoobi, Gums and Stabilisers for the Food Industry lo,
The Royal Society of Chemistry, Cambridge, 2000, p. 303.
2. R.K. Richardson and S. Kasapis, Instrumental Methods in Food and Beverage
Analysis, Elsevier, Amsterdam, 1998, p. 1.
3. A.V. Tobolsky, J. Appl. Phys., 1956,27,673.
4. S. Kasapis, Functional Properties of Food Macromolecules, Aspen Publishers,
Gaithersburg, 1998, p. 227.
5. M.C. Shen and A. Eisenberg, Progress in Solid State Chemistry, Pergamon Press,
Oxford, 1967, p. 407.
6. M.L. Williams, R.F. Landel and J.D. Ferry, J. Amer. Chern. SOC.,1955,77,3701.
7. R.G.C. Arridge, Mechanics of Polymers, Oxford University Press, Oxford, 1975, p.
24.
8. A. Tsoga, S. Kasapis and R.K. Richardson, Biopolymers, 1999,49,267.
9. S. Kasapis, I.M.A. Al-Marhoobi and A.J. Khan, International Journal of Biological
Macromolecules, 2000,27, 13.
10. E.R. Fitzgerald and J.D. Ferry, J. Colloid Sci., 1953,8, 1 .
1 1. S. Kasapis and S . S . Sablani, International Journal of Biological Macromolecules,
2000, 27, 301.
12. J.D. Feny and M.L. Williams, Journal of Colloid Science, 1952, 7,347.
13. J.D. Ferry, Viscoelastic Properties of Polymers, John Wiley, New York, 1980, p. 1.
14. G. Kraus and J.T. Gruver, J. Polymer Sci., 1965,3A, 105.
15. J.D.Ferry and R.F. Landel, Kolloid-Zeitschrift, 1956,148, 1.
16. J.D. Feny, Viscoelastic Properties of Polymers, John Wiley, New York, 1980, p.
321.
17. P.E. Jr. Rouse, The Journal of Chemical Physics, 1953,21, 1272.
Structure. Characterisation and Interactions 53
18. I. Pezron, T. Herning, M. Djabourov and J. Leblond, Physical Networks, Polymers

and Gels, Elsevier, London, 1990, p. 231.
19. T.G. Jr. Fox, P.J. Flory, J. Appl. Phys., 1950,21, 581.
20. J.D. Ferry, Viscoelastic Properties of Polymers, John Wiley, New York, 1980, p.
264.
21. M. Peleg, Critical Reviews in Food Science and Nutrition, 1992,32, 59.
22. A.H. Clark, Gums and Stabilisers for the Food Industry lo, The Royal Society of
Chemistry, Cambridge, 2000, p. 91.
23. J. Perez, E. Muzeau and J.Y. CavaillC, Plastics, Rubber and Composites Processing
and Applications, 1992,18, 139.
24. W.C. Jr. Child and J.D. Ferry, Journal of Colloid Science, 1957,12,327.
25. E.R. Morris, M.G.E. Gothard, M.W.N. Hember, C.E. Manning and G. Robinson,
CarbohydratePolymers, 1996,30, 165.
26. S. Kasapis, P. Giannouli, M.W.N. Hember, V. Evageliou, C. Poulard, B. Tort-
Bourgeois and G. Sworn, CarbohydratePolymers, 1999,38, 145.
27. D.J. Plazek, Journal of Rheology, 1996,40,987.
28. J.D. Ferry, Viscoelastic Properties of Polymers, John Wiley, New York, 1980, p.
224.
29. G. Sworn and S . Kasapis, CarbohydrateResearch, 1998,309,353.
30. V . Evageliou, S. Kasapis and M.W.N. Hember, Polymer, 1998,39,3909.
31. K . Deutsch, E.A.W. Hoff and W. Reddish, Journal of Polymer Science, 1954,13,565.
FLUORESCENCE ANISOTROPY MEASUREMENTS OF THE DYNAMIC
BEHAVIOUR OF BIOPOLYMERS
T. A. Smith, L. M. Bajada and D.E. Dunstan*
School of Chemistry and *CRC for Bioproducts,

Department of Chemical Engineering,
University of Melbourne,
Victoria, 30 10,
Australia
ABSTRACT
The local viscosity experienced by the chain backbone of fluorescein-labelled
hydroxypropylguar (HPG) in aqueous solution has been studied using time-resolved and
steady-state fluorescence polarization spectroscopy. These measurements have been
undertaken over a range of HPG concentrations and in a series of glycerol/water
solutions of varying Newtonian viscosity. The results indicate that despite the bulk
viscosities varying over several orders of magnitude as a finction of HPG concentration,
the rotational dynamics of the fluorophore attached to the backbone are only measurably
affected at concentrations of -10 wt%. Samples of the HPG at concentrations as high as
10 wt% therefore contain regions with viscosity similar to the aqueous phase as sensed
by the probe molecule.
The rotational correlation time of the probe attached to the HPG backbone shows a
square root dependence on the solvent viscosity in the glycerol/water mixtures. This is
attributed to the restricted motion of the probe molecule around the central axis in two
dimensions.
INTRODUCTION
The aims of this study were to investigate the dynamic rotational behaviour of covalently
attached fluorescein pendant to the backbone of a hydroxypropyl-modified form of guar
(HPG) over a range of solvent viscosities and as a hnction of polymer concentration far
exceeding C*. This has been undertaken in order to assess the degree to which the probe
senses the microenvironment (local viscosity) in which the chain The results
have been used to interpret the degree of molecular interactions between the chains and
to correlate these with the macroscopic rheological behaviour. A series of steady-state
and time-resolved fluorescence anisotropy measurements have been conducted on five
systems. Firstly, solutions of free fluorescein in water/glycerol mixtures were studied as
a reference for the experiments with HPG. The same measurements were carried out on
a series of glycerol/water solutions containing free fluorescein with 0.5 wt% unlabelled
HPG added. Thirdly, measurements from glycerol/water solutions of a small amount of
the fluorescein-labelled HPG with 0.5 wt% unlabelled HPG added were performed. The
fourth series of experiments were time-resolved fluorescence anisotropy measurements
on the labelled HPG as a function of concentration in water. Finally, several
measurements were undertaken on HPG which was cross-linked using borate ions. The
gels were produced at high HPG concentrations where the labelled HPG was added as a
tracer at a concentration optimised for the fluorescence measurements. Much of the prior
work conducted using fluorescence anisotropy has focused on synthetic polymer
systems5-18 with the work of Biddle and Pardham conducted on labelled
hydroxyethylcellulose. l 9
EXPERIMENTAL SECTION
Solution Preparation
HPG, the derivatised form of Guar gum, was obtained from Rhodia, North American
Chemicals, Cranbury, NJ (Jaguar HP-8000, guar hydroxypropyl ether). The substitution
of the hydroxyl groups on the sugar rings with hydroxypropyl groups is achieved by
reaction between guar and propylene oxide under alkaline conditions.20p21The HPG
studied here is such that a small percentage (molar substitution of 0.3 to 1.6)20922
of the
hydroxyl groups on the sugar rings are substituted with hydroxypropyl groups (Figure 1).
The weight average molecular weight ofthe HPG used was of the order of 3 milli0n.2~
A sample of fluorescein-labelled HPG was also provided by Rhodia, Complex Fluids
Laboratory. The fluorescent labelling was achieved through a process of partial
dissolution of natural guar in DMSO and pyridine, followed by the addition of
fluorescein isothiocyanate (FITC) and dibutyltindilaurate (DTDL). Approximately 3% of
the sugar units along the polysaccharide are labelled with the fluorophore.
Absorption and Fluorescence Measurements

Absorption spectra were recorded on a Cary 50 Bio absorption spectrometer.
Fluorescence measurements were carried out on both the fluorescently-labelled HPG,
and solutions of unlabelled HPG to which free fluorescein was added. Glycerol-water
mixtures ranging from 0-100 wt% glycerol (BDH AnalaR) were also used. Water was
purified using a Milli-Q system. Bulk solution viscosities were varied through the
addition of unlabelled HPG (Rhodia Jaguar HP-8000) to solutions of either fluorescein-
labelled HPG or free fluorescein in water/glycerol solutions. Solutions were stirred and
heated gently to promote dissolution of HPG in the glycerol/water medium.
Steady-state fluorescence polarization measurements were carried out on either a Hitachi
4500 or a Varian Eclipse fluorimeter with polarization accessory (excitation and
emission bandpass settings were both 5 nm, he,, = 450 nm and h e , = 500 nm). Time-
resolved fluorescence polarization measurements were carried out using the time-
correlated single photon counting method,24 employing as the excitation source the
frequency doubled output of a mode-locked titaniumxapphire laser (Coherent Mira
900F) (h,,,=436 nm, he,=507 nm), pumped by a 12.5W Ar laser (Coherent Innova 400).
This system provided 872 nm pulses of approximately 90 fs duration at 76 MHz. The
repetition rate of these pulses was reduced to 4 MHz with a home-made single pass pulse
picker based on a TeO2 Bragg Cell (Gooch and Housego) driven by a CAMAC CD5000
electronics unit. The detection electronics were as described elsewhere.25 Briefly, the
fluorescence was collected by anfl quartz lens (Melles Griot), spectrally selected using a
monochromator (Jobin Yvon, H20) and detected with a microchannel plate
photomultiplier (Hamamatsu R1564U-0 1). Time-resolved fluorescence anisotropy
measurements were carried out by collecting decays with the emission polarization
analyser set parallel and then perpendicular to the direction of the (vertical) polarization
of the excitation light. The emission polarization analyser was switched between the
parallel and perpendicular positions for preset periods (usually 30 second dwell times)
and the two decays collected in separate channels of a multichannel analyser (Tracor
Northern ECON 11) over total collection periods of approximately an hour in order to
enhance the signal to noise ratio of the anisotropy function.
The time-resolved fluorescence anisotropy function, r(t), is calculated using Equation 1
in which Ill(t) and I (t) are the individual decays collected with the polarization analyser
set parallel and perpendicular to the vertically polarized excitation light. The G-factor
is included to account for any polarization bias of the detection system. The influence of
this term was minimised by arranging the polarization analyser to be the first element in
the detection system and using a polarization scrambler immediately prior to the
emission monochromator.
The analysis of time-resolved fluorescence anisotropy data can become complex for even
relatively simple systems. Various analysis methods, appropriate for specific analysis
functions, have been discussed in the l i t e r a t ~ r e ?but
~ ~due
~ to the complexities involved
with analysing data from polymer systems in general, comprehensive analysis using
particular analysis fbnctions was not attempted here. In this work, data analyses were
conducted in a number of ways depending on the relevant time-scale. For the rapidly
depolarizing systems (up to -30 wt%), the method of autoreconvolution26was employed,
whereas for the longer time-scale processes, the raw anisotropy decay was fitted without
reconvolution using single or double exponential decay functions. A double exponential
decay function was required in the case of the fluorescein-bound HPG over the range of
HPG concentrations used.
RESULTS AND DISCUSSION
Steady-State Fluorescence Measurements

1. Unbound Fluorescein Systems
Steady state and time resolved fluorescence anisotropy measurements have been
conducted on unbound fluorescein in solutions over a range of glycerol/water
concentrations (Newtonian solvent viscosity). The steady-state anisotropy values as

plotted in Figure 1 show an interesting trend. There is an approximately linear
dependence of anisotropy on viscosity at low viscosities while at higher viscosities the
anisotropy values asymptote towards -0.32 which is close to the limiting (intrinsic)
anisotropy of fluorescein.28The intrinsic anisotropy is related to the angle between the
absorption and emission transition dipoles of the fluorophore. This asymptotic behaviour
reflects the time-scales over which the fluorescence lifetime of fluorescein is appropriate
to report on rotational behaviour; at high viscosities, the probe rotation is slowed beyond
the usefbl range of this probe, and thus limits to the intrinsic anisotropy of the probe.
The addition of unlabelled HPG (0.5 wt %) to the solutions of free fluorescein in
glycerol/water made no measurable change to the steady-state fluorescence anisotropy
values, despite this HPG concentration being well above c*. This indicates that the free
fluorescein behaves as if it were in the pure glycerol/water mixtures, and that the rotation
of the fluorescein is not hindered by the presence of the polymer network assumed to be
operative at this polymer concentration. These results also indicate that there is no
binding or affinity of the fluorophore for the HPG in solution, that the HPG is not
preferentially solubilized by the glycerol and that the solution is homogeneous in
glycerol/water. This is despite the observation that the HPG coil reduces in size in the 90
wt% glycerol solution.
2. Bound Fluorescein Systems
The steady-state fluorescence anisotropy values obtained from the bound fluorescein
with 0.5 wt% unlabelled HPG added as a fbnction of glycerol/water concentrations can
be compared with the corresponding values obtained from the free fluorescein system
(Figure 1). The higher anisotropy intercept of the bound form compared to the free
fluorescein reflects the restricted motion of the bound probe. It is seen that there is a
vastly different dependence of the steady-state fluorescence anisotropy on viscosity for
the two systems. The anisotropy values at high viscosity are essentially identical, while
there is a linear dependence on the logarithmic scale of the anisotropy with viscosity for
the bound probe. This dependence leads to a power law behaviour of the form: r aqa;in
this case a is found to have a value of -0.3. The implications of this finding are
discussed below.
For the labelled HPG in water at a range of unlabelled HPG concentrations, the steady
state anisotropy is insensitive to the polymer concentration at concentrations up to -1
wt%. Measurement at an HPG concentration of 10 wt% showed a small increase in the
steady-state anisotropy (r-0.12) relative to the lower concentrations.
This reflects the first noticeable evidence of interactions between the polymer chains
reported by the fluorescent probe. The onset of the interaction between polymer chains
as sensed by the fluorescent probe occurs at concentrations significantly above c*. At 10
wt% concentrations the probe molecule size is of the order of the mesh size (correlation
length) of the polymer solution, as will be discussed later.
58 Gums and Stabilisers for the Food Industry 1I
Time-Resolved Fluorescence Measurements

1. Unbound Fluorescein Systems
The fluorescence decays of unbound fluorescein with 0.5 wt% unlabelled HPG
added collected at the magic angle (54.7') are essentially exponential over the range of
glycerol content used. The fluorescence anisotropy from these solutions decays in an
approximately single exponential manner (not shown here) although there was some
minor contribution by a long (-10 ns) decay component in the anisotropy from the low
viscosity solutions which was disregarded. The rotational correlation times derived from
these anisotropy decays, are plotted in Figure 2 as a function of solvent viscosity. The
fluorescence decay times, rotational correlation times and initial anisotropy values agree
well with those reported by Lakowicz et a1.28
Log (viscosity)
0.5 1 15 2 2.5
-2.3
Figure 1. Steady-state fluorescence anisotropy values versus bulk viscosity (a) free
fluorescein (large squares) and (b) bound fluorescein-HPG (small diamonds) both in the
presence of 0.5 wt% HPG, with the viscosity varied through the glycerol/water content.
The IinearJit for (a) is through the lower five data points only.
The rotational correlation times, T ~ determined

~ , from the time-resolved fluorescence
anisotropy measurements of free fluorescein in a range of water/glycerol mixtures show a
simple linear relationship with solvent viscosity (Figure 2). An empirical form of the
relationship between the free rotor time (i.e. in the absence of solvent), z, and the
viscosity, q, is given be lo^.^'
z, = c q + z, (2)
Structure, Characterisation and Znteractions 59
In the case of fluorescein, the limiting rotor time, -o, would be expected to be negligibly
small as has been reported for molecules such as DODC13' and is neglected in the
following discussion for simplicity. The factor c is determined by the shape of the
molecule and the hydrodynamic boundary conditions. For the case of a simple spherical
rotor with stick boundary conditions, the Debye-Stokes-Einstein theory can be
implemented and c is given by V/kT. Under such conditions, the free probe rotational
diffusion constant can be described by:
where k is Boltzmann's constant, T the absolute temperature, q the solvent viscosity and
r the rotational hydrodynamic radius of the spherical diffusing object, and V, the
hydrodynamic volume.
The radius of the fluorescein calculated from the rotational diffusion coefficient
assuming spherical geometry ( z =1/6D)~ ~ is approximately equal to the radius of the
fluorescein in its planar orientation (0.5 nm).3' The free rotation in solution will be an
average of rotation about the different axes of the molecule. The fluorescein is relatively
planar and the thickness of the disc would be expected to be of order several Angstroms.
The data indicate that the dominant rotational modes occur with the free fluorescein
rotating as a disc in the plane.
The addition of unlabelled HPG (again to 0.5 wt%) to the unbound
fluorescein/glycerol/water solutions does not change the fluorescence decay or
fluorescence anisotropy characteristics to any measurable degree consistent with the
steady state measurements above.
2. Bound Fluorescein in GlycerolMater Mixtures
The fluorescence decay of the fluorescein-labelled HPG is, unlike the unbound case, non-
exponential. Non-exponential decay behaviour is commonly observed for probes
attached to polymers and is often attributed to a variation in the local environments of the
range of potential sites of the attached probe molecules on the polymer backbone. This
complicates subsequent interpretation of the time-resolved fluorescence anisotropy
measurements, and for this reason a simple sum-of-exponentials analysis function was
used in the following data interpretation.
The fluorescence anisotropy from the bound fluorescein glycerol/water systems also
decays in a non-exponential manner, with a double exponential decay being adequate to
describe the decay to a first approximation. The rapid component was constant over the
concentration range and only the longer component varied, so in the results reported
below, only the longer component is considered. Using the second rotational correlation
time as an average rotational time and plotting this as a function of the bulk solution
viscosity results in several interesting findings. It should be noted that the simple
interpretation used yields similar trends to the steady-state anisotropy data vindicating
the straightforward approach used.
Figure 2 shows that in the labelled HPG in glycerol/water mixtures the probe is sensitive
to the solvent viscosity in a manner not predicted by equation 2, but rather is described
by equation 4 with a-0.5 (again neglecting any limiting rotor time, cc0, term). The value
60 Gums and Stabilisersfor the Food Industry 1I
of log(z,,) as log(q) tends to zero is related to the effective volume taken up (swept out)
by the rotating fluorophore through the factor c. This volume would be expected to
differ between the free and bound forms of the fluorescein due to the reduced degrees of
freedom of the bound form.
To, = cq" (4)
The bound fluorescein probe is restricted to rotation in a plane perpendicular to the
central (linkage) axis. The larger rotational size interpreted for the bound form from the
rotational correlation time using equation 3 (0.94 nm in water) is greater than the size of
the calculated planar radius of fluorescein3' (0.5 nm) in contrast to the value interpreted
for the free fluorescein.
The key finding illustrated in Figure 2 is the non-linear ( ~ ~ - dependence q~) of the
rotational relaxation time of the probe on the solvent viscosity, where a-112. This is in
contrast to the observed behaviour of the unbound fluorescein where a-1. The value of
a from the time-resolved measurements, while somewhat different in magnitude to that
concluded from the steady-state measurements (a-0.3),is consistent with the steady-
state data (Figure 1) in showing a non-linear viscosity dependence. Similar power law
dependence of the rotational correlation time with viscosity has been reported for other
systems where a values of 0.75 and 0.41 have been r e p ~ r t e d . ~ * ' ~ ~ ' *
0 0.5 1 1.5 2
Figure 2. Rotational correlation times versus bulk viscosity ( a ) free fluorescein in

presence of 0.5 wt% HPG (diamonds) and ( b ) bound fluorescein-HPG
(squares)as a finction of viscosity varied through glycerolfwater content.
Both the a values and hydrodynamic sizes inferred from the data for the bound
fluorescein in this study are presumably due to the reduced rotational degrees of freedom
associated with the attached chromophore. The unbound fluorophore rotates randomly in
three dimensions while the pendant fluorophore must rotate around its central axis
parallel to the covalent bond linking the probe to the polymer backbone. It is of interest
that the steady state data for the free fluorophore shows a linear dependence on the
viscosity at low concentrations and plateaus to the limiting inherent anisotropy for the
fluorescein at high glycerol concentrations. As such the power law dependence must
derive from the intrinsic anisotropy of the chromophore, the rotational degrees of
freedom, the solvent viscosity and a component due to the bulk viscosity from chain-
chain interactions. Park and W a l d e ~ khave
~ ~ considered the isomerisation dynamics of
stilbene and conclude that a values less than unity arise from the reduced dimensionality
of the diffusion.
3. Bound Fluorescein in Water as a Function of Polymer Concentration
The anisotropy decays of the bound fluorescein at the different HPG concentrations are
shown in Figure 3. There is little difference observed in the fluorescence anisotropy
decay curves up to 1 wt% in solution. The 10 wt% solution shows that a long rotational
correlation time component is present with the relative contributions to the total
anisotropy decay of the rapid and slow processes being approximately equal. The probe
molecule is therefore sensing the presence of the neighbouring molecules to a measurable
degree at this concentration. It should be noted that while the viscosity data is not shown
for the 10 wt% solution, this solution is effectively a free standing solid.
The mesh size or correlation length, 5, is defined as a function of the polymer fraction (cp,
5 = (p)
314
defined as the fraction of sites occupied by monomers), by: where cp* is the
critical polymer fraction which is related to the critical concentration by cp*=c*a3 where
a3 is the volume of the unit cell in a cubic lattice.' The value of 6 for the polymer solution
at 10 wt% concentration is calculated to be of the order of 5 nm.' This correlation length
is still considerably larger than the size of the probe molecule of -0.5 nm as discussed
above. The interaction between the HPG molecules which gives rise to the solution
viscosity should be strong at concentrations of 10 wt% which is significantly higher than
the critical overlap concentration of 0.2 wt%.23
In order to assess the technique for measuring the interaction between the polymer chains
in gels, cross linking of the HPG using borate ions was ~ n d e r t a k e n This
. ~ ~ was done with
trace amounts of the fluorescein labelled HPG incorporated into the gel. The data for the
HPG at three concentrations with the same ratio of borate ions as the crosslinker
(maintained at pH 12) are shown in Figure 4.
n
Y
W
&
Time (ns)
Figure 3 . Time-rcsolvedfluorescence anisotropy decays for lubelled HPG as a funciion

of HPG concentration in water.
1.
I-0.5% HPG 2 0 g L borate
1- I % HPG 40gL borate
i- 1.50% HPG 6 o g n borate
h
a
2
* 0.1
0,
.m
E
c
0.01
10 15 20 2s
0 5
Time (ns)
Figure 4. Time-resolvedJluorescence anisotropy projles of HP 3 cross-linked by use of

borate ions.
There are observed two lifetimes with the rapid component being faster than observed for
the bound fluorescein in solution with the long component being slower than that
observed in solutions of concentrated HPG. The anisotropy decay curves show a minima
then appear to increase after long times. This behaviour has been observed previously
for heterogeneous systems where dye molecules were attached to colloidal particles,27
and other polymer systems,34 however, the reasons for the rise in anisotropy at longer
times in the present system is not fully understood.
The gelation of the HPG causes a separation of the rotational lifetimes into both a very
rapid and slow component. The probe molecules close to crosslinking sites will exhibit
slowed rotational motion while the others will be located far from the crosslink points
and experience only the free solvent.
CONCLUSIONS
The fluorescence anisotropy method has been shown to be a useful technique for
measuring the local chain environment in biopolymer systems.
Free fluorescein in glycerol/water mixtures senses the solution viscosity in a
manner predicted by the Stokes-Einstein-Smoluchowski equation. The bound
fluorescein shows a z-q dependence in the glycerol/water solutions due to restricted
motion of the probe. The rotational correlation time of the bound probe is relatively
insensitive to the concentration of the polymer in solution up to -1 wt%. The onset of
measurable polymer chain interaction (as sensed by the rotational motion of the probe) is
only observed to occur at concentrations far in excess of C*. At 10 wt% concentration of
HPG in solution, the probe senses the presence of the other polymer chains and a slowing
of the rotational diffusion of the probe molecule is observed.
Gelling the HPG using borate ions as the cross-linking agent yields a rapid and a
slow rotational correlation time. The observed anisotropy decay curves are consistent
with the system becoming heterogeneous at a molecular level under these conditions.
ACKNOWLEDGMENTS
We gratefully acknowledge the provision of the fluorescein-labelled HPG by Dr.
Jeannie Chang of Rhodia, Complex Fluids Laboratory, Cranbury USA.
REFERENCES
1 . DeGennes, P.-G. Scaling Concepts in Polymer Physics; Cornell University Press:
London, 1979.
2. Morawetz, H. Polym. Prepr,, 1986,27, 3 16-317.
3. Daoud, M.; Cotton, J. P.; Farnoux, B.; Jannink, G.; Sarma, G.; Benoit, H.;
Duplessis, J. P.; Picot, C.; DeGennes, P.-G. Macromolecules, 1975, 8, 804-8 18.
4. Oster, G.; Nishijima, Y . Fortschr. Hochpo1ym.-Forsch, 1964, 3 , 3 13-331 .
5. Hu, Y.; Horie, K.; Ushiki, H. Macromolecules, 1992,25,6040-6044.
6. Adolf, D. B.; Ediger, M. D.; Kitano, T.; Ito, K. Macromolecules, 1992, 25, 867-
872.
7. Johnson, B. S.; Ediger, M. D.; Kitano, T.; Ito, K. Macromolecules, 1992, 25, 873-
879.
8. Soutar, I.; Swanson, L. Macromolecules, 1994,27,4304-431 1 .

9. Jarry, J. P.; Erman, B.; Monnerie, L. Macromolecules, 1986, 19,2750-2755.
10. Monnerie, L. In Photophysics of Synthetic Polymers; D. Phillips and A. Roberts,
Ed.; Science Reviews: Northwood, 1982; pp 70-8 1.
1 1 . Valeur, B.; Monnerie, L. J. Polym. Sci., 1976, 14, 11-27.
12. Viovy, J. L.; Monnerie, L. Polymer, 1986,27, 181- 184.
13. Kasparyan-Tardiveau, N.; Valeur, B.; Monnerie, L. Polymer, 1983,24,205-208.
14. Bur, A. J.; Lowry, R. E.; Roth, S. C.; Thomas, C. L.; Wang, F. W.
Macromolecules, 1991,24,3715-3717.
15. Bur, A . J.; Lowry, R. E.; Roth, S. C.; Thomas, C. L.; Wang, F. W.
Macromolecules, 1992,25, 3503-35 10.
16. Ono, K.; Okada, Y.; Yokotsuka, S.; Sasaki, T.; Yamamoto, M. Macromolecules,
1994,27,6482-6486.
17. Clements, J. H.; Webber, S. E. J. Phys. Chem., 1999, 103,25 13-2523.
18. Ushiki, H.; O m , M. Eur. Polym. J., 1986,22,835-839.
19. Biddle, D.; Pardhan, S. Arkiv For Kemi, 1970, 32,43-53. (20) Guar, Locust
Bean, Tara, and Fenugreek Gums; 3rd ed.; Maier, H.; Anderson, M.; Karl, C.;
Magnuson, K., Ed.; Academic Press: San Diego, 1993.
21. Lapasin, R.; Pricl, S. Rheology of Industrial Polysaccharides: Theory and
Applications; Blackie Academic & Professional: London, 1995.
22. PrudHomme, R. K.; Constein, V.; Knoll, S. In SPE 18210, 63rd Annual Technical
Conference and Exhibition; Houston, Texas, 1987.
23. Power, D. J. PhD Thesis, Univeristy of Melbourne, 1997.
24. OConnor, D. V.; Phillips, D. Time-Correlated Single Photon Counting; Academic
Press: London, 1984.
25. Smith, T. A.; Haines, D. J.; Ghiggino, K. P. J. Fluor., 2000, 10,365-373.
26. Rumbles, G.; Smith, T. A.; Brown, A. J.; Carey, M.; Soutar, I. J. Fluor., 1997, 7,
2 17-229.
27. Smith, T. A.; Irwanto, M.; Haines, D. J.; Ghiggino, K. P.; Millar, D. P. Coll.
Polym. Sci., 1998,276, 1032-1037.
28. Lakowicz, J. R.; Cherek, H.; Maliwal, B. P. Biochem., 1985,24,376-383.
29. Haugland, R. P. Handbook of Fluorescent Probes and Research Chemicals; 7th
ed.; Molecular Probes: Oregon, 1999.
3 0. Fleming, G. R. Chemical Applications of Ultrafast Spectroscopy; Oxford
University Press: New York, 1986.
31. Chem3D, Cambridge Scientific Computing, Inc., 1993.
32. Glowinkowski, S.; Gisser, D. J.; Ediger, M. D. Macromolecules, 1990, 23, 3520-
3530.
33. Courtney, S. H.; Fleming, G. R. J. Chem. Phys., 1985,83,215-222.
34. Park, N. S.; Waldeck, D. H. J. Chern. Phys., 1989, 91,943-952.
35. Chee, C. K.; Ghiggino, K. P.; Smith, T. A.; Rimmer, S.; Soutar, I.; Swanson, L.
Polymer, 2001,42,2235-2240.
TIME RESOLVED POLARISED ATTENUATED TOTAL REFLECTION
SPECTROSCOPY OF A PROTEIN FILM AT THE SILICNSOLUTION
INTERFACE
M. L. J a y ~ n e ' ~M.
~ ,L.
~ , Gee173,
D. D ~ n s t a n * , ~
I
Polymer and Surface Interactions Group, School of Chemistry, The University of
Melbourne, 3010 Victoria Australia
*CRC for Bioproducts, Department of Chemical Engineering, The University of
Melbourne, 3010 Victoria Australia
3Particulate Fluids Processing Centre, The University of Melbourne 30 10 Victoria
Australia
1 INTRODUCTION
The characterisation of protein adsorption requires an understanding of the kinetics of film

formation in a range of environmental conditions, the kinetics of re-arrangement and the
final orientation. Despite its importance and applicability, information on the relationship
between the stability of these films and their orientational evolution remain scarce.
In this study, we show that Time Resolved Attenuated Total Reflection Spectroscopy
(TRATR) can be used to access this information from lysozyrne adsorbed on a silica
surface. TRATR is an evanescent wave technique that utilises polarised UV-visible light to
probe the adsorbed structures of macromolecules. Data obtained from TRATR
measurements has the advantage of being model independent and relatively simple to
analyse in order to obtain structural information of adsorbed layers. The use of light from
the near ultra-violet region of the spectrum to probe these layers allow more accurate
measurements of the adsorbed amount as the depth of penetration of the evanescent wave
at a given angle of incidence is much smaller than that achieved in the infi-a-red, hence
significantly reducing the bulk solution contribution. By polarising the light incident on the
protein film, the film structure is simply and unambiguously determined.
2 THEORY
2.1 Dichroic Ratio Analysis
A totally internally reflecting beam of light within an optically dense medium transfers
energy into the rarer medium in the form of an evanescent wave. Attenuation of the beam
by a chromophore at the interface can be described by the interfacial form of Beer's Law,
viz:
1
A(8,h) = N(8) E(h)Ajp(z)e-zdpdz
cose 0
where A(8,h) is the absorbance, N(8) is the number of reflections at an angle of incidence
8, ~ ( his) the extinction coefficient, p(z) is the chromophore distribution function along a
distance z from the interface and d, the evanescent wave penetration depth:
where h, is the incident wavelength, nl and n2 the refractive indices of the optically denser
and rarer media respectively.
Polarisation of the incident beam results in polarisation of the evanescent wave either
parallel to the plane of incidence (P polarisation), or perpendicular (S polarisation) to the
plane of incidence (i.e. in the plane of the surface, Fig. 1).
Figure 1: Nomenclature for analysis with

Cartesian co-ordinates: 1) Evanescent wave, 2)
silica-water interface, 3) rarer (water) medium, 4)
denser (silica) medium 9) Angle of incidence of
light beam
The extenrto which the absorbing species interacts with the evanescent wave will
depend on the orientation of the transition moment M with respect to the plane of
polarisation of that evanescent wave. By measuring the species absorbance in the two
planes of polarisation, it is therefore possible to determine the orientation of M with
respect to the surface. The ratio of absorbances in the two planes of polarisation, AdA,,
referred to as the dichroic ratio [l-31, can be predicted for different orientations of M with
respect to the surface by relating it to the electric field amplitude components of the plane
of polarisation. In the case of M aligned in the plane of the surface, the predicted AdAs
ratio would is:
A, - Et
3)
As Er
In the case of M being normal to the plane of the surface:
In the case of M having no preferred orientation:

A, Et+E;
- 5)
A, EY
where Ex,E, and E, are the electric field amplitude components in the x, y and z directions
respectively.
2.2 Application to Hen Egg White Lysozyme (HEWL)
Lysozyme is an enzymatic protein (oblate 4.5x3.0x3.0 nm) that can be derived from tears
to the egg white of avians [4]. Here, lysozyme derived from hen egg white (HEWL) is
utilised.
HEWL naturally contains six of the highly conjugated amino acid, Tryptophan (Trp),
distributed more or less along the long axis of the protein in the native tertiary structure [4,
51. Trp displays a distinctive electronic absorption spectrum in the UV range [ 6 ] : highly
intensive bands are observed at 195-196 nm and 217-219 nm, and the longwave band
being observed around 280-290 nm ( ~ ~ 2 8=0 5600). These bands and Trps high dipole
moment are the result of incomplete 7c-electron delocalization around the indolic ring
system. Within HEWL, it is anticipated that the six Trp residues will display an average
electronic transition moment (M) relative to the long axis of the molecule [7].
The presence of four disulfide linkages via the Cysteine residues allows HEWL to
retain its tertiary structure under a variety of environmental conditions [8], hence the three
dimensional positions of each Trp residue within HEWL remains essentially constant [7].
As M orientation with respect to the long molecular axis remains essentially constant as an
extension, it then becomes possible to relate its optical dichroism to HEWLs orientation
with respect to the surface.
In this study, the Trp absorption band observed in the near ultra-violet (280 nm) is
followed using TRATR as de-convolution of the spectrum from possible contributions of
other amino acid residues (Tyrosine, Phenylalanine) in the shortwave region becomes
unnecessary.
3 EXPERIMENTAL SECTION
TRATR is a purpose built instrument [9] with a fused silica waveguide in contact with the
sample solution within a stainless steel cell. The solution is pumped through the cell by a
Pharmacia P-1 peristaltic pump at a low flowrate to minimise the effect of shearing along
the solid-solution interface.
The number of reflections N(8) is determined by a procedure which has previously been
described elsewhere [9].
HEWL from Sigma Aldrich, three times crystallized, dialyzed and lyophilized, was
used as provided. Protein solution of 100 ppm was prepared fresh at the beginning of the
experimental run by dissolving the flakes in a phosphate buffered Milli-Q water solution of
ionic strength 0.01 M and pH 7.0. Nitrogen gas is bubbled through the solution for 1 hour
to allow it to degas before pumping into the cell at a rate of 20 ml h- for a duration of ten
68 Gums and Stabilisersfor the Food Industry I I
minutes. During the experimental run, the solution remained under quiescent conditions.
Spectral acquisition was performed at one angle of incidence which gave a penetration
depth of 56 nm as calculated using equation 2. Under these conditions, the theoretical
dichroic ratios calculated from equations 3, 4, 5 are 0.13, 1.23 and 1.09 for M aligned in
the plane of the surface, normal to the surface and with random orientation respectively.
Figure 2: TRATR schematic diagram
4 RESULTS AND DISCUSSION
260 270 280 290 300 310 320

wavelength (nm)
Figure 3: Absorption spectra of
Tryptophan in HEWL at the silica-water
interface, in both the S and P polarisations.
The absorption spectrum of Tryptophan was monitored with time in both the P and S
directions with the penetration depth constant. A typical set of absorption spectra is shown
in Figure 3. The absorption spectrum displays the three peaks characteristic of its bulk
solution spectrum [6]. HEWL unfolding at the interface will result in the increased
exposure of the Tryptophan residues to a more polar environment, with a resulting red-shift
in the spectrum occurring due to increased hydrogen bonding of the indole group to water.
No noticeable shift occurred within the experimental timescale, indicating that HEWL does
maintain its tertiary structure at the interface.
0 500 loo0 1500 2000 2500 3000
time (min)
Figure 4: Time resolved total absorbance
of the 280 nm peak in Tryptophan. R=
rinsing with buffer solution.
The total time resolved absorbance, given in Figure 4, indicates that the adsorption
process is most rapid in the initial two hour period, slowing down significantly for the next
38 hours of adsorption, with no plateau observed during the measurement period. The
rinsing region surprisingly shows a somewhat slow and monotonic rise in the absorbance
in the absence of bulk solution protein, a feature that will be discussed further.
An electrostatic potential map of HEWL [7] indicates a molecule that contains a more
or less homogeneous distribution of 19 positively charged sites on its surface [4,lo], with
its active site cleft containing most of the negative charge. With an isoelectric point of 11,
HEWL is expected to have an overall positive charge under the experimental conditions
and hence would readily adsorb onto the negatively charged silica sufface (isoelectric point
-1.9, [ 1I]), confirmed by the rapid rise in total absorbance in the initial two hours of the
process.
The time-resolved dichroic ratio AdAs is shown in Figure 5. It is observed to drop
rapidly in the initial two hours from the theoretical value for a random orientation of the
transition moment with respect to the surface to a value close to an ordered, perpendicular
orientation at the end of this period. Due to the homogeneous distribution of positive sites
on HEWL and at low surface coverages, it can be concluded that HEWL initially adsorbs
with random orientation on the bare silica surface. The large area per molecule expected at
such initially low coverages means that very little protein-protein interaction through steric
exclusion or electrostatic repulsion is experienced.
I.
90"
0 500 1000 1 5 0 0 2 0 0 0 2500 3000 3 5 0 0
time (min)
Figure 5: Time resolved dichroic ratios.
Literature on Lysozyme adsorption kinetics is in general agreement in that this initial

adsorption region for the solution-silica interface follows first order kinetics, dependent on
how much has already adsorbed on the surface [ 12, 131. This set confirms the fit to a first
order kinetic process with a correlation coefficient of 0.93. A, was arbitrarily chosen to be
A40hrs(immediately prior to flushing of the cell with protein deplete buffer solution) as an
equilibrium total absorbance was not reached. Normalisation with A, then leads to a fit for
the following first order expression:
This initial adsorption process yielded a pre-exponential constant B of 0.4, with a kinetic
constant kl of 3 ~ 1 0min-'.
-~
The second adsorption region, which extends for the next 38 hours, displays a much
slower rise in the total absorbance. The dichroic ratio can be seen to remain relatively
unchanged throughout this period from just below the theoretical value for a surface
perpendicular orientation of the transition moment, indicating that a highly ordered film
has been formed after the initial adsorption period which maintains it's structure even
during the adsorption of extra protein, Interestingly and in contrast to literature, this region
is best fitted to a zero-order kinetic process (r2 = 0.99),
A -A
= -kt +C 7)
Am
yielding a kinetic constant k2 of 2 ~ l O min-',

-~ almost ten times slower than the initial
adsorption region. Further adsorption of the protein can be explained in that any ordered
adsorption will most likely occur with the molecule's most negative site-that is, HEWL's
active site cleft-oriented away from the surface, even though the molecule maintains an
overall positive charge. As the surface becomes more crowded, each molecule is forced to
adopt the perpendicular orientation to facilitate hrther adsorption to achieve maximum
packing. At the end of the first adsorption process, the surface would most likely be near
saturation with HEWL, and hence to a newly arriving HEWL, the surface that is presented
to it is no longer the bare or patchy silica surface, but a near homogenous protein film. The
formation of the film therefore creates a fresh surface. Further adsorption is made possible
by the film having had the localised negative charge of the active site cleft presented
towards the approaching HEWL, but due to the overall positive charge of the film, the
approaching HEWL at the same time experiences an electrostatic repulsion, slowing down
the adsorption process and hence the much lower value of the kinetic constant.
Perhaps the most interesting part of the curves is the area after t=40 hrs when the
buffered protein solution in the cell is exchanged for protein deplete buffer solution only. It
was initially expected that a drop in the total absorbance, indicating desorption, will be
experienced [13]. No drop is observed, with even a slow increase in the total absorbance
occurring. Again, the dichroic ratio remains relatively unchanged throughout this period,
with a value close to the theoretical for perpendicular orientation of the transition moment.
These indicate that the highly ordered film formed during adsorption is also highly robust.
The apparent rise in total absorbance cannot necessarily be attributed in this region to
the addition of extra protein from bulk solution as the bulk is now protein deplete. It could
be argued that the rise could be a result of film re-arrangement, although the dichroic ratio
remains relatively unchanged. A possible explanation is the migration of proteins adsorbed
on the surface of the protein film to within the film itself, moving to a more intense region
of the evanescent wave and hence contributing more to the beams attenuation. Whether
migration does occur can be determined by performing the same experiment using a
different penetration depth of the evanescent wave, the subject of further study.
5 SUMMARY
The adsorption and orientational re-organisation kinetics have been measured for
phosphate buffered Hen Egg White Lysozyme using in-situ TRATR. The average
orientation of the film formed was further probed using VIEWS. The initial adsorption
region was found to be rapid, lasting for a little over two hours, and can be fitted to a first
order adsorption process. During this period HEWL adsorbs in a random orientation at the
interface. The second adsorption region, followed for the next 38 hours, is slow and can be
best fitted to a zero order adsorption process. The film formed is highly ordered, and no
desorption was observed upon rinsing with protein deplete phosphate buffered solution,
indicating irreversibility of the adsorption process.
References
1. U. P. Fringeli and W. H. Gunthard, Infrared Membrane Spectroscopy, in

Membrane Spectroscopy, E. Gell, Editor.New York. 1981.p. 270-332
2. D. J, Neivandt, M. L. Gee, M. L. Hair and C. P. Tripp, J. Phys. Chem. B, 1998.
102: p. 5107-5114.
3. R. Zbinden, Orientation Measurements, in Infra-red Spectroscopy of High
PoZymers.New York. 1964.p. 167-251
4. P. Jolles, Lysozymes: Model Enzymes in Biochemistry and Biology. 1996,
Birkhauser Verlag: Basel, Switzerland. p. 449.
5. R. L. Hill, K. Brew, T. C. Vanaman, I. P. Trayer and P. Mattock, i%e Structure,

Function and Evolution of alpha-Lactalbumin, in Brookhaven Symposia in Biology no. 21:
Structure, Function and Evolution in Proteins. 1969. p. 139-154.
6. A. P. Demchenko, Ultraviolet Spectroscopy of Proteins. 1986, Berlin: Springer-
Verlag. 278+.
7. R. Diamond, D. C. Phillips, C. C. F. Blake and A. C. T. North, J M o l Biol, 1974.
82: p. 371.
8. T. J. Lu and J. R. Lu, Langmuir, 1998.14: p. 438 - 445.
9. D. J. Neivandt and M. L. Gee, Langmuir, 1995. 11:p. 1291-1296.
10. D. Takahashi, Y. Kubota, K. Kokai, T. Izumi, M. Hirata and E. Kokufbta,
Langmuir, 2000. 16: p. 3133-3140.
1 1. R. K. Iler, The Chemistry of Silica. 1979, New York: Wiley.
12. D. Sarkar and D. K. Chattoraj, J, Colloid Interface Sci., 1993. 157: p. 219-226.
13. T. A. Horbett and J. L. Brash, Proteins at Interfaces 2: Fundamentals and
Applications, in ACS Symposium Series, T. A. Horbett and J. L. Brash, Editors. 1995,
American Chemical Society: Washington DC. p. 561.
ELUCIDATION OF PROTEIN-LIPID INTERACTIONS
Nazlin K. Howell
School of Biomedical and Life Sciences, University of Surrey, Guildford, Surrey,

UK GU2 7XH
1. INTRODUCTION
The properties of gels, foams and emulsions are governed by the components present in
the mixture and their interactions with each other.' Pioneering studies on protein
interactions by the autho? highlighted the importance of the effect of mixed systems on
synergistic behaviour, phase separation phenomena and aggregation of proteins which
affect the texture and acceptability of food products. However, the effect of lipids on the
structure and functional properties of proteins has received little attention at the molecular
level.
Protein-lipid interactions occur commonly in vivo for example in membranes and in
blood by the lipid-binding protein serum albumin. In food processing, proteins and lipids
are mixed to produce emulsions such as ice cream, sausages, cake batters and desserts.
The structure and physico-chemical properties of proteins govern their ability to bind or
emulsify oil and dictate the unfolding and adsorption of protein molecules at the oil/water
interface to produce a stable emulsion. The role of proteins at the lipid interface has been
studied extensively, including studies on competitive adsorption of proteins and
rheological beha~iour.~ In general, most published studies relate to the surface and
rheological behaviour of proteins and less attention has been paid to interactions between
lipid and protein molecules at a molecular level.
In this paper the effect of lipids on protein structure is examined by FT-Raman
spectroscopy and differential scanning calorimetry and the changes in structure are
related to texture using small deformation rheology.
2. ELUCIDATION OF PROTEIN-LIPID INTERACTIONS BY RAMAN

SPECTROSCOPY
Raman spectroscopy measures inelastic scattering or changes in frequency or wavelength

of the excited incident beam, related to vibrational transitions of functional groups and
stretching and bending of bonds. With the possibility of exciting Raman spectra in the
near infrared range (NIR) (1 = 1064 nm), Raman spectroscopy has become a potentially
p o w e h l technique for food analysis applications. FT-Raman spectroscopy is not
affected by interference due to fluorescence, like conventional Raman spectroscopy or
due to water as evident in infrared spectroscopy. FT-Raman spectroscopy is suitable for
non-invasive analysis of emulsions, gels, solutions, oil, protein and turbid food samples
which are often difficult to analyse by circular dichroism and solution NMR
spectroscopy.
The application of Raman spectroscopy to food proteins includes studies by the author
on the effect of heat processing on a-lactalbumin, P-lactoglobulin and ly~ozyme.~ These
studies show a decrease in the a-helix structure and concomitant increase in the P-sheet
structure, as well as changes in hydrophobic and disulphide groups on heating proteins.
Pioneering studies on protein mixtures of lysozyme with a-lactalbumin and P-
lactoglobulin undertaken by Howell and Li-Chan4 showed the involvement of
hydrophobic interactions, indicated by changes in the bands assigned to CH and CH,
bending vibrations as well as tryptophan residues. Further investigations of the CH
stretch regions (2800-3000 cm-') confirmed changes in both hydrophobic groups and
those due to polar and hydrophilic amino acids.5 The model suggested was initial
electrostatic interactions6 followed by hydrophobic interactions and disulphide cross-links
in the aggregated complexes.
Lipids can also be characterised by Raman spectroscopy. The degree of unsaturation
of lipids can be analysed using the ratio of the C=C stretching band near 1660 cm-' to the
C=O stretching band near 1750 cm-' or the CH, scissoring band near 1445 ~ m " . ~ '
Furthermore, the relative amounts of saturated, monounsaturated and polyunsaturated
fatty acids can be examined using the 3013, 1663 and 1264 cm-' bands.' The ratio of cis
and trans isomers and conjugated products resulting from auto-oxidation can also be
f o l l ~ w e d . ~It is also possible to investigate chain packing of lipids;" polymorphism
behaviour" and transitions of crystalline phases.',
However, few studies have been undertaken on protein-lipid mixtures. Studies
reported are mostly on biological membrane structures with the main emphasis on
changes in the lipids rather than protein components. 1 3 , l 4 The region below 1200 cm-'
provides information on the trans and gauche conformation of the hydrocarbon chains
whereas the 2800-3000 cm-' region (C-H stretching) is more sensitive to structural
changes in lipid-water and lipid-protein-water phase sy~tems.'~ In particular, the ratio of
the intensity of the peak at 2850 to that at 2890 cm-' increases with the 'looseness' of the
lateral packing of disordered hydrocarbon chains and an increase in the 2930 cm-' band
suggests interactions of lipids with proteins or polypeptides.
Information available by Raman spectroscopy on individual protein and lipid systems,
have only been applied recently to studies on food emulsions.I6Preliminary investigations
of emulsions formed by either vortex action or sonication, using corn oil and lysozyme
egg white or P-casein indicated changes in the C-H stretching region suggested
hydrophobic protein-lipid interactions and greater ordering of the fatty acid chains.I7 A
more detailed investigation of the interaction of lysozyme with corn oil was undertaken
by Howell et al. l 6 using NIR-FT-Raman spectroscopy and is described below. Difference
spectra, obtained by subtracting the individual lysozyme and corn oil spectra from the
spectrum of the emulsion, were used to ascertain the groups involved in the interactions.
2.1 Materials
Deuterium oxide, lysozyme and corn oil were obtained from Sigma Chemical Company,
Poole, Dorset, UK.
2.2 Emulsion Preparation
Emulsions of lysozyme (25% (w/v) solutions in D,O) and 10% (w/w) corn oil, were
prepared by homogenisation for 5 min. using a Omni-mixer homogeniser Model 17106
(Waterbury Limited, USA supplied by Camlab Limited, Cambridge). Each 3 ml sample
separated into three layers, which were separately analysed by FT-Raman spectroscopy,
together with control samples of lysozyme (25% in D20) and corn oil on their own.I6
2.3 Raman Spectral Analysis
The Raman scattering of samples was undertaken on a Perkin-Elmer System 2000 FT-
Raman spectrophotometer with excitation from a Nd:YAG laser at 1064 nm. Frequency
calibration of the instrument was performed using the sulphur line at 2 17 cm-'. The laser
power was 2600 mW and 32 co-added spectra were collected at a resolution of 4 cm-'.
The recorded spectra were analyzed using Grams 32 software (Galactic Industries Corp.,
Salem, NH). The cream layer and lysozyme (25% in D20)spectra were baselined, and the
intensity was normalised on the phenylalanine peak at 1004 ~ m - ' . ~ Corn , oil was
normalised on the 2855 cm-' peak. The corn oil spectrum was subtracted from the cream
emulsion spectrum by zeroing the C=O ester band at 1750 cm-', which is present in oil
but not in proteins. The difference spectrum was used to detect any changes in the
residual lysozyme signal compared with the spectrum of lysozyme (25% in D20) on its
own. In addition, subtraction of the normalized spectrum of lysozyme (25% in D,O)
solution from that of the emulsion cream layer gave a difference spectrum which was
used to detect any changes in the residual oil spectrum. The bands in the spectra were
assigned to protein or oil vibrational modes based on the literature.I6
2.4 Results and Discussion
The emulsion separated into three layers, with a top layer of corn oil, a middle cream
layer, and an aqueous solution at the bottom of the tube.
2.4.1 The Cream Layer.

Figure l a shows the spectrum of the cream layer in the 500-1800 cm-' region. By
subtracting the normalised spectrum of corn oil (Figure lb) from the cream layer
spectrum (Figure la) the difference spectrum obtained (Figure lc) was found to be
different from the spectrum for lysozyme in D 2 0 (Figure 1d). indicating protein-lipid
interactions.l 6
The main changes which occurred in the difference spectrum (Figure lc) were a
decrease in the intensity of tryptophan bands (760, 1013, 1340 and 1360 cm-' ) indicating
exposure of tryptophan residues, possibly to a more hydrophilic environment, due to
denaturation of the protein molecule. In addition, increased intensity observed in the CH
bending region may also indicate protein-lipid interactions involving C-H groups.
The intensity ratio I,,,/I,,, of the tyrosine doublet at 850 cm-' and 830 cm-' decreased to a
value of 0.5 for the protein in the cream layer compared to a value of 1.O for lysozyme in
D20. A lower 1,,,/1,3, ratio indicates tyrosine residues in a buried and more hydrophobic
environment or those acting as hydrogen donors rather than acceptors." Conformational
shifts were also observed with the disulphide groups of lysozyme from 508 cm-' to 525
cm-' in the cream layer, similar to those observed in heated ly~ozyme.~
Although it was not possible to characterise the protein secondary structure in the
h i d e I region near 1660 cm-' in the cream layer, due to interference from corn oil bands,
an increase in the protein P-sheet structure in the 980-990 cm-' region was noted
indicating protein-lipid interactions or protein-protein interactions at the interface. An
increase in unordered structures was also indicated by the increased intensity of the
h i d e 111' band at 965 cm-'.16
Further changes in the hydrophobic and polar groups were evident in the CH
stretching band 2800-3100 cm-'. of the cream layer (Figure 2a). Subtraction of corn oil
(Figure 2b) from the cream layer (Figure 3a) gave a difference spectrum (Figure 2c)
which did not resemble lysozyme on its own Figure 2d), but showed negative signals at
2855 and 301 1 cm-', suggesting protein-lipid interactions involving CH groups.
This interaction was confirmed on subtraction of lysozyme (Figure 3b) from the cream
layer (Figure 3a) to give a difference spectrum (Figure 3c) which did not resemble corn
oil on its own (Figure 3d) but showed a decreased signal at 2855 cm" (lipid CH,
symmetric stretch) and at 301 1 cm-' (the =C-H stretch of the unsaturated fatty acid
hydrocarbon chains). In addition, the intensity ratios 1 2 9 0 8 2 9 3 3 and 129,,/1,,,, of the C-H
stretching bands at 2855,2900 and 2933 cm-l were both higher in the difference spectrum
than in the spectrum of corn oil alone. These changes suggested hydrophobic
interactions involving the C-H groups of the lipid with protein in the cream layer to give a
stable emulsion.'6
2.4.2 The Aqueous Phase and Oil Layer

The aqueous layer at the bottom of the tube contained protein only as indicated by the
spectrum which resembled that of 25% lysozyme in D,O solution. The main difference
was in the shoulder of the D,O signal, which was lower for the bottom aqueous layer
compared with lysozyme solution; this confirms the changes observed in the D,O signal
of the difference spectrum of the cream layer and corn oil (Figure 3c) which could be
affected by the corn oil or by the protein.I6 In a food emulsion the effect of oil on the
water molecules and their subsequent restructuring may also affect adjacent protein
groups and hydrophobic interactions.
The spectra of the top layer was similar to that of corn oil except for the presence of a
small D 20 peak confirming the interaction of lipid with D,O. Any traces of protein
present in the top layer which could affect the lipid-D,O interface, were not detected.
From the foregoing, Howell et a l l 6 concluded that the presence of lipids can alter the
molecular structure of proteins and result in the exposure of hydrophobic groups, changes
in the secondary structure and conformation of disulphide groups. The mechanism of
protein denaturation by lipids may be due to protein-lipid complexing or due to the
restructuring of water molecules surrounding proteins.
___-._
-
*bg
(a)
.d
v1
e
c1
E
tb)
.I
-8
v1
.1
m
( 0 -
9
5 <(
-(a)
*F
A . h a n m a - m m
Wavenumber cm-'
Figure 3. FT-Raman spectra (500-3400 cm-1 region) for a) cream layer b) lysozyme c) cream-
lysozyme difference spectrum and d) corn oil
3. ELUCIDATION OF PROTEIN-LIPID INTERACTIONS BY DIFFERENTIAL

SCANNING CALORIMETRY
Differential scanning calorimetry (DSC) can indicate conformational changes in proteins

that are exposed to or interact with oil. In this study the effect of methyl linoleate on the
light meromyosin fraction of the myosin molecule was examined. Myosin can be cleaved
enzymatically into subfragments; these are heavy meromyosin (HMM) comprising a
globular head region S1 and an S2 fragment as well as light meromyosin (LMM) which is
a helical coiled-coil rod fragment.
3.1 Materials and Methods
Myosin was prepared from fresh minced cod fillets according to Martone et a1.I'
Digestion o f myosin with chymotrypsin was undertaken by the method of Maita et el.*'
The LMM fraction was treated with 0.1% methyl linoleate and examined by DSC.
3.1.1 Differential scanning calorimetry. Changes in the thermodynamic parameters

of LMM were examined by DSC. The sample (0.750 g) was heated at OS"C/min, in
0 072 -
0 082 ..
0 0 5 2 --
0 042 ..
% 0 032 ..
Temperature OC
Figure 4. Differential scanning calorimetry spectra of (a) light meromyosin (LMM)

fraction and (b) LMM in the presence of 0.1% methyl linoleate (Badii and Howell")
Setaram MicroDSC VII with an equal weight of water as reference. The initial
temperature was 10C and final temperature 80C before cooling to 10C (cycle 1). The
sample was reheated and cooled (cycle 2) to establish the baseline. The transition
temperature (Tm) was noted and the enthalpy change (AH) was measured by integrating
the area under the peak.
3.1.2 Results. Figure 4a shows two main endothermic transitions in the LMM fraction
on heating from 20-80C. On heating the second time the protein was found to be heat
reversible and only partially denatured. In the presence of only 0.1% methyl linoleate the
transitions were considerably reduced indicating major conformational changes in the
native LMM structure (Figure 4b). On heating for the second time the protein was totally
denatured in the presence of methyl lino1eate.21The results indicate a major effect of oil
on an otherwise very stable protein.
4. EFFECT OF PROTEIN-LIPID INTERACTIONS ON RHEOLOGICAL

PROPERTIES
The effect of oil on the rheological properties of muscle and other proteins have been
investigated by the a ~ t h o r . ~ItI -is~interesting
~ to note that the addition of methyl linoleate
or fish oil can increase the G and G values of the protein especially if it is oxidised. In
this paper water-soluble proteins extracted from fish fillets exposed to methyl linoleate
are used to illustrate the effect of lipids on the protein texture.
4.1 Method
Water soluble sarcoplasmic proteins were extracted from fish muscle using 50mM
phosphate buffer pH 7.5. 21 The rheological properties of the proteins in the presence of
0.1% methyl linoleate of fish oil were assessed on a Rheometrics constant stress 100
rheometer using a parallel plate geometry with a gap of 0.3mm. A temperature sweep was
undertaken heating the proteins from 25 to 90C and cooling back to 25C. The stress
was 1 Pa and frequency 1 radsec.
4.2 Results
The results indicated that the increase in G was greater for proteins containing methyl
linoleate compared to untreated control samples on heating. However, on cooling to the
final temperature of 20C, the G values were lower than that of the control.2
The results obtained for proteins in the presence of fresh oil differed from those
exposed to oxidised oil. In the presence of oxidised fish oil, Saeed and Howell reported
an increase in G for fish muscle paste both on heating and cooling. The effect was
reduced if antioxidants C and E were added to the oxidised oil and fish muscle during
storage prior to analysis confirming that the increase was due to the presence of oxidised
lipid. The increase in the G was due to the transfer of free radicals from the oxidising
lipid to the protein during storage as evidenced by ESR spectroscopy; this resulted in
protein aggregation which was clearly indicated by fluorescence ~pectroscopy.~~~ 25
14.00
1m.w.
tm.oo.
mw .
n
BOD-
i3
4.00- El
El
""1
0.00
0.00 10.00 ao.00 a.w 40.00 50.00
Temperature O
c
moo 70.00 ~0.w
I
~0.w 10o.00
I+wsP G' ---wsP+l%Ol c?]
Figure 5. G' values for a) water-soluble proteins and b) water soluble proteins treated
with 0.1% methyl linoleate heated from 20 to 90C and cooled back to 20C.
5. CONCLUSIONS
The presence of lipids causes major changes in the structure of a variety of proteins
by direct interaction of lipids with proteins or by inducing protein-protein interactions
via free radicals and re-structuring water molecules around the proteins. The resultant
structural changes influence the rheological properties of proteins and texture of food
products.
ACKNOWLEDGEMENTS
Financial support from The European Commission and The Royal Society, UK and the
Natural Sciences and Engineering Research Council of Canada are gratefully
acknowledged. The author would also like to thank Dr. Farah Badii and Dr. Suhur Saeed,
School of Biomedical and Life Sciences, University of Surrey; Dr. Henryk Herman and
Nicola Walker, Chemistry Department, University of Surrey and Professor E. Li-Chan,
Food Science, University of British Columbia, Canada.
References
1. N. K. Howell, N. Protein-protein Interactions. In Biochemistry of Food Proteins.

Ed.B.J.F.Hudson. Elsevier Applied Science Publ.Ltd.,Essex, 1992. pp 35-74.
2. N. K. Howell in Gums and Stabilisers for the Food Industry 10. Ed. G.O. Phillips.,
P.A. Williams Royal Society of Chemistry, Cambridge UK, 2000 pp 3 17-327
3. E. Dickinson, An Introduction to Food Colloids, 1992, Oxford University Press,
Oxford.
4. N, K. Howell, E. Li-Chan, Int. J. of Food Sci. and Technol., 1996,31,439-451.
55 N. K. Howell, G. Arteaga, S. Nakai, E. C. Y. Li-Chan, J Agric. Food Chem. 1999,
47,924-933.
6. N. K. Howell, N. A. Yeboah, D. F. V. Lewis, Int. J of Food Sci. and Technol.,1995,
30,321-334.
7 . H. Sadeghi-Jorabchi, P.J. Hendra, R. H. Wilson, P. S . Belton, J. Amer. Oil Chem.
SOC.1990,67,483-486.
8. V . Baeten, P. Hourant, M.T. Morales, R. Aparicio, J Agric. Food Chem. 1998, 46,
2638-2646.
9 . J. K. Agbenyega, M. Claybourn and G. Ellis, Spectrochim. Acta, 1991, 47A (9/10),
1375-1388.
10. M. T. Devlin, I. W. Levin, J Raman Spectroscopy 1990,21,445-451.
11. R. J. Shannon, J. Fenerty, R.J. Hamilton, F. B. Padley, J Sci. Food Agri. 1992, 60,
405-417.
12. P. Tandon, G. Forster, R. Neubert, S . Wartewig, J. Molec. Structure 2000,524, 201-
215.
13. D. Aslanian, M. Negrerie, R. Chambert, Eur. J Biochem., 1986,160,395-400.
14. P. Carmona, J. M. Ramos, M. de Cozar, and J. Monreal, J Raman Spectroscopy,
1987,18,473-476.
15. K . Larsson, in Lipids Volume 2: Technology, eds. R. Paoletti, R, G. Jacini, R,
Porcellati, Raven Press, New York, 1976,355-360.
16. N. K. Howell, H. Herman and E. C.Y. Li-Chan, J. Agrk. Food Chem., 2001, 49,
1529-1533.
17. E. C. Y. Li-Chan, S. Nakai, M. Hirotsuka, in Protein Structure-Function
Relationships in Foods, eds. R. Y. Yada, R. L. Jackman, J. L. Smith, Blackie
Academic and Professional: London, UK, 1994, pp. 163-197.
18. A. T. Tu, in Spectroscopy of Biological Systems; Clark, R. J . H., Hester, R. E., Eds.;
John Wiley & Sons: New York; 1986, pp. 47-1 12.
19. C. B. Martone, L. Busconi, E. J. Folco, J, R. E. Trucco and J. J. Sanchez, J. Food
Sci., 1986,51, 1554-1555.
20. T. Maita, E. Yaajima, S, Nagata, T. Miyanashi, S. Nakayama and G. Matsuda, J.
Biochem, 1991,110,75-87.
21. F. Badii and N. K. Howell, J. Sci. Food Agric., 2001Submitted.
22. S. Saeed and N. K. Howell, J. Sci. Food Agric., 2001, In press.
23. F. Badii, F and N. K. Howell, Food Hydrocolloids, 2001, In press.
24. S. Saeed, S. A. Fawthrop and N. K. Howell, J Sci. Food Agric., 1999,79, 1809-18 16.
25. N. K. Howell, in Food Hydrocolloids I: Physical Chemistry and Industrial
Application of Gels, Polysaccharides and Proteins. Ed. K . Nishinari, Elsevier,
Amsterdam 1999. Pp 399-406.
Rheological Aspects
HELIX-COIL TRANSITION IN THERMOREVERSIBLE GELS
K.Nishinari, Y.Nitta, E.Miyoshi*, S.Ikeda and T.Takaya

Graduate School of Human Life Science, Osaka City University
3-3-138, Sugimoto, Sumiyoshi, Osaka, Japan 558-8585
*Present address: Osaka University of Foreign Studies
Abstract
The step-like change was found in the temperature dependence of mechanical loss
tangent, specific ellipticity in circular dichroism, and the endothermic and exothermic
peak were found at the same temperature range in heating and cooling DSC curves for
gellan gum gels. It was concluded that helix-coil transition occurred in a cylindrically
moulded gel of gellan gum on heating and on cooling while the size and shape of the gel
was kept. This transition was thermoreversible. The possible molecular mechanism
was discussed based on a traditional model where junction zones are linked by flexible
molecular chains and on a fibrous model where only stiff chains associate to build up a
three dimensional network. In both models, the number of junction zones which
contribute to the elasticity may be changed by the helix-coil transition. It was compared
with the increase of the elastic modulus of K. -carrageenan gels induced by the
immersion in alkali metal salt solutions.
Introduction
As is well known, solutions of gelatin, agarose, K -carrageenan, gellan form a gel on
cooling. The coil to helix transition is believed to be pre-requisite for sol to gel
transition. Although there have been many studies dedicated to the molecular
mechanism of thermo-reversible gel-sol transition, it remains to be a matter of debate.
Gellan gum is a microbial polysaccharide consisting of a tetrasaccharide unit, P
-D-glucose, p -D-glucuronic acid, P -D-glucose, Q -L-rhamnose, and its gelation
behaviour has been studied extensivelyqd. Gellan gum was chosen as a model
biopolymer to understand better the gel-sol transition, and the common sample has been
distributed to 15 laboratories with different techniques including light scattering,
osmometry, rheology, differential scanning calorimetry (DSC), circular dichroism (CD),
x-ray small angle scattering, dielectric measurement, nuclear magnetic resonance (NMR),
electron spin resonance. The results of the collaborative studies were published in
special issues of Food hydro colloid^^, Carbohydrate Polymers and Progress in Colloid
and Polymer Science'. On coollng a gellan gum solution, it was found that the
molecular weight doubled by light scattering" and osmotic pressure measurement' and
that the radius of gyration increased". The specific ellipticity in CD12 was found to
decrease steeply on cooling at the same temperature range. Storage shear modulus was
found to increase steeply13 and an exothermic peak a~peared'~, and the spin-spin
relaxation time in NMR12 decreased at the same temperature range on cooling. AH
these data suggest that the transition fiom a single chain to a double helix occurs on
cooling a gellan gum solution, and that this is a reversible change; helix-to-coil transition
occurs on heating.
In the present work, the temperature dependence of complex Young's modulus,
specific ellipticity in CD were examined and DSC measurement was performed to get
fbrther insight into the gel-sol transition and helix-coil transition in gellan gum.
Materials and Methods

Materials
Gellan gum used in the present study was the first common sample, potassium form
deacylated gellan gum KGG-1 used in the collaborative research group in Japan. The
molecular characteristics and gelling behavior were reported in the special issue of Food
Hydrocolloids'. The content of the inorganic ions was as follows: Na', 0.19%; K',
2.68%; Ca", 0.512%; Mg2+, 0.146%. The molecular weight was determined by
converting it into the tetramethyl ammonium form using light scattering ( K ~ b o t a ' ~and
)
osmometry (Ogawa15) as Mw=2.1 X lo5 and M n =0.5 X lo5.
Preparation of a cylindrical gel for rheological measurement

Gellan gum KGG-1 was swollen in distilled water at 40C for 20 hours and then
heated at 90C for 1 hour to dissolve completely. Hot solutions were poured into
teflon cylindrical (20mm diameter, 30mm height) moulds, and then the temperature was
lowered to a room temperature and then kept at 5C for 20 hours.
Methods
The complex Young's modulus of a cylindrically moulded gel was determined by the
observation of longitudinal vibrations, from the amplitude ratio and the phase difference
of the stress and strainI6.
The schematic representation of an apparatus for the determination of complex
Young's modulus is shown in Fig. 1. The gel is immersed in silicone oil to prevent the
evaporation of water and to control the temperature. The advantage of this method is
Rheological Aspects 87
that it is completely fiee fiom a notorious problem of slippage which affects quite often
the viscoelastic measurement in shear oscillational mode", I*. The temperature was
raised fiom 10C to 55C at an interval of 5C. The temperature was kept at each
temperature for 15 min. Then, the temperature was lowered from 55C to 10C at the
same rate. Since it is well known that the storage modulus of a gel does not depend so
much on the fiequency'", the fiequency is fixed as 3Hz and the amplitude was also fixed
as 100 1.1 m (strain 0.03) for this apparatus.
E"
II
Fig.1 Schematic diagram of the apparatus for determining complex Young's modulus
of gels (Rheolograph Gel, Toyo Seiki Seisakusho Ltd., Tokyo). G, cylindrical gel; 0,
silicone oil; V, vibrator; S, strain gauge for detection of stress, s, strain gauge for
detection of strain, P, phase sensitive circuit.
The specific ellipticity at 202 nm in CD was measured by JASCO-820A

spectropolarimeter (Jasco, Tokyo) in the temperature range fiom 5C to 55C at 0.5"C
/Inin.
DSC measurements were performed by a Setaram micro DSC-I11 (Caluire, France).
The temperature was raised fiom 5C to 55C at O.S"C/minand then lowered to 5C at
the same rate.
Results and Discussion

Figure 2 shows the storage Young's modulus E' of gellan gum gels of various
concentrations fiom 1.0 to 2.0 wt% on heating and on subsequent cooling. E' decreased
gradually with increasing temperature, and the decreasing tendency became more marked
with decreasing concentration as has been observed for agarose gels''. On cooling E'
did not recover the value during an equilibration time of 15 min above 25"c, and it
recovered the initial value at 5C.
1.2%
1%
0 20 40 60 80
Temperature / "c
Fig.2 Temperature dependence of storage Young's moduli E' of gellan gum gels of
various concentrations. Frequency, 3Hz; equilibration t h e , 15min (solid line) or 60min
(dotted line).
The temperature dependence of elastic modulus of thermo-reversible gels was

discussed by a reel-chain model2'. It was assumed that the junction zones consist of
aggregated ordered chains and some segments are released with increasing temperature
fkom junction zones and they are incorporated into it with decreasing temperature. A
junction zone is a reservoir of these segments. According to this model, elastic
modulus E of thermo-reversible gels as a h c t i o n of temperature depends on binding
energy E , mean end-to-end distance of rmof chains which connect junction zones, and
the ceiling number of segments Y , i.e. the upper limit of the number of segments which
can be released from a junction zone just before gel-to-sol transition occurs. According
to this treatment, E increases monotonically for large values of E , r,, or v , while E
decreases monotonically for small values of these three parameters. In the intermediate
values of these parameters, E increases up to a certain temperature and then decreases

with increasing temperature, which has been observed for concentrated gels of agarose,
gellan, and poly(viny1 alcohol) . The temperature dependence of gels of agarose",
gellan2' and poly(viny1 alcohol) 2o was interpreted by the reel-chaii model.
When the equilibration time at each temperature was prolonged to one hour instead
of 15 min, the storage Young's modulus E' slightly decreased at higher temperatures
than 35C whilst E' did not change so much below 35C both in the heating and cooling
processes (dotted line in Figure 2). It is suggested that some segments are released
from weak junction zones when the temperature is kept at a higher temperature and this
may decrease the number of junction zones hence the decrease in the number of
elastically active chains. It is quite possible that some weak junction zones are
disintegrated with increasing temperature, which is induced by helix-coil transition but
does not lead to gel-to-sol transition because these weak junction zones exist only locally.
On the other hand, when the temperature is lowered, a fiee chain which is linked to only
one junction zone may encounter another fiee chain, and since these chains are converted
into helices at lower temperatures, they may build up a new junction zone, hence the
number of elastically active chains increases. This will be discussed again later.
Temperature dependence of storage and loss Young's moduli, E' and E", of gellan
gum gels is shown in Fig.3 together with a heating DSC curve and the specific ellipticity
in CD measurements. Storage Young's modulus E' decreased gradually while loss
Young's modulus E" showed a step-like decrease around 26C. The temperature was
kept at each measurement temperature for 15 min before the measurement. The heating
DSC curve showed an endothermic peak at this mid-point transition temperature. The
specific ellipticity at 202 nm in CD measurements showed a steep increase at the same
temperature range.
On cooling at the same rate, storage Young's modulus E' increased gradually while
loss Young's modulus E" showed a step-like increase around 26C. The cooling DSC
curve showed an exothermic peak at this mid-point transition temperature, which was
sharper than the corresponding endothermic peak. This was discussed using a zipper
model approach22:the transition caused by the opening of zippers will start as soon as
the temperature arrives at a certain temperature which depends on the average rotational
fieedom in gel state on heating whilst the pair-wise coupling cannot start so easily on
cooling because of the diflticulty for a long molecule to find its partner in appropriate
positions for the zipper construction. The specific ellipticity at 202 nm in CD
measurements showed a decrease at the same temperature range which is steeper than
the corresponding increase on heating because of the same reason stated for DSC curves.
These experimental findings suggest that helix-coil transition occurs at 26C in a

cylindrically moulded gel. Although no drastic change in shape and size of the
cylindrical gel was recognised, the helix-coil transition occurred inside the gel.
Helix-coil transition can occur in different ways. Segment can be released fiom
junction zones on heating as was discussed in reel-chain model approach. On cooling,
these released segments are reeled into junction zones so that junction zones become
longer and/or thicker (Fig.4(a)). If the elastic modulus is mainly determined by the
5 0.1 5 0.1
.
Ld
a .
L
u
e
Y Lou Lo
- 4
-
8 .* 4 P
'Ld 'Ld
a a
9 9
3 1 I 1 1 I
0.01 3 0.01
-20 0 20 40 60 80 100 -20 0 20 40 60 80 100
0.6 0.6
- 0.4 - 0.4
'E 's
# 0.2 .-0 8 0.2
Q
5 mT
Ns 0 50 E
.
0 e-
2 ** Q
6 4
M
.
'0
E -0.2 1 2 -0.2
-
k
N
N
-0.4
-5 N
-0.4
-0.6 -0.6
-20 0 20 40 60 80 100 -20 0 20 40 60 80 100
Temperature / "c Temperature / "c
Fig.3 Temperature dependence of storage and loss Young's moduli, E' (A) and E"(A),
DSC curves (solid line) and the specific ellipticity T a t 202nm (+) for a
and tan6 (0).
1.2% gellan gum gel. Left, heating; right, cooling. Scan rate for DSC and CD
measurements was O.S"C/min, and the equilibration time for measurements of E' and E"
was 15min.
I :I !
Fig.4 (a) Change in the size of junction zones by helix-coil transition, (b) Change in the
number ofjunction zones linked by flexible molecular chains by helix-coil transition, (c)
Change in the number of supporting points in a fibrous model by helix-coil transition.
number of elastically active chains which connect junction zones as in the theory of
rubber elasticity, the increase in elastic modulus should be attributed to the increase in
the number of elastically active chaiis rather than to the thickening or lengthening of
junction zones. It is quite possible that long segments released fiom junction zones
form a new junction by helix formation (Fig.4(b)). Helix-coil transition does not
necessarily occur only at the cham ends, but it also may occur at the intermediate point
of long chains (Fig.4(b)).
There may be diferent junction zones with different thermal stabilities. Weak
junction zones with low thermal stabilities may disintegrate at lower temperatures which
are directly related to helix-to-coil transitions, and these chains produced fiom the
disintegration of helices may form helices on cooling only at lower temperatures. The
distribution of thermal stability may cause a broad endo- and exo-thermic peak in DSC
heating and cooling curves as well as gradual increase and decrease of the specific
ellipticity in heating and cooling CD measurements. Morris proposed recently a fibrous
model for a gellan gum gel based on atomic force microscopic observation23724.
According to this model, the elasticity of gels arises mainly fiom stretching and bending
of fibres that consist of aggregated helices. When some ends of these fibres are
converted into flexible chains by helix-to-coil transition, these ends will cease to
contribute to the elasticity and then the elastic modulus will decrease(Fig.4(~)). Even if
one of these conformational changes occurs, it does not necessarily leads to the
gel-to-sol transition when these changes are only local; only less ordered helices are
converted into coils, and the whole network structure is not broken down, and keeps the
size and shape of the gel.
It was reported that the complex Young's modulus of carrageenan gels immersed in
alkali metal salt solutions increased with the lapse of time25 (Figure 5). The
concentration of gels was found to increase by immersing gels in salt solutions (Table 1).
It is well known that the elastic modulus as a h c t i o n of concentration is written as E =
kc", where c is the concentration, k and n are constants. For this IC -carrageenan gel, it
was found that k=4.48 and n=3.16 respectively. The change in the elastic modulus was
estimated by the concentration change using this relation between the elastic modulus
and concentration. As is shown in Table 1, the observed change in the elastic modulus
was far larger than that estimated by the concentration change. The increase in the
elastic modulus induced by the structural change was estimated far larger than that
induced by the concentration change.
"I I
0L 1 2 3 - 4 5 6 3
time t Ih
Fig.5 Storage Young's modulus E' (solid line) and loss tangent tan6 (broken line) of 2%
K: -carrageenan gels immersed in 0.5M alkali metal salt solutions at 25C as a h c t i o n
oftime. 0 , CsC1; 0 , KC1; a,NaCb x, LiCl.

RheologicalAspects 93
Table 1 Change in storage Young's moduli E' at 2.5Hz of K. -carrageenan gels. The
initial concentration and size of gels was 2w/w %, and 30mm height, 18.8mm diameter.
Initial value of E' was 2.4 X lo4 Pa. Temperature, 25C.
LiCl NaCl KCl CSCl

Observed value of E'
after 6 hr of immersion 8.1 9.0 25.0 26.9
(lo4 Pa)
(height, diameter) (29.0, 18.5) (29.4, 18.5) (28.9, 17.9) (28.7, 17.8)
after 6 hr of immersion
(mm)
Concentration 2.09 2.13 2.30 2.34

after 6 hr of immersion
(W/W?h)
Estimated value of E' 5.6 4.9 6.2 6.6

fkom E' = 4.48 X lo4X c3.16
Therefore, the marked increase in the elastic modulus induced by the immersion of
gels in alkali metal salt solutions was attributed mainly to the structural change in the gels
and not to the concentration change. This structural change may be a similar change
which was observed in the present study. The introduction of cations into gels may
cause the coil to helix transition and make a new junction zones in the gel.
Concluding remarks
The creation and annihilation of a crosslink was discussed by Yamamoto26in 1957,
and by a reel-chain model2', and fkther by Tanaka27-29 using a transient network theory.
Since Tanaka has successhlly determined the junction multiplicity and the number of
sequential units on a chain participating in a network junction in thermoreversible
gelation for various polymers with different molecular weights, it is expected to extend
his theory to the present case to determine the molecular parameter of gellan gels. It is
urgently required to examine the gelling behaviour of gellan gum with different molecular
weights.
References
1. K.Nishinari, Colloid Polym. Sci., 1997,275, 1093.
2. K.te Nijenhuis, Adv. Polym. Sci., 1997, 130, 1.
3. A.H.Clark and S.B.Ross-Murphy, Adv. Polym. Sci., 1987,83, 57.
4. G.R.Sanderson, Food Gels, Ed., P.Harris, Elsevier, 1990, p. 201.
5. K.Nishinari, Gums and Stabilisers for the Food Industry 8, Eds. G.O.Phillips, P.A.Williams
and D.J.Wedlock, IRL Press, Oxford, 1996, p. 371.
6. M.Rinaudo, Novel Macromolecules in Food Systems, Eds., G.Doxastakis and V.Kiosseoglou,
Elsevier, 2000, p.239.
7. Food Hydrocoll., 1993, 7,Special issue on gellan gum, 361-456.
8. Carbohydr. Polym., 1996,30, Special issue on gellan gum, 75-207.
9. Prog. Colloid Polym. Sci., 1999, 114, Special issue on gellan gum 1-135.
10. R.Takahashi, M.Akutu, K.Kuboka and K.Nakamura, Prog. Colloid Polym. Sci., 1999, 114,
1.
1 1. E.Ogawa, Prog. Colloid Polym. Sci., 1999, 114, 8.
12. S.Matsukawa, Z.Tang and T.Watanabe, Prog. Colloid Polym. Sci., 1999, 114, 15.
13. E.Miyoshi and K.Nishinari, Prog. Colloid Polym. Sci., 1999, 114, 68.
14. T.Okamoto, K.Kubota and N.Kuwahara, Food Hydrocoll., 1993, 7, 363.
15. E.Ogawa, Food Hydrocoll., 1993, 7, 397.
16. K.Nishinari, Jpn. J. Appl. Phys., 1976, 15, 1263.
17. R.K.Richardson and F.M.Goycoolea, Carbohydr. Polym., 1994,24,223.
18. H.Zhang, M.Yoshimura, K.Nishinari, M.A.K. Williams, T.J.Foster and I.T.Norton,
Biopolymers, 200 1,59,
19. K.Nishinari, M. Watase, K.Kohyama, N.Nishinari, S.Koide, K.Ogino, D.Oakenfbll,
P.A.Williams and G.O.Phillips, Polymer J., 1992, 24, 871.
20. K.Nishinari, S.Koide and K.Ogino, J.Phys. (France), 1985,46, 793.
2 1. M.Watase and K.Nishinari, Food Hydrocoll., 1993,7,397.
22. K.Nishinari, S.Koide, P.A.Williams and G.O.Phillips, J.Phys. (France), 1990, 51, 1759.
23. V.J.Morris, A.R.Kirby and A.P.Gunning, Prog. Colloid Polym. Sci., 1999, 114, 102.
24. V.J.Morris, A.P.Gunning, A.R.Kirby, A.R.Mackie and P.J. Wilde, In "Hydrocolloids, Part 1:
Physical Chemistry and Industrial Application of Gels, Polysaccharides, and Proteins",
K.Nishinari Ed, Amsterdam, Elsevier, 2000, p. 99.
25. M.Watase and K.Nishinari, Colloid & Polym.Sci., 1982,260,971.
26. M.Yamamoto, J.Phys.Soc.Jpn., 1957, 12, 1148.
27. F.Tanaka and S.F.Edwards, J.Non-Newtonian Fluid Mech., 1992, 43,247, 273,289.
28. F.Tanaka and K.Nishinari, Macromolecules, 1996, 29, 3625.
29. K.Nishinari, M.Watase and F.Tanaka, J.Chim.Phys., 1996,93, 880.
MICROSTRUCTURAL ORIGINS OF THE RHEOLOGY OF FLUID GELS
Frith W.J., Garijo X., Foster T.J., Norton I.T.
Unilever Research Colworth, Colworth House, Sharnbrook, Bedfordshire, MK44 1LQ,

U.K.
1. SUMMARY
Fluid, or sheared, gels have considerable potential for use as stabilisers, or rheology
modifiers in products from both the food and personal products areas. Consequently, it is
important for us to understand the origins of their behaviour in order both to improve our
ability to produce the materials and to develop new applications. To this end, studies of the
rheological properties of two sheared agars, at a range of dilutions, are discussed. These
results are compared with data obtained on suspensions of spherical microgels, produced
from the same agar. Such a comparison allows the interpretation of the behaviour of the
fluid gel in terms of the shape of its component microgel particles. It turns out that it is
possible to model the rheological behaviour of the spherical microgels based on an
interparticle potential. Comparison of this model with the fluid gel suspensions serves to
highlight the influence that particle morphology has on the rheology
In addition, the model allows conclusions to be drawn regarding the influence of
processing conditions on the sheared agar microstructure. For instance, it appears that at
sufficiently high impeller speeds, continuous throughput and stirred pot production routes
for sheared agars produce a product microstructure that is largely independent of the
concentration of agar used in the process. At lower impeller speeds however, the stirred pot
route appears to produce material that has a microstructure that is quite dependent on agar
concentration.
2. INTRODUCTION
The ability of certain biopolymers to form fluid gels when gelled whilst being subjected
to an appropriate shear field has been known for some years, and their application in foods
and personal products was patented in 19901. The importance of these materials in food
and personal product applications lies in the novel nature of their rheology, and hence in
their utility as stabilisers. It is the flow behaviour, and its origins in that fluid gel
microstructure, that is the subject of this study.
Fluid gels are formed by applying an appropriate flow field to a biopolymer solution
whilst it gels. They can be produced by a variety of gelation mechanisms, although this
work is restricted to those formed via thermal gelation. In general, the essential feature of
the process seems to be that the sample should experience a uniform flow history, that is it
should either be well mixed or subjected to a uniform flow field, and the flow should be
sufficiently vigorous. An additional requirement, in the case of thermal gelation, is that

temperature gradients in the sample should be kept to a minimum, either by mixing or
sufficiently slow cooling. This is necessary in order to prevent the build-up of unsheared
gel regions in the sample. The use of stirred pot and continuous throughput routes has been
investigated in the both of which employ complex well mixed flow fields.
However, it has also been shown that such flow fields are not required in order to form
fluid gels, and that simple shear fields in couette or cone & plate flow are sufficient4.
It is now known that the materials formed in this manner consist of a highly
concentrated suspension of gel particles of irregular shape in a continuous phase that is
essentially pure (i.e. free from gel forming polymer) aqueous medium4. However, the
detailed characterisation of the size, morphology, mechanical properties and volumes of
the gel phase has proved extremely challenging, precisely because of the complex shape of
the particles. Light scattering and optical microscopy have been employed with some
S U C C ~ S S ~ and
- ~ , indicate that the gel particles seem to consist of non-spherical primary
particles that are also (apparently irreversibly) aggregated into larger secondary particles.
Confocal images of fluid gels4, and centrifugation studies3, appear to indicate that the
volume fractions of gel particles are in the region of 0.5-0.8. However, these results do
suffer from a high degree of uncertainty arising from artefacts such as the resolution of the
microscope and sample preparation methods.
Taking into account the studies discussed above, there is still little published on
systematic investigations of the microstructural origins of the rheological properties of
fluid gels. Considerable work has been done, however, and it is the aim of this study to
attempt to draw past work together and so allow some conclusions to be drawn regarding
the flow behaviour of fluid gels. The irregular shape of the fluid gel particles has long been
believed to play an important role in the flow behaviour of fluid gels3, and the aim of this
work is to test this hypothesis and, if possible model the behaviour.
The approach taken here is to compare the behaviour of fluid gels with that of model
suspensions of spherical microgel particles, made from the same biopolymer. In this way it
is hoped to keep the particle deformability and interactions constant whilst altering particle
shape in a controlled manner. Comparison of the rheological behaviour of the two types of
system (fluid gel and model microgel) should then allow investigation of the influence of
particle shape on the flow properties of the fluid gels.
Studies are reported here of the linear viscoelastic behaviour of both fluid gels and
model microgel suspensions, as a function of dilution of the sample. The behaviour is
compared with published models for the behaviour of model suspensions with soft
repulsive interactions between the particles.
3. MATERIALS
3.1. Fluid Gels
Fluid gels are produced by a variety of routes, these include continuous processes, stirred
pots and rheometric flows as discussed in the introduction. The materials used in this study
were produced from Luxara agar (type 1253) using a continuous throughput process. Fluid
gels were made with two different agar concentrations, 0.75% and 1.75%, using a pilot line
as described below:
1. Agar was dispersed in cold, deionised water using a Turrax high shear mixer.
2. A 95' C water jacket was used to heat the agar dispersion / solution to 90' C.
3. The solution was processed through three Contherm scraped surface heat
exchangers (SSHEs), the units were set as follows:
Unit 1 Product temp. 90' C, speed 350 rpm.

Unit 2 Product temp. 18' C, speed 350 rpm (+2"C wall temp.)
Unit 3 Product temp. 0-lo C, speed 350 rpm (-25C wall temp.).
Throughput ca. 2.5 kg.min-'.
The temperature profile was very similar for the 0.75% and 1.75%.
4. Collected at 4' C and stored at 2-5' C.
The materials produced from this process were utilised without further purification.
Lower concentrations were produced by diluting the sample with de-ionised water, higher
concentrations were produced by centrifuging the samples at successively higher speeds
and decanting off the supernatant.
3.2. Model Microgel Suspensions

The spherical microgel suspensions used in this study were prepared by an emulsion route
that has been described previou~ly~'~. 200ml of 2% or 5% agar (Luxara 1253) was
dissolved by heating in a boiling water bath. This solution was then emulsified in
groundnut oil at -80C using a Silverson homogeniser set at minimum or maximum speed
for the 2% and 5% solutions respectively. A suitable surfactant (2% Hypermer B246, ex.
ICI) was dissolved in the oil phase. This procedure resulted in a water in oil emulsion with
a drop size in the region of 1-1Opm. On cooling, the agar gelled, forming the microgel
particles. An aqueous solution of surfactant (1% Tween 20 ex. ICI) was then added with
the Silverson set on maximum speed until phase inversion was achieved, this resulted in
the majority of the microgel particles being released into the aqueous phase, though some
remained trapped within oil droplets. The microgel was then separated by centrifugation.
As produced, the microgel had a significant contamination of surfactant and soluble
salts, the material was subsequently purified by repeated centrifugation and redispersion in
de-ionized water.
3.3. Particle size measurements

The particle size distributions of the microgels were determined by static light scattering
using a Malvern Mastersizer X (Malvern Instuments, UK). These sizes were also
confirmed by microscopic examination. The microgel sizes were lOpm with an span of
+5pm and 2pm with an span of f l p m for the 2% and 5% microgels, respectively
An attempt was also made to measure the particle size distribution of the sheared gels
using the Malvern Mastersizer. Obtaining reliable size distributions proved difficult
because of the small refractive index difference between the particles and the suspending
fluid. However, microscopic examination of the fluid gels indicated that that structure was
similar to that found in previous These studies employed higher concentrations
of agar, and hence had improved contrast, allowing particle sizes to be estimated to be
between 5 and 5Opm.
3.4. Rheological techniques

All measurements on the sheared agars were carried out on a Weissenberg
Rheogoneometer ex. TA Instruments. The geometry used was 5Omm parallel plate with a
gap of 2mm. Plate surfaces were roughened by attaching P600 emery paper to them with
double sided tape and the sample was covered in mineral oil to prevent evaporation.
The model microgel suspensions were measured using a Rheometric scientific SR200
controlled stress rheometer. As with the sheared agar, a parallel plate geometry was
. . . .. , . . . . . . .., . . . . . . .. , . . . . . . ..
0.01 0.1 1 10
Frequency (Hz)
Figure 1 Frequency dependence of the storage modulus for the 0.75% sheared agar.
1.75%
444A+444433 -f- 1.575 Yo
-A- 1.4 %
t 1.225 %
-+- 1.05 %
+0.875 70(33)
Frequency (Hz)
Figure 2 Frequency dependence of the storage modulus for the 1.75% sheared agar.
employed (40mm diameter, 0.5mm gap) the surfaces of which were covered with p600
emery paper. The sample was covered with dodecane in order to prevent evaporation.
4. RESULTS AND DISCUSSION

Figures 1 and 2 show the frequency dependence of the storage modulus for the 0.75% and
1.75% sheared agar systems at a range of dilutions. In virtually every case, the samples
show a gel-like response, which persists over a broad range of dilutions given the
particulate nature of these suspensions. Such behaviour is characteristic of samples that are
either flocculated or have highly anisotropic particles. The primary particles in these
systems are known to be highly non-spherical and often possess long tails, a shape that is
reminiscent of droplets that have been trapped in the process of breaking up4. These
primary particles also to form aggregates that are very resistant to break-up and do not
form a continuous network throughout the suspension, which is demonstrated by a lack of
any pronounced thixotropic response. Therefore, it seems more appropriate to regard the
materials as a suspension of extended and highly irregular particles.
h
cp
4 1000 7
N
50 0
0
c.
-
v)
z
100:
'0:
.. w 0
0
7
.
U 0
fa, 1:
rn 0
w
.
!
w
0.1 , 0
1
Agar Concentration (wt. YO)
Figure 3 Dependence of the plateau storage modulus on % agar concentration,

when diluted, for the 0.75% and 1.75% sheared agars.
0
0 .
0
ziri 0.01
0
O ' l L
1
Normalized Concentration
Figure 4 Dependence of the normalised plateau storage modulus (G k2)on the
normalised concentration (c/cp)for the 0.75% and 1.75% sheared agars.
The overall result is that the individual particles extend much further through space than
would a spherical particle with the same volume. They can thus interact with one-another
at lower concentrations than is the case with spherical particles, leading to a gel-like
response over a broad concentration range, even though the sample is not aggregated.
The plateau modulus of both sheared agar samples is shown in figure 3, as a function of
the weight concentration of agar. In this log-log plot, it is apparent that both samples show
a similar response to dilution with water, being approximately power law. The 1.75%
sheared agar was also concentrated by centrifugation and removal of supernatant and these
data are also shown in figure 3. In this region, deviation from the power law behaviour is
observed, with the dependence of the modulus on concentration being weaker than at lower
concentrations. Interestingly, the 1.75% sheared agar systems display a lower modulus at a
given agar concentration than do the 0.75% sheared agar systems. This supports the
accepted model of these materials being suspensions of microgel particles that possess a
similar structure, which has been determined by the processing history as much as by the
. polymer concentration. It might be assumed that the sheared agar particles produced in the
1.75 and 0.75% systems have similar shapes and occupy similar volumes at their
respective preparation concentrations, though the 0.75% microgel particles should be more
deformable than the 1.75% particles.
If we assume that the particles in the two sheared agars do have similar volumes when
first produced, then a more appropriate concentration comparison is the overall
concentration divided by the starting agar concentration. Further, it is reasonable to assume
that the sheared agar moduli should scale with the particle moduli which, in turn, should
follow that of a quiescently cooled agar (i.e. a concentration squared law7). To this end the
data are replotted in figure 4 as G/C,2 (Pa.%-*)as a function of c/cp,where c is the overall
agarose concentration and cp is the concentration of agarose at preparation. Both sets of
data now superimpose to a first approximation, which seems to indicate that the structure
and volume fraction of the microgels produced in both samples are indeed very similar,
with only the stiffness of the particles being different. This conclusion can only be an
approximation, however, as it is evident that the concentration dependence of the two
samples on dilution is not identical, which implies a slight difference in particle structure.
This behaviour has also been observed for materials prepared in a stirred pot process
when stirring conditions are similar to those used in the pilot These data are
reproduced in figure 5, and demonstrate that when the tip speed of the impellers in both the
stirred pot and the pilot line are similar (1000rpm and 350rpm respectively) then the
properties of the sheared gels produced are very similar. This implies that, at sufficiently
high tip speeds the sheared gel particles have a microstructure that is largely independent
of concentration and hence show a similar G(c) slope to quiescently cooled agar.
However, at lower tip speeds in the stirred pot, a different concentration dependence is
seen, implying that particle microstructure is quite concentration dependent under such
conditions.
The form of the relationship between sheared gel modulus and concentration, when the
sample is diluted, is characteristic of the sample microstructure. Factors influencing this
relationship include particle deformability, shape, orientation and long range forces. If we
assume that long range repulsive or attractive forces are absent, and that the particle
orientation is isotropic, then it is possible to interpret the dependence of modulus on degree
of dilution in terms of the particle shape. This can be illustrated by comparing the G(c)
behaviour of the sheared gels with that of spherical microgel particles produced from the
same material, this is illustrated in figure 6. The concentrations in this plot have been
normalised, for convenience, so that the preparation concentrations of the sheared agar and
the maximum packing fraction of the spherical microgel are equal to 1.0. In this figure the
maximum packing fraction has been determined from the fit of equation 4 to the data (see
below).
It is clear from figure 6 that the concentration dependence of the sheared and spherical
agar particles is quite different. Since both of these materials are produced from the same
agar, the interactions between the particles can be assumed to be the same. Hence, the
difference in behaviour can only be due to the differing particle shape in the sheared and
spherical agar systems. It is also worth noting that as the concentration increases, the G(c)
behaviour in the samples becomes more similar. It appears then, that as the free water in
the suspensions is progressively removed, the influence of particle shape becomes less
important and increasingly we see the elastic response of the agar gel as it is becomes more
concentrated.
1 1
0 Quiescently gelled agar
1 0 Sheared agar from stirred pol (100rpm)
0 Sheared agar from stirred pol (1000rpm)
Sheared agar from pilot line
h
--- 1.88 power law
.....1.47 power law
L
Y
b loooo 1 P'
Id
loool
100
0.1
0
0
,o-
m .
I
1
m
10
Agar Concentration at Preparation (% wlw)
Figure 5 Comparison of the dependence of sheared gel modulus on the

concentration of agar used in the preparation. Results are shown for samples from
the stirred pot and from the pilot line. Also shown is the modulus-concentration
dependence of quiescently cooled agar.
If the particles in the microgel suspension are spherical and have uniform internal
structure and modulus, then it is possible to develop a predictive model for the modulus as
a function of particle volume fraction. This was done, for sephadex microgels, by Evans
and Lips', and tested for spherical agar microgel particles by Frith and Lips'. The approach
is based on the relationship, due to Zwanzig and Mountain", that relates the modulus of a
dispersion of particles to the pairwise interactions between them e.g.
Where N is the number density, kb is the Boltzman constant, T is the absolute temperature,
g(R) is the radial distribution function and dV(R)/dR is the force that particles exert on one
another as a function of their separation R . Given a relationship for V(R) one may predict the
modulus as a function of concentration. For elastic spheres V(R) is approximated by the
Hertz model ,''
Where ROis the rest radius of the particles, Gpis the modulus of the particles and CT is their
Poisson ratio. If we assume that only the nearest particles interact g(R) can be approximated
by a delta function, thus G is given by:
Here n is the number of nearest neighbours and & is the maximum packing fraction.
Combining eqns. 2 and 3, we get:
. . . . -I
1
I
I
10000
a 2% spherical agar
microgel
5% spherical agar
microgel
0 1.75% Sheared agar
v 0 75% Sheared agar
- Hem' model for 5%
agar microgel
l1
: -- -Hem' model for 2%
a- /' agar microgel
ra' .... G=aQ for 1 75%
Sheared agar
/*
-.-- As above tor 0.75%
d' Sheared agar
0.1 I
0.4 0.5 0.6 0.7 0.80.9 1 2 3
Nominal Concentration
Figure 6 Comparison of the concentration dependence of the plateau moduli

of the sheared agars and the model spherical agar suspensions. Also shown in
this figure are the prediction of the model based on Hertzian interactions
between the particles (equation (4), assuming a particle modulus of 33 Wa).
Where @ed = @'&.When compared with the results obtained from the agar suspensions
(see figure 6), the prediction from the above equation shows a similar form to the spherical
agar particles. The agreement is only qualitative, however, implying either that the particles
are not ideal elastic spheres or that the Hertz model is imperfect. Nonetheless, the similar
forms of the results from the experiments on the spherical agar microgels and from eqn. 4
demonstrate that the observed response arises from the elastic interactions between the
particles. Further, it is also clear that the difference between the sheared and spherical agar
systems does indeed arise from particle shape effects and not from other factors, such as
aggregation.
5. CONCLUSIONS
Comparing the viscoelastic behaviour as a function of concentration of sheared agars with
that of spherical microgel suspensions has highlighted the important role that particle shape
plays in the rheology of these systems. In the case of the spherical microgel suspensions,
this can be related to the interactions between elastic spheres arising from their
deformation. The fluid gels, however, display a very different viscoelastic response, with a
more gradual increase in G' with concentration, that arises from the complex nature of the
particles.
Interpreting the behaviour of the materials in this way allows interesting conclusions to
be drawn regarding influence of processing conditions on the sheared agar microstructure.
It appears that material produced on the continuous throughput pilot line has a particle
microstructure that is largely independent of the concentration of agar used in the process.
However, results obtained using a stirred pot indicate that impeller speed is important in
determining the particular microstructure. At high impeller speeds, materials with similar
microstructures are produced regardless of agar concentration or whether the continuous or
stirred pot process is used, whilst at low speeds a quite different behaviour is observed.
Here, the material microstructure seems to be quite dependent on the concentration of agar
used in the preparation. The reasons for this are still unclear, and certainly merit further
study.
Acknowledgements
The authors gratefully acknowledge the help of C.R.T. Brown, P. Knight and S . Daniels
for providing the fluid gel samples.
References
1. C.R.T. Brown, A.N. Cutler, and I.T. Norton, inventors, 1990, pat. no. EP0355908
2. I. Norton, T. Foster, and C.R.T. Brown, Special Publications - Royal Society of
Chemistry - Gums and Stabilisers for the Food Industry 9, 1998,218, 259-268.
3. G. Cassin, I.A.M. Appelqvist, V. Normand, and I.T. Norton, Colloid and Polymer
Science, 2000,278,777-782.
4. I.T. Norton, D.A. Jarvis, and T.J. Foster, International Journal of Biological
Macromolecules, 1999,26, 255-26 1.
5. W.J. Frith, A. Lips, J.R. Melrose, and R.C. Ball, In: 'ModernAspects of Colloidal
Dispersions', eds. R.H. Ottewill and A.R. Rennie, Kluwer, 1998, 123-132.
6. W.J. Frith, A. Lips, and I.T. Norton, In: 'Structureand Dynamics of Materials in
the Mesoscopic Domain', eds. M. Lal, R.A. Mashelkar, B.D. Kulkarni and V.M.
Naik, Imperial College Press, 1999,207-218.
7. A.H. Clark and S.B. Ross-Murphy, British Polymer Journal, 1985,17, 164-168.
8. I.D. Evans and A. Lips, Journal of the Chemical Society-Faraday Transactions,
1990,86,3413-3417.
9. W.J. Frith and A. Lips, In: 'Proceedings of the XIIth International Congress on
Rheology ', eds. A. Ait-Kadi, J.M. Dealy, D.F. James and M.C. Williams, Lava1
University Press, 1996, pp. 558.
10. R.D. Zwanzig and J. Mountain, Journal of Chemical Physics, 1965,43,4464.
11. H. Hertz, Gesammelte Werke, 1895,1, 155.
DYNAMIC MECHANICAL CHARACTERISTICS OF RED ALGAL
POLYSACCHARIDE GELS COMPOSED OF SOLUBLE STARCH AND STARCH
GRANULES
V.M.F. Lai', H.-J. Liang2 and C.-y. Lii2
'Department of Food and Nutrition, Providence University, Shalu, Taichung 43301,

Taiwan (e-mai1 address : mflai@pu.edu .tw)
Institute of Chemistry, Academia Sinica, Nankang, Taipei 1 1529, Taiwan
1 INTRODUCTION
Generally, the rheological properties of starch-containing systems depend on the

individual properties of the continuous (containing water-soluble materials from
polysaccharide gums or starches) and dispersed (starch granules) phases and the molecular
interactions between these phases. Recently, the interactions between chemically
dissimilar polysaccharide molecules such as red algal polysaccharide gums and starches
have been the subject of numerous studies on rheological Nonetheless, the
observable rheological changes caused by polysaccharide interactions are governed by
both kinetic and thermodynamic factors and, in practice, by sample preparation and
gelling Accordingly, the purpose of this study was to investigate the
potentially individual effects of soluble and granular starch components on the structural
features of red algal polysaccharide gels from the viewpoint of activation energy of
mechanical relaxation by dynamic mechanical analysis. Agarose and K-carrageenan were
used because of typically different rheological changes with the addition of ~tarches.~
2 METHODS AND RESULTS
2.1 Sample preparation
Agarose (type IA, low EEO, A0169), Ic-carrageenan (type 111, from Eucheuma cottonii,
C1263), wheat starch (S2760) and corn starch (S9679) were obtained from Sigma Co. (St.
Louis, MO, USA). Rice starch from indica rice (Kaoshiung Sen 7 variety) was prepared
with an alkaline steeping method.' Hot-water soluble and insoluble (i.e . granular) fractions
of the above starches at 80 or 9OoC for 30 min were separated by centrifugation (8000 g,
15 min) and consecutive freeze-drying. The resultant starch fractions were marked as SOG,
SOS, 90G and 90S, according to the heating temperatures used and sample type (granular
or soluble).
Aqueous agarose or Ic-carrageenan solutions and starch component solutions were
separately prepared by heating in boiling water for 20 min to ensure complete
solubilisation and at 80 OC for 30 min to disperse the soluble and granular fractions. These
two solutions were then mixed at equal portions (v/v) to give the final concentrations of
1.5 wt% and 1.0-4.0 wt% for red algal polysaccharides and starch components,
respectively. After gelation in cylindrical teflon moulds and ageing at room temperature
for 24 h, the gels (17 mm ID x 3 mm H) were immediately removed from the moulds for
dynamic compression measurements on a Dynamic Mechanical Analyzer (DMA 2980, TA
Instruments Ltd., Surrey, England). A layer of silicone oil was applied to the exposed edge
of sample to avoid moisture evaporation. All the data present in Figures and Tables were
means of three replicate measurements.
2.2 Changes in phase properties of the mixtures
Since the influences of starch components on the rheological properties of agarose and
K-carrageenan gels may be through changing the local concentration of gelling
polysaccharides by the exclusion effect of dispersed granules or through interfering with
the formation of junction zones by starch soluble in the continuous The
swelling powers (Q in g wet starch insoluble/g dry starch insoluble) and water solubilities
(S in g soluble/100 g dry starch or wt%) of the starch fractions at the mixing temperature
of solutions (80C) for 30 min were therefore analyzed according to the procedure
described previ~usly.~.'For facilitating the understanding, only the properties of 80G and
80s fractions with various soluble and granular compositions were discussed herein. The
average Q-S data of three replicated measurements were 15.6-5.0 wt% and 15.8-52.5
wt% for rice 80G and 80S, 14.2-16.0 wt% and 9.9-37.1 wt% for corn 80G and 80S, and
14.5-17.6 wt% and 23.4-68.9 wt% for wheat 80G and 8OS, respectively.
By assuming that the red algal polysaccharides are entirely excluded from swollen
granules, the local concentrations of red algal polysaccharide (CX,red) and starch (CX,,,) in
continuous matrices can be estimated according to equations ( 1)-(4).2,7
4Y = C t , S t ( W Q (1)
@X=l-& (2)
C x , s t = ct,st.s/@X (3)
CX,red = ct,red/& (4)
Where Ct,st and Ct,red (in wt%) are the total concentrations of starch and red algal
polysaccharides in the whole systems; and & and & are the volume fractions of
continuous and dispersed phases, respectively. The resultant &, CX,red and Cx,stof the
mixtures with 80G and 80s fractions are tabulated in Table 1. When the rice 80G
concentration increased up to 4.0 wt%, the &, C X , @ and Cx,strose up to 0.593, 3.68 wt%
and 0.49 wt%, respectively, different from those of corn and wheat 80G (-0.48,2.87 wt%
and 1.2-1.4 wt%). In the case of 4.0 wt% 80s fractions, the &, CX,red and Cx,stwere in the
ranges of 0.25-0.30, 2.0-2.2 wt% and 2.0-3.9 wt% (and wheat > rice > corn),
respectively, for the starches examined.
2.3 Rheological properties of polysaccharide gels
Figure 1 indicates that the strain dependencies of storage (E') and loss (E") moduli of 1.5
wt% agarose and K-carrageenan gels changed with the addition of 4.0 wt% 80G fractions
from various starches. Typically, the gel systems showed a decreased E' and increased E"
with increasing the strain examined (0.1-2.5 %). Such changes with strain, particularly for
Table 1. Phase volume fractions and local concentrations of the red algal polysaccharide
and starch components in the mixtures studied a
Starch Conc Rice corn Wheat
fraction @Yb cX,ret CX,stb $% CX,red CX,st @Y CX,red CX,st
(WtYO) (wt%) (wt%) (wt%) (wt%) (wtY0) (wt%)
80G 1 0.148 1.76 0.06 0.119 1.70 0.18 0.119 1.70 0.20
2 0.296 2.13 0.14 0.239 1.97 0.42 0.239 1.97 0.46
3 0.445 2.70 0.27 0.358 2.34 0.75 0.358 2.34 0.82
4 0.593 3.68 0.49 0.477 2.87 1.22 0.478 2.87 1.35
80s 1 0.075 1.62 0.57 0.062 1.60 0.40 0.073 1.62 0.74
2 0.150 1.76 1.24 0.125 1.71 0.85 0.146 1.76 1.61
3 0.225 1.94 2.03 0.189 1.84 1.37 0.218 1.92 2.64
4 0.300 2.14 3.00 0.249 2.00 1.98 0.291 2.12 3.89
a Calculated using the swelling powers ( Q in g/g)-water solubilities ( S in wt%) data detected at
80C = 15.6-5.0 wt% for rice 80G, 15.8-52.5 wt% for rice 80S, 14.2-16.0 wt% for corn 80G,
9.9-37.1 wt% for corn 80S, 14.5-1 7.6 wt% for wheat 80G, and 23.4-68.9 wt% for wheat 80s; red
algal polysaccharide concentration = 1.5 wt%.
Subscripts X and Y indicate the continuous and dispersed phases, respectively; @Y = the volume
fraction of dispersed phase; C X , r e d and CX,st
= the theoretically local concentrations of red algal
polysaccharides and soluble starches in the continuous matrices.
160 I 100
3 120
%
Y 80
1 40
30 ' ,++++ + + t + + + + + + + + + :: I I I I
5
n
20
W
10
0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
Strain (YO) Strain (YO)
Figure 1. Strain dependencies of storage (E') and loss (E") moduli of agarose and
K-carrageenan gels mixed with and without 80G starch fractions. The total concentrations
of red algal polysaccharides ( G r e d ) and starch fractions (C,,,,) are 1.5 and 4.0 wt%,
respectively. (+ = pure gel, 0 , A and = the gels with rice, corn and wheat starch fractions,
respectively; 80G = granular starch fractions isolated at 80C; measured at 3OoC,
frequency = 1 Hz)
E", appeared to level off for all samples examined at a strain greater than 0.5-1.0%. The
slopes of E' against strain for the agarose or K-carrageenan gels containing 4.0 wt% 80G
decreased or were likely invariant, respectively, over the strain range examined, in contrast
to the gels without starch. The magnitudes of E for the starch-containing agarose gels at
low strains (50.5%) followed the order of agarose alone rice > corn > wheat; and for E",
agarose alone, wheat > corn > rice. Accordingly, the elastic property suggested by tan 6 (=
FIE') was: agarose alone > rice > corn > wheat. However, the orders at high strain
regions (> 2%) changed to: rice, corn > agarose alone > wheat for E' and agarose alone >
rice, corn > wheat for E'. In the case of K-carrageenan gels, the E' and E" followed the
sequence of carrageenan alone and corn > wheat > rice over the strain range examined.
The strain of 0.5% within the linear viscoelastic region was applied in the following DMA
measurements.
Figure 2 shows the E and tan 6 changes of 1.5 wt% agarose gels with applied
frequency cf= 0.5-80 Hz) or temperature ( T = 1O-5O0C at each 5 "C increment) during
multi-frequency temperature ramps. The rice 80G and 80s at 4.0 wt% were illustrated as
examples. Typically, the E values of pure agarose gels (Figures 2A-C) maximized,
accompanying with a tan 6 minimum, at f = 20 Hz and T = 15OC. In contrast to pure
agarose gels, the gels with 4.0 wt% 80G (Figures 2D-F) and 80s (Figures 2G-I)
generally had lower E' and greater tan 6 values, both more scattered within the frequency
and temperature ranges applied. The critical frequency and temperature responsible for E
maximum and tan 6 minimum appeared to be elevated ( c a . 50 Hz and 20C).
In the case of K-carrageenan gels, the criticalfand T responsible for E' maxima and
tan 6 minima and the effects of added rice 80G and 80s at 4.0 wt% on the shifts of the
critical f and T and on the changing tendency in E' or tan 6 were similar to the results of
the agarose gels (Figure 2). However, the addition of the starch fractions, particularly 8OS,
reduced the scattering, i.e. temperature dependency, of E' data at the same frequency. The
scattering, i.e. frequency dependency, of E' at the same temperatures increased (for 80G)
or decreased (for 80s) in the presence of starch. The E' increases and decreases within low
and high frequency (or temperature) ranges, may be respectively attributed to the enthalpic
and entropic arrangements of molecular chains for mechanical relaxation of gel
networks,' '-I2 agreeing with exothermic and endothermic evidence respectively obtained
by differential scanning calorimetry.
2.4 Activation energies for mechanical relaxation of gels in terms of phase property
The E data sets obtained by multi-frequency temperature ramps were treated with
time-temperature superposition software (TTS Data Analysis, TA Instruments Ltd., Surrey,
England) using the data at 298.2 K as reference. Based on the viewpoint of Arrhenius-type,
first-order kinetics, the activation energies (&T in J mol-' ) responsible for the mechanical
relaxation of gel networks were obtained from the temperature dependencies of the
horizontal shift factors ( a ~as
) follows (eqn (5)):'*,13
where R is the gas constant (8.314 J K-' mol-I), and T is the experimental temperature (K).
The Arrhenius rather than William-Landel-Ferry (WLF) model is empirically operative
over the temperature and frequency regimes examinedI2-l4 because the gels experienced a
transition from the rubbery state to flow during the measurements. In this study the
-1 0 1 2 - 1 0 1 2 - 1 0 1 2
L o g frequency/Hz
0.10" ' ' ' 1 1 4 ' ' ' 1 ' 1 ' ' ' 1 1
10 20 30 40 50 10 20 30 40 50 10 20 30 40 50
Temperature (OC)
Figure 2. Changes in E and tan 6 (= E'/E)values of pure agarose gels (A-C), agarose
gels mixed with rice 80G (D-F) and 80s (G-I) with multi-frequency temperature ramps.
= 4.0 wt%, measured at a strain = 0.5 YO).
(Ct,red= 1.5 wtYo, Ct,st
correlation coefficients (r2)for the linear regressions of log UT against T I were in the
range of 0.80-0.98.
Figure 3 indicates that the Ear values of pure agarose gels were in the range of 86-108
kJ mol-' in a linear relationship with its concentration (CX,, in wt%):
E ~ =T 5.54*Cx,a+ 82.91 (r2=0.522) (6)
This tendency agrees with its changes in gelling and melting temperature." In contrast,
the E a ~values of pure K-carrageenan gels decreased from 172 to 55 kJ mol-' in two
concentration ( c X , k in wtY~)
dependencies (equations (7-8)). This decrease may be linked
to a greater proportion of ineffective chains in highly viscous polysaccharide system^.'^,'^
EaT = -14.26.Ck -k 193.16 for c X , k = 1.5-3.0 wt% (r2=0.999) (7)
E a =~ -95.5.Ck + 436.8 for c X , k = 3.0-4.0 wtY0 (8)
The E a values
~ for K-carrageenan gels at < 3.5 wt% were obviously higher than those of
agarose gels, perhaps due to the electrostatically repulsive interactions between
carrageenan chains. The E a values
~ for 1.5 wt% K-carrageenan els are in a similar order
to those by the same TTS-Arrhenius treatments on creep curves," but much greater than
Figure 3. The activation energies (&T)

of pure agarose (filled symbols) and
1 2 3 4 K-carrageenan (open symbols) gels in
Concentration (wt%) terms of concentration.
those obtained by Arrhenius treatments on the temperature dependence of elastic modulus

(10 k~ mo1-l).l4
In the presence of 80G, 80S, 90G and 90S, the E u ~changes with total starch
concentration (&) are illustrated in Figures 4 and 5 for agarose and K-carrageenan gels,
respectively. To facilitate differentiating the exclusion effects of granules from the
interfering effects of soluble starch, the theoretical E u changes
~ estimated by introducing
the CX,red values of Table 1 in equations (6) (Cx,a for agarose) and (7)-(8) (CXJ for
K-carrageenan) are also indicated in the Figures in solid and dashed lines, respectively, for
the systems with 80G and 80s. Interestingly, addition of starch fractions at 1.O wt% (& 5
0.15) caused a notable E a ~increment for both agarose or rc-carrageenan systems, as
compared with the theoretical lines. From the differences (A&T) between theoretical &T
and the maximal E u at ~ 1.0 wt% starch fractions, the increasing effectiveness between
various starches were differential in agarose gels: wheat > corn > rice (A&T = 70, 58 and
3 8 kJ/mol, respectively) and soluble fractions > corresponding granular fractions,
particularly for the high-solubility wheat samples. For the K-carrageenan systems with 1.O
wt% starches, the results different from the above for agarose gels were the notable
influences by 80G rather than 80s except the wheat fractions with high solubility, and the
A&T values (1 14-1 28 kJ/mol) much higher than those for agarose. Between soluble starch
fractions, addition of high-temperature isolates generally resulted in a greater E u ~
increment at Ct,st= 1-2 wt% and greater E u decrement
~ with increasing Ct,st,than did their
low-temperature counterparts.
By comparing the results of Figures 4 or 5 and Table 1, the critical Ct,stfor the E u ~
values below the theoretical lines were 2-3 and 3-4 wt% for agarose and K-carrageenan
gels, respectively. Accordingly, the critical #y values (#y*), at which E u started ~ to be
lower than the theoretical values, tended to reduce with increasing CX,red and decreasing
the function of #y* = 0.295 Cx,a - 0.0023 C X ,-~0.342
C X ,as~ ~ ~ (R2= 0.994) for agarose
and &* = 0.207 CX,k - 0.0030 - 0.145 (R* = 0.972) for K-carrageenan.
The addition of 1.O-2.0 wt% starch components generally resulted in the 1.5 wt% red
algal polysaccharide gels with decreased E' (Figure 2), reduced gel strengths7, but
increased &T. Because the E u concerns
~ the energetic requirement for overcoming chain
deformation in el networks, the E u increments
~ can be linked to the reinforcing effects of
starch fillers 3*4'3i17 on the gel structure. On the other hand, the successively reduced &-I-
values of both gel systems with increasing (& (2.0-4.0 wt%) could be mainly
200
Corn Wheat
-
n
0
160
T
f
.
2 120
W
80
40 ,
,,I
1 2 3 4 1 2 3 4
Figure 4. Effects of total starch concentration (Ct,st)on the E,T values of agarose gels with
various starch fractions. SOG, SOS, 90G and 9 0 s are the granular or soluble starch
fractions isolated at 80 or 9OoC for 30 min. Solid and dashed lines without data points
were the theoretical E u changes
~ with the red algal polysaccharide concentration.
380
-
n
-.-310
-
3
240
170
100
1 2 3 4 1 2 3 4 1 2 3 4
Figure 5. Effects of total starch concentration (Ct,J on the Ear values of K-carrageenan
gels with various starch fractions. The experimental conditions are described in Figure 4.
explained as the decreasing effects due to soluble interferences7 or molecular

in~ompatibility~-~ because the increasing effects on Ear by granular exclusion indicated by
the theoretical lines (Figures 4 and 5 ) were insignificantly changed with Ct,St,except for
the ~-carrageenan-4.0wt% rice 80G mixture. The interferences can be further explained
from the decreases in gelation rate, in gelling and melting temperatures and in the
concentration dependency of storage moduli of these gels by adding of starches7.
Nonetheless, starch r e t r ~ g r a d a t i o n ' and
~ ~ ' ~possible phase ~ e p a r a t i o n ~may
- ~ be partially
responsible for the above E' and E,T changes.
3 CONCLUSION
The presence of 4.0 wt% soluble and granular fractions from various starches usually
resulted in the decreases in E' and E'' of 1.5 wt% agarose as well as K-carrageenan gel
systems at low strain regions. The dependencies of E' on frequency and temperature may
be increased or decreased by adding starch fractions, depending on the concentration and
sample type (granular or soluble) of starch fractions interested. At the gum concentrations
< 3.5 wt%, the activation energies (&T) of mechanical relaxation of agarose gel networks
were lower than the K-carrageenan. The effects of added starch fractions on increasing the
E u ~values at 1-2 wt% starch fractions and on the E u ~decrements with starch
concentrations appeared to be greater for the agarose than for K-carrageenan gels. Starch
samples with higher water solubilities showed greater influences on the Ear changes of
agarose gels, but not K-carrageenan. Generally, the interferences of soluble starch in the
continuous matrices and partially the exclusion effects of swollen granules would be
responsible for most of the starch effects on the mechanical characteristics of agarose and
K-carrageenan gels.
ACKNOWLEDGMENT
This work was financially supported by the National Science Council, Executive Yuan,
Taiwan (NSC-89-2313-B-001-008). The authors thank Miss Li-Hwa Yu, research assistant
of Institute of Chemistry, Academia Sinica, for performing DMA measurements.
References
1. A.C. Eliasson, J Texture Stud., 1986, 17,253.
2. J.L. Doublier, G. Liamas, and M. Le Meur, Carbohydr Polym., 1987, 7,25 1 .
3. H. Liu and J. Lelievre, Cereal Chem., 1992,69, 597.
4. P. Rayment, S.B. Ross-Murphy, and P.R. Ellis, Carbohydr Polym., 1995,28, 121.
5. D. Eidam, W.-M. Kulicke, K. Kuhn, and R. Stute, StarcWStarke, 1995,47,378.
6. Z.H. Mohammed, M.W.N. Hember, R.K. Richardson, and E.R. Morris, Carbohydr
Polym., 1998,36,27.
7. V. M.-F. Lai, A.-L. Huang and C.-Y. Lii, Food Hydrocoll., 1999, 13,409.
8. N.A. Abdulmola, M.W.N. Hember, R.K. Richardson, and E.R. Morris, Carbohydr
Polym., 1996,31,65.
9. C.C. Yang, H.-M. Lai, and C.-y. Lii, FoodSci. (Chinese), 1984, 14,212.
10. T.J. Schoch, in Methods in Carbohydrate Chemistry IV, ed. R.L. Whistler, R.J. Smith,
J.N. BeMiller and M.L. Wolfrom, Academic Press, New York, 1964, p. 106.
1 1 . M. Watase and K. Nishinari, Carbohydr Poi'ym., 1989, 11,55.
12. E.E. Braudo, I.G. Plashchina, and V.B. Tolstoguzov, Carbohydr Polym., 1984,4,23.
13. J.J. Aklonis and W.J. MacKnight, in Introduction to Polymer Escoelasticity, 2nd Edn.,
John Wiley and Sons, New York, 1983, ch. 6, p. 130.
14. J. Zhang and C. Rochas, Carbohydr Polym., 1990,13,257.
15. M.-F. Lai and C.-y. Lii, Int. J Biol. Macromol., 1997,21, 123.
16. A.H. Clark and S.B. Ross-Murphy, Adv. Polym. Sci., 1987,83,60.
17. S. G. Ring and G. Stainsby, Prog. Food Nutr Sci., 1982,6,323.
18. M. J. Miles, V. J. Morris, P. D. Oxford and S. G. Ring, Carbohydr Res., 1985, 135,
257.
19. M.-L. Tsai and C.-Y. Lii, StarcWStiirke, 2000,52,44.
MORPHOLOGY CONTROL IN DISPERSE BIOPOLYMER MIXTURES WITH AT
LEAST ONE GELLING COMPONENT
B. Wolf, W. J. Frith, and I. T. Norton
Unilever Research Colworth, Colworth House, Sharnbrook, Bedford MK44 1 LQ, UK
1 INTRODUCTION
The rheological behaviour of biopolymer mixtures can be influenced in a number of ways,

which typically include the choice of biopolymers, their concentrations, the solvent
conditions, and the phase behaviour. For systems with a discrete dispersed phase, the
volume of this phase, and its morphology can be used as an additional factor for control,
and, manipulation of the rheological behaviour.
The control of dispersed phase morphology requires a suitable process, which has been
discussed previously.' This process is based on deformation of the droplet phase in a two
phase liquid biopolymer mixture in shear flow, and kinetic trapping of the deformed
droplet phase through superimposed temperature induced gelation. This paper is concerned
with a more in-depth understanding of the process, and of the generated particle
morphologies.
Previously, only a steady shear process has been discussed. The micrographs of the
resulting gel particles showed morphologies ranging from uniform ellipsoids to particles
with less well defined shapes. In this paper, a stress step-up process for producing long
extended particles in a controlled manner is introduced.
The model system studied is composed of the two gelling polysaccharides, gellan and
K-carrageenan. A gelling continuous phase allows removal of the sample as a bulk gel
from the processing geometry. This ensures that only sample from the shear gap is
collected, and analysed. Additionally, the orientation of the gelled particles within the
matrix can be visualised.
The influence of particle morphology on rheo-mechanical properties of biopolymer
mixtures is demonstrated with the rheological characterisation in steady shear for
suspensions with an internal phase volume, OV,of 0.2.
2 MATERIALS AND METHODS
2.1 Biopolymer mixture
Aqueous mixtures of the two polysaccharides gellan (Kelco gel F, Kelco) and K-
carrageenan (Genuvisco X-0909, Hercules) were prepared from stocks of phase separated
material as described previously.' Phase volumes of gellan-rich dispersed phase from 0.0 1
to 0.2 were included in this study.
The steady shear material properties of the two phases prior to the onset of gelation, and
the gelation behaviour of the two phases while subjected to the same time-temperature
history as applied in the shear-cooling process (see below for details) were quantified. In
their liquid state, both phases display shear-thinning rheology with an extended Newtonian
plateau for low shear stresses, and the gellan-rich phase gels at higher temperatures than
the K-carrageenan-rich phase. For the batch of material used in this study the gelation
temperature for the gellan-rich phase, defined as the temperature at which tan 6 = 1, gave
T, = 47.1 k 0.7OC. The K-carrageenan-rich phase gelled at a slightly lower temperature.
An important system parameter for the analysis of droplet shapes under the influence of
flow is the viscosity ratio, h, defined as the ratio of the dispersed phase viscosity to the
continuous phase viscosity. At 60C, the starting temperature of the discussed process, a
value for h = 2.12 was found. The viscosity ratio at gelation was h, = 2.83.
The interfacial tension represents another important material parameter for the
discussed process. A value of G = 7.5.10-6 f 1.4~10-~ N.m-' for 6OoC was previously
reported.
2.2 Deformation process
The liquid mixtures were placed in the concentric cylinder gap of a stress controlled
rheometer (SR 200, ex Rheometrics) at 6OoC.They were pre-sheared for 10 min at 1 Pa in
order to equalise morphological differences generated by sample preparation. In some
cases, an additional pre-shearing step at lower shear stresses was used in order to increase
droplet size by coalescence. The mixtures were then subjected to the actual structuring
process while cooling at a rate of 0.9"C till after the continuous K-carrageenan-rkh phase
had gelled.
The structuring process previously discussed involved shearing at a constant shear stress
for generation of ellipsoidal gelled particles whereby the event of droplet break-up limits
the maximum possible degree of anisotropy.' Creation of larger aspect ratios is still
possible, however, using non-steady state conditions. A sudden increase of deformation
-
stress in shear flow for systems with h 1 is known to lead to the formation of large aspect
ratio droplets prior to break-up? We make use of this mechanism to form gelled
biopolymer particles of cylindrical shape by trapping the stretched droplet shapes through
gelation before break-up O C C U ~ S . ~
For both types of shear process, deformation conditions were characterised by the
Capillary number, Ca:
Z.X
Ca=-
2.0
with I: = shear stress, x = droplet diameter, and (T = interfacial tension.
2.3 Particle morphology
The dimensions of the gelled particles were quantified by image analysis (software
package KS400 Imaging System, ex Imaging Associates Limited, Thame, UK).
Microscopic images were obtained using phase contrast illumination and were directly
digitised. Care was taken to ensure that the preparation of the microscope slides had not
damaged the particles. For quantification of particle dimensions the deformation
parameter, D, and cylindrical particles were quantified using the particle aspect ratio, r:
D=-L - B r = -L
L+B B
with L and B = length and width respectively of the deformed particle.
Diameters of deformed particles were calculated as diameters of volume equivalent
spheres and volume equivalent cylinders for ellipsoidal particles and cylindrical particles
respectively.
2.4 Steady shear behaviour
The steady shear behaviour for some mixtures with gelled gellan-rich particle phase
volume, Ov,of 0.2 was quantified as follows. The mixtures were processed as described
above, and left in the concentric cylinder geometry. The continuous K-carrageenan-rich
phase was liquefied by raising the temperature slowly to 6OoC to transfer the gel composite
into a gelled particle suspension. At 60C the sample was sheared for 30 minutes at 100 Pa
in order to make sure that particle aggregates or networks are broken. 100 Pa corresponds
to the maximum stress that was applied to all samples during their rheological
characterisation by decreasing the shear stress in a step-wise fashion.
3 RESULTS
Figure I shows the morphology of a mixture processed at a constant shear stress of 0.2 Pa
denoted deformed A, and a mixture denoted deformed C for which the stress was
stepped up from 0.05 Pa to 0.5 Pa prior to gelation. Image analysis of these two samples
gave a number average deformation for sample A of D50,o = 0.2 which is equivalent to a
number average aspect of r50.0= 1.5 1. The orientation of high aspect ratio particles formed
in the step-up process (under different conditions then sample C in Figure 1) is visualised
in Figure 2.
200 microns
Figure 1: Micrographs of a steady shear processed mixture (left), and a step-up processed
mixture (right)
200 microns
Figure 2: Particle orientation in the gel composite
3.1 Steady shear processed mixtures
Figure 3 shows the average deformation of mixtures processed under steady shear
conditions in form of a 3D graph. The data demonstrate that the deformation of the gelled
particles increases with increasing particle diameter, and increasing structuring stress. This
correlation is expected when compared with literature on deformation of single droplets in
steady shear. There are, however, a few exceptions that are not obvious in the
representation of the data in Figure 3. They become clear when the data are analysed using
a recently published model for droplet deformation, see reference for the The
model is strictly valid only for Newtonian fluid phases, a condition which is valid for the
system investigated prior to gelation. It follows that comparison of deformation values
predicted by this model against values measured for the gelled particles analysed here
should give an indication for the influence of gelation on the particle morphology.
A comparison of predicted values for D with the experimental data is shown Figure 4.
There are two observations to be made in Figure 4. First of all, the measured deformation
values for the gelled state are significantly smaller than the values predicted for the liquid
state. Secondly, the experimental D values seem to show little correlation with increasing
Capillary number, whereas the model predicts a continuous increase. However, neglecting
the experimental data marked with an asterisk in Figure 4, leaves a set of data that follow
the predicted D(Ca) increase. The marked data points result from experiments at shear
stresses 3 0.5 Pa. An explanation why these data do not follow the trend of the data
obtained at shear stress < 0.5 Pa requires further experimental investigations.
Figure 3: Particle deformation for mixtures processed in steady shear flow
0.5
.
I
,
solid line: model p&cticm for liquid state
synbds measuremerffsfor gelled state (D=,J
0.4 - cpv=O.O1 A av=0.05 + cpv=0.2
-
0" 0.3--
5
.-
CI :
sz 0.2-
.
- 0 -
0.1 -
Mi
A
0.0: 1 I ' I ' I ' I '
0.0 0.1 0.2 0.3 0.4 0.5 ( 6
caphlay nurber, q m [-I
Figure 4: Deformation of gelled particles (steady shear) and model predictions for liquid
state (Ca-number calculated with c = 7.5.10-6N.m-')
The data presented suggest that the droplet shape relaxes during the gelation process.
Indeed, other material systems that have been investigated showed complete relaxation
during gelation of the initially deformed droplet phase. One example of such a material
system is a mixture of gelatin and dextran6 Relaxation during gelation is driven by the
changing rheological material properties, in particular the increase in elasticity. However,
there seems to be some evidence that the rheological properties of the continuous phase are
also of importance. This has been demonstrated for mixtures of gelatin with guar which
allow trapping of a shear flow induced anisotropic morphological state.' Additionally, the
interfacial properties are hypothesised to be influenced by g e l a t i ~ n . ~
3.2 Mixtures processed by stepping-up the shear stress prior to gelation
Results for mixtures treated in the step-up process are plotted in Figure 5 . As expected an
increase in step-up stress leads to higher aspect ratios, as does an increase in particle
diameter. The latter is implied by the results for @V = 0.2 assuming that the phase volume
itself does not significantly influence the aspect ratio.
Figure 5: Particle aspect ratios for mixtures gelled after a stress step-up
The results were analysed using a predictive model for particle deformation under affine
deformation conditions as given by Janssen.' This model requires that the deformational
strain experienced by the droplets between the step-up of the stress and gelation is known.
This proved to be a problem since the point of gelation is not known with suficient
accuracy from rheological measurements. Values for the strain associated with this scatter
range over one decade leading to calculated aspect ratios ranging from under 10 to a few
hundred.
However, it is still possible to show that the model has the potential to be predictive, if
the gelation temperature is known with sufficient accuracy. This observation follows from
application of the model to the experimental data rather than using it as a predictive model.
It was found that this returns the same gelation temperature for mixtures with the same
dispersed phase volume.
3.3 Suspension rheology
Steady shear rheological results for a suspension with @V = 0.2 and particle morphology
varying from spherical to cylindrical (see Figure 1 for microstructures of the deformed
samples) are shown in Figure 6. The data are represented as relative viscosities. The
influence of particle shape is obvious. At high shear stresses particle orientation leads to
decreased viscosity with increasing particle aspect ratio. The influence of the particle
aspect ratio on viscosity is more pronounced at low shear rates. Higher aspect ratio
suspensions show higher viscosities.
10
-A- deformed A
+defOrmedC
2
.-
51
.-
.-P
-lu
U
er
1
0.01 0.I 1 10 100
shear stress [Pa]
Figure 6: Steady shear flow behaviour of structured suspensions
4 CONCLUSION
The present study has shown that the previously reported shear cooling process could be
refined with respect to higher degrees of particle anisotropy. Quantification of particle size
and shape parameters lead to a more in-depth discussion of process - morphology
relationships. The reported rheological data on the gel particle suspensions give a
preliminary insight in rheology - morphology relationships.
References
1 B. Wolf, R. Scirocco, W.J. Frith and I.T. Norton, Food Hydrocolloids, 2000, 14 (3),
217.
2 F. Rumscheidt F and S. Mason, Journal of Colloid Science, 1961,6,238.
3 B. Wolf, W.J. Frith, S. Singleton, M. Tassieri and I.T. Norton, Rheologica Acta, 2001,
40 (3), 238.
4 P.L. Maffettone and M. Minale, Journal of Non-Newtonian Fluid Mechanics, 1998, 78
(2-3),227.
5 P.L. Maffettone and M. Minale, Journal of Non-Newtonian Fluid Mechanics, 1999, 84
(l), 105.
6 private communication with P. Ding and A. Pacek, 2000.
7 B. Wolf, W.J. Frith, Journal of Rheology, 2000, submitted.
8 J.M.H. Janssen, in Materials science and technology A comprehensive treatment,
Volume 18: Processing ofpolymers, VCH, eds: R.W. Cahn, P. Haasen and E.J. kamer,
vol.ed. H.E.H. Meijer, VCH, 1999, ch. 3, p. 113.
DETERMINING THE YIELD STRESS OF XANTHAN SOLUTIONS VIA
DROPLET RISE EXPERIMENTS.
H.C.Mielke'$ and D.E. Dunstan'"

'CRC for Bioproducts, Department of Chemical Engineering, The University of Melbourne
3010 Victoria Australia
2Particulate Fluids Processing Centre, Department of Chemical Engineering, The
University of Melbourne 3010 Victoria Australia
1 ABSTRACT
Droplet rise experiments have been developed as a novel technique for rheologically
characterising the weak gel system xanthan gum. A drop rise column has been designed in
which hexadecane is injected into the gum and allowed to rise. Velocity profiles are
established as a function of the square of droplet radius (R2). Linear regression of these
yields a finite positive intercept for v=O which corresponds to drops at the point of incipient
failure. The radius taken from v=O may be used to calculate an apparent yield stress at each
gel concentration. The yield stress of the xanthan weak gel is found to depend on the
concentration according to T, = -0.12493 + 0.51643~.The yield stress is found to increase
over several days and varies according to drop rise time, due to visco-elastic effects in the
solutions.
2 INTRODUCTION
Xanthan gum is a high molecular weight, anionic, extracellular polysaccharide produced by

the bacterium Xunthumonus campestis. Its unusual rheological properties are subject to
extensive study due to the vast range of applications xanthan has in the food (e.g., desserts,
salad dressings, syrups etc), pharmaceutical, agricultural (e.g., pesticides sprays) and oil
(e.g., drilling) industries [ 11. The most important properties xanthan gum aqueous solutions
are considered to have include high viscosity at low shear, high psuedoplasticity even at
low concentrations, significant resistance to shear degradation and stability over wide salt
content, pH and temperature ranges [2].
Fluid-like food products often contain dispersed particles, which are stabilised against
sedimentation or creaming of these particles by xanthan gum, which gives the medium a
yield stress. The literature available on the yield stress of weak and fluid gel systems is
primarily derived from direct measurements employing a controlled stress rheometer.
Although it is not reasonable to suppose that a weak gel will fracture, it should be plausible
to assume two regions at the point of flow, an elastic yielding followed by a creep yield
stress as the gel moves between the behaviour of elastic response to viscous solution.
Much of the flow characterisation of xanthan has been carried out using the Ostwald
model [3]. Unfortunately this model provides no information on flow plasticity, a
characteristic directly related to xanthan's stabilising and emulsifying ability [2]. Yield
stress values, in particular, have been postulated by extrapolating the shear rate to zero
shear stress [4]. However, a log-log plot of this data shows significant curvature at low
stresses, providing inconclusive evidence of a yield stress. Another method employed in
determining the yield stress is by taking the square value of the intercept of the regression
line obtained by regressing the values of 0'' versus f'. Experimental data were fitted to the
Casson model [ 5 ] . The results show a shear rate dependence of the yield stress, which is a
linear function of the gum's concentration.
The presence of a yield stress in xanthan has received a lot of attention in the literature
and still is the source of much debate, both in terms of if a yield stress exists and what a
yield stress represents [6]. Complete understanding of the yield stress procedure is
necessary for the successful application of measurement results from the rheometer to
industrial food processing.
The aim of this study is to provide a simple method for determining the yield stress in
the fluid gel system xanthan gum and elucidating some of the controversies represented in
literature. A drop rise method is employed for this.
Oil drops will only rise through the solution if they are large enough to overcome the
yield stress of the material; otherwise they will remain stationary in the fluid gel. By
plotting the drop rise velocity as a function of bubble size, a linear regression may be fitted,
enabling the determination of the x-axis intercept. This coincides with the largest bubble,
which will remain stationary in the fluid. This in turn can be related to the yield stress of
the material via theory derived for the creeping flow of a sphere through a Bingham fluid
[71.
3 THEORY
The theory for a solid sphere of radius &, considered to be moving in creeping flow
through an unbounded Bingham plastic medium is valid for fluids with Herschel-Bulkley
behaviour at very low shear rates where the velocity is linear with respect to bubble
volume. This can be extrapolated reasonably to the point of incipient failure. It has been
shown, that a sphere will only move through solution, when the dimensionless value of the
critical yield stress (Yg)
is below 0.143 [7]. Below this value, the material acts as a solid.
The external force, F, acting upon a sphere rising freely through a fluid is
Where psis the density of the sphere
The yield-stress parameter Y, is defined from the ratio of the yield stress to the externally
applied force [71 :
Y,= 22,lt R: / F
Under the conditions of incipient failure, where

Y, = 0.143,
Now the yield stress of the material can be calculated in terms of the radius of a sphere at
the point of incipient failure,
4 EXPERIMENTAL
4.1 Methods
Figure 1 illustrates the drop rise apparatus, which was developed. It consists of a graduated
cylinder into which the xanthan is poured. The cylinder dimensions are 60 x 4.5 cm and it
is encased in a square glass jacket to allow for complete temperature control. The oil drops
are injected into the cylinder through a septum via a micro syringe. The syringe is
stabilised against any horizontal or lateral movement. The rise velocity of the oil drops is
monitored with a scope, designed to eliminate parallax when viewing at short range. A
camera is set up to capture images of the drops, ensuring they are spherical enough to be in
accordance with theory.
Figure 1: Schematic of the drop rise

apparatus
4.2 Characterisation
The natural pH of all solutions was established and the ionic content of the gums was
determined by running an ashed sample dissolved in acid through an Inductively Coupled
Argon Plasma (ICP) unit. These data are reported in Table 1. It was found that the pH of
the solutions was in the range of 4.65to 4.89which is constant for each concentration. The
dominating counterions in the gum were Na and K.
Stress sweeps of the xanthan gums were determined in order to establish the degree of
plasticity and any hysteresis associated with the solutions. Experiments were carried out in
concentric cylinder geometry on the Rheologica StressTech stress controlled rheometer.
Counterion Concentration (ppm)

Na 83.89
I K+ I 365.89 I
Ca 3.06
Mg2+I3+ 2.14
trace
Table 1. Results of ICP analysis for Xanthan
4.3 Materials
The xanthan gum (Keltrol) was commercially provided by Kelco AIL and came supplied in
the powdered form. It is reported to have a molecular weight of 3.4 x lo6 g/mol[8]. Sodium
azide (~0.02%)was used as a biocide in the solutions prepared with 10 M NaCl as a
background electrolyte to aid the dispersion and subsequent hydration of the xanthan. It is
reported, that low salt levels and high polymer concentration (>0.25%) reduces shear
thinning and the addition of salt increases the consistency [ 5 ] . Five samples were prepared
in the concentration range of 0.4 to 0.8 w/w%
Complete hydration of the samples without further purification was ensured by a cold
dissolution technique to allow the gum to swell and heating to 8OOC. They were then
tumbled for several hours and a subsequent heating to 80C ensured air bubbles trapped in
the gum were removed before testing. The cylinder was filled with the hot solutions and
left to equilibrate at 2 1C for 12 hours.
Hexadecane with 0 red dye was used for the drops rising through the xanthan. This was
the most density-matched alkane with the xanthan in solution state at room temperature.
The dye was used to enhance visibility.
5 RESULTS AND DISCUSSION:
5.1 Shear Rheology

Figure 2 depicts the data for the shear stress versus shear rate for a range of xanthan
concentrationsand illustrates the limitations of the shear control rheometer with respect to
giving accurate low shear rate viscosity data. These observations confirm previous studies
for which it is proposed that repeatable results for xanthan shear viscosity measurements
can only be obtained above shear rates of 0.01 s-' [9].
loo I l l l l l l l l l
t
10 i
0.1 j
0.01 1
lo6 0.0001 0.01 1
Shear Rate (1 /s)

100 1o4
Figure 2: Stress sweep data for xanthan gum using

cone and bob geometry.
5.2 Rise Velocity
At low shear rates, where (3 = V/d for a sphere, the velocity of the sphere must be linear
with respect to the R2.
V = (2Apg) / (9q) R2 4)
A range of shear rates for each concentration was probed until the shear rate was low
enough to achieve linearity. Due to the greater shear thinning nature at higher
concentrations of xanthan, it was found that much lower shear rates were required at the
larger concentrations in order for the data to be in the linear regime. (Fig. 3 ) At higher
shear rates, deformation of the bubbles also occurred. Only the lowest shear rate data was
employed in the calculations for the yield stress due to this.
Velocity profiles for all measured concentrations at the lowest shear rates plausible are
shown in Figure 4.
A regression was used to calculate the values from figure 4. The 0.7w/w% data was
discarded as the values for the velocity profile were unexpectedly high with respect to the
remaining data. Yield stress values were obtained from equation 1.2. The data shows a
linear dependence of the yield stress on xanthan concentration (Fig. 5.).
These measurements are consistent with those of Pastor et al. 1994 where the yield
stress decreases with decreasing shear rate [S]. The data presented in the present paper
shows a much lower yield stress as a function of concentration due to achieved shear rate
regimes of a thousand orders of magnitude lower then those obtained in the above
mentioned publication.
I I I I I I
0.01
0.m-
3
E.
0.m-
B m
3
f 0.004-
l!
0.m-
Dlopkt Radlru (anA2)
Figure 3: Rise velocities for 0.6 w/w 95 Xanthan. Shear

rate range: A. 0.0041-0.0075s~',B. 0.0217-0.0547d'
I I I
0.01 i I I I I I
I
I 0
m.
....
v
t
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Droplet RediusV (anh2)
Figure 4: Rise velocity of hexadecane in Xanthan

as a function of radius2.
It is interesting to note that stress sweep data obtained via the rheometer does not
indicate the presence of a yield stress, evident as a plateau at low stresses, however, droplet
rise experiments unequivocally show xanthan has one. The magnitude of the yield stress as
determined via droplet rise is comparable to the accessible range of the StressTech
rheometer. The absence of a yield stress here is consistent with typical attempts to
determine the yield stress from controlled shear rheometers. Perhaps even the smallest
shear rate in steady flow can lead to deformations large enough to damage the structure
prior to testing as is the case for waxy crude oils [lo]. In addition, effects such as
instrument inertia, damping characteristics of the rotation body and sensitivity to external
disturbance may contribute to poor reproducibility and a lack of conclusive data using
classical rheological methods[111.
0.3
0.25
--2 0.2
t
m
0.15
t
0.1
I J
-y = -0.12493 + 0.51643~ R= 0.99471
Figure 5: Yield stress of xanthan as a function of

polymer concentration
The thixotropy of xanthan is also noted through an increase in yield stress as a function
of time. This may introduce slight error into the calculations of yield stress values at
concentrations above O.~W/W%, where velocities profiles take up to 7 days to develop.
0.032
0.03
1- I I I
A
I I
A A
-
0
0
A
A m
2
0
0.028
w
I
.-%
0
-0 0.026
>
Q m
0.024
0.022
2 4 6 8 10 12 14 16
Height increment (cm)
Figure 6: Rise velocity of drops as a function of

distance travelled through the column for 0.6 w/w%
xanthan with 0.7, 0.8 and 0.9pL hexadecane
droplets.
The rise velocity of the drops as a function of distance travelled up the column varies, as
can be seen for the 0.6w/w% (Figure 6). There is a very slight trend of increase in velocity
as a function of height, which is also seen at other concentrations. This increase, however,
is only of the order of about 5%, and would not significantly affect the yield stress
calculations. Data collected for the calculation of the yield stress was taken from a constant
sample height at all concentrations in light of this. It is of interest to determine the origin of
the acceleration of the droplets. Thixotropic effects would cause a decrease in velocity with
time. Concentration as a function of sample height has been investigated via mass balance
and no trend was observed. Likewise G was measured as a function of sample with no
variations within the limits of the experiment. Further investigations need to be conducted
to assess the reasons for such an increase in rise velocity.
6 CONCLUSIONS
Xanthan weak gel solutions are found to have yield stresses which show a linear
dependence of zy = -0.12493 + 0 . 5 1 6 4 3 ~on the concentration. We have demonstrated that
the method is able to measure yield stresses in these fluid gel systems.
References
1. WE Rochefort, S Middleman, Journal of Rheology, 1987.31: p. 337-369.
2. GR Sanderson, 1 in Gums and stabilisers for thefood industry, GO Phillips, DJ Wedlock, PA
Williams, Editor.New York. 1982.p. 77-88
3. RA Speers, MA Tung, J. Food Sci., 1986.51: p. 96-98, 103.
4. TW Schenz, Food Technology, 1997.51: p. 83-85.
5. MV Pastor, E Costell, I Izquierdo, I Dran, Food Hydrocolloids, 1994.8: p. 265-275.
6. D C-H Cheng, Rheologica Acta, 1986.25: p. 542-554.
7. AN Beris, JA Tsamopulos, RC Armstrong, RA Brown, J. Fluid Mech., 1985.158: p. 219-244.
8. AB Rodd, DE Dunstan, DV Boger, Carbohydrate Polymer, 2000.45: p. 159-174.
9. AB Rodd, Characterisation of xanthan based, polymer solutions, physical gels and permanent
networks, in Chemical Engineering. 2001, University of Melbourne: Melbourne. p. 125.
10. C Chang, DV Boger, Znd. Eng. Chem. Res., 1998.37: p. 1551-1559.
11. A Magnin, JM Piau, Journal of Non-NewtonianFluid Mechanics, 1990.36: p. 85-108.
PREDICTING THE RHEOLOGY OF WATER-IN-WATER EMULSIONS
J.R. Stokes and W.J. Frith
Unilever Research Colworth, Colworth House, Sharnbrook,Bedford MK44 lLQ, UK
1 INTRODUCTION
A class of liquid-liquid mixtures commonly found in food systems is aqueous solutions of

mixed biopolymers that phase separate to form two aquesous phases. These systems are
viscoelastic and display very low interfacial tensions (- N/m)'. The purpose of this
work is to establish what role this low interfacial tension has in determining the rheological
behaviour of the mixtures and to test existing theories of emulsion behaviour.
Rheological emulsion models have recently been used to predict the morphology of
polymer blends. In particular, linear viscoelastic models24 have shown that emulsions
exhibit elastic properties due to interfacial tension effects. The Palierne4 constitutive
model for semi-dilute emulsions assumes small deformation and incorporates
hydrodynamic interactions, dispersed droplet size and polydispersity, and the linear
viscoelastic behaviour of the phases. The model predictions are valid for the description of
linear viscoelastic behaviour as given in equation (l), where the complex modulus is a
function of the modulus of the individual phases, phase volume, interfacial tension, and the
average droplet size (or distribution). The predictions using such linear viscoelastic
emulsion models have been validated for a variety of systems, in particular for polymer
blends and emulsions with high viscosity phases6-'. The models can be used to gain an
understanding of the changes in the measured storage modulus as a function of pre-shear
rate. The morphology or droplet size is then determined by the equilibrium between break-
up and coalescence such that an understanding of the effect of shear history can be gained*.
The following paper examines the use of the linear viscoelastic emulsion model for the
prediction of the rheology - morphology relationship of water-in-water emulsions
consisting of phase-separated biopolymer mixtures. In contrast to polymer blends, the
individual phases typically have a low viscosity (< 1 Pas) and often have negligible
viscoelasticity in their non-gelled state. The interfacial tensions of such systems am also
typically of the order of a few micro-Newtons per metre'. The system under investigation
here is a mixture of maltodextrin and gelatin, which phase separate at elevated
temperatures in their non-gelled state. Although the system is not an ideal one, it is
directly relevant to applications in the food industry for its textural and structuring
properties.
2 EXPERIMENTAL
2.1 Mixture Preparation and Rheology Measurements
The biopolymer mixture under investigation here consists of solutions of gelatin (lime hide
gelatin LHle) and maltodextrin (Paselli SA2). For a temperature of 60"C, mixtures of
these two biopolymers phase separate at particular concentrations, characterised by a
binodal curve, into gelatin-rich and maltodextrin-rich phases9-". The biopolymer mixtures
were prepared by first making individual solutions of both gelatin and maltodextrin in
0.1M NaCl and 0.02wt% NaN3 at elevated temperatures. Appropriate amounts of each
biopolymer solution were then mixed at just over 60C for one hour to give a mixture
consisting of 8wt% gelatin and 20wt% maltodextrin. The mixture was then centrifuged to
separate the phases and remove contaminants. The phases were extracted and re-mixed in
appropriate proportions to give the required phase volumes of 10% and 30%. The samples
were used and stored at 60 "C. The interfacial tension was previously mesured to range
between 25 and 80 W/m, with an average value of approximately 46 W / m 13.
Rheological experiments were carried out using the Rheometrics ARES (constant
strain) rheometer with a 50 mm diameter 0.04 radian cone and plate. Experiments
consisted of pre-shearing the emulsions for half an hour before measuring the linear
viscoelastic properties. The following protocol was followed. The pre-mixed emulsion
was loaded into the rheometer and an initial pre-shear rate of 100 s-l applied. A frequency
sweep from 100 rads to 1 rads was used to measure the linear viscoelastic properties, with
the strain typically set at 0.1 to be in the linear viscoelastic region. Testing followed a 30
minute preshear at each of the following sequence of shear rates: lOOs-', lose', 100s-'(2),
ls-', lOOi'(3).
The linear viscoelastic properties of each phase was measured and described using a 2-
mode Maxwell model12 with the model parameters listed in Table 1. The shear viscosity
was constant with shear rate and was 66 mPas and 61 mPas for the gelatin-rich and
maltodextrin-rich phases respectively. However, some variation was observed in the phase
rheology between batches of solutions. The specific gravity of the gelatin-rich phase was
1.070 while for the maltodextrin-rich phase it was 1.113.
Table 1: Material parameters for phases of a gelatin-maltodextrin mixture using the 2-

mode Maxwell model.
Gelatin-rich Phase Maltodextrin-rich Phase
rll (Pas) 0.0016 0.0014
rh (Pas) 0.0637 0.0593
A1 6) 0.0374 0.0857
A2 6) 0.0017 0.0020
rls (Pas) 0.001 0.00 1
2.2 Emuslion Model
]
The emulsion complex modulus using Palieme's model is given by the following:
G'=GZ[ 1+ 3
c &Hi 1
1- 2 C 4 H i
Here, Hi is a function of frequency (w) and given for each particle size Ri. G,* and G; are
the complex modulus of the continuous and dispersed phases respectively, a is the
interfacial tension, @ is phase volume distribution (i.e. qj is the phase volume for a mono-
dispersed emulsion), and the equivalent volume average radius (R,) is used to describe the
droplet size6. The value of the parameter d R was predicted using the Palierne model to
give the measured linear viscoelastic properties of the emulsions with minimum error.
The zero-shear viscosity for an emulsion of two Newtonian fluids using the Palierne
model is given by the following5:
1O(p + 1)+ 3@(5p+ 2 )
(3)
O
' = "'1O(p + 1) - 2$(5p + 2)
In the above, p is the viscosity ratio (p = 'Jqc) where "d is the dispersed phase viscosity
and rcis the continuous phase viscosity. At low shear or frequencies, the two biopolymer
phases have negligible viscoelasticity and hence can be considered Newtonian. It was
noted on several occasions that individual phase viscosities could change over time and
between batches of solutions. These changes arise from such phenomena as degradation of
gelatin and crystallisation of maltodextrin. It was also thought that there is a possibility
that the phase composition may also change on re-mixing when in the emulsion state. To
combat changes in the phase rheology, it was necessary to adjust the continuous phase
viscosity such that the zero-shear viscosity of the emulsion is accurately predicted using
equation (3).
3 RESULTS
The droplet size distribution of the emulsions was measured in an optical shearing cell
(Linkam shear cell CSS450) with a typical example shown in Figure 1 . A summary of the
morphology results for the set of preshear rates at different phase volumes is listed in Table
2. The average droplet size varied for different emulsions but was typically of the order of
around 5 um, 25 um, and 75 um for preshear rates of 100 s-', 10 s-', and 1 s-' respectively.
Figure 1 Morphology of 30% maltodextrin-rich phase emulsionfor a preshear of 10 s'.

Image width is 700 p.
Rhealogical Aspects 131
Table 2 Summary of results for wR. a i s calculated from measured values of R,. The
predicted values of R, are calculated assuming a = 50 ,uN/m.
SYSTEM Preshear cdR, a RV R,
rate Measured Predicted
(s-9 (m/m) (pm) (pm)
10%gelatin-rich 100 15 3.4
10 3.7 70 19 14
1 1.7 47 27 29
30%gelatin-rich 100 6.4 7.8

10 2.0 59 30 26
1 0.27 24 88 185
10%maltodextrin- 100 17 92 5.3 2.9

rich 10 3.2 51 16 16
1 1.3 74 55 37
30%maltodextrin- 100 9.3 65 7 5.4

rich 10 1.4 44 31 35
1 0.34 25 72 147
The small-amplitude oscillatory rheology for two of the emulsions is shown in Figs. 2
and 3 with a summary of all of the results shown in table 2. The lines in the figures are the
predicted curves using the emulsion model in equation (1) with fitted CdR values. As
opposed to polymer blends, where small-amplitude oscillatory measurements are obtained
across several decades of frequency, the rheology of the biopolymer mixture here could
only be measured with accuracy over two decades of frequency with the most accurate
measurements above .a frequency of 5 rads. Despite this, distinct differences are observed
for different preshear rates for all the emulsions tested.
Figure 2 shows the experimental measurements for G' and q' for the 10 vol% gelatin-
rich phase with comparisons to the emulsion model predictions. The experimental results
are visually similar with only slight variations, and there is a distinctive cross-over in G'
just above a frequency of 10 rads between each different preshear rate. The values for r,f
decrease at the middle range of frequencies as the preshear rate is decreased. This
lowering of q' measurements is due to the increased average droplet size. In this current
set of experiments, the measurements for each preshear rate of 100 s-' overlap each other
indicating that coalescence, sedimentation and creaming have not substantially altered
during the sequence of tests in this case. The predictions for the rheology of the 10 vol%
gelatin-rich phase are shown as lines in Figure 2. Values for cdR of 15 N/m2, 3.7 N/m2 and
1.7 N/m2 show accurate predictions for the measurements at preshear rates of 100 s-', 10 s-l
and 1 s-l respectively. The optical shear cell indicated that the volume average droplet
radius at a preshear of 10 s-' is Rv = 19 f 5 pm. Therefore, from the CdR value for a
preshear of 10 s'l, the interfacial tension is estimated to be 70 m / m . This value for the
interfacial tension fits within the range of values determined from drop deformation
e~perirnents'~.Using an interfacial tension of 50 w / m , the predictions indicate average
droplet radii (volume basis) of 3.4 pm, 14 pm, and 29 pm for preshear rates of 100 s-', 10
s-l and 1 s-' respectively. Whilst the morphology for 100 s-' could not be easily determined
using the optical shear cell, the droplets did appear to be of the order of 3 - 10 pm and
hence the prediction of 3.4 pm is reasonable. For a preshear of 1 s-I, the optical shearing
cell indicated a droplet radius of 27 f 8 pm, which is only marginally smaller than the
predicted value of 29 pm.
(a)
1
0.1
n
0.01
0.001
0.0001 1
-- afR = 1.73 N/m2 1
1 10 100
w (rads)
0.12
0.10
n
;a
& 0.08
F
0.06
0.04 b.
1 10 100
o (rads)
Figure 2 Linear viscoelastic measurementsfor 10 vol% of gelatin-rich phase in terms of
(a)storage modulus G and (b)dynamic viscosity q
Figure 3 shows the rheology of the 10 vol% maltodextrin-rich emulsion. The

measurements could be accurately predicted for each preshear rate using the emulsion
model. For the reshear rates of 100 s-l (initial), 10 s-l and 1 s-, the value for d R was 17
N/m2, 3.2 N/m;P and 1.3 N/m2 respectively. From the droplet size measurements, the
interfacial tension is estimated to be 92 pN/m, 51 pN/m and 74 pN/m respectively, which
are once again in the vicinity of the values determined from drop deformation experiments
and those determined for the 10 vol% gelatin rich phase. Using a value of 50 pN/m for the
interfacial tension, the average droplet radii by volume are predicted to be 2.9 pm, 16 pm
and 37 pm respectively. These predictions of the droplet radii are similar to those
measured in the optical shear cell, indicating that the model is reasonably accurate in
predicting the morphology of the emulsion. It should also be noted that a lower zero-shear
rate dynamic viscosity is observed for the final set of measurements at a preshear rate of
100 s'l in comparison to the two previous tests at 100 s-'. These differences in the
measurements between the tests at 100 s-l could be associated with changes in the phase
properties, but also sedimentation or creaming effects. In the model, the lower zero-shear
rate viscosity could be accounted for by either lowering the continuous phase viscosity
from that used for the previous predictions, or lowering the phase volume from 10% to
about 3.4%. If the phase volume is lowered, a value for cdR of 1.6 N/m2 is determined,
indicating a drop size of about 40 pm which is unrealistically large. If the continuous
phase viscosity is lowered by less than lo%, a realistic value for cdR of 20 N/m2 is
obtained, as shown in figure 3.
10
6 0.1
a
0.01
A -- CUR= 1.34 N/m'

0.001 , -. CUR = 19.82 Nlm
1 10 100
0.16
0.14
V
n 0.12
6 0.10
0.08
0.06
I 10 I00
o (rad/s)
Figure 3 Linear viscoelastic measurements for 10 vol% of maltodextrin-richphase in
terms of (a)storage modulus G' and (b) dynamic viscosity 7'.
Further experiments were conducted for phase volumes of 30% with similar accuracy
found between predictions and experimental results as those for a phase volume of 10%.
The summary of the results for all four emulsions is shown in Table 2. An average value
for the interfacial tension of 55 pN/m was determined using the fitted d R values and the
measured morphology from the optical shear cell. The interfacial tension from the
rheology measurements ranged in value from 24 pN/m to 92 pN/m, which are similar to
those measured using droplet deformation experiments where the average interfacial
tension was about 46 pN/m. Therefore, due to the large range of values for the interfacial
tension measured using both techniques, a value of 50 pN/m was considered a reasonable
value to use when comparing the predicted morphology to that measured experimentally.
4 DISCUSSION
The small-amplitude oscillatory properties of a phase-separated biopolymer mixture

consisting of gelatin and maltodextrin are able to reflect morphological changes. This is
despite the very low viscosity of the system (17 c 0.1 Pas) and low G' values (G' < 10 Pa).
Due to the slight difference in density between the two phases and the low viscosity of the
system, sedimentation and creaming effects are more important than in more viscous
systems such as polymer blends. The phase diagram of phase-separated biopolymer
mixtures is also very complicated and susceptible to small fluctuations in composition and
temperature". Despite the complexities of the system, and the aforementioned difficulties,
the comparison of the dynamic data to the emulsion model predictions is reasonably
accurate and hence extremely useful for determining morphological changes. It also seems
to be a sensitive technique for picking up effects and differences due to such phenomena as
sedimentation and creaming, phase inversion, and changes in composition of the phases.
While it appears surprising that it is possible to measure and predict the linear
viscoelastic properties for water-in-water emulsions in comparison to polymer blends, the
similarity can be understood by examining the time scale for the processes. The longest
relaxation time for an emulsion is related to the time required for a deformed droplet to
recover its spherical shape, and is proportional to Rqla. Assuming a droplet size of 10
- -
pm, a typical polymer blend with q 100 Pas and a 1000 pN/m will have relaxation
- -
time of h 1 s. Similarly, for a typical biopolymer mixture with q 1 Pas and a 10 -
-
pN/m, the relaxation time is also h 1 s. Hence, the relaxation process is of the same
order of magnitude for the two very different systems and you would expect to detect this
at a similar frequency range in the rheometer. For comparison, the relaxation time for
- - -
typical oil-water emulsions (q 1 Pas, a 1000 pN/m) is h 0.01 s and hence large
frequencies are required to detect the relaxation process using linear viscoelastic
measurements for these systems.
It is also found that transient step-shear tests are also able to detect deformatiodbreakup
of droplet, another commonly used technique for polymer blends14. Figure 4 shows such a
test for the 30% gelatin-rich phase emulsion at the following sequence of shear rates 1 s-',
10 s-', 50 s-' and 100 s-'. As for polymer blends, an undershoot is observed for the shear
stress and viscosity. This undershoot has been associated with the breakup of elongated
dropsl5. It is therefore expected that models for the steady and transient shear behavior,
which are often used for polymer blends16-18,will also be applicable to phase-separated
biopolymer mixtures.
8 1J L 0.12
0.10
6-
n
- 0.08
&
4 - 0.06 p
8
0.04 5
2-
- 0.02
0 -
+ 0.00
0 50 100 150 200 250
time (s)
Figure 4 Transient step shear rheologyfor 30~01%of gelatin-richphase emulsion.
5 CONCLUDING REMARKS
The emulsion model commonly used to examine polymer blends has been shown to predict
the rheological measurements of a phase-separated biopolymer mixture (gelatin-
maltodextrin) reasonably accurately. The morphology predicted using an interfacial
tension of 50 pN/m in the emulsion model was similar to that observed in the optical shear
cell. Despite the complexities of the maltodextrin-gelatin system in particular, and of
phase-separated biopolymer mixtures in general, including their low viscosity and low
viscoelasticity, the viscoelastic emulsion model appears to be suprisingly useful for the
examination of the rheology-morphology relationship of phase-separated biopolymer
mixtures and water-in-water emulsions. The techniques and methods used to study the
rheology and morphology of emulsions, and particularly those being developed for
polymer blends, are applicable to phase-separated polymer solutions.
6 REFERENCES
1. Wolf, B., R. Scirocco, W.J. Frith & Norton I.T., (2000) Shear-induced anisotropic
microstructures in phase-separated biopolymer mixtures, Food Hydrocolloids 14,pp.
2 17-225.
2. Oldroyd, J.G., (1950) On the formation of rheological equations of state, Proc. Roy.
SOC.A200,pp. 523-541.
3. Oldroyd, J.G., (1953) The elastic and viscous properties of emulsions and
suspensions, Proc. Roy. SOC.A218,pp. 122-132.
4. Oldroyd J.G., (1955) The effect of interfacial stabilising films on the elastic and
viscous properties of emulsions, Proc. Roy. SOC.A232,pp. 567-577.
5. Palierne, J.F., (1990) Linear rheology of viscoelastic emulsions with interfacial-
tension, Rheol. Acta 29,pp. 204-214.
6 . Graebling, D., Muller, R., & Palierne, J.F., (1993) Linear viscoelastic behaviour of
some incompatible polymer blends in the melt - Interpretation of data with a model of
emulsion of viscoelastic liquids Macromolecules 26,320-329
7. Vinckier, I., P. Moldenaers, & Mewis, J., (1996) Relationship between rheology and
morphology of model blends in steady shear flow J. Rheol. 40,pp. 613-631
8. Minale, M., P. Moldenaers, & Mewis, J., (1997) Effect of shear history on the
morphology of immiscible polymer blends, Macromolecules 30,pp. 5470-5475
9. Normand, V., P.D.A. Pudney, Aymard, P., & Norton, I.T., (2000) Weighted-average
isostrain and isostress model to describe the kinetic evolution of the mechanical
properties of a composite gel: Application to the system gelatin : maltodextrin, J.
Appl. Polym. Sci. 77 ,pp. 1465-1477
10. Aymard, P., M.A.K. Williams, Clark, A.H., & Norton, I.T., (2000) A turbidimetric
study of phase separating biopolymer mixtures during thermal ramping Lungmuir 16,
pp. 7383-7391
11. Taylor, G.I. (1934) The formation of emulsions in definable fields of flow Proc. Roy.
SOC.A146,pp. 501-523
12. Bird, R.B., Armstrong, R.C., and 0. Hassanger. (1987) Dynamics of Polymeric
Liquids. Volume 1: Fluid Mechanic. Wiley Interscience, 2nd Edn..
13. Stokes, J.R., Wolf, B., and Frith, W.J. (2001) Phase-separated biopolymer mixture
rheology: Prediction using viscoelastic emulsion model, Journal of Rheology, In
Press.
14. Takahashi, Y., S. Kitade, Kurashima, N., & Noda, I., (1994) Viscoelastic Properties
Of Immiscible Polymer Blends Under Steady And Transient Shear Flows, Polymer
Journal 26,pp. 1206-1212.
15.Doi, M. & Ohta, T., (1991) Dynamics And Rheology Of Complex Interfaces .1.
Journal Of Chemical Physics 95, pp. 1242-1248.
16. Lee, H.M. &. Park, O.O., (1994) Rheology And Dynamics Of Immiscible Polymer
Blends, Journal Of Rheology 38, pp. 1405-1425.
17. Grmela, M. & Aitkadi, A., (1994) Comments On The Doi-Ohta Theory Of Blends,
Journal Of Non-Newtonian Fluid Mechanics 55, pp. 191-195.
18. Lacroix, C., M. Grmela, & Carreau, P.J., (1998) Relationships between rheology and
morphology for immiscible molten blends of polypropylene and ethylene copolymers
under shear flow, Journal of Rheology 42,pp. 41-62.
INFLUENCE OF PENTOSANS ON THE RHEOLOGICAL AND THERMAL
PROPERTIES OF STARCH ISOLATED FROM TWO DIFFERENT WHEAT
VARIETIES
Dora M.J. dos Santos & J.A. Lopes da Silva
Departamento de Quimica, Universidade de Aveiro, 38 10-193 Aveiro, Portugal
1 INTRODUCTION
Pentosans, a heterogeneous mixture of non-starch polysaccharides, account for a small

fiaction of wheat flour composition, but are known to affect the physical properties of
gluten and doughs and thereby the baking performance of wheat They are
characterized by a high water-absorbing capacity and high viscosity, and, therefore, are
expected to compete with starch for water.
The properties of starch pastes and gels have been shown to be dependent on
concentration, size and rigidity of starch granules, swelling degree and amount of amylose
and amylopectin leached out during heating, amylose-amylopectin, granule-to-granule,
amylose-granule and amylopectin-granule interactions.6-0In addition, water is known to
affect gelatinisation, the gelation mechanism and retrogradation of starch.
The gelatinisation and retrogradation of starch is also affected by the presence of
several hydrocolloids, including pentosan~.~~ The quality of baked products and starch-
based ingredients can, thus, be affected by the level and properties of these
polysaccharides. The present research assessed the role of wheat water-soluble pentosans
on the rheological and thermal behaviour of starch dispersions during gelatinisation and
after gel formation. Two different water levels were studied, considered as representative
of intermediate and excess hydration, using both rheological and DSC measurements.
2.1 Samples
Prime starch and the water-soluble fiaction were obtained fiom defatted flours of two
Portuguese wheat varieties, Amazonas (soft wheat) and Sorraia (hard wheat), as
previously de~cribed.~ Water-soluble pentosans (WSP) were extracted from the water-
soluble fiaction by a modified method fiom the one described by Izydorczyk et al. 4, based
on denaturation of water-soluble proteins by heat treatment followed by enzymic (a-
amylase) digestion of soluble starch, ethanol precipitation and washing of the WSP
fiaction using graded volatile organic solvents. WSP were oven-dried at 35C.
138 Gums and Stabilisers for the Food Zndustry 11
2.2 Physico-chemical analysis
Moisture, protein, and crude fat content were determined using AACC approved
methods.I6 Total starch, amylose content and the level of damaged starch were determined
using enzymatic kits (Megazyme, Wicklow, Ireland), following the methods of McCleary
and c o - ~ o r k e r s . ' ~ The
- ' ~ relative viscosities of WSP aqueous solutions were measured
with a Cannon-Fenske type capillary viscometer at 25.OoC+0.1"C, and the intrinsic
viscosities were determined by the double extrapolation method of the Huggins and
Kraemer equations2' The relative amounts of monosaccharides were determined by GLC
after derivatisation to alditol acetates.21
2.3 Viscoelastic Behaviour
The effect of water-soluble pentosans on starch gelatinisation was studied at two different
levels of hydration (50% and 8O%w/w) and at different WSP concentrations (0.2-
l%,w/w). The aqueous starch/WSP dispersions were left for hydration under gentle
stirring, for two hours. Gelatinisation was allowed to occur "in situ" on the rheometer-
measuring device. Rheological tests were carried out using a controlled-stress rheometer
(AR-1000, TA Instruments), fitted with cone-and-plate geometry (angle 2", 4 cm
diameter). The water loss was minimised by placing low-viscosity mineral oil on the
exposed edge of the sample. Structure development during heating and cooling (20-80-
20C) was followed by temperature sweep experiments (2"C/min) at constant oscillatory
fiequency (0.5 Hz) and low strain amplitude (0.5% strain), and the viscoelastic behaviour
of the gels evaluated by fiequency sweeps right after the end of the cooling ramp (20C,
1% strain). All tests were performed at least in triplicate.
2.4 Differential scanning calorimetry
DSC assays were made on a DSC-50 (Shimadzu, Japan). An aliquot ( 1 0 s mg) of the
starchlwater or starch/WSP/water dispersions, prepared as above, was transferred into a
high-pressure aluminum pan, accurately weighed, and hermetically sealed. The sample
pans were heated fiom 25C to 140C at a scanning rate of 2"C/min, in a nitrogen
atmosphere. Then the sample pan was cooled slowly to the room temperature and stored at
4C for 7 days. A second heating run was performed after this period of time. An empty
pan was used as the reference. All tests were performed at least in triplicate. For each
endotherm, transition temperatures and enthalpies (AH) were evaluated using the TA
Work Station software (Version 1.O 1, Shimadzu, Japan).
Amazonas ( A M ) starch showed an amylose content ~ 4 % lower than the Sorraia (SOR)
starch (Table 1). Sorraia starch showed a somewhat higher lipid content, The higher
damage starch level present in the Sorraia starch is likely associated with the higher
mechanical damage during the flour milling process, since Sorraia corresponds to a hard
wheat variety. In fact, several reports (see e.g. Zeng et a1.22)have shown that grain
hardness is highly correlated to the level of damage starch. Both starch samples showed
similar protein content.
Rheological Aspects I39
Table 1 Main physico-chemical characteristicsa of the

starch and water-soluble pentosanfractions.
AMA SOR
Starch
Total starch, % 93.0 f 0.3 92.2 f 0.4
Amylose, % 22.3 f 1.4 26.2 f 0.4
Damaged starch, % 0.65 k 0.01 1.10 f 0.05
Residual protein, % 0.38 f 0.04 0.41 f 0.07
Crude fat, YO 1.30 f 0.02 1.55 f 0.06
Internal polar lipids, %b 0.22 f 0.03 0.29 f 0.05
WSP
Total neutral sugars, % 77.9 f 0.6 81.5 k0.4
Ara/xyl 0.83 f 0.01 0.84 f 0.01
Residual protein, % 13.3 f 0.4 12.4 f 0.3
h l , dL/g 2.24 k 0.01 3.78 k 0.01
a Mean f (standard deviation) for triplicate measurements (except
for amylose - five replicated measurements); dry weight basis
ButanoVwater extracted lipids, according to Eliasson et d.25
Both pentosan samples presented similar physical-chemical characteristics (Table 1),

being the main observed difference the higher intrinsic viscosity of the sample extracted
fiom the Sorraia wheat variety, which must be associated to a higher molecular weight.
3.1 Effect of WSP at Intermediate Water Level
Under the experimental conditions that have been used, two endothermic transitions have
been observed for the wheat starch aqueous dispersions. The first, corresponding to the
starch gelatinisation, and the second to the phase transition within amylose-lipid
complexes23y24. Table 2 shows the values obtained of the peak temperatures and transition
enthalpies associated with the gelatinisation and amylose-lipid endothermic transitions, for
50% water level. Also, there are shown the results obtained fiom the second heating run,
after storing the samples for 7 days at 4C.
At 50% starch, SOR starch gelatinises at lower temperatures than the A M starch, but
no significant differences were found between the gelatinisation enthalpies. For both
starch samples, the gelatinisation peak in the DSC thermograms shifted slightly to higher
temperatures with increasing WSP concentration (0.2-1%, w/w), whilst the endothermic
enthalpy of gelatinisation were not significantly affected (at ~60.05).
Generally, the retrogradation peaks in the second heating run,after 7 days storage at
4"C, shifted to lower temperatures than the gelatinisation peaks, and are associated with
lower enthalpies, whilst the endothermic peak associated with the amylose-lipid
disintegration shifted to higher temperatures and also showed lower enthalpies. It is
known12 that once the starch granules are gelatinised, thus losing their order and
crystallinity, the ordered structure cannot be completely recovered even after storage at
low temperatures for a long time.
Enthalpies of the starch-WSP-water mixtures stored for 7 days were similar or slightly
smaller than those of the starch-water systems, but the presence of WSP shifted the
retrogradation endothermic peaks to higher temperatures and broadened the temperature
range of the transition endotherm.
Table 2 Peak temperatures (Tp) and transition enthalpies @

evaluated
If) for starch-
WSP mixtures at 50% hydration level."
A M Starch I SOR Starch

O%WSP O.S%WSP I%WSP I O%WSP O.S%WSP I%WSP
Gelatinisation
T@/OC 57.0 k 0.6 57.5 f0.7 58.9 f 1.0 53.4 f 0.2 54.3 k 0.7 55.9 f 0.3
/Jg-'(b) 2.2 f 0.2 2.3 k 0.2 2.3 f 0.3 2.9 f 0.6 1.9 f 0.1 2.6 f 0.2
Amylose-lipid
dissociation
TpMIoC 1 1 1 . 1 f 1.0 110.5 k0.8 109.5 k 1.0 106.6 f0.6 109.8 f 1.0 110.9 k 0.2
AHAL/Jg-'(b) 2.0 f0.2 1.2 & 0.1 1.6 f 0.4 1.3 f 0.4 0.73 & 0.2 0.76 & 0.06
Retrogradation
T p ~ I 0 C 49.0 k 0.6 50.6 f 0.9 53.0 f 0.8 44.3 k0.8 48.4 k0.6 51.8 +_ 1.3
/Jg-'(b) 1.2 f0.1 1.o k 0.3 0.50 k 0.07 0.48 f 0.09 *
0.33 0.07 0.28 f 0.04
TdAmyl. -lipid
dissociation
Tp$~IOC 113.2f0.8 112.8kO.9 113.2fO.S 110.3f2.3 107.6f0.4 108.8k0.4
miL/Jg-'(b) 0.4 1 k 0.12 0.40 f0.20 0.44 f 0.12 0.57 f 0.04 0.53 f0.1 0.47 f0.06
a Mean f (standard deviation) for triplicate measurements; Based on dry weight of starch
The results obtained fiom the second heating run must be regarded with caution, due
to the small endothermic peaks, and given the relatively low sensitivity of the calorimeter.
However, they might help to conclude that under these conditions, retrogradation is not
promoted by the presence of the WSP.
The changes of the storage modulus (G') during heating and cooling (20-80-20C) of
starch/WSP dispersions at 50% hydration level are shown in Figure 1. When starch is
placed in water, the granules start to swell slightly due to the water penetration, leading to
an increase of their volume. When a starch dispersion is heated, no significant changes
occur until a certain critical temperature is reached. This critical temperature,
corresponding to the rapid increase in the storage modulus, is related to that point when
irreversible changes occur. Upon heating, a larger additional swelling of the starch
granules will bring about an increase of the granule-to-granule interactions leading, thus,
to a rapid increase of the viscoelastic parameters.
The A M starch showed an onset of gelatinisation at higher temperature than the SOR
starch (Figure l), as defined by the initial rapid increase in the storage modulus (G'Pa), in
concordance with the DSC results. The lower onset temperature of the SOR starch may be
related with its higher amount of starch damage. In fact, it was shown that for certain
wheat varieties a high correlation exists between starch damage and the onset temperature
of gelatinisation, as measured by viscosimetric methods.22
The A M starch also showed a higher peak modulus. The peak modulus corresponds
to the maximum apparent modulus reached during the heating process. High negative
correlations between peak paste viscosity and the amylose content of starch have been
reported.22This effect was attributed to the greater swelling occurring for starches with
higher amylopectin (lower amylose) content.22,25326However, Miller et ~ 1 . attributed
2~ the
most important cause of higher starch peak viscosity to the greater amylose leaching in
reduced-amylose starch such that, in combination with reduced fiee water, a more viscous
network is formed. In the absence of shearing and for the relatively low temperature that
was reached during the experiments (SOOC), the most likely explanation for the observed
results is the higher swelling degree for the AA4A starch granules, the one with lower
amylose content. Supporting this interpretation, we have also observed a very low increase
in structure during cooling, similar to both samples, indicating that a small amount of
amylose was leached out fiom the granules under these conditions.
The presence of WSP shifted the onset temperature of the rheological events to higher
values (Figure 1 - inserts). The peak temperature is not significantly affected, but it is
observed a decrease in the peak modulus with increasing WSP concentration. During the
cooling process back to 2OoC, the changes in G' are small and only slightly affected by the
presence of WSP. Mechanical spectra performed at 20C revealed that the viscous
character of the starch gels increases with increasing WSP concentration.
350 I 4 0
7 30
300 -
- 20
250 I 10
0
cb 200 I 45 - 50 55
$
3 150 I
50
loo
0
1
1
10
I I
20
I I
30
1 I
40
I I
50
temperature PC
// I
60
, (7
70
I-
80
1 -
90 20 30 40 50 60
temperature PC
70 80 90
Figure 1 Effect of WSP on viscoelastic properties of (A) Amazonas and (B) Sorraia
starch during temperature sweeps (20-80(10 min)-2O0C, y =O. 5 %, 2"C/min) at 50% (w/w)
- -
hydration: I - 0%, 2 - 0.2%, 3 0.5%, and 4 1 % WSP.Inserts show details on the initial
rapid increase in the storage modulus (G '/Pa).
3.2 Effect of WSP in Excess of Water
Table 3 shows the values obtained of the peak temperatures and transition enthalpies
associated with the gelatinisation and amylose-lipid endothermic transitions, for 80%
water content. From the second heating run,after storing the samples for 7 days at 4"C, no
endothermic peaks have been detected, because of the expected lower retrogradation rate
under these conditions andor due to the low sensitivity of the measuring system.
Table 3 Peak temperatures (Tp) and transition enthalpies (M)evaluatedfor starch-

WSP mixtures at 80% hydration level."
A M Starch SOR Starch

O%WSP O.S%WSP I%WSP O%WSP O.S%WSP I%WSP
Gelatinisation
T ~ G P C 59.2 f0.5 59.8 f 0 . 4 60.9 f 0 . 7 56.1 fO.2 57.7 f 0 . 2 58.0 +O.2
&&/Jg-'(b) 5.8k0.4 3.4f0.3 2.2k0.2 11.1k1.2 3.7k0.7 1.9k0.2
Amyl.-lipid diss.
T p/"C~ ~ n.d. ('I n. d. n.d. 102.2 f 0 . 9 100.0 f 1.5 n.d.(')
A& /Jg-'(b) n.d. n.d. n.d. 2.4 f 0.3 1 . 1 f0.4 n.d.
a Mean k (standard deviation) for triplicate measurements; Based on dry weight of starch
Not detectable
The gelatinisation onset and peak temperatures, and transition enthalpies were higher
at this higher hydration level. A different behaviour was observed for the transition
temperatures associated with the amylose-lipid dissociation within the SOR granules;
when the water content increases, the endotherm is shifted towards lower temperature,
although the transition enthalpy also increases. SOR starch also gelatinises at lower
temperatures than A M starch, but gave higher values of gelatinisation enthalpies.
The endothermic enthalpy of gelatinisation decreased with WSP concentration. A
more pronounced decrease in AHG was observed for the SOR starch-WSP mixtures. The
peak temperatures increased with WSP, especially for the SOR samples, whilst those of
the disintegration of amylose-lipid complex decreased.
The changes of the storage modulus (G') during heating and cooling (20-80-20C) of
starcWSP dispersions are shown in Figure 2. As expected, hydration of starch to 80%
(w/w) lead to higher gelatinisation temperatures and lower G' values. As the volume
fraction of starch granules decreases, the interactions between them are weaker, leading to
a weaker gel network, and higher temperatures are necessary to bring about the rapid
increase of the viscoelastic parameters. At this water level the AMA starch also showed an
onset of gelatinisation at higher temperature than the SOR starch (Figure 2, inserts A.l and
B.l), and also a higher peak modulus, although the differences between samples were less
pronounced at this higher water content. At higher water level, more extensive swelling
and higher amylose leaching occurred, compared to what was observed for 50% starch,
thus leading to a higher increase in G' during cooling, meaning a higher structure
development. During the cooling process (80-20C) the difference between starch samples
is small, being the structure development profile very close for both starches (Figure 2).
Generally, the effect of the WSP is much more pronounced at this higher water level,
in agreement with that observed in the DSC experiments. Addition of WSP increased the
gelatinisation temperature and decreased the highest value of the storage (G') and loss
(G') moduli reached during heating to 80C. The effect was much more pronounced for
the SOR starch-WSP mixtures, which may be attributed to the higher intrinsic viscosity of
this WSP sample. Clearly a more drastic effect is observed for the SOR sample, with the
WSP hindering the building of the starch network during cooling (Figure 2, insert B. 1)
AAer heating to 80C and cooling down to 20C, an increase in pentosan concentration
resulted in higher tan 6 (G"/G') and lower G' values of the starch gels, and the effect was
more pronounced at higher water levels (Figure 3). This effect may be attributed to a less
efficient network formation in the presence of WSP, for the experimental conditions that
60
4
20 40 60 80
OCL i r
10 20 30 40 50 60 70 80 90 20 30 40 50 60 70 80 90
temperatme PC temperam PC
Figure 2 Effect of WSP on viscoelastic properties of (A) Amazonas and (B) Sorraia
starch during temperature sweeps (20-80(10 min)-2O0C, y =0.5 %, Z"C/min) at 80% (wlw)
-
hydration: 1 - O%, 2 0.2%, 3 - 0.5%, and 4 - 1% WSP. Inserts (A.1) and (B.1) show
details on the initial rapid increase in the storage modulus (G 7, and (B.2) details on the
G ' changes during cooling 8O-2O0C,for the Sorraia samples.
0.0 1 0.1 1 10 0.0 1 0.1 1 10

frequency /Hz
frequency /Hz
Figure 3 Effect of WSP on viscoelasticproperties of (A) A M and (B) SOR starch at 80%
(w/w)hydration. Storage modulus (G ', squares) and tan S(diamonds) measured at 20C
after the heating and cooling steps. a
0-0% WSP; 4 +-1%(wh) WSP.
have been analysed, which may have an inherent and important technological implication
in many starch-based food processes.
Due to their high water-absorbing capacity, WSP compete with starch for water
afinity, lowering the water availability for swelling and gelatinisation of the starch
granules. This may explain the higher temperatures for starch gelatinisation, observed in
both DSC and rheological tests. However, competition for water and decrease in water
availability due to the presence of WSP could not explain the observed effects during
144 Gums and Stabiiisers for the Food Industry 11
cooling, especially the detrimental effect for gel network development associated with the
SOR WSP. Changes in the rheological results during cooling and DSC results of starch-
WSP with storage time, and the more pronounced effects observed at excess water,
suggest that the WSP molecules, besides competing with starch for water, might also
interact with the amylose and amylopectin and prevent reorganisation of both
macromolecules within the starch network.
Acknowledgements
We thank the ENMP (Elvas, Portugal) for providing the wheat flour samples, Ms. Claudia
Gama for assisting during starch and pentosan extraction, Ms. Susana Cardoso for helping
with the sugar analysis, and the FCT (Lisboa, Portugal) for financial support (Project
PRAXIS PCNA/BI0/0703/96 and PRAXIS XXI/BD/14770/96).
References
1 T. Sasaki, T. Yasui and J. Matsuki, Food Hydrocolloids, 2000, 14,295.

2 A.L. Faurot, L. Saulnier, S. Berot, Y. Popineau, M.-D. Petit, X. Rouau and J.-F. Thibault,
Lebensm.-Wiss.u.-Tehnol., 1995,28,436.
3 J. Michniewicz, C.G. Biliaderis and W. Bushuk, Cereal Chem.,1990,67,434.
4 M.S. Izydorczyk, C.G. Biliaderis and W. Bushuk, J. Cereal Sci.,1990,11,153.
5 A.C. Gama, D.M.J. Santos and J.A. Lopes da Silva, in Wheat Gluten, ed. P.R. Shewry
and A.S. Tatham, Royal Society of Chemistry, Cambridge, 2000, p. 503.
6 M.J. Miles, V.J. Morris, P.D. Orford and S.G. Ring, Carbohydr. Rex, 1985,135,271.
7 S.G. Ring, StarcWStarke, 1985,37,80.
8 A.-C. Eliasson, 3: Texture Stud., 1986,17,253.
9 J.O. C d and Y. Zhou, J Rheol., 1998,40,221.
10 L.M. Hansen, R.C. Hoseney and J.M. Faubion, Cereal Chem., 1991,68,347-351.
11 A.-C. Eliasson, StarcWStarke, 1980,32,270.
12 C.G. Biliaderis, Food Technol., 1992,46(6), 98.
13 V.P. Yuryev, I.E. Nemirovskaya and T.D. Maslova, Curbohydr. Polym., 1995,26,43.
14 S.K. Kim and B.L. DAppolonia, Cereal Chem., 1977,54, 150.
15 M. Gudmundsson, A.-C. Eliasson, S. Bengtsson and P. Aman, StarchiStarke, 1991,
43,5.
16 American Association of Cereal Chemists. Approved Methods of the American
Association of Cereal Chemists. AACC, St. Paul, MN, USA, 1995.
17 B.V. McCleary, V. Solah and T.S. Gibson, J. Cereal Sci., 1994,20,5 1.
18 T.S. Gibson, V. Solah and B.V. McCleary, J. Cereal Sci., 1997,25, 111.
19 T.S. Gibson, H. Al Qalla and B.V. McCleary, J Cereal Sci., 1991,15, 15.
20 E.R. Morris and S.B. Ross-Murphy, in Techniques in Carbohydrate Metabolism, ed.
D.H. Northcote, Elsevier, Amsterdam, 1981, B310, p. 1.
21 A.B. Blakeney, P.J. Harris, R.J. Henry and B.A. Stone, Carbohydr. Rex, 1983, 113, 291.
22 M. Zeng, C.F. Morris, I.L. Batey and C.W. Wrigley, Cereal Chem., 1997,74,63.
23 C.G. Biliaderis, T.J. Maurice and J.R. Vose, J Food Sci., 1980,45, 1669.
24 J.W. Donovan, Biopolymers, 1979, 18,263.
25 G.B. Crosbie, J. Cereal Sci., 1991,13, 145.
26 H.N. Dengate, in Advances in Cereal Science and Technology, ed. Y. Pomeranz,
American Association of Cereal Chemists, St. Paul, MN, USA, 1984, vol. VI, p. 49.
27 B.S. Miller, R.I. Derby and H.B. Trimbo, Cereal Chem., 1973,50,271.
GELATINISATION PROPERTIES OF SAGO STARCH IN THE PRESENCE OF
SALTS
Fasihuddin B. Ahmad and Peter A. Williams"
Faculty of Resource Science & Technology, Universiti Malaysia Sarawak, 94300 Kota
Samarahan, Sarawak, Malaysia.
*Centre for Water Soluble Polymers, The North East Wales Institute, Plas Coch, Mold
Road, Wrexham LL11 2AW, United Kingdom.
Abstract
The effect of various salts on the gelatinisation properties of sago starch has been studied
using differential scanning calorimetry. The presence of salts affected the gelatinisation
temperature, Tge, and gelatinisation enthalpy, AH depending on salts used and their
concentrations. Generally in the presence of salting-out ions such as sulphate, Tgel
increased with increasing salts concentration, while in the presence of salting-in ions such
as thiocyanate and iodide, Tgel and AH decreased. For other salts such as sodium chloride,
potassium chloride and sodium nitrate Tgel increased at low salt concentration and then
decreased as the salt concentration increased while AH decreased with concentration. For
salts such as calcium chloride, magnesium chloride and lithium chloride, at low
concentration, Tgel increased slightly, then decreased to a minimum (- 3M for CaC12, -
3M for MgC12 and - 7M for LiCl) and increased again at higher salt concentrations and
the transition became exothermic. The effect of salt concentration on the gelatinisation
properties of other starches was also studied. Metal chloride salts appear to have similar
effect on the gelatinisation of other starch types although the decrease in Tgel at high
concentration was more pronounced for B-type starches (potato) compared to A-type
starches (corn and tapioca). Sago starch which has a CA-type structure behaved typically
as A-type starches.
Introduction
Starch is composed of amylose, which is a linear molecule consisting of (1 +4)-linked a-

D-glucopyranose units and amylopectin which is a highly branched molecule consisting
of short chains of (1 +4)-linked a-D-glucose with (1 +6)-a-linked branches. Although
amylose accounts for only 20-30% of the total in the case of many common starches, it
makes a major contribution to the overall properties. Starch occurs in the form of
granules which are insoluble in cold water and when aqueous suspensions are heated
above a certain temperature, swelling and dissolution occurs. The degree of swelling,
granule disintegration and release of amylose depends on various factors such as type of
starch, starch concentration, temperature and the presence of solutes (Biliaderis, 1991).
Sago starch is isolated from sago palm, MetruxyZun spp. especially Metruxylun
sagu which is distributed throughout Southeast Asia. A mature palm can produce 100-
550 kg of sago starch. Previous studies have shown that it consists of -30% amylose.
The granules have an average diameter of 30 pm, and exhibit a C-type X-ray diffraction
pattern. The gelatinisation temperature is -70 "C (Ahmad et al., 1999).
Salts have a different and rather complex effect on the gelatinisation of starch. It
has been reported that salts can cause an elevation or depression of gelatinisation
temperature and gelatinisation enthalpy depending on the type of salt and their
concentration (Wooton and Bamunuarachchi, 1980; Evans and Haisman, 1982;
Chungcharoen and Lund, 1987; Oosten, 1990; Paredes-Lopez and Hernandez-Lopez,
1991 ; Jane, 1993). For example in the presence of NaC1, To, Tgeland T, increased initially
with salt concentration and then decreased as the concentration was increased hrther for
various types of starches (Evan and Haisman, 1982; Lii and Lee, 1993). The
gelatinisation enthalpy showed a similar behaviour to the gelatinisation temperature.
Certain salts such as sodium sulfate caused a progressive increase in the gelatinisation
temperature of potato starch, from about 62 "C to 80 OC, while sodium bromide brought
about a reduction to about 44 "C (Takahashi and Wada, 1992).
The objective of this study is to investigate the effects of various salts and their
concentration on the gelatinisation properties of sago starch.
Materials and Methods
Materials- Sago starch used in this study was a food grade sample and was a gift from
Netsei Sago Industry, Sarawak, Malaysia. It was found to have moisture content of
13.9% and an amylose content of 31%, average particle size around 30 pm and molecular
mass of the amylose fraction 7.9 x lo5 dalton. The sample was used as provided without
any further treatment. Pea, potato, corn and tapioca starches (a gift from Cerestar,
Trafford Park, UK) were used as received for comparison.
All salts used in this study were of analytical grade and purchased from Sigma
Chemical co.
Differential Scanning Calorimetry (DSC). DSC measurements were performed using a
micro DSC (Setaram, Lyon, France). Starch samples were prepared on a dry weight
basis, and 10% (w/v) starch was used for all experiments. Salt solutions at the required
concentration were prepared on a molar basis. An appropriate amount of starch was
added to the salt solutions and the pH of the suspension adjusted to 5.5 by adding 0.02 M
HC1 or NaOH. The suspensions were shaken thoroughly before being weighed into the
DSC cell. An aliquot of the sample (-0.95 g) was transferred to the DSC measurement
cell, the top sealed, and the weight recorded. An equal mass of solvent was added into the
reference cell. The cells were heated from 0 to 99 "C at a rate of 0.5 "C/min. The initial
gelatinisation temperature (To), the peak temperature which corresponds to gelatinisation
temperature (Tgel),the conclusion temperature (Tc), and the gelatinisation enthalpy AH
were determined. Two measurements were performed for all samples and results are
given as an average. The results were very reproducible and agreed within *O. 1%.
Results and Discussion
Table 1 shows the gelatinisation temperature and gelatinisation enthalpy of sago starch in
the presence of various salts. In the presence of NaCl, KC1 and NaN03 at low
concentration, Tgel increased slightly to a maximum value, and at much higher

concentration Tgel decreased with concentration. For Na2S04, Tgel increased significantly
with increasing concentration, while for iodide and thiocyanate Tgel decreased
significantly. In the presence of NaI, NaSCN, KI and KSCN at a concentration of - 2M,
the starch sample was gelatinized at room temperature. AH increased slightly at low salt
concentration for Na2S04, NaC1, KC1 and NaNO3, but at high salt concentration AH
decreased with salt concentration. For KSCN, NaSCN, ISI and NaI, AH decreased with
salts concentration. Similar observations were observed for corn, potato and tapioca (data
not shown).
Table 1: Effect of salt concentration on the gelatinisation temperature, Tgel and

gelatinisation enthalpy, AH for 10% sago starch
Notes: na- not available
Figure 1 and 2 show DSC endotherms for sago starch in the presence of CaC12
and LiCl respectively, while Table 2 shows the effects of LiCl, CaC12 and MgC12
concentration on the gelatinisation properties of sago starch. For CaC12, initially Tgel
increased at low salt concentration to a maximum value, then decreased to pass through a
minimum and increased again at concentration -4 M where the transition became
I 0.5 m W ( for 12 ivl LiCl and 15 bl LiCl = 1 mbv)
I I I I I I I i
Temperature/C
Figure 1 DSC heating curves for 10% sago starch. a) sago starch alone and in the
presence of b) 0.25M c) 2M d) 4M e) 5M f) 6M g) 12M and h) 15M LiCl
,
b
I, 0.5 mW 30(for 6.5 M 40

20 50
CaClZ = 1 mW) 60 70 80 90
100
Figure 2 DSC heating curves for 10Y0sago starch. a) sago starch alone and in the
presence of b) 0.2m c)lM d) 2.m e) 3M f) 3.5M g) 4M h) 5M and I) 6.5M CaC12
exothermic. Similar observations were noted for MgC12. In the presence LiCl, Tgel
increased initially and at a concentration of > 4 M, Tgel decreased to pass through a
minimum and increased again at concentration > 12 M and the transition became
exothermic. Similar observations were recorded for corn, potato and tapioca starch (data
not shown).
Table 2: The effect of LiC1, MgC12 and CaC12 on the gelatinisation temperature, Tgel and
gelatinisation enthalpy, AH for 10% sago starch
Notes: na-not available
Generally the influence of ions was in accordance with the Hofmeister series.
Thus, for salt concentrations up to lM, for the sodium salts of the various anions Tgel and
AH decreased in the order SO:-> C1- > N O i > I- > SCN-. For the cations Tgel decreased
in the order Mg2" > Ca2' > Li" > Na" > K" while AH did not show any specific trend.
Anions seem to have a greater effect on the gelatinisation process compared to cations.
Various suggestions have been given to explain the effects of salts on Tgeland AH (Evans
and Haisman, 1982; Oosten, 1982, 1983, 1990; Jane, 1993, Ahmad and Williams 1999).
It seems that the influence of salts on the gelatinisation properties of starch can be
attributed to various factors especially the influence on polymer-solvent interaction, the
effects on water structure and the electrostatic interaction between starch and the ions.
According to Jane (1993), ions of high charge density such as SO:- increase the structure
of water, stabilize the starch granules, reduce the fraction of free water molecules,
increase solution viscosity, retard diffusion and subsequently increase Tgel and AH. On
the other hand, ions of low charge density such as SCN- and 1- interact strongly with
starch molecules which induces formation of a single helical conformation as inferred
from 13 C-NMR studies (Chinachoti et al., 1991), which facilitates chain dissociation.
According to Jane (1993), KI and KSCN solutions have an increased fraction of free
water and will reduce the solvent viscosity, facilitate the diffusion into the granules and
hence reduce Tgel and AH. The negative potential of the starch granule attracts the cations
and repels the anions. The attraction or repulsion will be proportional to the charge
density of cations and anions. For the anions, an increase of charge density causes more
structured water and higher repulsion with the starch -OH groups and the effects will
stabilize the starch granules (Jane, 1993). Due to this effect, gelatinisation temperature
will increase as the charge density and concentration of salts increases. On the other hand
for cations, increase in charge density will result in more structured water, stabilize the
starch granules and increase the gelatinisation temperature. At high salt concentrations
the interaction with starch granules was sufficiently exothermic to induce the crystalline
regions of the starch granule to melt and subsequently reduce Tgel and AH. For LiCl,
CaC12 and MgC12 which were studied to much higher concentrations complex behavior
was observed where at high concentrations an exothermic peak was observed. Although
the crystalline structure of the starch granules is disrupted at room temperature in the
presence of these salts, some crystalline regions remain as noted by observation with a
polarizing microscope. On heating, these remaining crystalline regions are disrupted but
an endothermic peak is not observed, because at the same time the amylopectin chains
disaggregate through disruption of intermolecular hydrogen bonds. The hydroxyl groups
which became exposed to the solvent interact exothermically with the cations present in
solution. The fact that the exothermic process is non-reversible supports this view.
Figure 3 shows DSC endothems in the presence of KCl and NaCl for sago and
pea starches. For pea starch the DSC endothems exhibited two peaks which became
apparent at salt concentrations > 0.5M. The first peak corresponds to the peak obtained in
the absence of salts (-57 "C) and the second peak occurs at higher temperature - 70 "C.
Other starches (corn, potato and tapioca) behaved similarly to sago starch and gave a
single DSC peak.
Figure 4 shows the effect of NaCl on Tgel for various starches. Tgelwas reduced to
a greater extent for potato and pea starch. Almost similar behaviour was observed for the
effect of NaCl on sago, corn and tapioca starches. For pea starch, the first peak showed
similar behaviour to potato starch, while the second peak showed similar behaviour to
sago, corn and tapioca starches.
Metal chloride salts appear to have a similar effect on the gelatinisation of other
starch types although the decrease at high concentration is more pronounced for B-type
starches (potato) compared to A-type starches (corn and tapioca). Sago starch with a CA-
type structure behaves as an A-type, whereas pea starch with a CB-type structure shows
two endothermic peaks. One closely follows the behaviour of A-type starches while the
other follows the behaviour of B-type starches (Refer to Figure 4). In pea starch the outer
part of the granules contain A-type and will be disrupted at higher temperature, while the
central part contains the B-type which will be disrupted at lower temperature
(Bograchewa et al., 1998). The structure of A-type starches is more compact than that of
B-type. For A-type starches double helices occupy the central open space while in the B-
types, the central space is occupied by water molecules (Sarko & Wu, 1972; Imberty &
Perez, 1988). Since B-type starches are less compact, ions can penetrate easier into the
granules and reduce Tgel to greater extant than for A-types. Similar observations were
reported for pea starch in KCl (Bogracheva et al., 1998) and lotus starch in NaCl (Lii &
Lee, 1993). Although sago starch is a C-type starch, the properties are different to other
C-types such as pea and lotus starches most probably because the crystallinity within
individual granules is more homogeneous and their properties are close to other A-types
starches.
40 50 60 70 80 90 LOO
Temperature /C
Figure 3 DSC heating curves for 10% starch a) sago starch alone b) sago starch/
1.5M KCI c) sago IlSM NaCl d) pea starch alone e) pea/ 1.5M KCI f) peal 1.5M
NaCl
85
80
75
70
65
60
55
50
+Tapioca Jt Pea 1
45
+Pea 2 +Potato
40
0 2 4 6
Sa1t Concentration /M
Figure 4 Effect of NaCl concentration on the gelatinisation temperature of various
starches
Conclusions
The effects of various salts on gelatinisation properties of sago starch have been studied.
Salts have been found to have complex effects on the gelatinisation properties and the
effect was found to depend on the type of salt and their concentration. There are two
main factors that might contribute to the effect on gelatinisation properties i.e. the effects
of salts on starch-solvent interaction which is influenced greatly by anions and the
interaction of cations with the hydroxyl groups of the starch molecules to form
complexes. Generally in the presence of salting out ions (such as sulfate) Tgel increase
while in the presence of salting in ions (such as iodide and thiocyanate) Tgel decreases.
References
Ahmad, F.B. and Williams P.A., 1999. Effect of salts on the gelatinisation and
rheological properties of sago starch. J.Agric and Fd Chem. 47 3359-3366.
Ahmad, F.B., Williams, P.A., Doublier, J.L., Durand, S. & Buleon, A. 1999. Physico-
chemical characterization of sago starch. Carbohydrate Polymers 38: 36 1-370.
Biliaderis, C.G. 1991. The structure and interaction of starch with food constituents. Cad.
Physiol. Pharmacol. 69: 60-78.
Bogracheva, T.Y., Morris, V.J., Ring, S.G. & Hedley, C.L. 1998. The granular structure
of C-type pea starch and its role in gelatinisation. Biopolymers 45: 323-332.
Chinachoti, P., White, V.A., Lo, L. & Stengle, T.R. 1991. Application of high resolution
carbon-13, oxygen-17 and sodium-23 NMR to study the influence of water,
sucrose, and sodium chloride on starch gelatinisation. Cereal Chem. 68: 238-244.
Chungcharoen, A. & Lund, D.B. 1987. Influence of solute and water on rice starch
gelatinisation. Cereal Chem. 64:240-243.
Evans, I.D. & Haisman, D.R. 1982. The effect of solutes on the gelatinisation
temperature range of potato starch. Starch 34: 224-23 1.
Imberty, A. & Perez, S. 1988. A revisit to the three-dimentional structure of B-type starch
polymorph. Biopolymers 27: 1205-1212.
Jane, J.L. 1993. Mechanism of starch gelatinisation in neutral salts solutions. Starch 45:
161-166.
Lii, C.Y & Lee, B.L. 1993. Heating A-, B- and C-type starches in aqueous sodium
chloride: Effects of sodium chloride concentration and moisture content on DSC
thermograms. Cereal Chem. 70: 188-192.
Oosten, B.J. 1982. Tentative hypothesis to explain how electrolytes affect the
gelatinisation temperature of starches in water. Starch 34: 233-239.
Oosten, B.J. 1983. Explanation of phenomena arising from starch electrolyte interactions.
Starch 35: 166-169.
Oosten, B.J. 1990. Interaction between starch and electrolyte. Starch 42: 327-33 1.
Paredes-Lopez, 0. & Hernandez-Lopez, D. 1991. Application of DSC to amaranth starch
gelatinisation: Influence of water, solutes and annealing. Starch 43:57-61.
Sarko, A. & Wu, H.C.H. 1972. The crystal structure of A-, B-, and C- polymorph in
amylose and starch. Starch 30: 70-78.
Takahashi, K. & Wada, K. 1992. Reversibility of salts effect on the thermal stability of
potato starch granules. J. Food Sci. 57: 1141-1142.
Wooton, M. & Bamunuarachchi, A. 1980. Application of DSC to study starch
gelatinisation. Starch 32: 126-132.
EFFECT OF K+AND Ca2+ CATIONS ON GELATION OF K-CARRAGEENAN
J. Doyle,' P. Giannouli,' K. Philp2and E.R. Morris'*
'Department of Food Science, Food Technology and Nutrition, University College Cork,
Cork, Ireland
2QuestInternational Ireland Limited, Kilnagleary, Carrigaline, Co. Cork, Ireland
1 INTRODUCTION
The carrageenans are linear, sulphated polysaccharides obtained fiom various species of
red seaweed.' Their main use is as thickeners and gelling agents: predominantly in the food
industry. The principal gelling carrageenans are iota (I) and kappa (K). Both have structures
approximating to a disaccharide repeating sequence of P-D-galactose glycosidicdly linked at
positions 1 and 3, and 3,6-anhydro-a-D-galactoselinked at positions 1 and 4. The idealised
disaccharide repeat unit of K-carrageenan contains a single sulphate group, at O(4) of the
D-galactose residue; t-carrageenan is also sulphated at this position, but has an additional
sulphate group at O(2) of the anhydrogalactoseresidue.
The main deviation fiom these idealised structures is the occasional absence of the
anhydride 'bridge' in the 4-linked residues, which introduces a change in chain geometry
('kink').3 Treatment with alkali enhances structural regularity by converting most of the
kinking residues to the anhydride form. Using this procedure of 'alkali modification',
preparations with -95 % of the idealised t-carrageenan or K-carrageenan repeating structure
can be obtained from, respectively,Euchema spinosum or Euchema ~ o t t o n i i . ~
X-ray fibre diEaction studies have shown conclusively that the ordered structure of
t-carrageenan in the solid state is a co-axial double helix, with parallel strands each having
3-fold symmetry! The difhction patterns for K-carrageenan are of poorer quality, but a
broadly similar double helix structure gives best agreement with the experimental data.'
Carrageenan gels form on cooling and melt again on heating. The sol-gel and gel-sol
transitions are accompanied by a co-operative change in chain conformation, which can be
monitored by a range of physical techniques, including optical rotation, NMR linewidth
and differential scanning Although other models have been proposed,' the
most generally accepted interpretation' is that both r-carrageenan and K-carrageenan exist
in solution as disordered coils at high temperature, and that the primary process in gel
formation on cooling is interchain association by formation of the double helix structure seen
in the solid state.
As anticipated fiom polyelectroyte theory, the conformational transition is displaced to
higher temperature by addition of simple electrolytes (salts).' For samples approximating
closely to the idealised r-carrageenan structure, the variation in transition temperature with
varying concentrations of different salts can be explained by the non-specific effect of ionic
strength." For K-carrageenan,by contrast, there is clear evidence of binding interactions with
specific cations. In particular, K+, Rb' and Cs+ give much higher transition temperatures9
than equivalent molar concentrations of Li+ and Na', and specific binding of these large
Group I cations to the ordered conformation of K-carrageenan has been demonstrated by
NMR11-13, with no evidence for any such binding of Li' or Na'.
The cations that have been shown to bind directly to the K-carrageenan helix are also
much more effective than other monovalent cations in promoting gel f~rmation,'~~'~ and
commercial K-carrageenan is therefore normally produced in the K+ (rather than the Naf)
salt form.* Site-binding of K+ cations suppresses electrostatic repulsion between the
carrageenan helices, and allows formation of aggregated 'superstrands' which have been
visualised directly by electron microscopy.l6 Helix-helix association increases gel strength,
and is also believed to be responsible for the thermal hysteresis observed between formation
and melting of K+ K-carrageenan gels, with the aggregated assemblies remaining stable to
temperatures higher than those at which individual helices will form on cooling.*'
Divalent cations are less effectivegthan the 'specific' Group 1 cations (K+, Rb+ and Csf)
in raisirig the temperature of the ic-carrageenan disorder-order transition; they are, however,
more effective than equivalent molar concentrations of the 'non-specific' Group I cations
(Li+ and Na'), which can be explained by their greater charge."
Although the effect of divalent cations on transition temperature seems f m l y
established, their ability to promote aggregation and geIation of K-carrageenan has attracted
little research effort, and the few results available are contradictory. In an investigation
of K-carrageenan converted to specific salt forms by ion exchange, Morris and Chil~ers'~
concluded that Ca2+ cations give stronger gels than K', but the opposite conclusion was
reached in a study of mixed salt forms by Hermansson and co-workers.20
In the present work we have explored the effect of progressive addition of K+ and Ca2+
cations to non-gelling solutions of K-carrageenan in the Na+ salt form.
2 MATEFUALSANDMETHODS
The Na' K-carrageenan used was an alkali-modified extract from Euchema cottonii,
provided by Quest. The polymer was dispersed in distilled deionised'water at ambient
temperature, dissolved by mechanical stirring at 80"C,and brought to the required ionic
environment by addition of the appropriate volume of a concentrated (2 N) solution of
KC1 or CaC12 (AnalaR grade from BDH). All experiments were carried out using a fured
carrageenan concentration of 1 .O wt %.
Samples for compression testing were prepared by filling the hot solutions into
cylindrical moulds (13.5 mm height; 12.35 mrn internal diameter), sealing with lubricated
cover-slips, and storing for 24 h at 5C.Compression curves (3 replicates for each sample)
were recorded at I mm/s using a cylindrical probe (50 mm diameter) on a TA-XT2 Texture
Analyser fkom Stable Microsystems.
Small-deformation oscillatory measurements of storage modulus (G') and loss modulus
(G") were made at 10 rad s-I and 2 % strain, using parallel-plate geometry (40 mm diameter;
0.5 mm separation) on a Carri-Med CSL-100 rheometer. Samples were loaded in the
solution state at 70"C,coated around their periphery with light silicone oil to minirnise
evaporation, cooled to 3C at l"C/min, held at 3C for 2 h, and re-heated to 70C.
Differential scanning calorimetry @SC) measurements were made at a scan rate of
0.3"Umin on a DSC I11 microcalorimeter from Seraram, using water as reference. Sample
and reference pans were balanced to within -t 0.05 mg.
Changes in turbidity accompanying gel formation were characterised by measurement of

absorbance at 400 nm on a Cary IE W-visible spectrophotometer fiom Varian. Solutions
(70C) were filled into a jacketed cell of pathlength 10 mm; the temperature of the sample
was then reduced by manual adjustment of a Haake circulating water bath, and measured
using a thermocouple positioned within the cell, but above the light path. Readings were
taken after thermal equilibration for -10 min at each temperature.
Figure 1 shows the changes in G' and G ' observed for 1.0 wt % Na' K-carrageenan with
5 mM KCl on cooling and heating at l"C/min. The onset of gel formation on cooling is
marked by an abrupt increase in log G', followed by a slower progressive increase at lower
temperatures, with an accompanying, smaller, increase in log G".The gelation process is
fully reversible, but, even at this comparatively low concentration of K,' there is substantial
thermal hysteresis, with final loss of gel structure on heating occurring at a temperature
-15C higher than the onset temperature for gel formation on cooling.
While the gelation and melting processes shown in Figure 1 are smooth and progressive,
similar studies of the gelation of K-carrageenan with higher concentrationsof K+ have shown
an initial increase in modulus, followed by an abrupt decrease.I6 Although this effect has
been interpreted in terms of changes in the structure of the carrageenan network,I6
comparative studies using different measuring geometrie?'J? strongly indicate a slippage
artefact. For systematic comparison of gel strength over a wide range of concentrations of
K+ and Ca2+ we therefore used measurements of Young's modulus (E) from compression
testing, where no such artefacts can occur. The results obtained are shown in Figure 2.
loooo F-----7 -
Triangles cooling
0 10 20 30 40 50 60
Temperature ("C)
Figure 1 Changes in G' Cfilled symbols) and G" (open symbols) on cooling (triangles) and
heating (circles) at I"C/min for 1.0 wt % Na+ *carrageenan with 5 m M added KCI.
Measurements were made at a fiequency of 10 r a d d and 2 % strain.
1 10 100 1000
Cation concentration (mM)
Figure2 Variation of Young's modulus with concentration of J? (a) or Cd' (0)in

mixtures with 1.0 wt % Na+ ficarrageenan.Measurements were made a$er 24 h at 5C.
0.8 I 1 ' 1 I I ' 1 1 -

600
Turbidity
500
0.6
3
0
400
0.4 300
200
0.2
I I 100
t I
0.0 0
0 10 20 30 40 50
Temperature ("C)
Figure3 DSC cooliplg curve (0.3"C/rnin) for 1.0 wt % Na' *carrageenan with 50 rnM
added CaCZ2, in comparison with the increase in turbidity on cooling characterisedfor the
same sampZe by measurements of absorbance at 400 nm.
Progressive addition of KC1 to solutions (1 .O wt YO)of Naf K-carrageenan caused a

progressive increase in modulus of the resulting gels, up to the highest Kf concentration at
which samples could be prepared (500 a) .
For Ca2+, by contrast, the moduli showed an
initial increase with increasing cation concentration, but then decreased sharply with fUrther
addition of CaC1,. The disaccharide repeating unit of K-carrageenan in the Na' salt form
has a molecuIar mass of 408 D. Since each unit contains 1 sulphate group, the sulphate
concentration in the 1.0 wt % solutions used in this investigation is -25 mM. Exact charge
balance with Ca2' ions would therefore be reached at a Ca2+ concentration of -12.5 mM,
which, as shown in Figure 2, is close to the concentration at maximum gel strength,
suggesting the possibility of stoichiometricbinding of Ca2+ions to the K-carrageenan helix.
100
+E 40
20
1 10 100 1000
Cation concentration (mM)
E
s
C'
0
1 10 100 1000
Cation concentration ( m i
Figure 4 Variation of (a) transition-midpoint temperature (Trrs from DSC, and (3) the
reciprocal of Tm , with concentration of CaZf (circles) or (triangles) in mixtures with
1.0 wt % Nu' K-carrageenan during cooling (open symbols) and heating filled symbols).
As discussed above, neutralisation of the charge on the individual helices by site-bound

counterions would be expected to promote helix-helix aggregation, and indeed formation
of aggregated assemblies during Ca2+-induced gelation of K-carrageenan was strongly
indicated by the other investigative techniques used. As shown in Figure 3, conformational
ordering on cooling is accompanied by a well-defined exotherm in DSC. Over the same
temperature range, there is a large, sigmoidal, increase in turbidity, consistent with formation
of large aggregates capable of causing substantial scattering of light.
Figure 4a shows the effect of K+ and Ca2+ concentration on transition-midpoint
temperature (T,) from DSC. As found in previous studies by other techniques: K+ ions
give higher transition temperatures than equivalent molar concentrations of Cazf, and
the T, values recorded on heating in the presence of Kf are substantially higher than the
corresponding values for conformational ordering of the same samples on cooling. Similar
thermal hysteresis, however, is also evident for the samples incorporating Ca2+ counterions,
indicating they too are capable of stabilising helix structure by promoting aggregation.
As shown in Figure 4b, the reciprocal of T, for the disorder-order transition on cooling
varies linearly with the reciprocal of IS+ or Ca2+ concentration, consistent with previous
investigationsg.The corresponding plots for the T , values observed on heating also show
reasonable linearity. For both Kf and Ca2+, the heating and cooling plots are roughly
parallel, but the separation between them (i.e. the degree of thermal hysteresis) is greater
for Ca2+ than for K+, suggesting that the divalent cations are more effective in promoting
aggregation of K-carrageenan helices.
4 CONCLUSIONS
As shown in Figure 2, the maximum gel strengths attained with K+ and Ca2+ are roughly
the same, but occur at grossly different cation concentrations, which may explain,
at least partially, the conflicting conclusions reached in previous s t ~ d i e s . ' ' ~At
~ low
cation concentrations, Ca2+ gives stronger gels than K,' but at high concentrations the
converse is true.
Since Ca2+ cations give maximum gel stren@ at around stoichiometric equivalence,
the most likely interpretation is that they act by direct bridging between adjacent helices,
in a mechanism somewhat analogous to the 'egg-box' model postulatedU for Caz+-induced
gelation of alginate or low-methoxy pectin. Kf ions appear to act in a less direct
way, by binding to individual helices and thus suppressing the electrostatic barrier
to aggregation, and the resulting gel strength increases progressively (Figure 2) with
increasing concentrationof. 'K
Another marked difference is that K+ cations cause a large elevation in sol-gel transition
temperature, whereas the increases induced by Ca2+ and other divalent cationsg are no
greater than would be expected from their charge." A possible interpretation is that K+ ions
can stabilise short stretches of helix structure by binding to them as they form, thus 'driving'
conformational ordering by suppressing the competing back-reaction (unwinding), but that
binding of Ca2+ ions between helices is a co-operative process, which can occur only once
the individual helices have grown to a stable length, and therefore has no discernable effect
on the temperature course of the disorder-order transition.
An initial increase and subsequent decrease in gel strength with increasing concentration
of Caz+, and other divalent cations, has been reported previously by Watase and Nishinari.''
In that investigation, however, the carrageenan used was in the Kf salt form,'' with
interpretation therefore being complicated by the possibility of specific binding of Kf ions
to the carrageenan helices, in addition to any specific interactions with the divalent cations.
The closely similar results obtained in the present study, where the counterions to the
polymer were 'non-specific' Na' ions, indicates that any such binding of Ks was swamped,
or eliminated, by stronger binding of divalent cations.
Since the work reported here was completed, and presented at the international
conference "Gums and Stabilisers for the Food Industry l l " , we have carried out similar
studies at higher polymer concentrations (1.5, 2.0 and 3.0 wt %), and in each case observed
maximum gel strength at stoichiometric equivalence of carrageenan and Ca2'. Further work
is in progress, and will be reported elsewhere.
References
1 T.J.Painter, in The Polysaccharides, ed. G.O. Aspinall, Academic Press, New York,
1983, vol. 2, p. 195.
2 M. Glicksman, in Food Hydrocolloids, ed. M. Glicksman,CRC Press, Boca Raton, FL,
1983, vol. 2, p. 83.
3 N.S. Anderson, T.C.S. Dolan and D.A. Rees, J. Chem. SOC.,Perkin Trans. I, 1973,2 173.
4 S. Arnott, W.E. Scott, D.A.Rees and C.G.A. McNab, J. Mol. Biol., 1974,90,253.
5 RP. Millane, R. Chandrasekaran, S. Arnott and I.C.M. Dea, Carbohydr, Res., 1988,
182, 1.
6 D.A. Rees, E.R. Morris, D. Thorn and J.K. Madden, in The Polysaccharides, ed. G.O.
Aspinall, Academic Press, New York, 1982, vol. 1, pp. 195
7 0. Smidsrad, I.-L. Andresen, H. Grasdalen, B. Larsen and T. Painter, Carbohydr, Res.,
1980,80, C l l .
8 L. Piculell, in Food Polysaccharides and their Applications, ed. A.M. Stephen, Marcel
Dekker, New York, 1995, p. 205.
9 C. Rochas and M. Rinaudo, Biopolymers, 1980,19,1675.
10 L. Picullel, C. H b s o n and S. Nilsson, Int. J. Biol. Macromol.,1987,9,297
1 1 H. Grasdalen and 0. Smidsrad, 198 1,14,229.
12 P.S. Belton, V.J. Moms and S.F. Tanner, Macromolecules, 1986,19, 16 18.
13 L. Picullel, S. Nilsson and P. Strom, Carbohydr.Res., 1989,188, 121.
14 E.R. Morris, D.A. Rees and G. Robinson, J.Mol. Biol., 1980,133,349.
15 M. Watase and K. Nishinari, Colloid Polym. Sci., 1982,260,971.
16 A.-M. Hemansson, Carbohydr. Polym., 1989,10,163.
17 E.R. Morris and I.T. Norton, in Aggregation Processes in Solution, e d E. Wyn-Jones
and J. Gormally, Elsevier, Amsterdam, 1983, p. 549.
18 S. Nilsson, L. Picullel and B. Jonsson, Macromolecules, 1989,22,2367.
19 V.J. M o m s and G.R. Chilvers, Carbohydr. Polym., 1983,3,129.
20 A,-M. Hermansson, E. Eriksson and E. Jordansson, Carbohydr. Polym., 1989,10,163.
21 P. Moldenaers, J. Mewis and H. Berghams, in Progress and Trends in Rheology II,
ed. H. Giesekus and F.F. Hibberd, Steinkopff Verlag, Darmstadt, 1988,420.
22 RK. Richardson and F.M. Goycoolea, Carbuhydr.Polym., 1994,24,223.
23 G.T.Grant, E.R Morris, D.A. Rees,P.J.C. Smith and D. Thorn, FEBS Letts., 1973,32,
195.
24 M . Watase and K. Nishinari, in Gums and Stabilisersfor the Food Industry 3,
ed. G.O. Phillips, D.J. Wedlock and P.A. Williams, Elsevier, London, 1986, p. 185.
RHEOLOGICAL,PROPERTIES OF K-CARRAGEENANWEAK GELS
S. Ikeda,' K. Nishinari,' and V. J. Morris2
Department of Food and Nutrition, Osaka City University, 3-3-138 Sugimoto,

Sumiyoshi-ku, Osaka 558-8585, JAPAN
Institute of Food Research, Norwich Research Park, Colney, Norwich NR4 7UA, UK
1 INTRODUCTION
The term weak gel is used for differentiating a gel-like polymer dispersion from a true gel.
A weak gel exhibits gel-like dynamic viscoelasticity against oscillating linearly small
strains but can steadily flow without breaking under a steady stress that exceeds the yield
stress value of the system.' An aqueous solution of xanthan is perhaps the best-known
weak gel example that is actually utilized in the food industry.2 Since the existence of the
yield stress can lead to the retardation of quality-deterioration processes such as a
coalescence of emulsion droplets, superb stabilities can be achieved by incorporating
xanthan into liquid food products.2 Additionally, a weak gel exhibits strong
shear-thinning behavior, which even facilitates handling of a weak gel material at a
relatively high shear rate. The origin of such favorable weak gel rheological properties
has been believed to be due to polymer chain entanglements and even association that do
not relax or break as long as the magnitudes of applied stresses or deformations are
sufficiently small.1-3 However, recent studies using Atomic Force Microscopy (AFM)
have revealed that xanthan molecules in commercially available xanthan powders are
partially denatured during their production, leading to micron-sized insoluble aggregates,
called microgels, which are recognized in an aqueous dispersion of xanthan.4 Therefore,
a xanthan weak gel may be considered as a dispersion of xanthan microgels that sterically
interact each other. Aqueous dispersions of insoluble particles of other polysaccharides,
such as curdlan5 or cellulose,6 and indeed a dispersion of micron-sized polysaccharide gel
particles7 are known to exhibit weak gel-type rheological properties as well.
Kappa-carrageenan is another important category of polysaccharide in the food
industry due to its thermoreversible-gel forming ability.8 On cooling an aqueous
dispersion of x-carrageenan, the random coil forms of =carrageenan molecules transfer
into double-helical conformer^.^ Those helices are believed to fibrously aggregate fbrther
in the presence of specific gel-promoting cations such as K', eventually forming a network
occupying the entire space at a sufficiently high concentration.10 Specific anions, on the
other hand, are known to bind to marrageenan helices, prevent aggregation of the helices,
and thus prevent successive gelation.l 1 Recently, Piculell and colleagues have shown that
a dispersion of K-carrageenan helices in the presence of an aggregation-impeding anion I-
can exhibit weak gel-type rheological properties at a sufficiently high concentration of the
polymer.12-14 This is an important finding since it suggests that aggregation of helices is
not necessarily required for a K-carrageenan helical system to exhibit an elastic nature.
The goal of the present study was to elucidate detailed rheological properties of
mxrrageenan weak gels, formed under non-aggregating conditions in the presence of
iodide ions. We have analyzed mechanical responses of K-carrageenan weak gels against
linearly small and destructively large deformations, and compared them with those of
conventional true K-carrageenan gels, formed in the presence of a gelation-promoting
potassium salt. Additionally, some preliminary AFM observations were conducted in
order to compare K-carrageenan helical networks formed under aggregating and
non-aggregating conditions.
2.1 Materials
Kappa-carrageenan (C-1263) was purchased from Sigma Chemicals (St Louis, MO, USA)
and used without fkther purification. Other chemicals were of reagent grade quality.
2.2 Rheological Measurements
Sample solutions were prepared by dissolving the K-carrageenan into distilled water,
containing either 0.2 moVdm3 KCl or 0.2 moYdm3 NaI, under stirring for 30 min at
90-95C. The ionic concentration of the added salts in the test samples was intended to
be high enough to mask effects of pre-existing ions in the K-carrageenan as impurities
(several percent by weight at most). It has also been shown that the modulus of
K-carrageenan aqueous dispersions containing NaI increases with increasing concentration
of NaI up to a maximum at 0.2 moVdm3.'3
Dynamic viscoelastic and steady flow properties of K-carrageenan dispersions were
measured using a stress-controlled rheometer (RheoStress 1, Haake, Germany), equipped
with a double-gap cylinder test fixture (DG41; inner gap, 0.25 mm; outer gap, 0.3 mm;
height, 55 mm; Haake, Germany). A sample solution was loaded into the test cup of the
instrument at 80 or 90C, immediately covered with silicone oil, and cooled down to 20C
at a rate of 1"Clmin. Frequency sweep measurements of the storage modulus G' and the
loss modulus G" were performed at 20C at a strain of 0.01. In order to determine the
linear viscoelastic strain region, oscillating controlled-shear stresses z, ranging iiom 0.1 to
100 Pa (1 rads), were applied at 20C and instrumental output values of G' (G'effechve) and
those of G" (G"effective) were monitored (dynamic stress sweep measurements). Steady
stress sweep measurements were made at 20C in order to determine the steady shear
viscosity at steady shear stresses from 1 to 100 Pa.
2.3 AFM Observation
The K-carrageenan was dissolved into either 0.1 molldm3 KCl or 0.1 moVdm3 NaI aqueous
solutions (ca. 100 pg/mL) under gentle stirring at 85C for an hour. After being cooled to
the room temperature (ca. 20"C), the samples were diluted into distilled water to 10 pglmL.
Drops (2 pL) of the samples were deposited onto fieshly cleaved mica, air-dried for 10 min,
and then imaged under butanol. Images were taken in direct current contact mode using
an ECS AFM located at the Institute of Food Research (Nonvich, UK). Details of the
instrument and the imaging procedure are described elsewhere.15-17
Figure 1 represents the fiequency dependence of the moduli of a true K-carrageenan gel,
formed in the presence of KCl, and that of a weak gel, formed in the presence of NaI.
Both gels exhibited typical gel-type mechanical spectra; i.e. G'> G" in the entire fiequency
range examined, while the moduli of the weak gel were slightly frequency dependent.
The value of tan6 (= G"/G') was less than 0.1 for the true gel, suggesting that the gel was
predominantly elastic, while that of the weak gel was ca. 0.4 at all fiequencies examined.
In spite of only one-tenth polymer concentration, the true gel exhibited even larger values
of the storage modulus than the weak gel.
1000 t
1
1 10 100
F req ue ncy [radls]
Figure 1 Frequency dependence of G' (yolid) and G" (open) of 0.15% w/w
K-carrageenan in 0.2 rnol/dm3 KCl (true gel, square) and 1.5% w/w
K-carrageenan in 0.2 mol/dm3 NaI (weak gel, circle).
100 .
0.01 0.1 1 10 I00 1000

10' (I) [radls]
Figure 2 Frequency dependence of G'(solid) and G" (open) of 0.5% w/w (circle), 0.7%
w/w (triangle), and 1% w/w (square) K-carrageenan in 0.2 rnol/drn-lNal. Data
were shifted horizontally to avoid overlapping.
Chronakis et al. observed gel-like behavior of K-carrageenan aqueous dispersions

containing NaI only at relatively high concentrations of K-carrageenan (i.e. above 0.9%
w/w).13 The concentration dependence of the mechanical spectra of it-carrageenan in a
non-aggregating condition is shown in Figure 2. The 1.O% w/w K-carrageenan dispersion
exhibited a gel-like spectrum, similar to that of the 1.5% w/w K-carrageenan dispersion
(Figure l), while the 0.5% w/w dispersion appeared as a semi-dilute polymer solution; that
is, predominant G over G at low frequencies and a crossover of G and G in the highest
fiequency range examined. By increasing the concentration up to 0.7% w/w, the
crossover fiequency, which corresponds to the average relaxation time of entanglements or
the lifetime of association, was lowered close to the lowest end of the accessible fiequency.
The linear viscoelastic strain region of a weak gel is considered to be narrower than a
true strong gel. However, this was not the case with K-carrageenan weak gels examined.
Since the extent of the linear strain region depends on the concentration of the polymer,1
gels with similar values of the moduli, rather than those with similar polymer
concentrations, are compared in Figure 3. The linear viscoelastic strain region of a true
gel was observed up to strain of ca. 0.1 and the gel fractured at strain of ca. 0.2. The
linear viscoelastic strain region of a weak gel also extended up to ca. 0.1 but the gel never
fractured, even when the sample was stirred by rotating the measuring device very rapidly
under the dynamic shear stress of 100 Pa. The values of the moduli (Geffective and
Geffeaive in the linear strain region) determined at the second-run measurements on the
weak gel were slightly less than those at the initial (first-run) measurements (Figure 3b).
Thus, it may be suspected that (1) the fiiction at the interface between the sample and the
measuring device was somehow reduced by intensive stirring, (2) entangled helices were
oriented in the direction of the flow, or (3) some structural breakdown took place.
However, no m h e r change in the spectrum was noticeable between the second and the
third-run measurements (Figure 3b), confirming that gel-like mechanical characters, i.e.
G(w) > G( w), were not lost on the application of extremely large deformations.
1000
CI
(P
P
Y
:100
.g
CI
u
2
- 3 10
b
3
b
i1
0.1
0.001 0.01 0.1 1 0.001 0.01 0.1 1 10 100 1000
Strain Strain
Figure 3 (a) Strain dependence of GLficrhe (solid square), GeDcthe (open square), and z
(triangle)for 0.15% w/w K-carrageenan in 0.2 moVdm3 KCI.
(6) Strain dependence of Gifictbe (solid circle) Gefic1ive (open circle), and z
(diamond)for 1.5% w/w =carrageenan in 0.2 moVdm3 NaI. Results of
the second- and third-run measurements of the moduli are represented by
upper and lower triangles, respectively
Steady and dynamic rheological properties of a K-carrageenan weak gel are compared
in Figure 4. Under the steady shear flow, a Newtonian flow region was observed at the
low limiting shear rate and a strong shear thinning at a higher shear rate. Values of the
dynamic viscosity rf were larger than those of the steady shear viscosity q at the
comparable angular fiequency and the shear rate. By repeating the measurements,
reproducible results were obtained, suggesting that yielded structures in the weak gel under
larger deformation healed at rest. Similar rheological features have been found for an
aqueous dispersion of xanthan, a typical weak gel.'-'J*J9
Gel-like mechanical spectra of a =carrageenan weak gel can also survive severe shear
treatments. In Figure 5, the first-run measurement was made after shearing a weak gel at
the rate of ca. 1000 s-1 for at least 30 s, and likewise the second- and third-run
measurements followed after respective shear treatments. Applications of extremely large
deformations did not influence the gel-like dynamic spectra at all. It is now safely
concluded that the dispersion of rigid helical conformers of K-carrageenan exhibited
gel-like rheological behavior, under dynamic small strains in an ordinary available
frequency range, while it was able to flow without fracture, under extremely large strains.
Figure 4 Shear strain rate dependence of the steady shear viscosity (circle) and the
steady shear stress (square) andfrequency dependence of the dynamic viscosity
(triangle) of 1.5% w/w K-carrageenan in 0.2 moVdm3 Nal. Solid and open
symbols represent thefirst- and the second-run measurements, respectively
1 10 100
Frequency [radls]
Figure 5 Frequency dependence of G' (solid) and G" (open) of 1.5% w/w K-carrageenun
in 0.2 mol/dm3 Nal. Circles, triangles, and squares represent thefirst-,
second-, and third-run measurements, respectively. Each measurement WUS
done after applying extremely large strain (at the shear rate of ca. 1000 s l f i ) r
at least 30 s) to the sample.
A fluid gel is another material that is known to show both gel-like linear viscoelasticity
and liquid-like flow proper tie^.^ A fluid gel is a dispersion of closely packed small gel
particles (-p) that can be formed by shearing a polysaccharide gel.7 Therefore, one
might expect that a fluid gel emerged by shearing K-carrageenan dispersions in this study.
It is known, however, that small gel particles in a fluid gel coalesce if the extent of
aggregation of polysaccharide helices is insufficient and that the entire system heals back
to a single large gel.7 Since investigated K-carrageenan dispersions with NaI are
considered to be essentially aggregation-fiee, their weak gel behavior is unlikely to be due
to the formation of a fluid gel.
The examined K-carrageenan formed gels even at 0.05% w/w in 0.1 mol/dm3 KC1 at
room temperature. At a slightly lower concentration (0.01% w/w), no macroscopic gel
formation was recognized at room temperature. This solution was further diluted to
0.001% w/w and drops were taken from this diluted solution, deposited onto freshly
cleaved mica, air-dried, and then imaged under butanol. The AFM image (Figure 6a)
clearly reveals a local network (microgel) of K-carrageenan aggregates formed in the
aqueous system whose polymer concentration was below the critical concentration of
macroscopic gelation. No such aggregate was observed when a hot solution of 0.01%
w/w ti-carrageenan was quenched, namely directly diluted into water at room temperature
before being cooled down (image not shown). AFM images of K-carrageenan that can be
found in literatures always show preferentially oriented networks (like the one that will be
shown in Figure 6b), which are considered to arise as a consequence of distortion during
deposition and drying on mica.l5 In this study, perhaps because an excess amount of
gel-promoting salt (KCl) was added, the local networks appeared rigid enough to avoid
being orientated.
In the presence of NaI, aggregation of K-carrageenan helices was presumably
prevented and thus familiar orientation was observed (Figure 6b). However, many
branches are recognized in the image, suggesting that K-carrageenan can form a network
even without aggregation of helices, even though such a network may not be so rigid as the
one formed in the presence of aggregating cations. A textbook explanation of the branch
(or junction zone) formation by aggregating helices should not be considered here.
Rather, branches must be formed by multiple polymer chains that are involved in the
formation of a double helix.l6 Since it is highly unlikely that only molecules with
perfectly matching lengths associate into a double helix, branches can occur. It may be
Figure 6 K-Carrageenan networkformed in 0.1 mol/dm3 KCl (a) or in 0.1 mol/dm3 NaI
(b) at 20 '17. Image size is 2 p n x 2 p.
suspected that a small amount of potassium ions preexisting in the commercial

K-carrageenan sample as impurities may be also responsible for the branch formation,
while our preliminary studies on purified Na-form K-carrageenan revealed the formation of
weak (oriented) networks even in the absence of potassium ions (data not shown).
Consequently, a K-carrageenan weak gel can be considered as a weak network
structure that breaks under large deformation but heals at rest. If the network formation
occurs only locally, the entire system can be regarded as a dispersion of branched polymers
that sterically interacting each other.
Acknowledgement
S.I. and K.N. are gratefbl to Mr. Toshiaki Yazaki (EKO, Japan) and Mr. Wolfgang H.
Marquardt (Haake, Germany) for lending them the Haake rheometer. S.I. is indebted to
Nestle Science Promotion Committee, Japan for finding and also to Osaka City University
for supporting his stay at the IFR, Nonvich, UK.
References
1 S. B. Ross-Murphy, J Rheol., 1995,39, 1451.

2 V. J. Morris, in Food Polysaccharides and Their Applications, ed. A. M. Stephen,
Marcel Dekker, New York, 1995, ch. 1 1, p. 34 1 .
3 E. R. Morris, in Food Polysaccharides and Their Applications, ed. A. M. Stephen,
Marcel Dekker, New York, 1995, ch. 16, p. 5 17.
4 V. J. Morris, A. P. Gunning, A. R. Kirby, A. R. Mackie and P. J. Wilde, in
Hydrocolloids-Part I , ed. K. Nishinari, Elsevier Science B. V., Amsterdam, 2000, p.
99.
5 M. Hirashima, T. Takaya and K. Nishinari, Thermochim.Acta, 1997,306, 109.
6 D. Tatsumi, S. Ishioka and T. Matsumoto, Nihon Reoroji Gakkuishi, 1999,27, 243.
7 I. Norton, T. Foster and R. Brown, Gums and Stabilisersfor the Food Industry, 1998,9,
259.
8 L. Piculell, in Food Polysaccharides and Their Applications, ed. A. M. Stephen,
Marcel Dekker, New York, 1995, ch. 8, p. 205.
9 C. Viebke, J. Borgstrom and L. Piculell, Carbohydr Polym., 1995,27, 145.
10 C. Viebke, L. Piculell and S. Nilsson, Macromolecules, 1994, 27,4160.
1 1 H. Grasdalen and 0. Smidsrod, Macromolecules, 1981 , 14, 1842.
12 J. Borgstrom, L. Piculell, C. Viebke and Y. Talmon, Int. J Biol. Macromolecules, 1996,
18,223.
13 I. S. Chronakis, L. Piculell and J. Borgstrom, Carbohydz Polym., 1996,31,215.
14 L. Piculell, J. Borgstr&n, I. S. Chronakis, P.-0. Quist and C. Viebke, Int. J Biol.
Macromolecules, 1997,21, 141.
15 A. R. Kirby, A. P. Gunning and V. J. Morris, Biopolymers, 1996,38,355.
16 A. P. Gunning, A. R. Kirby, M. J. Ridout, G J. Brownsey and V. J. Morris,
Macromolecules, 1996,29,6791; 1997,30, 163.
17 V. J. Morris, A. R. Kirby and A. P. Gunning, Atomic Force Microscopyfor Biologists,
Imperial College Press, London, 1999.
18 S. B. Ross-Murphy, V. J. Morris and E. R. Morris, Faraday Symp. Chem. Soc., 1983,
18, 115.
19 M. Milas, M. Rinaudo, M. Knipper and J. L. Schuppiser, Macromolecules, 1990,23,
2506.
MIXED SYSTEMS OF FISH GELATIN AND K-CARRAGEENAN
I. J. Haug, H. Devle, K. I. Draget and 0. Smidsrod
Department of Biotechnology/ Norwegian Biopolymer Laboratory (NOBIPOL),

Norwegian University of Science and Technology (NTNU), Trondheim, Norway.
1 INTRODUCTION
The importance of gelatin is reflected in the widespread use in food, pharmaceuticals and
photographic industries. Bovine and pork have traditionally been the sources for gelatin
production, but the outbreak of Bovine Spongiform Encephalopathy (BSE) in the 1980s
accelerated the search for a replacement of mammalian gelatin.
Fish gelatin could be an alternative to, but is not directly exchangeable with mammalian
gelatin due to low gel strength and low gelling and melting temperature. A possible
approach to overcome this problem could be to mix fish gelatin and marine
polysaccharides which may give gelling systems with higher gel strength, gelling and
melting temperature.
2.1 The Carrageenan Sample
The K-carrageenan (K-CG) sample, kindly provided by FMC Biopolymer A / S , Drammen,

Norway, was a proximately 91 % (w/w) pure with 9 9% (w/w) in the iota-form, as
P
determined by H-NMR analysis'. The sample contained 90 % (w/w) CG, 8 % moisture
and 2 % salts. This sample contained 6 % (w/w) potassium ions, 2 % (w/w) calcium ions
and 1 % (w/w) sodium ions and the weight average molecular weight was about 600 kDa
(FMC Biopolymer). From the given data it can easily be found that there is no significant
surplus of inorganic salts in the sample.
2.2 The Fish Gelatin Sample
The fish gelatin (FG) sample was produced from skins of cold water fish species and was
kindly provided by Norland Inc., USA (lot 9187). The weight average molecular weight,
Mw,as obtained from SEC-MALLS analysis was 170+17 kDa (averagefS.D., n=9). The
second-viral coefficient, A2, used in the light-scattering was 1. lo4 and the refractive index,
dn/dc, was 0.190. A solution of 0.05 M Na2S04/ 0.01 M EDTA at pH 9.6 was used as
solvent. The isoelectric point, IEP, was found to be 8.4.
2.3 Solution Preparation
The CG solutions were made by heating for 30 minutes at 90C in a water bath with
stirring. The concentrations were varied to obtain a given concentration in the final gels (%
(w/v)). After dissolving, the solutions were incubated at 60C for 30 minutes.
The FG was mixed with distilled water and dissolved at room temperature with stirring.
The FG solution was then pre-heated to 60C for - 30 minutes and mixed with CG solution
until the desired concentration of both polymers was reached.
The ionic strength of the mixtures was adjusted with 1.0 M NaCl or 1.0 M KCl in the
CG solutions before mixing with FG. The ionic strength is reported as extra addition of
salt, and does not include the intrinsic ionic content. The pH in the mixed solutions was
7.0.
2.4 Gel Preparation
The solutions of CG or CGIFG were filled into wells of macro well plates (Costar) with
care to avoid air bubbles. Each well was 16 mm in diameter and 18 mm in height and with
24 wells/plate. The pure CG gels and the mixed gels of CG and FG were cooled to room
temperature (- 22C) and stored over-night. Some of the mixed gels of CG and FG were
cooled to room temperature and stored over-night at 4C.
All pre-treatments of the gels stored at 4C were done in the refrigerator and the gels
were compressed immediately after they were taken from the fridge. The compression
measurements were performed at ambient temperature.
2.5 Compression Measurements
The compression measurements were performed on a Texture Analyzer, TA.XT2i (Stable

Micro Systems, Surrey, UK). The gels were positioned in the centre below the probe (P25
Aluminium- 25 mm in diameter), and the compression was performed at 0.1 m m / s until
breakage of the gel. The mechanical properties of the gels were evaluated by measuring the
initial linear slope of the force-deformation curve of the gels (N/m) giving Young's
modulus, E2, and the breaking strength (g). The breaking strength of the gels is in this
connection defined as the distance at which the applied force makes the gel break, giving
the maximum force and deformation. Deformation above the breaking point gives an
irregular shape of the stress-strain curve.
3 RESULTS AND DISCUSSIONS
3.1 The Carrageenan System
Compression measurements of the gels with increasing concentration of K-CG were

performed at room temperature. A minimum concentration of 1% (w/v) was required to
give measurable gels. This concentration results in a solution with 13.8 mM K+-ions, 3.9
mM Na+-ions and 4.5 mM Ca2+- ions. Figure 1 shows a double-logarithmic plot of the
Young's Modulus (E) as a function of K-CGconcentration, and the gel strength increases
approximately linearly.
At concentrations larger than the critical concentration of gelation the elastic moduli of
K-CGgels, in excess salt, follow an exponential law of the form E = kc" . There has been
reported values of the exponent close to 2374,while others found it to be 2.45.The cations
in the CG sample were only the counterions and their concentration therefore depended on
the CG concentration. No extra salt was added, and the system had no significant excess of
salt. This may probably be the reason why the slope in figure 1 is different to the expected
value of 2-2.4 . The breaking strength of the carrageenan gels increased linearly (linear
plot) from 500 g to almost 2000 g between 1 and 2 %, while the strain at breaking was
almost constant and approximately 45 % (data not shown).
ion 4
Y
m
1 ' I I I I1 1
2 3 1 5
1
Concentration K-CG ( % ( w l v ) )
Figure 1 Compression measurements for gels of increasing concentration of K-

carrageenan (average -tS.D., n > 3).
It has previously been found that the modulus is independent of the molecular weight of
the K-CG,as long as the molecular weight exceeds about 200 As a standard for
further experiments, a concentration of 1 % (w/v) K-CGwas chosen, later referred to as 1%
K-CG,which has a gel strength of approximately 8 kPa in its pure form.
It is well known that the gelation of K-CGcan be induced by both nonspecific (eg. Na',
Li') and specific (e.g. K', Cs') monovalent cations. The specific cations generally give the
strongest gels under comparable ionic ~ t r e n g t h ~Figure
- ~ . 2a) shows the moduli for K-CG
gels with and without KCI, at room temperature and at 4C. At room temperature the
modulus of the K-CGgels was more than twice as high when 20 mM KCl was added. The
gels of 1 % K-CGwith 20 mM KCl at 4C had a modulus which was approximately 20 %
higher than for the gels at room temperature (bar B vs. bar D in figure 2a)). There was no
significant increase in the modulus for the gels at 4C without KCl (bar A vs. bar C).
Many biopolymer gels do not follow the theory of rubber elasticity, since the main
iissumpt ions of random-coil behaviour of the elastic chains and point-like cross-linking of
the chains are not valid. Biopolymer gels have areas of junction zones, which bind the
polymer chains together making a polymer network. When the temperature is decreased
the flexibility of the polymer chains in the network is also decreased due to less freedom of
rotation and movement. This gives a stiff chain network where the elastic modulus is
increasing with decreasing temperature.
40 -
1
35 -
30 -
-
3
3 25
5
v 20-
w 15 -
10 - 10
T
A
5 - 5
0 1
D F 1
Figure 2 Young's moduli for a ) K-carrageenan gels: A: 1 % (w/v) K-CG at room

temperature, B: 1 % (w/v) K-CGwith 20 mM KCl at room temperature, C: I % (w/v) K-CG
at 4C' D: 1% (w/v) K-CG with 20 mM KCl at 4C b) mixed systems of K-CG and FG: E:
I % (w/v) K-CGand 2% (w/v) FG with 20 mM KCl at room temperature and
F: 1% (w/v) K-CGand 2% (w/v) FG with 20 mM KCl at 4C (averageS.D.,n > 4).
3.2 The CarrageenadFish Gelatin System
FG does not gel at room temperature, but has to be cooled to typically below 8C before
gelation can occur". It was therefore decided to quantify the gel strength of mixtures of K-
CG and FG at both room temperature and at 4C. At room temperature only the CG is in
the ordered conformation, while at 4C both CG and FG are in the ordered state. The pH of
the solutions was approximately 7.0, which gives oppositely charged biopolymers, since
pH was lower than PI for the fish gelatin.
We may now compare the results for K-CG with the data given in figure 2b) where the
gels consist of 1% K-CG and 2% FG with 20 mM KCl. At room temperature the Young's
modulus, E, was approximately 25 kPa, while at 4C the modulus was approximately 37
kPa. It is obvious that the increase in the gel strength at 4C is not a result of the
contribution from the K-CGalone, since the 1% K-CG with 20 mM KCl gels at 4C have a
gel strength of approximately 26 kPa. Gels made from 2 % FG were almost not able to
carry their own weight, and the gel strength was expected to be very low (but not
measurable because the gels melt immediately at room temperature). The increase in gel
strength at 4C is therefore expected to be the result of a synergetic effect between FG and
K-CG.
Electrostatic interactions between oppositely charged proteins and polysaccharides are,
in most cases, the primary interactions in associative mixed biopolymer systems' I . Mixing
of K-CG and FG results in turbid solutions and gels, and the turbidity was probably a result
of the formation of complexes due to associative phase separation. The turbidity was
measured as the difference in the optical density at 405 nm in 100 p1 gels between pure 1 %
K-CG gels and the mixed gels. The solutions were filled into micro well plates and were
allowed to form a gel before the optical density was measured. The
turbidity in the mixtures was found to be dependent on the concentration of the
biopolymers, the ionic strength, the nature of the added salt and the pH. When KCI was
added the turbidity decreased, while the moduli increased. The turbidity was near
maximum for the mixture of 1 % K-CG and 2% FG with 20 mM KCl, meaning that the
phase separation was minimum in this mixture. When the turbidity was high, the Young's
modulus was low.
At 4C both polymers are in ordered conformation. In segregative phase separation it is
expected that the biopolymer gelling first mainly determines the structures of the final gel,
since it develops its network before the gelation of the second polymer". This will
probably also be the case for the mixed gels of K-CG and FG, even though this system
probably undergoes associative phase separation. At room temperature the CG gels,
forming helices and a network. The FG will be in random conformation and is probably
distributed in the CG-network as a random coil. When the temperature is lowered the FG
will retrieve its triple helix conformation over a few amino acids in isolated areas of the
coils and the FG will contribute to the gelling network, giving an increase in the Young's
moduli. The stiff chain network, as discussed previously, will also contribute to give an
increase in the elastic modulus at decreasing temperatures.
Small-strain oscillatory measurements of the mixed gels exposed that the gelling and
melting temperature of the mixtures followed the gelling and melting for K-CG. This may
be because the gelling network in the mixtures was dominated by the CG, since this
biopolymer, as previously mentioned, gelled at the highest temperature and determined the
structure of the final gel. The small-strain oscillatory measurements also revealed that the
FG contributed to the gel strength. The elastic modulus, G', increased considerably at 4OC,
but no changes were detected in the phase angle.
4 CONCLUSION
The gels made from 1% K-CG and 2% FG with 20 mM KCl were strongest at 4C. The
increase in modulus is probably due to a synergistic effect between CG and FG when both
molecules are in the ordered conformation. The system undergoes associative phase
separation, giving opaque gels. The degree of phase separation was decreased when KCl
was added, giving gels with higher moduli. The gelling and melting temperature of the
mixtures were the same as for pure K-CG gels under the same conditions.
References
1 D. Welti, J. Chem. Research ( S ) , 1977,312.

2 0. Smidsrgd, A. Haug and B. Lian, Acta Chemica Scandinavia, 1972,26,71.
3 0. Smidsrgd and H.Grasdalen., Hydrobiol., 1984,116/117, 19.
4 C. Rochas and S . Landry, in Gums and stabilisers for the food industry 4 , ed. G.O.
Phillips, P.A. Williams and D.J. Wedlock, IRL Press, Oxford, 1988, p. 445.
5 E.E. Braudo, I.G. Plashchina and V.B.Tolstoguzov, Carbohydr. Polym., 1984,4.
6 C. Rochas, M. Rinaudo and S. Landry, Carbohydr. Polym., 1990,12,255.
7 A-M. Hermansson, E. Eriksson and E. Jordansson., Carbohydr. Polym., 1991,16,297.
8 E.R. Morris, D.A. Rees and G. Robinson, J. Mol. Biol., 1980,138, 349.
9 M. Watase and K. Nishinari, Colloid Polym. Sci., 1982,260,971.
10 R.E. Norland, in Advances in Fisheries and Biotechnology for Increased Profability, ed.
Voigt and Botta, Technomic Publishing Company Inc., Lancaster, 1990, p. 325.
11 J.-L. Doublier, C. Gamier, D. Renard and C. Sanchez, Current Opinion in colloid &
interface science, 2000,5,202.
EFFECT OF GELATINKARRAGEENAN INTERACTIONS ON THE STRUCTURE
AND THE RHEOLOGICAL PROPERTIES OF THE MIXED SYSTEM
C. Michon, K. KonatC, G. Cuvelier and B. Launay
Laboratoire de Biophysique des Matkriaux Alimentaires, ENSIA, 1, avenue des

Olympiades, F - 91744 MASSY Cedex
1 INTRODUCTION
Mixed systems of biopolymers are often used in the complex formula of food gels. In such
mixture two concommitant phenomena occur and contribute to the structure of the system.
The first one is phase separation, with the classical distinction between associative and
segregative phase separation. In an associative phase separation one of the separating
phases is enriched in both polymers, whilst, in a segregative phase separation, each phase
is enriched in one of the polymers.' The underlying reasons for the phase behaviour of
mixed biopolymer systems are often treated as being due to short-range interactions and
polyelectrolyte effects, which can be rationalized within the framework of the Flory-
Huggins model. The second phenomenon is gelation of polymers, depending on the type
of interaction involved between them. However, to understand the mechanisms of
interaction and the effect of ordering on those interactions is of great interest for predicting
the structure and the properties of the mixed system.
In this paper, we present results obtained on the gelatin / iota-carrageenan mixed
system. Iota-carrageenan is a sulphated polysaccharide. Gelatin is an animal protein
mostly extracted by an alkaline or by an acid process. Two types of gelatin are obtained
and mainly differ by their PI : about 9 and 4.5 respectively. Both carrageenan and gelatin
for carrageenan ''

are able to undergo a coil to helix transconformation when the temperature decreases. The
classical models roposed for the ordered conformation is double helical junction zones
and triple helical junction zones for gelatin! Interactions in coil
conformation have first been studied through phase diagrams for both gelatin samples
mixed with iota-carrageenan. Interactions at the molecular level are studied using a
spectrophotometric method and the effect of conformation on carrageenan interactions is
discussed in terms of local interactions and, at a supramolecular level, in terms of
rheological properties and of organisation of these systems.
2 MATERIAL AND METHODS
2.1 Sample Preparation

The biopolymers were all industrial grade samples provided by SKW Bio-Systems,
France. The gelatin samples were a first-extract alkaline process sample (PI about 4.5)
obtained from limed hide (called LH) and a first-extract acid process sample (PI about 9)
obtained from pig skin (called PS). Gelatin was dispersed in cold de-ionized water and
then dissolved at 65C for 30 min under stirring. The iota-carrageenan sample, called CI,
extracted from the red seaweed Euchema Denticufatum, was in the sodium form and
showed a very low amount of kappa and lambda << impurities B. The iota-carrageenan was
dispersed in cold water and then dissolved at 70C and then at 90C for 30min under
stirring. Sodium azide (0.04wt%) was systematically added as an anti-microbial agent.
Solutions of gelatin and iota-carrageenan in de-ionized water were mixed to obtain
compositions covering the whole practically accessible range. Hot solutions were mixed
and stirred at 65C for 15 min. pH was adjusted to a value of 7 at 60C by addition of 2M
NaOH. The amount of NaCl to be added was calculated taking into account the total ionic
strength brought about by the counter-ions of the polymers, the sodium azide and the
NaOH solution. Following NaCl addition at 60"C, solutions were then stirred for 10 min
at these temperature.
2.2 SpectrophotometricMethod
When a system did not show a macroscopic phase separation at rest or following l h
centrifugation, at 60C and 3800g, the optical density was measured at 800 nm and 60C
using a spectrophotometer with a spectral resolution of 1 nm (Perkin Elmer h3, France).
The optical density of the mixtures was compared to the sum of optical densities of gelatin
and iota-carageenan alone at the same concentration. When the optical density of the
mixture was higher than the sum, this was considered to be due to a microscopic phase
separation.
The study of the interactions between gelatin LH and iota-carrageenan versus
temperature was studied with a spectrophotometer Cary 1 0 0 (Varian, France), using a
spectral resolution of 2nm. Cuvettes were thermostated at +/- 0.2 "C using a Peltier block.
Temperature ramps were performed from 60C to 20C at 0.1"C/min. The study of the
interactions by the methylene blue method was performed at 554nm and 663nm as
previously described.s
3.1 Interactions in Non Gelling Conditions

In the study of gelatin PS / iota-carrageenan mixtures at 60"C, three types of behavior
were observed as a function of the total concentration of biopolymers.
When the total concentration of biopolymers is lower than about 0.1 wt%, a
macroscopic phase separation is observed whenever iota-carrageenan is added to a gelatin
solution. The analysis of the composition of the two phases after shows that the upper
phase contains practically no polymer. The concentration domain of phase separation is
smaller at l00mM (eq. NaCl) than at 25 mM (results not shown), a typical behavior for
associative interactions : the higher the salt concentration, the more screened are the
interactions.' This behaviour is in good agreement with the fact that the PI of gelatin PS is
about 9. At pH 7, the net charge of gelatin chains is positive, they associate to the
negatively charged carrageenan chains and form insoluble complexes (complex
coacervation). However, even if this type of interaction is classicaly described in the
literature," the stoichiometry of the associative interactions between oppositely charged
biopolymers is not well understood.
When the total concentration of polymers in the mixture is higher than 0.1 wt%, no
niacroscopic phase separation is observed but some of the systems may be turbid, which
corresponds to the formation of large complexes having enough affinity to the solvent to
CI (wt%) CI (wt%)
Gelatin PS (wt%) Gelatin LH (wt%)

Figure 1 Phase diagrams of gelatin / iota-carrageenan mixtures in 1OOmM eq. NaCl, at
p H 7 and at equilibrium at 60C.( a )gelatin PS, (b)gelatin LH.
form solvated aggregates. Using turbidity measurements, the total scattering behaviour of
mixed biopolymer systems was measured in order to determine the phase state of the
mixtures. The concentration domain for which no additivity is observed for mixtures
compared to the polymers alone is reported in Figure la (turbid systems domain). At a
fixed iota-carrageenan concentration, the turbidity decreases as the concentration of
gelatin in the mixture increases, until the additivity is recovered (transparent systems
domain). Complexes become soluble as their composition tends to be dominated by
gelatin. Soluble complexes may be obtained when opposite charges carried by the two
macro-ions are not equal in number.77sWhen the gelatin concentration is higher than 5 to 6
wt%, systems are transparent. Even if electrostatic interactions still exist, they are altered
depending on the amount of each macromolecule available to interact. High
macromolecular concentration supresses complex coacervation."
The phase behaviour of mixtures containing gelatin LH (PI 4.5)mixed with iota-
carrageenan is illustrated by Figure 1b. No macroscopic phase separation was obtained
whatever the total polymer concentration, the stoichiometry, the salt concentration (25mM
or lOOmM NaCl). The turbidity measurements show that increased scattering compared to
the polymer solutions was observed in a large range of gelatin concentrations even at very
low carrageenan concentrations. The higher the salt concentration the smaller the turbid
domain of concentrations (result not shown), which is also, as for PS/CI mixed systems, a
typical behaviour for associative interactions. However, an opposite conclusion was
expected as the net charge on the protein was now negative. Further evidence for the
electrostatic nature of the association was obtained from absorption spectroscopy,'
showing that associative interactions exist between LH chains and CI chains for salt
concentrations lower than 300mM at 60C. This result can be attributed to the non-
uniform distribution of the charges : i.e. there are still regions along the amino-acid chain,
called patches, which are positively charged.'
Comparing Figure l a and b, we see that, in the same pH and salt conditions, more
gelatin LH than gelatin PS is needed to promote the formation of aggregates leading to a
loss of the turbidity additivity in mixtures. The positive charge density is much lower for
LH and electrostatic interactions can only occur on the limited regions corresponding to
the patches. The parts of gelatin chains not involved in the interaction with carrageenan
chains are well solvated in the water and the complexes are soluble. So for such systems
associative interactions are governed by density and distribution of charges. Furthemore,

the initiation of complexation is independent of the polymer concentrations, but the
formation of insoluble complexes is dependent on the polymer weight ratio.'
3.2 Evolution of the rheological properties during melting
Rheological properties of the gelatin / iota-carrageenan mixed gels during melting has
been studied in order to see if the type of gelatin (LH or PS) has an influence on it." Two
types of storage modulus G' versus temperature curves have been obtained and
represented diagrammatically on Figure 2. For gelatin LH / iota-carrageenan systems
(Figure 2b), G' decreases in two steps, called I and II. Step I corresponds to the melting of
the gel of gelatin dominant phase, step II corresponds to the melting of the gel of iota-
carrageenan dominant phase. Even if associative interactions have been shown to exist,
they do not influence the macroscopic rheological behavior of the mixed gel which is
typical of a gel structured by two independent bi-continuous networks." For gelatin PS /
iota-carrageenan systems, G' decreases in three steps, called Ia, Ib and II. Steps Ia and 11
correspond to the same two melting phenomena as those described above for LH/ iota-
carrageenan systems but an additional step, called Ib, exists between Ia and II. As step Ib
ends at the temperature of beginning of helix to coil transition of the gelatin (40"C), it is
attributed to a stiffening of the iota-carrageenan network by the gelatin chains in helical
conformation." At temperatures above 40C at which gelatin is in the disordered state,
associative interactions still exist but the stiffening effect of gelatin chains in the coil
conformation is too weak to influence the rheological behavior of the iota-carrageenan
dominant network.
So for both gelatin LH and PS, associative interactions exist in the mixed gels but if
they influence greatly the behavior of the mixed gel at temperatures below the
transconformation temperature of the gelatin for the PS / iota-carrageenan mixture, it is
not the case for the LH / iota-carrageenan mixtures in which they are probably not
numerous enough. Stiffening effect is not observed too for both gelatins at temperatures
(a) log G' (b) log G'

t, i Gelatin helix to coil
I i Ttransconformation
Ttransconformation
I LL.-
I
:i
-
I1
20
,
I
, 1
30 40 50 60
11; I
, 1 i,
\
\
I
Temperature ("C) Temperature ("C)

+
Figure 2 Typical evolution ofthe storage modulus G ' observed as a function of increasing
teniperulrrri>jilt- iotti-carrageenan / gelatin mixed gels, previously formed at about 25C.
( ( I ) Rdcititi PS, (b) Relatin LH. Vertical dotted lines : helix to coil transconformation
lerriiwrcrture of gc>lrtitiPS mid LH.
the temperature of end of gelatin coil to helix transition from which the gelatin chains
become too flexible to reinforce the iota-carrageenan network, whatever the interaction
density. Phase behavior and rheological properties during melting are really different
when using PS or LH. If the formation of a two interacting bi-continuous networks is clear
for gelatin PShota-carrageenan, a study of the evolution of the interactions versus
temperature and specifically through the successive transconformation of iota-carrageenan
and gelatin LH, at the molecular level, is interesting to understand how this type of
systems is structured.
3.3 Influence of the Chain Conformation on the Interactions
The attractive electrostatic interactions between the chains of a sulfated polysaccharide
and methylene blue molecules modifies the visible spectrum of the methylene blue.'* A
methylene blue solution shows a maximum optical density at 662-664 nm corresponding
to the absorption of the free methylene blue molecules (Figure 3). In the presence of
carrageenan chains, the peak at 663 nm decreases and a shoulder appears at 554 nm which
can be inferred to correspond to the absorption of the methylene blue interacting with the
sulfated groups of carrageenans chains (Figure 4). In this case, methylene blue molecules
are relatively close to each other. Then, they can interact through long range forces
involving water molecules and form metachromatic complexes of methylene blue
molecules in parallel absorbing at 554 nm (Figure 4). These long range forces can only
take place if methylene blue molecules are close enough i.e. between 0.35 and 0.77 nm.I2
When carrageenan chains are in the coil conformation the distance between the sulfated
groups from one disaccharide to the following one is 1 nm. This distance is theoretically
too high to allow the formation of such long range forces, but, as carrageenan chains are
flexible in the coil conformation, a few of them take place, resulting in a small decrease of
the optical density at 663nm. Consequently, when carrageenan chains are in the coil
conformation, interactions between methylene blue molecules and carrageenan chains
exist without resulting in a large decrease of the optical density at 663nm.
N S
Figure 3 Methylene blue molecule (a) and its diagrammatic representation (b)
182 G u m and Stabilisers for the Food Industry 1 I
-01s -0-
Figure 4 Molecular representation of the methylene blue / carrageenan interactions
When the temperature decreases, the OD663nm obtained with the methylene blue / iota-
carrageenan solution decreases continuously (Figure 5 ) which corresponds to a slow down
of the chain movements : the life time of the methylene blue / CI interactions is higher
and, statistically, a higher number of methylene blue molecules are aligned and absorb at
554 nm instead of 663 nm. This result is in good agreement with those of Schoenberg and
Moore'* obtained with toluene blue and hyaluronic acid. However, a sharper decrease of
OD663nm is observed between 43 and 38"C, the temperature domain where the coil to helix
transition of carrageenan chains occurs at this ionic strength. When carrageenan chains are
in the helix conformation the sulfated groups (at a distance of 0.22 nm) are positioned
around the helix, pointing in three directions, thus two aligned sulfated groups are
separated by 0.66 nmI3 allowing the formation of long range forces with methylene blue
molecules. The chain is more rigid and a higher number of metachromatic complexes are
formed : the OD554nm increases and the OD663nm decreases proportionally.
The addition of gelatin LH to the methylene blue / iota-carrageenan solution at 60C
leads to an increase of the OD663nm and a decrease of the OD554nm (Figure 5). As
associative interactions between carrageenan and LH exist,5 this phenomenon corresponds
to a competition of LH positive patch and methylene blue molecules for the negatively
charged sulfated groups of the carrageenan and, in consequence, to a release of methylene
blue molecules in the solution. When the temperature decreases, the OD663nmstarts to
decrease whatever the LH concentration but at a lower rate than for the methylene blue /
CI solution, as if temperature had a lower effect on the rigidity of carrageenan chains.
At temperatures corresponding to carrageenan chains coil to helix transition, a much
sharper decrease of OD663nm is observed in the presence of gelatin LH. The OD663nm
curves tend to join the one of methylene blue / iota-carrageenan solution. The first
hypothesis that can be proposed is a disappearance of the LH-CI interactions which could
allow more methylene blue molecules to interact with carrageenan. This would correspond
to a decrease of the affinity of LH chains for carrageenan chains when they are in helical
conformation. However, associative interactions still exist : blue complexes are formed at
low temperature in the presence of gelatin depending on the stoichiometry of the mixture,
the higher the gelatin concentration, the bigger the complexes. Note that the OD663nm is at
the same level for 0.25 wt% and 0.75 wt% of LH (Figure 5a). On the contrary, the
OD554nm level measured for the mixture containing 0.25 wt% LH is equivalent to the level
of the methylene blue / carrageenan solution but twice the level of the mixture containing
0.75 wt% LH. At the higher gelatin concentration (0.75 wt%), blue complexes are visible,
they are composed of LH / CI / methylene blue aggregates in which some of the
2.5
0.7
2
.-0 0.25 wt% LH

.-3 0.6
.
I
0
$ 1
1.5
'3
2
0.5
'il
0 0.4
0.5 j
0.3
0
-20
OW% LH
I
25
I I
30 35 40 45 50
I I I I
55
I
60
0.2'
20
I
25
I
30
'
35
I
40
'
45
I
50
I
55
'
60
Temperature ("C) Temperature ("C)
Figure 5 Evolution of the optical density measured at 663 nm (a) and 554nm (b) versus
decreasing temperature for methylene blue / carrageenan mixtures containing various
concentrations of gelatin LH. Rate of temperature decrease :0.1 OC/min.
methylene blue molecules are hidden: they are not free in solution and do not form
metachromatic arrangements. Another hypothesis can be proposed to explain that, in
presence of 0.25wt% LH, results are very similar to that of a methylene blue/carrageenan
solution alone. A few gelatin chains are fixed on carrageenan chains, so there is a higher
- - of
number
... -. - methvlene blue molecules free in the mixture than in thP mPthvlPne
J -
I.. C'n" """'J1V""
blue/carrageenan solution. However, the fixed LH chains stiffen the carrageenan helices
so that their flexibility is reduced. Then, the movements of the methylene blue molecules
around their parallel position in the metachromatic complexes are reduced, resulting in an
increase of the absorption at 554nm for the same number of methylene blue molecules in
the complex. In consequence, the mixture containing 0.25wt% absorbs more at 663 p cJ at
554nm than the methylene blue/ carrageenan solution alone.
4 CONCLUSION
In summary, the results show that associative interactions may occur between highly
charged sulfated carrageenans and gelatin below and above the PI of the protein. The
supramolecular behavior is different depending on the density of positive charges on the
gelatin chains. The rheological properties and the phase behavior are cleary not the same
too, which is in agreement with the fact that they are not introduced for the same food
applications. It seems that for carrageenan / gelatin systems, the biopolymers interact
immediately as they are mixed at pH 7. However, soluble carrageenan / gelatin aggregates
are formed as a critical level of associating protein is reached. The association of gelatin
chains to the carrageenan chains seems to increase the rigidity of the later whatever their
conformation but more specifically in the helical one. For gelatin LH this increase in
rigidity does not influence the rheological behavior of the mixed gel as interactions are
probably not numerous enough. For gelatin PS, a reinforcement of the mixed gel is
observed at temperatures at which the gelatin network is already melted but chains are still
in the helical conformation.
184 G u m and Stabilisers for the Food Industry 1I
5 ACKNOWLEDGEMENTS
A part of the results presented in this paper has been obtained with financial support from
the Commission of the European Community, Agriculture and Fisheries, specific program
FAIR CT961015 entitled << Mixed Biopolymers - Mechanism and Application of Phase
Separation>>.They do not necessarily reflect its views and in no way anticipate the
commissions future policy in this area.
The authors thank SKW Bio-Systems for helpful discussion, practical work and
financial support.
6 REFERENCES
1 L. Piculell, K. Bergfeld and S. Nilsson,. in Biopolymer Mixtures, eds. S . Harding, S.E.
Hill and J.R. Mitchell, Nottingham University Press, Nottingham, 1995, p. 13.
2 I.C.M. Dea, A.A. Mc Kinnon and D.A. Rees, J. Mol. Biof., 1972,68, 153.
3 R.M. Abeysekera, E.T. Bergstrom, D.M. Goodall, 1.T. Norton and A.W. Robarts,
Carbohyd. Res., 1993,248,225.
4 A. Veis, The Macromolecular Clzernistry of gelatin, Academic press, New-York,
1964,433 p.
5 C. Michon, F. Vigouroux, P. Boulenger, G. Cuvelier and B. Launay,. Food
Hydrocolloids, 2000,14, 203.
6 C. Schmitt, C. Sanchez, S. Desobry-Banon and J. Hardy, Critical. Rev.Food Sci.
Nutri., 1998, 38, 689.
7 V.B. Tolstoguzov, in Food Proteins and their Applications, eds. S. Damodaran and A.
Paraf, Marcel Dekker, New-York, 1997, p. 171.
8 K.W. Mattison, W. Yingfan, K. GrymonprC and P.L. Dubin, Macromol. Synzp., 1999,
140, 52.
9 J. Xia and P.L. Dubin, in Macromolecular Complexes in Chemistry and Biology, eds.
P.L. Dubin, J. Bock, R. Davis, D.N. Schultz and C. Thies, Springer-Verlag, Berlin,
1994, p. 247.
10 C. Michon, G. Cuvelier, B. Launay and A. Parker, in Food Colloids - Proteins, Li1Jid.s
and Pofysaccharides, eds. E. Dickinson and B. Bergenstahl, Royal Society of
Chemistry, Cambridge, 1997, p. 316.
11 S . Kasapis, E.R. Morris, I.T. Norton, and A.H. Clark, Carbohydr. fOlyl72., 1993, 21,
261.
12 M.D. Schoenberg and R.D. Moore, Biochim. Biophys. Acta, 1973,83,42.
13 S. Nilsson, L. Piculell and B. Jonsson, Macromolecules, 1988,22,.2367.
HYDROCOLLOIDS IN REAL FOOD SYSTEMS
I.T. Norton and T.J. Foster
Unilever Research and Development Colworth, Colworth House, Sharnbrook, Bedfordshire,

UK.
ABSTRACT
Hydrocolloids (polysaccharides and proteins) are used in the Food Industry to impart texture
and appearance properties to fabricated foods. It is quite often the case that single
hydrocolloid solutions or gels are incapable of delivering all the desired product properties.
Although in the chemical industries this problem can be overcome by chemical and
biochemical modification of the polymeric species, this is not an option that is readily open to
the Food Industry due to constraints on food additives.
It is with this background that we developed the idea of kinetic trapping to design
microstructures from single and mixed hydrocolloid systems in order to deliver dynamic
material properties. This paper will discuss the mechanism of structure formation and the
influence of kinetic trapping and flow modification on the creation of composite structures and
also cover some of the mechanical and sensorial consequences of the microstructures formed.
1 INTRODUCTION
Hydrocolloids are used in many fabricated food products in order to impart the required
quality in terms of stability (hence the name stabilisers), texture (in-product and in-mouth) and
appearance. The business in hydrocolloids has increased significantly over the last two to three
decades as the demand for high quality, lower fat (even fat free) products escalated, together
with the divergence of product types that people in developed societies demand. On a similar
health angle the image of hydrocolloids is again on the increase thanks to the trend in
functional foods, being initially focussed on dietary fibres. Along with the positives come
some negatives. One notable example is the way gelatin has fallen out of favour in some
societies. Therefore, a drive for gelatin replacement has been witnessed in a number of
fabricated food product types, mirrored by a range of commercial gelatin replacers offered
by alternative ingredient suppliers. Other issues, such as climate effects on crop yield, can
influence the hydrocolloid marketplace. This can lead to higher material costs or cost
fluctuations, which become unattractive for the food manufacturers. This then requires the end
user to have the ability to interchange ingredients, whilst retaining the required product
attributes. In order to accomplish this, a detailed knowledge of why the initial ingredient /
hydrocolloid was in the product in the first place is required, ie. What was its functionality?
The vast majority of food grade hydrocolloids are water soluble and therefore control the
aqueous phase properties of the manufactured products (many of which are multicomponent,
like emulsions, foams and frozen products). Therefore, in order to derive how the hydrocolloid
in question functions in the final product we need to understand:
its viscosity / gel properties,
the kinetics of gelation,
0 how it is influenced by the presence of other hydrocolloids (mixed systems),
how the manufacturing process (cooling, shearing, dehydration, rehydration) influence the
structures created,
the complex composite properties of these final structures.
The approach taken is highlighted in Figure 1.
~~
Product Properties
StabiIity
Microstructure
Texture
1 Process 1 I Appearance
Figure 1 Approach taken in the design of food microstructures.
2 SINGLE HYDROCOLLOID SYSTEMS
2.1 Structure / Function
The objective is, therefore, to develop an understanding of hydrocolloid structure/function

relationships in real food systems for interchangeability and development of new textures.
This requires an understanding of the hydrocolloid primary structure and how this influences
the nature of the polymeric interactions and interaction rates. Past and recent work in this area
has investigated the structure /functionality of galactomannansIp2,pectins3, alginate~~.~,agars6,
gellan gum and carrageenans. Indeed, as measurement techniques are developed further, the
ability of the analytical chemist to fingerprint the molecular architecture of water soluble
hydrocolloids will improve further. As testament to the current developments in this area a
number of papers in this edition are dedicated to the Material properties are then
measured to relate the molecular fingerprint to the functionality of the chosen hydrocolloid. Inl
this way the influence of primary structure variation and changes in environmental conditions,
(solvent, co-solute, temperature and process effects) can be tested.
As many reviews have identified, there is considerable knowledge on single systems.
However, the end user needs to know how to manipulate the properties and how to
interchange materials. For instance, regarding the solution properties of viscosifying
hydrocolloids eg. the galactomannans, these properties are dictated by hydrodynamic

properties of the hydrocolloid molecule ie. the volume it sweeps out in solution and also
describes how the molecule effects the shear thinning behaviour at higher shear rates12.
However, using the galactomannans as an example, the molecular architecture does determine
the functionality in terms of gel formation (typically induced by freezing and thawing the
ample)'^. With a ribbon-like backbone consisting of p-1,4 D mannose with varying degrees
of a-1,6 D galactose sidechains, LBG (23% galactose) behaves differently to Guar Gum (40%
galactose). Recent developments have seen guar gum functionality match that of LBG using
enzyme technology to prune the galactose sidechains. However, an in-depth study of LBG
structure / functionality has highlighted the fact that a single sample of LBG is heterogeneous
in its molecular composition. The galactose content is broad average and the inter- and intra-
molecular distribution of galactose is important in determining the overall fun~tionality~.
LBG can be fractionated with respect to solubility. Work on these fractions has been used in
the past to study the synergistic interactions of LBG with xanthan gum5 and K-
carrageenan16. The structure / hnctionality of these fractions as single systems is equally as
illuminating. Cold soluble LBG has a high galactose content and does not gel. High
temperature soluble LBG has a galactose content of 16% and gels at ambient temperature.
Therefore consideration of the distribution of galactose sidechains is required when choosing
or designing the functionality of galactomannans in the final food product. The gelation rate of
LBG is dependant on concentration, co-solute content (solvent quality) and temperature. The
self-association is therefore kinetically controlled, as a function of the number of available
junction zones.
Typical gel forming hydrocolloids (eg. agar, carrageenan, gellan and gelatin) undergo a
molecular rearrangement as a prerequisite to gel formation. Under favourable thermodynamic
conditions the system lowers its free energy by ordering the molecule from a random coil to a
helical state. This is often achieved by decreasing temperature. These ordered helices then
aggregate to form through space filling polymer networks. The structures created are,
however, under kinetic control and the rate at which gelation is forced via temperature, or salt
level (for those systems where gelation is ion mediated) ramps determine the final gel
properties17 I9 . This is an area of academic research that is not fully exploited.
As all hydrocolloids are derived from natural sources the inherent biodiversity of the fine
structure seems to be a common occurrence. The suppliers tend to smooth these out by
blending. However, this may be the way forward for new materials, for instance Figure 2
shows the gel properties of three different pectins. These pectins do not differ in their degree
of esterification (DE), however they do differ in their inter- and intra-molecular DE
distribution.
Therefore, although the number of available food grade hydrocolloids is limited (Table l),
their biodiversity means that the structural possibilities offered are still being developed.
Recent work on alginates2*and gellan gum21has shown that slight changes in the polymer fine
structure, as a function of varying material source or how the polymers are extracted and
subsequently processed, can also affect the gel properties of the single hydrocolloid systems.
Other new polymers are also emerging, particularly in developing markets, where they have
been used in local manufacture and consumption. These hydrocolloids have the potential for
world-wide exploitation, with examples being konjac glucomannan and curdlan, which have a
long history of use, particularly in Japan, and offer potential for delivering new fhctionalities
in Western food products.
190 Gums and Stabilisersfor the Food Industry I1
0 20 40 60 80 100 120 140 160 180 200

Time (minutes)
Figure 2 Showing the gelation of three pectins with time and temperature. All pectins have
with an average DE of 45 and are gelled with Calcium at an R-value Of 0.5.
As well as the effects of primary structure on the functionality of specific hydrocolloids,

the solution and process conditions have an equally important role to play.
2.2 Salt Effects
In foods the level of additives is determined by consumer needs. Thus we need to be aware of
and understand how these effect thermodynamic and kinetic properties of hydrocolloid
structuring. Table 1 shows that a single polymer eg. pectin can be used as either a gelling or
thickening agent. This depends on the ionic (or pH) environment that the pectin molecule
finds itself in. Calcium ions are known to facilitate pectin gelation, whereas in pHs above
pectin PI and at relatively low NaCl concentrations the same molecule acts as a typical random
coil polymer. However, if the pH is lowered, or the NaCl concentration increased then gel-like
characteristics are produced. The gel properties are, however, very different to those formed in
the presence of calcium ions. Indeed, the level of calcium ion used to induce gelation effects
the final gel properties. Similar effects are also seen for alginate22 and carrageenan23 gel
systems. Figure 3 shows the difference in gel molecular architecture of gellan gels formed in
the presence of different salt systems, at constant polymer type and level.
2.3 Sugar Effect
The examples given above are important when considering the use of hydrocolloids in
savoury food products. However, they are also used extensively in sweet food products eg.
confectionery items, ice cream, desserts etc. Therefore, the influence of added sugars on the
properties of hydrocolloid structures is required. This can be via direct binding or a result of a
change in solvent quality. Confectionery items can be termed high solids systems, whose
properties have been shown to be highly dependent on the microstructure of the end products.
Table 1 Showing a range offood hydrocolloids and their in-productfunctionalites.
Figure 3 Electron micrographs of 2.5% deacylated gellan in the presence of 70mM KCI (lep)
and 35mM CaCl2.2H20.Magnification x 40,000.
These are typically mixed hydrocolloid systems and will be covered later on in this article.
Past work on agar, carrageenan and gelatin24y25 has shown that the properties of gels are
significantly modified in the presence of high levels of added sucrose. Current work on
gellan26 and ~c-carrageenan~is highlighting similar trends in terms of modification of gel
structures. The level and type of sugar used is important in the structural modifications
-
witnessed. On a molecular level intermediate levels of added sugars (up to 40%) appear to
stabilise the ordered, helical structures of agar, carrageenan and gellan gum, imparting higher
gelling and melting temperatures, along with an increase in the amount of order in the system
(ie. higher AH values). As the amount of sugar is increased the extent of order is decreased,
inducing lower gel strengths and producing gels with much more elastic properties. Structure
loss is also dependant on the type of sugar used, which has been associated with the number of
equatorially attached OH-groups present on the added sugar (in the order ribose < mannose <
fructose < glucose < sucrose < maltose < raffinose). The effects here are due to the changes in
solvent quality, related to the solvation of the hydrocolloids and how they move into rubbery /
glassy states, which influence the molecular structures attainable by the hydrocolloid itself.
Therefore in an unperturbed system the possibilities of creating different properties appear

to be endless! The art is to be able to do this in a controlled way.
2.4 Process Effects
In typical food manufacture the systems are subjected to thermal treatment and shearing.
Recent work28has highlighted the importance of understanding the effects of these process
conditions on the final structures created. This can be known as disrupted gelation, where gels
are formed in the presence of a shear field. Small gel particles, in the region of 1-50 pm, are
produced. These particles interact with one another, in the form of a particulate dispersion,
promoting textures varying from pourable fluids to thick pastes. The composite properties are
dependent upon the number and size of the particles produced. This in turn is dependent upon
the polymer used, the polymer concentration, the shear field and the temperature profile in
which the particles are formed. Models have been developed for particle formation and the
phenomenon seems to be valid for hydrocolloids that gel via a process of aggregation28.Fluid
gel composite structures have been made from agar, carrageenan, pectin, alginate, gellan and
gelatin gel systems. The particles reported previously, however, have been formed under
typical process conditions ie. fast cooling and shearing. This does not enable adequate control
of particle shape. However, by considering the rates of ordering and gelation, it is possible to
distort droplets of hydrocolloids and then kinetically trap them by gelation to form asymmetric
particles29.These particles with altered morphologies impart quite dramatic effects on the
solution viscosities in which they reside (Figure 4). It has only been possible, thus far, to
create such particles in an immiscible, phase separated system and may therefore remain a
problem to be solved in single hydrocolloid systems3'.
10
Figure 4 Different rheologies from dispersed gellan particles of different aspect ratios in a
continuous phase of K-carrageenan (+ elongated, I spherical) Dispersed phase volume of
20%, total polymer concentration is 2%, no added salt and measured at 60C.
3 MIXED HYDROCOLLOID SYSTEMS
Mixed hydrocolloid systems have been the subject of a lot of research over recent years. The
consequence of mixing hydrocolloids is the creation of complex composite structures (Figure
5). These are very often a result of polymer incompatibility, forming at least two phases, each
rich in one of the constituent hydrocolloids. There are, however, exceptions to this, where
either phase separation occurs with both polymers separating into one of the phases or the
hydrocolloids remain miscible. When this occurs the polymers can either interact with one
another or remain as individual polymers, coexisting together. The former are classed as
synergistic systems, the latter as interpenetrating networks. Examples of synergistic
systems are formed when xanthan gum interact with either LBG or konjac glucomannan.
These systems are finding wide us in food systems such as sauces, dressings and ice cream,
while work progresses in the understanding of the molecular basis for the synergistic
effect ,31.
Figure 5 Confocal micrograph of a Rowntrees Fruit Pastille, indicating the complexity of

microstructure.
3.1 Phase Separation
However, in this article we will concern ourselves with the much more common and useful
occurrence of immiscible systems. These systems are used, by design or fortune, in many food
products. The required properties are a result of the microstructure of the immiscible system,
which depends on the relative phase volumes of the phases, the phase compositions and
properties and how the different phases interact with one another.
Such phase separated systems can be described by what are known as phase diagrams,
where compositions inside the binodal phase separate along a tie-line to give different
compositions of the respective phases. Points on the same tie-line give phases with the same
phase composition, but the composite structures differ in their relative phase volumes. Figure
6 shows a typical phase diagram. The accompanying measuring cylinders show the differences
in phase volume of the two phases as a result of changing the composition along a tie-line.
This is the resultant structure of the system if it is allowed to 'bulk' phase separate. However,
when the systems are mixed, in the solution state, the microstructure resembles that of an
emulsion. Indeed, such systems behave as emulsions and have been called water-in-water
emulsions, whose properties can be described using conventional emulsion r n ~ d e l s ~ * - ~ ~ .
4 8 16 23
+
MaI ode i irin" O
"'
Figure 6 Phase diagram of gelatin : maltodextrin, with measuring cylinders showing bulk
phase separation.
Phase diagrams themselves do not supply all the answers that one needs to be able to
design the correct microstructures for specific applications. They only describe a starting
point, as they represent an equilibrium position under certain experimental conditions. As has
been described in previous sections, many hydrocolloids change their hnctionality when
solution conditions are changed ie. changes in temperature or solvent conditions. It is therefore
important to understand how these effects change the thermodynamic nature of immiscible
solutions. When the concept of gelation was considered for single hydrocolloid systems, the
kinetic aspects were shown to be important in determining the final gel structure and
properties. Kinetic aspects are also very important when considering the microstructures
produced from immiscible, mixed systems. Recent work has shown that as one or more of the
hydrocolloids start to gel the ordering shifts the thermodynamic balance of the system, either
inducing phase separation or further phase separation35z6.Therefore, a system is under kinetic
control both in terms of the driving forces for further phase separation and gelation, trapping it
away from its final equilibrium state ie. where it wants to be at the end of the environmental
shift. Figure 7 shows a schematic of how a phase diagram moves with changing temperature
and figure 8 shows the evolution of microstructure during a quench to a nominal temperature
from a miscible starting solution. It is clearly seen in figure 8 that the microstructure is trapped
in the early stages of gelation. Recent work has shown that such microstructures are created as
a result of either spinodal decomposition or nucleation and growth3'. This, in turn, is a
function of where, within the phase diagram, the system is quenched to ie. inside the spinodal
or into the metastable region, between the binodal and spinodal. When considering the kinetic
aspect already discussed, this concept is important, as the different mechanisms of structure
formation are critical in determining initial phase volumes and phase compositions. The
relative rates of phase separation and gelation therefore need to be born in mind when
designing microstructures for various functionalities.
Real food systems that take their functionality from immiscible hydrocolloids are, to a
greater or lesser extent, governed by these criteria. Ice cream is a real food system that fits this
line of argument. Model studies in recent years have shown that mixtures of galactomannans
and milk proteins are immiscible m i x t ~ r e s ~Earlier
~ ' ~ ~ in
. this article we have highlighted the
ability of certain galactomannans to form gels. Therefore, it seems obvious that the
microstructures of many food products are under kinetic control, and the ability to create novel
textures will be with the groups that can take this control.
12 -l ' l ' l ~ l ~ l ~ l ~ l ~ l
Isothermal Binodal Evolution Owing to Ordenng

10 -
- 58 -.
3
6(SA2]1st = 0 4 % / min
-
\
7
6 -
54- 0 .
2 -
O', , , , , , I , , , 1 , I , ,
Figure 7 Schematic showing the movement of the binodal after quenching to a nominal
temperature.
There are occasions, however, where gelation of the hydrocolloids does not induce phase
separation. The individual hydrocolloids seem oblivious to the fact that the other polymer is
present and gel as they would have done in their absence. These systems are called
'interpenetrating networks' and have been the subject of recent studies4042.
3.2 Process Effects
We have previously reported the effect of shear on the microstructure of immiscible systems,
and the consequence that this has on the properties of the final composite material. Rules were
developed stating that when cooling in a shear field the first gelling hydrocolloids forms the
dispersed phase43. At the time, the second hydrocolloid did not gel in the shear field (eg.
maltodextrin or oxidised starch), or rehealed post shear (eg. gelatin). It is evident from figure 9
that this general rule is still under phase volume control. It can also be assumed that if both
polymers gel in the shear field, and create particles which are stable, the composite structure
will resemble that of a fluid gel made from particles of different polymers gels, whose relative
phase volumes are determined by the initial polymer concentrations. The example given in
figure 9 is a result of very high shear rates and stresses. However, if these are better controlled,
then microstructures can be designed, as previously mentioned and shown in figure 4, to give
very different textural properties.
2200
2WO
1800
1600
-
Q
E
C
1400
1200
800
600
400
200
0 50 1W 150
tl min
Figure 8 Time evolution of estimated droplet size of a 4% gelatin :4% maltodextrin
system (0 quench to 25C; 0 quench to 20"C), morphology when
quenched to 20C with time.
Figure 9 Phase diagram of gelatin : oxidised starch and the resultant microstructures created
when gelling in the presence or absence of a shearfield
4 WHY BOTHER? DONT WE ALREADY KNOW WHAT WE NEED TO KNOW?
4.1 Changes in material properties
We have spent some time considering the emulsion-like microstructure of immiscible

hydrocolloid mixtures. If we are to design material properties, then we need to control and
manipulate the interface between phases. This is well known and applied in synthetic
polymers and ceramics.
Recent work in hydrocolloid mixed systems has shown that, depending on the strength of
the interface between the two phases, very different properties are achieved. Debonding has
been seen to occur in a gelatin : maltodextrin system, where the gelatin phase is the continuous
phase44(figure 10). Debonding therefore produces a weak interface, which has a considerably
different breakdown pattern to a system that has a much stronger interface, which, in this case,
is a gelatin : agarose immiscible composite (figure 11). Again, advances in measurement
techniques are improving our understanding of such systems. Environmental scanning electron
micrographs show maltodextrin gelled droplets popping out of a gelatin continuous phase,
upon deformation4. If the particles themselves are ruptured, the same technique can give a
fascinating insight into the crystalline structure of the maltodextrin gel (figure 11).
Figure 10 (above) CSLM image taken

during a tensile test showing debonding
around maltodextrin particles.
Figure 11 (right) Differences in stress

strain response of a debonding interface
and a stronger interface.
ESEM micrographs showing maltodextrin Strain
Particle debondingfrom a gelatin matrix
(top right) and the fine structure of the inside of a fractured rnaltodextrin particle.
4.2 Controlled Release
In the introduction we discussed the growth of the hydrocolloid industry as the market in
healthy, low fat food products increased in size. One problem that faces food manufacturers is
achieving parity of taste and flavour when creating high quality low-fat products, in
comparison to the high fat equivalent. The major problem is the early loss of lipophilic
flavours, due to the limited reservoir of fat in which they are solubilised. However, as the
environment in the mouth is dynamic, with the food product undergoing shear and temperature
198 Gums and Stabilise,psfor the Food Industry 11
changes, it is possible to use the product microstructure to control flavour release46.Thus, by

incorporating the lower level of fat inside a hydrocolloid gel particle the release of lipophilic
flavours is slowed down, via a mechanism of mass transfer control in the dynamic
environment. Figure 12 shows an example of the hydrocolloid gel emulsion particles and their
effect on flavour release (as measured by volatile detection mass spectrometry, while being
chewed by a volunteer). It can be seen that, using this technology, a 3% fat emulsion can be
given the same flavour release profile as a much higher fat product.
Time (seconds)
Figure 12 Flavour intensity (measured by volatile detection mass spectrometry) of water

continuous emulsions containing 40% fat (- - - - -), 3% fat (-- -- --) and 3% fat in a
1.
hydrocolloid gel emulsion particle (--------
5 CONCLUSIONS AND WHERE NEXT?
The requirement from the consumer for high quality fabricated food products will remain, and
has the potential to grow, as the world population becomes increasingly demanding. In the
hurry to formulate the product developer might incorporate a wide range of food hydrocolloids
in the quest to achieve the quality desired and required. The problem faced here is that very
quickly they can not reasonably justify the inclusion of a certain hydrocolloid (if not most) in
the stew pot of aqueous phase structuring agents. It is therefore important to know where a
specific hydrocolloid is, within a complex microstructure, and what role it is really playing
(answering the questions Where is it? and Why is it there?). In this article we have tried to
indicate that a good knowledge of hydrocolloid fine structure and structuring rates, how it
mixes with others (kinetic trapping, interfacial design etc), linked to the choice of process
used, can enable microstructure design to provide product structure and the required product
properties.
As the market for these food products becomes more globalised, we also predict that the
following three areas will begin to play an increasingly important role, and will probably be
featured in this set of proceedings in years to come.
0 Use of alternative hydrocolloids from around the world, for local manufacture of food
products with quality parity from region to region.
0 The use of biomaterials (natural structuring), incorporating the area of biornimetics, to

give fabricated food products increased health perception and authenticity.
0 The use of hydrocolloids in nutrition and nutrient control as the consumer increases their
drive towards healthy living, under the increased awareness of 'You are what you eat'.
References
1 B. V. M'Cleary, A. H. Clark, I. C. M. Dea and D. A. Rees, Carbohydrate Research, 1985,

139,237-260.
2 I. C. M. Dea, A. H. Clark and B. V. M'Cleary, Carbohydrate Research, 1986,147 (2), 275-
294.
3 E. R. Moms, M. J. Gidley, E. J. Murray, D. A. Powell and D. A. Rees, International Journal
of Biological macromolecules, 1980,2 (5), 327-330.
4 D. Thorn, I. C. M. Dea, E. R. Morris and D. A. Powell, Progress in Food and Nutrition
Science, 1982,6 (1-6), 97-108.
5 K. I. Draget, G. S. Braek, 0. Smidsrod, Carbohydrate Polymers, 1994,25 (l), 31-38.
6 M. J. Gidley, N. D. Hedges, R. A. Hoffmann and C. P. Dawes, in Gums and StabiEisers for
the Food Industry 7, eds. G. 0. Phillips, P. A. Williams and D. J. Wedlock, Oxford University
Press, 1994,3-14.
7 E. R. Moms, M. G. E. Gothard, C. E. Manning and G. Robinson, Carbohydrate Polymers,
1996,30 (2-3), 165-175.
8 R. A. Hoffinann, M. J. Gidley, D. Cooke and W. J. Frith, Food Hydrocolloids, 1995, 9 (4),
281-289.
9 H. A. Schols, P. J. H. D. Daas, K. Grolle, T. van Vliet and H. H. J. de Jongh, this edition.
10 M. A. K. Williams, G. M. C. Buffet and T. J. Foster, this edition.
1 1 Handbook of Hydrocolloids, eds G.O. Phillips and P.A. Williams, Woodhead Publishing
Limited and CRC Press LLC, 2000.
12 E. R. Morris, A. N. Cutler, S. B. Ross-Murphy and D. A. Rees, Carbohydrate Polymers,
1981,1, 5-21.
13 P. H. Richardson and I. T. Norton, Macromolecules, 1998,31 (5), 1575-1583.
14 P. H. Richardson, A. H. Clark, A. L. Russell, P. Aymard and I. T. Norton,
Macromolecules, 1999,32 (5), 15 19-1527.
15 R. 0 Mannion, C. D. Melia, B. Launay, G. Cuvelier, S. E. Hill, S. E. Harding and J. R.
Mitchell, Carbohydrate Polymers, 1992,19 (2), 91 -97.
16 L. Lundin and A.- M. Hermansson, Carbohydrate Polymers, 1995,28, 91-99.
17 J. P. Busnell, S. M. Clegg and E. R. Morris, in Gums and Stabilisers for the Food Industry
4, eds G. 0. Phillips, D. J. Wedlock and P. A. Williams, IRL Press, 1988, 105-1 15.
18 P. Aymard, D. R. Martin, K. Plucknett, T. J. Foster, A. H. Clark and I. T. Norton,
Biopolymers, 2001, 59 (3), 13 1 - 144.
19 I. T. Norton, D. M. Goodall, E. R. Morris and D. A. Rees, Journal of the Chemical Society
- Chemical Communications, 1979,22,988-990.
20 B. T. Stokke, K. I. Draget, 0. Smidsrod, Y. Yuguchi, H. Urakawa and K. Kajiwara,
Macromolecules, 2000,33 (5), 1853-1863.
21 T. A. Talashek, Abs, American Chemical Society, 1998,215.
22 K. I. Draget, K. Steinsvag, K. Onsoyen and 0. Smidsrod, Carbohydrate Polymers, 1998,
35 (1-2), 1-6.
23 L. Lundin and A.- M. Hermansson, Carbohydrate Polymers, 1997,34 (4), 365-375.

24 K. Nishinari, M. Watase, K. Kohyama, N. Nishinari, D. Oakenfull, S. Koide, K. Ogino, P.
A. Williams and G. 0. Phillips, Polymer Journal, 1992,24 (9), 871-877.
25 K. Nishinari, M. Watase, E. Miyoshi, T. Takaya and D. Oakenfull, Food Technology, 1995,
49 (lo), 90.
26 G. Sworn and S. Kasapis, Food Hydrocolloids, 1998,12 (3), 283-290.
27 S. Kasapis, I. M. A. Al-Marhoobi and A. J. Khan, International Journal of Biological
Macromolecules, 2000,27 (l), 13-20.
28 I. T. Norton, D. A. Jarvis and T. J. Foster, International Journal of Biological
Macromolecules, 1999,26 (4), 255-26 1 .
29 B Wolf, W. J. Frith and I. T. Norton, Journal of Rheology, 2001,45 (5), 1141-1 157.
30 B. Wolf, W. J. Frith, S. Singleton, M. Tassieri and I. T. Norton, Rheologica Acta, 2001,40
(3), 238-247.
31 C. E. Cronin, P. Giannouli, E. R. Morris, B. V. M'Cleary and M. Brooks, this edition.
32 C. R. T. Brown, T. J. Foster, I. T. Norton and J. Underdown, in Biopolymer Mixtures, eds.
S. E. Harding, S. E. Hill and J. R. Mitchell, Nottingham University Press, 1995,65-83.
33 J. R. Stokes, B. Wolf and W. J. Frith, Journal ofRheology, 2001,45 (5), 1173-1191.
34 J. R. Stokes and W. J. Frith, this edition.
35 N. Loren, A.- M. Hermansson, M. A. K. Williams, L. Lundin, T. J. Foster, C . D. Hubbard,
A. H. Clark, I. T Norton, E. T. Bergstrom and D. M. Goodall, Macromolecules, 2001, 34 (2),
289-297.
36 L. Lundin, K. Odic, T. J. Foster and I. T. Norton, in Supramolecular and colloidal
Structures in Biomaterials and Biosubstrates, eds. M. Lal, P. J. Lillford, V. M. Naik and V.
Prakash, Imperial College Press, 2000,436-449.
37 M. F. Butler, Macromolecules, in press.
38 S. Bourriot, C. Gamier and J. L. Doublier, Food Hydrocolloids, 1999,13 (l), 43-49.
39 C. Schorsch, A. H. Clark, M. G. Jones and I. T. Norton, Colloids and Surfaces B-
Biointerfaces, 1999,12 (3-6), 3 17-329.
40 A. H. Clark, S. C. E. E y e , D. P. Ferdinando and S. Laganigue, Macromolecules, 1999,32
(23), 7897-7906.
41 E. Amici, A. H. Clark, V. Normand and N. B. Johnson, Biomacromolecules, 2000, 1 (4),
721-729.
42 E. Amici, A. H. Clark, V. Normand and N. B. Johnson, Carbohydrate Polymers, 2001, 46
(4), 383-391.
43 T. J. Foster, C. R. T. Brown and I. T. Norton, in Gums and Stabilisers for the Food
Industry 8,eds G. 0. Phillips, P. A. Williams and D. J. Wedlock, Oxford University press,
1996,297-308.
44 K. P. Plucknett, V. Normand, S. J. Pomfret and D. Ferdinando, Polymer, 2000, 41 (6),
2319-2323.
45 K. P. Plucknett, F. S. baker, V. Normand and A. M. Donald, Journal of Material Science
Letters, 2001, 20 (16), 1553-1557.
46 1. A. M. Appelqvist, C. R. T. Brown, J. E. Homan, M. G. Jones, M. E. Malone and I. T.
Norton, 2001, EP 1102548 Al.
INTERACTION OF CARRAGEENAN WITH OTHER INGREDIENTS IN DAIRY
DESSERT GELS.
Joop de Vries
Danisco Cultor Innovation, Edwin Rahrs Vej 38, 8220 Brabrand, Denmark
1 ABSTRACT
Interaction of carrageenans with starches and protein plays an important role in the main
food applications of carrageenan, especially in dairy products. The minimum
concentration required for gelling is much lower in milk compared to other solvents, the
so-called milk-reactivity of carrageenans. All gelling carrageenan types show this
behaviour. Carrageenan blends show unexpected synergisms. The synergistic effect of
casein micelles on carrageenan gelation is higher compared to any other protein. Most
likely, the interactions are not specific but based on general hydrocolloid interaction
phenomena. Adsorption probably is the key phenomenon. Starches and also locust bean
gum have an antagonistic effect on kappa carrageenan gelation, but not on iota
carrageenan gelling. Also the negative effect of cocoa particles on kappa carrageenan
gelation is still unexplained.
2 INTRODUCTION
Chocolate milk and gelled dairy desserts are among the well-established application areas
for carrageenan. The literature about carrageenan and milk protein interaction is
substantial (1 - 7), but nevertheless a generally accepted theory for the so-called milk
reactivity of carrageenan has not yet been formed. A new - but still hypothetical - model
will be put forward here. Many carrageenan containing dairy desserts contain starch as
well as other polysaccharides like locust bean gum and other ingredients such as cocoa
powder. In industrial practice many interaction effects of these ingredients are well-
known, but little scientific studies have been published that study the whole system and
not just the interaction of two components in model systems. This paper describes some of
the interaction effects of carrageenans in dairy desserts. Hopefully the industrial
experience reported here will stimulate scientific studies.
3 CARRAGEENAN PROPERTIES
Carrageenans have been described in several reviews, e.g. reference 8. Gelling

mechanisms of carrageenan have been reviewed in e.g. reference 9. The next table
summanses:
I Types I Kappa, iota, lambda, hybrid = 'kappa-2' + precursors I

Kappa: K
Metal required
Iota: Ca
for gelling
Hybrids: both
Gel strength: kappa > > hybrid > > iota
Brittleness: kappa > hybrids >> iota
Gel
(hybrids look like iota with Ca, and like kappa with K)
characteristics
Shear reversibility: iota > hybrids >> kappa
Lambda: does not gel
Gelling Kappa forms much larger strands of aggregated helices
mechanism compared to the other types
Table 1. Properties of carrageenans.
Food-physical literature on carrageenan very much focuses on kappa and iota

carrageenan, extracted from Kuppaphycus ulvurezii (old name: Euchema cottonii) and
Euchema denticulatum (old name: Euchema spinosum,) respectively, and almost neglects
hybrid carrageenan. Hybrid carrageenan molecules consist of kappa as well as iota
disaccharide structures. The fact that hybrids form iota-like gels with Ca and kappa-like
gels with K suggests that the two types of structures occur in long sequences (above the
minimum required for junction zone formation). They can be distinguished from blends of
pure iota and pure kappa by Smith degradation if the precursor is available (lo), by
fractionation in KC1 solutions, by viscosity profiles, by enzymatic degradation (kappa
carrageenase) and by electrophoresis.
Capillary electrophoresis of carrageenans
- hybrid
__
c
0 5 10 15 20
Figure 1. Capillary Electropherogram of the basic carrageenan types.

(Running buffer: 5 mM phtalic acid, pH 5.5, 40C,15 kV,24.5 cm uncoated capillary
column 75p,Y-as: detection UV 215 nm, X-as: time in minutes).
Figure 1 shows capillary electrophoresis of some carrageenans and clearly illustrates that
hybrid carrageenans are not blends of pure kappa and pure iota carrageenan. This
technique can be used to study the charge density distribution of hybrid carrageenans. The
kappa and iota sequences perhaps occur in heterogeneous distribution, but more work is
needed to elucidate the fine structure of hybrid carrageenans. It is suggested that the
precursor of Chondrus crispus is a v carrageenan (1 1); this suggests that de-sulphatation
takes place after polymerisation during biosynthesis.
4 MILK REACTIVITY
Carrageenans are the preferred gelling agents in milk products because they gel in much
lower concentrations in milk compared to any other gelling agent. The next figure shows
that they also gel in lower concentrations compared to any other solvent: the so-called
milk reactivity.
140
120
g 100
--e F(milk)
$ 80
-g 60
-f- F(KCL)
40
20
0
0 0.5 1 1.5 2
Gum%
Figure 2. Breaking strength (g/cmZmeasured by SMS TA-TX2)

of hybrid carrageenan from Gskottsbergii) gels in milk compared
to 0.25% KCI solution.
Carrageenans show milk reactivity only at low concentrations. At carragcenan

concentrations of above 0.5% there is not much difference between gel strengths in milk
compared to salt solutions. Carrageenans show milk reactivity in skim milk as well as in
whole milk, which suggest that the proteins are the main ingredients involved. The
following phenomena could play a role in milk reactivity:
- Chargekharge interactions between negatively charged carrageenan and positively

charged segments of (overall negatively charged) kappa casein (1).
- Incompatibility (3, 12).
- Depletion flocculation (3, 12).
- Adsorption (7).
- Bridging flocculation (7).
- Specific interactions like in enzymehbstrate complexes.
It is not that easy to prove the exact mechanism. Incompatibility certainly occurs: when
hydrocolloids are added to milk, phase separation can be seen after several hours at
temperatures above the setting temperature of the hydrocolloid used. Pectin, carrageenan,
LBG, and CMC cannot be added in concentrations exceeding 0.1% or so, whereas
gelatine (a protein) and starch (not completely dissolved) are more compatible (more than
0.5% required for phase separation). Adsorption has been shown to occur in
carrageendmilk protein systems (3, 13). However, in a milk gel both carrageenan and
milk proteins are insoluble and this raises the question: Which polymer plays the soluble
part and which polymer offers a surface to adsorb on?
Comparing rheological behaviour of carrageenan in milk and KCVCaC12 solutions

suggests that carrageenan is the dominant gelling agent in carrageendmilk gels, and that
any contribution from protein is less important. If the system were a coupled network
(with a continuous network of carrageenan and casein bound to each other) one would
expect this to affect rheological behaviour. Table 2 (rheology of gels at more or less the
same gel strength in milk compared to water) points to similarity more than to difference.
The slopes of the log G vs. log frequency are similar (about 0.06) for all four gels,
indicating similar gel structures (14). This is in agreement with sensory evaluation: milk
and water gels of carrageenan have quite the same texture (the typical kappa and iota
textures show up in milk as well as in water). In addition, the shear reversibility of milk
gels made from the different carrageenan types is the same as in water gels. This suggests
that carrageenan gels in milk are not coupled network gels. However, if casein micelles
only act as fillers occupying space then other fillers like swollen starch granules would be
expected to behave similarly and this is not the case (see below). The table confirms that
kappa carrageenan has higher gel breaking strength compared to hybrid carrageenan (from
Sarchotalia crispata), although it has been suggested that the so-called kappa-2
carrageenan has higher milk reactivity (1 5).
Breaking G 1.5 Hz G 1.5 Hz Tan 6

strength G 0.1 Hz G 0.1 Hz (average)
0.2% hybrid 240 27

24 0.123
in milk 207 28
0.2% kappa
in milk
0.6% hybrid
I 73 1 ::v380
1 109
117
40
0.119
27 0.105
in water 388 41
1 1 I
~~~
0.3% kappa 200

0.103
in water 140 lg50
1870 195
Table 2. Large and small deformation rheology of carrageenan gels i-- --ilk and
KCl/CaClt solutions (SMS TA-TX2and Bohlin VOR).
Hydrocolloih in Real Food Systems 205
5 CARRAGEENAN - PROTEIN INTERACTIONS
If milk protein is the component in milk that is responsible for carrageenans milk
reactivity, then the next question obviously is: Which of the many components in milk
proteins is the most important one? The following table compares some (commercial)
milk proteins from different sources:
Breaking strength in Breaking strength in

0.25% KCl 50% milk
Protein (5% pure protein) (3% pure protein)
g/cm2 g/cm2
Acid casein 35 22
Na caseinate No gel 20
Ca caseinate 10 37
Whey protein
12 12
isolate
Control No gel No gel
I Milk I 29 I 29 I
Table 3. Effect of dgeerent milk proteins on carrageenan gelation.
The fact that soluble caseinate does not increase carrageenan gelling suggests that the
proteins either serve as fillers or to adsorb carrageenan. Insoluble Ca caseinate shows low
gel strength in water, but in diluted milk it can restore milk reactivity. The explanation is
that in KCl solution the protein precipitated before onset of carrageenan gelation. Whey
protein gives a slight increase in gel strength, but far less than caseins. The results suggest
that the casein micelles in milk are responsible for carrageenans milk reactivity. The fact
that carrageenan milk gel strengths are reduced to half at increased milk pH (7.0) indicates
that intact micelles are required. It has been shown that compatibility of locust bean gum
and milk increases fkom pH 6.6 to pH 6.9 and this suggests a decreased physical effect of
casein micelles at around pH = 7: the micelles partly disintegrate. Addition of CaC12 to
milk in amounts that precipitates the casein reduces carrageenan gel strength.
There seems to be a critical minimum concentration for the micelles in order to show
effect on carrageenan gelation (figure 3). Milk can be diluted to about 75% without much
loss of gel strength; on the other hand the presence of extra casein micelles does not
increase gel strength much.
The results shown here indicate that the typical micellar structure of the milk proteins is
important for carrageenan gelling of milk.
It is interesting to note that in gels made by a cold-filled type of process (stirring while
cooling) milk proteins have comparable effects as shown for regular carrageenan-milk
gels: they result in increased viscosity compared to controls without protein.
206 Gutns and Stabilisers for the Food Industry I I
Milk gel strength vs. milk concentration
" I
0 50 100 150 200 250

Milk concentration
Figure 3. Casein micelle concentration and 0.25 % hybrid carrageenan milk gelling.
(Milk was diluted in 0.3% KC1 I 0.2% CaC12. Concentration was done by dialyzing concentrated milk
against normal milk - 100 % = pure skimmed milk. The gel strength figures represent breaking strength
expressed as g/cm2 measured by SMS TA-TXZ).
6 CARRAGEENAN-CARRAGEENAN INTERACTIONS IN MILK
Figure 1 showed that milk reactivity can be characterised by low minimum concentration
required for gelation, and by a certain plateau level for the gel strength. Table 3 shows
some typical values of these parameters for the basic carrageenan types.
Minimum concentration 'Plateau'

for milk gelling* breaking strength
Iota 0.15% 25 g/cm2
Kappa 0.05% 120 g/cm2
50% iota / 50% kappa 0.05% 120 g/cm2
Hybrid 0.07% 80 g/cm2
Hybrid in 0.25% KCl 0.35%
Table 4. Comparison of carrageenan types in milk gels.
* To obtain gels which do not deform more then 10% when unsustained.
All types of carrageenan show milk reactivity, but kappa carrageenan gives highest gel
strengths. Blends of kappa carrageenan and iota carrageenan show a remarkable
synergism as one would hardly expect the iota part to contribute. In hybrid carrageenans
extracted from different seaweeds using the same procedure (modification and extraction
in 5% KC1/5% KOH at 90C for 1 hour, followed by neutralisation, filtration and
isopropanol precipitation) there is a positive correlation between the amount of kappa
structures and the milk gel strength (table 5).
Many dairy desserts are made in a process where filling of the product into the final
container takes place after cooling. In these processes gel formation is hindered by the
shear applied during cooling and filling. Partial re-setting of the gels determines many of
the cold-filled products textural characteristics, and gel strengths in those products
generally decrease with increasing kappdiota ratio. Also the effect of molecular weight on
gel strength and the interference of other ingredients with carrageenan gelling (see below)
results in less clear relationships of gel strengths and kappdiota ratio in carrageenans used
in dairy desserts. Nevertheless, milk gel strength at around 0.2% carrageenan
concentration and water gel strength (at higher concentrations giving the same gel
strengths) shows the same dependency on carrageenan basic type.
Carrageenan source Kappdiota ratio* Milk gel breaking

(WW strength at 0.2%
Chondrus crispus 0.35 35 g/cm2
Sarchotalia crispata 0.60 45 g/cm2
Gigartina chamissoi 0.95 85 g/cm2
Chondrus canaliculata 1.75 105 g/cm2
Kappaphycus alvarezii k 20 110 g/cm2
Table 5. Milk gel strength and proportion of kappadimeric structures in hybrid
carrageenan.
* Determined by FT-IR.
7 INTERACTIONS OF CARRAGEENAN AND OTHER INGREDIENTS IN

DAIRY DESSERT GELS.
Most dairy dessert gels contain starch - mainly to improve mouthfeel and to increase
nutritional value. Some products (mainly cook-up powders) still contains only starch as
thickening or gelling agent, but most liquid dairy desserts use a combination of
carrageenan and - frequently non-gelling waxy maize starch as gelling system.
Carrageenan has the advantage of a low processing viscosity, reducing negative effects of
too high starch levels on flavour and mouthfeel, and adding new textures to dairy desserts.
Mouthfeel of carrageendstarch milk gels is different from pure carrageenan milk gels
(less brittle, less gelled) even if breaking strength and distance-at-breaking as measured by
large-deformation rheology do no differ much.
The next table shows that starch has unexpected interactions with carrageenan:
10.2%hybrid I 41 I 45
~~~
I 45 1 44
0.2% kappa 100 84 75 89
0.7% iota 38 50 49 49
~~~ ~~~~~
Table 6. Effect of addition of 2% starch on carrageenan milk gel breaking strength

(measured by S M S TA-TX2).
Kappa carrageenan
0.2% in milk
Wm2)
60 1
0.5% in milk
(g/cm2)
250
0.5% in
0.3 % KCl
(!dcm2)
230
1
Kappa carrageenan I
50 260 385
LBG 65 135
Table 7. Breaking strength of gels from carrageenan and carrageenan/LBG in milk and
KCI.
In 0.5% milk, there is not much difference between gel strength in milk compared to a
KC1 solution, suggesting that the gel is a pure carrageenan gel and not a
caseidcarrageenan coupled network; nevertheless there is no synergism with LBG.
Modifying the texture and reducing syneresis may justify the use of LRG in dairy
desserts. Perhaps the most difficult interaction effect to explain is the well-known
antagonism of cocoa powder and carrageenan milk gelling:
Milk + 2% starch + 3% cocoa

Wm2) (g/cm2) (dcm 1
Powder Both
(dcm3
0.15% kappa 60 43 26 20
0.25% hybrid 52 50 22
0.7% iota 38 42
-
The interaction effects with starch and cocoa shown here have never been reported in
scientific literature, although they are well-known in industry. It may be worthwhile
studying the effects in more detail.
8 CARRAGEENAN IN ACIDIFIED MILK
In acidified milk, carrageenan does not show milk reactivity, which - again - suggests that
the intact micellar system in milk is important for milk gelling with carrageenans, because
gelling of carrageenan is not pH dependent (apart from the hydrolysis that may occur on
heating the relatively acid labile carrageenan at low pH). But of course the opposite
charge of carrageenan and milk proteins causes complex coacervation to occur. In
yoghurt, flocculation resulting in graininess and syneresis increases from hybrid - through
kappa - to iota carrageenan. It does not completely follow charge density, perhaps related
to kappa carrageenan's higher gelling capacity. Hybrid carrageenans show the least
negative effects in yoghurt. The dependency of the viscosity on the concentration of
carrageenan in acidified milk drinks shows an interesting curve (figure 4) that can be
interpreted as follows: at low concentrations viscosity increases because of neutralisation,
but increasing carrageenan concentration results in charge reversal. This does not lead to
stabilisation as in the case of e.g. pectin, because carrageenan starts to gel at these
concentrations and this results in bridging flocculation. The type of curve as shown in
figure 4 is not specific for carrageenan, but can be observed for CMC, PGA and pectin as
well (16). The difference between kappa carrageenan and pectin is that carrageenan can
form a gel in yoghurt and pectin (at least the types used for stabilisation of acidified milk)
cannot.
Brookfield viscosity in acidified milk drink
0 0.1 0.2 0.3 0.4 0.5

Conc. %
Figure 4. Kappa carrageenan concentration vs. viscosity in acidified

milk drink. (Using lactic acid bacteria acidified 50% milk added 10% sugar).
9 CONCLUSION
The focus in carrageenan food-physical literature is on pure kappa and iota carrageenan.
Many dairy desserts contain carrageenans of the hybrid type, and these carrageenans
deserve more attention from the scientific community. Most interaction effects of
carrageenans in dairy desserts (with starch, locust bean gum, cocoa powder) still go
unexplained.
Although much progress has been made in explaining the milk reactivity of
carrageenans over the last few years, there is still no generally accepted theory. The data
presented here could be explained by the following hypothesis: on cooling a carrageenan
solution in milk carrageenan molecules adopt ordered conformations resulting in strands
of carrageenan helices. The kappa casein molecules on the outside of the micelles adsorb
to the strands resulting in a coupled network forming a gel with junction zones of
carrageendmicelle, and carrageendcarrageenan. In this model no micelle/micelle
bonds as in flocculated systems exist. Microscopic techniques could perhaps show
whether or not aggregated micelles occur in carrageenan milk gels.
References
1. T.H.M. Snoeren. Ph.D. Thesis Agricultural University Wageningen, 1976.

2. J.-L. Doublier, C. Garnier, D. Renard and C. Sanchez, Current Opinion Colloid
Interface Sci., 2000,5,202.
3. V. Langendorff, G. Cuvelier, C. Michon, B. Launay, A. Parker and C.G. de Kruif,
Food Hydrocolloids, 2000,14,273.
4. E. Dickinson, Trends Food Sci. & Technol., 1998,9,347.
5 . A. Tziboula and D. S . Horne, Int.Dairy J., 1999,9,359.
6. D.D. Drohan, A. Tziboula, D. McNulty and D.S. Horne, Food Hydrocolloids, 1997,
11,101.
7. C. Schorsch, M.G. Jones and I.T. Norton, Food Hydrocolloids, 2000, 14,347.
8. G.O. Phillips and P.A.Williams (eds), Handbook of Hydrocolloids, Woodhead
Publ., 2000.
9. L. Piculell, Current Opinion Colloid Interface Sci., 1998,3,643.
10. J. de Vies, in J.P. Roozen, F.M. Rombouts and A.G.J. Voragen, Food Science:
basic research for technological progress, Pudoc Wageningen, Holland, 1989, 179.
11. S.H. Knutsen. Ph.D. Thesis Univ. of Trondheim, 1992,23.
12. S. Bourriot, C.Garnier and J.-L. Doublier, Carbohyd.Polyrners, 1999,40, 145.
13. T.H.M. Snoeren, Miss P. Both and D.G. Schmidt, Neth. Milk Dairy J., 1976, 30,
132.
14. B. Egelandsdal, K. Freitham and 0.Harbitz, JSci.Food Agric., 37, 944.
15. S. Boumot, 5thInt. Hydrocolloid Conf., Raleigh USA September 10-15,2000.
16. P.E. Glahn and C. Rolin, Conference Proceedings FIE 1994,252.
EMULSIONS AS INGREDIENTS IN FOODS - SURFACE COMPOSITIONS AND
THEIR MODIFICATIONS
Douglas G. Dalgleish
Danone Vitapole, 15 avenue GalilCe, 92350 Le Plessis Robinson, France*
1 INTRODUCTION
If we take a somewhat over-simplified view of oil-in-water food emulsions, we may define

two broad classes, depending on their end uses. The first type of emulsion is intended to
be used as it is. It is produced fiom a set of ingredients, packaged and sold. Simple
emulsions such as coffee creamers are of this type. These "ready-to-use" emulsions need
not be simple in composition; for example mayonnaises are emulsified by a highly
complex mixture of proteins and phospholipids. However, the demands on these types of
emulsion (leaving aside organoleptic attributes) are mainly those of stability, during the
processing they need to achieve and maintain sterility, and during their subsequent shelf-
life. This type of emulsion is by these simple requirements relatively easy to produce.
The second broad category contains emulsions that are used as ingredients in foods.
As the tendency to reduce the quantity of fats in food products becomes more important, it
is necessary to "functionalise" the fat or oil which is left in the formulation, so that it plays
its part in the generation of texture or organoleptic qualities of the food. Because of the
increased function demanded of them, these emulsions may be required not only to
maintain stability in the presence of other food constituents but also to participate in
forming structures. They may be required to destabilize in a controlled way so as to
contribute to the structure of the product. The classic case of this type of secondary use of
emulsions is that of ice cream, where the original "homogenized cream'' emulsion is
destabilized by interaction with the air bubbles whipped into the mixture. In ingredient
emulsions, long-term stability may not be essential, but the destabilization must be closely
controlled. In contrast, we may also consider how emulsified fat contributes to the
structures and textural properties of products such as yoghourts, in which the emulsion has
to remain stable, although it is incorporated into the structure of a food product.
2 THE EMULSION SURFACE
To at least a first approximation, the behaviour of an oil-in-water emulsion as it is

incorporated into a more complex food is expected to be governed by the material that is
adsorbed at the oil-water interface.' By changing the amount and nature of this material,
the interactions between individual emulsion droplets can in principle be modified, and in
*Presentaddress: University of Guelph, Guelph, Canada
addition the interfacial rheology can be changed? Conceivably, even the way in which
molecules are arranged at the interface may play a role in the stability and interactions of
emulsion droplets, although, few general rules have been developed to describe this.
Emulsions are used as ingredients because of certain functional properties, mainly
that they help to create the structure of a food, and that they produce desirable organoleptic
effects (these latter effects are outside the scope of this paper). The interactions between
the (surfaces of) the emulsion droplets and the matrix of the food (i.e., the proteins,
polysaccharides, inorganic ions and other small molecules contained in it) may be critically
important for the definition of the properties of that food. That is, the emulsion droplets
become part of the matrix-forming components of the food. This type of behaviour is
made clear by the importance of homogenization on the properties of different casein gels.
Homogenization of milk causes some of the casein micelles to be spread around the
surfaces of droplets of milk fat, which are in turn created from the break up of the milk fat
globules of the original milk. In the manufacture of cheddar cheese, the quality of the
rennet gel is adversely affected if the cheese milk is homogenized. That is, somehow the
homogenized fat globules, which interact with the casein gel, disrupt the rennet curd in a
way that native milk fat globules do not, even though the latter do not interact directly with
the casein gel. Exactly in an opposite manner, the homogenization of the fat in a yoghourt
mix is considered essential for the successful formation of desirable texture. Evidently in
this second case, the properties of the curd are enhanced by the interactions between the
casein matrix and the fat. Exactly why these rather contradictory results should be
obtained is not clear. It appears that a strong interaction between fat globules and the
protein matrix is favourable in one case but not in the other. It must be pointed out,
however, that the structures of the two gels and the properties required of them are quite
different, since the rennet-induced gel is expected to undergo syneresis and wheying-off, in
contrast to the acid gel, where the aim is to reduce syneresis to a minimum and encourage
water-holding capacity.
It is possible to control the composition of the interfacial layers of oil-in-water
emulsions by performing the homogenization under different conditions. The interfacial
composition of the emulsion is mainly defined by the composition and the concentration of
the emulsifier mix (mixes of different proteins and small-molecule ern~lsifiers).~-~ Other
factors such as temperature and homogenization pressure may have an effect, but these
variables are generally more effective in determining the overall efficiency of the
homogenization step. Thus, by controlling the amounts of the different emulsifying
materials, we may control the size and the interfacial composition of the emulsion droplets.
We can obtain a fairly good idea of the overall composition of an emulsion interface by
directly measuring the protein adsorbed! We may also obtain an idea of the gross
geometry of adsorbed protein layers (for example, their thickness or how far they extend
from the emulsion interface).697The composition of the interfacial material can be in some
cases controlled by subsequent addition of surfactant material, and certain rules appear to
apply. For example, it is in general difficult to displace one protein by another (but see
below), but it is generally possible to replace adsorbed proteins by water-soluble small
molecule emulsifiers. On the other hand, it is difficult to add oil-soluble surfactants after
the formation of an emulsion, and so they cannot be used to change the interfacial
composition at that stage. However, even these rules do not always apply; for example,
emulsions made with whey proteins and then aged tend to form networks of adsorbed
proteins which are difficult to displace.
It is not at all easy to draw a direct line between the interfacial composition and
structure and the functional behaviour of an emulsion. Especially, this is the case when the
interfacial material is complex, as is generally the case in real food emulsions, rather than
the models used for laboratory study. In this respect, all food emulsions stabilized by
proteins can be considered as complex, because even the simplest industrially-used
proteins (e.g., sodium caseinates, whey protein concentrates and isolates, soy protein
concentrates) are mixtures of several types of protein molecules. In addition, even these
"simple'' proteins are likely to have undergone a certain degree of heat treatment during
their isolation, and therefore may be partly denatured. As a consequence, it is not always
certain that the emulsions made using industrial proteins will behave in exactly the same
way as model laboratory emulsions, which tend to be prepared under stringent conditions
from fairly carefully purified and native proteins.
One of the possibly important aspects of interfaces in emulsions containing mixtures
of proteins, and one that we know relatively little about, is how the particular surfactant
molecules are distributed. For example, it has recently been shown that the displacement
of proteins by Tween fiom interfaces does not yield a homogeneous structure of the mixed
interface.* The presence of the small molecule surfactant causes changes in the
conformation of the adsorbed protein, as well as the creation of protein-rich and surfactant-
rich regions of the interface. In effect, a 2-dimensional phase separation occurs on the
planar interface. It is probable that these results can be extrapolated to the surfaces of
droplets of emulsions containing proteins and small-molecule surfactants, where in most
cases the local curvature is not large on the scale of molecular dimensions. Some time ago,
it was suggested on the basis of the measured thicknesses of adsorbed protein layers in
emulsions that there were at least two effects of Tween during the displacement of caseins
fiom an oil-water interface, although at that time it was not possible to define specifically
what these effects were.g If this type of phase separation occurs in emulsion interfaces, it
may be possible to create or identify "hot spots" on the surfaces of the emulsion droplets
via which the interactions of the droplets with their environment can be controlled.
Moreover, if the phase-separated protein adhering to the interface has a modified structure,
this may also change its reactivity with its surroundings.
If there are likely to be phase-separation effects on the surfaces of emulsion droplets
coated with mixtures of protein and small-molecule surfactant, we may also speculate
whether mixtures of proteins would in their own way phase-separate when they adsorbed.
For instance, could the four proteins of which sodium caseinate is composed form
differentiated patches on the oil-water interface? It appears that in emulsions made using
sodium caseinate the proteins compete only weakly among each other for the interface," in
contrast to purified a,l and p-caseins, which compete strongly." This suggests that the as2
and/or K-caseins may have a function in maintaining the structure of the caseinate
complex, both on and off the interface. In solution, the proteins of sodium caseinate form
aggregates containing some tens of proteins;" so it is evident that these aggregates must
dissociate and spread around the oil-water interface during homogenization, since they
appear to form a monolayer of protein, as evidenced by electron microscopy and other
measurements of the surfaces of the droplets in the emulsions,13'l4 as long as the amounts
of caseinate are not too high. That is, the adsorption creates the opportunity for the
proteins to separate one from another, Whatever may be the outcome of this adsorption
and spreading, we should at least be aware of the possibility that emulsion droplets with
patches of material at their surfaces may not interact in the same way as those with more
homogeneously-covered interfaces. The situation will increase in complexity with the
number of available proteins, and will also depend on the natures of the proteins involved.
Perhaps the ultimate in heterogeneity for an "ingredient" type of emulsion is fat or oil
homogenized with milk. In this case, there are fat droplets of varied sizes, whose
interfaces are covered with variably-sized fragments of casein micelles. Small amounts of
2 14 G u m and Stabilisers for the Food Industry I1
whey protein may also be present. But in these particles, the possibility of "hot" spots, of
casein-rich and casein-poor regions of the interface, may be important. This complexity of
the material is enhanced by heat treatment of the milk before homogenization.
3 PATH DEPENDENCE AND THE MODIFICATION OF THE INTERFACE

DURING PROCESSING
This raises an important point on the path-dependence of processes. It has been

demonstrated that the surfaces of particles in homogenized milk have different
compositions and structures depending on whether the milk was heated before or after
homogenization.l 5 Qualitatively, we can explain this. Heating the milk before
homogenization causes the denaturation of the whey proteins and their binding to the
casein micelles. Such whey proteins as are not bound to the casein micelles aggregate in
solution. Thus, when the milk is homogenized, the whey proteins end up adsorbed to the
fat droplets because they form part of the casein micelles. In addition, since larger
emulsifiers are favoured during homogenization, the whey protein aggregates in solution
will also be more readily brought to the interface compared with unaggregated whey
proteins. When the milk is homogenized before heating, basically only the casein micelles
are adsorbed, and subsequent heating attaches some of the whey proteins to these adsorbed
micelles. However, the whey protein will also bind to non-adsorbed casein micelles as
well as aggregating in solution. The net result is that there is a path-dependence of the
detailed compositional properties of the emulsion.
The effects of this have not been investigated in detail. However, industrial practice
does show that in concentrated milks the stability does depend on the relative order of
heating and homogenization in the process (although here we are talking about
forewarming of unconcentrated milk rather than rather than heat treatment of the
concnetrate).
In the case of homogenized milks, the denaturation of the whey proteins causes them
to bind to the fat globules that are already emulsified by casein micellar fragments. The
net effect of this is to increase the protein Ioad on the interface. A more complex
phenomenon, and one in which path-dependence is very manifest, is the heat-induced
displacement of caseins b whey proteins from the surfaces of emulsion droplets prepared
using sodium caseinate." Experiments at room temperature have shown that there is
generally little exchange between adsorbed caseins and dispersed whey roteins on the one
hand and adsorbed whey proteins and dispersed caseins on the other.' If anything, the
balance favours the adsorption of caseins at neutral pH. However, studies of emulsions
stabilized by whey proteins which were then subjected to heating in the presence of sodium
caseinate, showed that no replacement of the adsorbed whey proteins occurred. In
addition, it was not possible to detect any reaction between adsorbed whey protein and the
cysteine-containing caseins, such as would normally be expected at elevated temperatures.
In summary, heating of whey-protein stabilized emulsions in the presence of caseinate had
no visible effect (Figure 1).
Emulsions that were prepared using sodium caseinate and then treated with whey
proteins did not show exchange between the interfacial and dispersed proteins at room
temperature. However, when the emulsions were heated, it was seen that much of the
casein was displaced by adsorbing whey proteins (Figure 1). Even at relatively low
temperatures (40-50C) substantial amounts of caseins were displaced by the whey
proteins, which suggests that the whey proteins do not need to be denatured (in the
traditional sense) to participate in the displacement reaction.l8 The displacement reaction
.
(0.75%)
Oil (20%) +
caseinate (0.75%)
Homogenize
Heat (SOOC, 2 min)

I
Caseinate-stabilized
Addition of WPI
emulsion
No displacement of
adsorbed casein by WP
I
I
1 Homogenize
Whey protein-stabilized
emu1sion
adsorbed W by casein
I
Addition of caseinate
No displacement of
(0.75%)
Heat (SOOC, 2 min)
No displacement of
adsorbed WP by casein
Figure 1. Path-dependence of surface concentrations of proteins in emulsions of the same

composition (20% oil, 0.75% sodium caseinate, 0.75% whey protein isolate (WPI), but
which differ in the order of addition of the constituents.
appears to be rather specific, because a , ~

215
and p-caseins were displaced and P-lactoglobulin

and a-lactalbumin became adsorbed. The as2and K-caseins remained on the interface and
appeared to be unaffected by the exchanges of the other proteins. In the particular case
being studied, it could also be seen that the two whey proteins a-lactalbumin and p-
lactoglobulin played very different roles in the displacement reaction, with the latter
protein playing the more active function and the former simply acting more as a "filler" of
the interfacial layer. The reaction was also seen to be independent of ionic strength but
dependent to an extent on pH. Although the caseins which remain on the surfaces of the
emulsion droplets are those which contain cysteine, and are therefore capable of
participating in the formation of disulphide bonds with the whey proteins, this does not
seem to be an important factor in the displacement reaction.
Whatever may be the precise mechanism of the displacement reaction, it is further
evidence that emulsions may completely change their composition in the course of being
processed. Not only is it small molecule emulsifiers which can cause a change in the
properties of emulsions during processing, but proteins may also play a role. It is also
clear from these kinds of experiment that the production and modification of a food
emulsion is a one-way process; there is no possibility for equilibria between adsorbed and
non-adsorbed proteins to exist, and so any modifications of hctionality are irreversible.
4 CONCLUSIONS
We wish to define and optimize the functionality of the emulsions that we use in foods. It
seems clear that we need to establish the changes that take place in real emulsions (i.e., in
complex food mixtures) during their use as ingredients, because they may exhibit
synergistic effects that are not evident from the study of simpler model systems. It is
possible to some extent to explain the differences in behaviour of emulsions stabilized by
purified caseins,'. l9 but it is difficult to extrapolate the general approach to the behaviours
of emulsions made with mixtures of proteins, especially those containing whey proteins.
It is possible at least in principle to control the textures of foods containing
emulsions by specifically controlling the properties of the emulsions. This may mean the
definition not only of the compositions but also the structures of their interfaces. We will
then need to ask what types of new model systems need to be defined for this purpose.
Clearly, such systems need to be more complex than the models that have been used
hitherto. There seem to be several challenges before us: first, to define in more detail the
interfacial structures of emulsion droplets containing mixtures of emulsifiers; second, to
demonstrate that changing the structure does modify the functional properties of the
droplets; third, how to control the properties of real systems by or during processing. In
addition, we need to develop more complex methodologies needed to look at such real
systems. This means the harnessing of the ever more sophisticated analytical techniques
used in other sciences to the study of food emulsions, at appropriate concentrations, in real
food matrices.
References
1 E. Dickinson, Colloids Surf B, 200 1,20, 197
2 E. Dickinson, S.E. Rolfe and D.G. Dalgleish, Int. J. Biol. Macromol., 1990,12, 189.
3 E. Dickinson, S.E. Rolfe and D.G. Dalgleish, Food Hydrocolloids, 1989,3, 193.
4 J.A. Hunt and D.G. Dalgleish, Food Hydrocolloids, 1994,8, 175.
5 E. Dickinson and S. Tanai, J Agric. Food Chem., 1992,40, 179.
6 Y. Fang and D.G. Dalgleish, J Colloid Interface Sci., 1993,156, 329.
7 E. Dickinson, D.S. Horne, J. Phipps and R. Richardson, Langmuir, 1993,9,242.
8 A.R. Mackie, A.P. Gunning, P.J. Wilde and V.J. Morris, J Colloid. Interface Sci., 1999,
210, 157.
9 D.G. Dalgleish, M. Srinivasan and H. Singh, J Agric. Food Chem., 1995,43,2351.
10 E.W. Robson and D.G. Dalgleish, J Food Sci., 1987,52, 1694.
1 1 E. Dickinson, S.E. Rolfe and D.G. Dalgleish, Food Hydrocolloids, 1988,2,397.
12 H.S. Rollema, in Advanced Dairy Chemistry - 1 - Milk Proteins: Molecular,
Physicochemical and Biological Aspects, ed. P.F. Fox, Elsevier Applied Science,
Barking, p. 1 1 1 .
13 D.G. Dalgleish, S.J. West and F.R. Hallett, Colloids Surf. A, 1997,123 145.
14 F.A.M. Leermakers, P.J. Atkinson, E. Dickinson and D.S. Horne, J. Colloid Interface
Sci.,1996, 178,681.
15 S.K. Sharma and D.G. Dalgleish, J. Dairy Res., 1994,61,375.
16 J.M. Brun and D.G. Dalgleish, Int. Dairy J., 1999, 9, 323.
17 J.A. Hunt and D.G. Dalgleish, Food Hydrocolloids, 1996, 10, 159.
18 D.G. Dalgleish, H.D. Goff, J.M. Brun and B. Luan, Unpublished results; submitted for
publication
19 E. Dickinson, D.S. Horne, V.J. Pinfield and F.A.M. Leermakers, J Chem. SOC.
Faraday Trans., 1997, 93,425.
INFLUENCE OF HYDROCOLLOIDS ON FLAVOUR RELEASE AND SENSORY-
INSTRUMENTAL CORRELATIONS
Ross Clark
CP Kelco, 8225 Aero Drive, San Diego, CA 92 123 United States
1 INTRODUCTION
Hydrocolloids of various types have long been used to modify the texture of foods. It is
now being realized that these ingredients can also have a significant effect on the flavour
and aroma of the food as well.
Sensory evaluation is commonly used to quantify the texture and flavour effects of
hydrocolloids and other food ingredients. Correlation of food sensory attributes with
instrumental measurements of flavour, aroma, or texture is useful to speed product
development and provide a more basic understanding of factors controlling the sensory
properties.
Taking these two ideas together, it is now possible to understand how to better quantifL
and understand the changes that occur when hydrocolloids are added to foods. Flavour
and aroma changes may be due to simple viscosity increases, adsorption of the flavours
compounds, changes in diffusion rate, or surface active properties of the hydrocolloid.
2 PREVIOUS WORK DONE
2.1 Historical
In the area of sensory / instrumental correlation, one of the most significant contributions
occurred in 1963 with the introduction of the Texture Profile Method by Dr. Szczesniaks
group at General Foods (1). This method provided for a multidimensional rating of food
texture. Furthermore, it established a link between sensory properties and instrumental
measurements. The value of this contribution is proven by its wide application in our
industry today, nearly 40 years after its development.
A little later, Wood2 conducted studies to determine what range of instrumentally

measured shear rates gave the best correlation with senso perception of viscosity during
7
swallowing. With the foods used, he concluded that 50s was a reasonable figure. Still
later, Shama and Sherman3refined this early data to conclude that low viscosity foods
tend to be perceived under a more or less constant shear stress of 100 dynes/cm*. Thick
foods, on the other hand, showed the best correlation when an instrumental shear rate of
10 5.' was used. When the limited viscosity range of Wood's original samples is taken
into account, these two studies corroborate one another.
Among the first efforts to try and link the viscosity of food samples thickened with
hydrocolloids and with their basic taste intensities was study published by Pangborn et a14.
The authors concluded that low levels of gums did change the perception of basic
flavours, They found that the change in perception was more dependent upon the gum
used that the level of the gum (within reasonable limits). This provided an early
indication that hydrocolloids change more than the texture of the foods to which they are
added. Higher gum levels suppressed sourness of citric acid and bitterness of caffeine.
The sweetness of saccharin was increased somewhat by addition of gums.
2.2 Current work
Adsorption of flavour compounds on surfaces such as starch or MCC (microcrystalline

cellulose) is one possible form of hydrocolloid / flavour interaction. Work by Boutboul'
et al. has shown that starch can selectively adsorb and release flavour compounds. Inverse
phase gas chromatography was used to look for adsorption in starch columns controlled to
various relative humidity values. They concluded the relative adsorption by functional
group was: ester < ketone < aldehyde < 2" alcohol < 1" alcohol. Adsorption also increased
as the of total number carbon atoms in the chain increased. At higher water contents,
adsorption of water-soluble components was stronger. Linear materials were adsorbed
less than cyclic ones and single bond containing materials adsorbed less than similar ones
with a double bond. The result is clear; in a complex mixture of food flavours, starch will
interact more strongly with some components more than others and change the overall
flavour balance.
Starch is not the only material capable of changing flavours. In the work by Braudo and
co-workers6, pectic materials were shown to interact through hydrophobic interactions and
hydrogen bonding. By using the technique of equilibrium dialysis, they were able to
discern how various flavour compounds would partition in solutions of high or low
methoxyl pectin. Low methoxyl pectin is postulated to adsorb normal aliphatic 2-ketones
through van de Waals interaction with hydrophobic regions in the pectin molecule. In
acid media, heterocyclic flavour compounds are thought to hydrogen bond directly to the
undissociated carboxyl groups of the LM pectin. There is an optimal pectin concentration
for interaction of about 0.6% by weight. At higher gum levels, pectin-to-pectin
interactions began to dominate instead of pectin-to-flavour interactions. Lower pectin
levels did not provide as much reactive surface and so showed less interaction.
Dufour and Bayonove' found the volatility of aroma substances in red wine was modified
by the hydrocolloids present naturally in the wine. Polysaccharides can occur at levels
from 500 to 1500 mg/l in red wine. These relatively low molecular weight materials can
be divided into two broad categories. The grapes contribute, arabinoglactans (AG) and
arabinoglactan protein (AGP). During fermentation, organisms like Saccharomyces
cervisiae can produce mannoproteins. The authors concluded that protein rich
hydrocolloids led to adsorption of esters. On the other hand, uronic acid rich gums tended
to salt out esters. Increasing ethanol levels disrupt hydrophobic gum interactions by
modification of solvation state. On the other hand, pH had little effect on adsorption.
Diffusion of flavour compounds through a gel or liquid can influence the sensory
properties, especially the temporal aspects or impact of a flavour. In a recent
publication by Bayarri and co-workers this effect is detailed using gels made with LA
gellan gum and K-carrageenan. Both aspartame and sucrose were used as model
sweeteners. Diffusion rate was measured at 37OC. The soft, nearly melted, K-carrageenan
gels showed faster diffusion rate than LA gellan gum. In gellan, diffusion was nearly
independent of gum concentration. On the other hand, higher levels of rc-carrageenan
slowed diffision significantly. This shows once again how important the property of
melting in the mouth is when predicting the overall flavour properties of a gelled
system. Despite the negative health issues being associated with gelatine, it is exceptional
in this regard. As might be expected based upon molecular size, aspartame showed faster
diffusion rate than sucrose.
Finally, the topic of the optimal shear rate for instrumental correlation with human
perception continues to attract creative researchers. Houska and co-workers established
proposed relationships with shear
rate and five sensory attributes
related to thickness. These
parameters were; mixing with a
spoon, pouring, slurping,
swallowing, and pressing between
the tongue and palate. In each
case, the relationship was
established using a power-law
model. The goodness of fit was
reasonable and the relationships
Viscosity (Pas) established agreed in broad terms
with previously accepted work.
Figure 1 Postulated shear rates for sensory processes
3 EXPERIMENTALWORK
3.1 Methods and Materials
Quantitative Descriptive Analysis (QDA) was used for the sensory work described
below. The process began with panel screening based upon sensitivity. Twelve panel
members spent two or three sessions developing the language that was used to describe
the samples. They began with what was noticed first (appearance) and ended with the
f d sensation (aftertaste). For each attribute, the panel also provided a working
definition and anchor words for the scales used to measure intensity. In addition to the
sensory language, a testing protocol was developed for evaluating each product; i.e., the
order and method of attribute evaluation were specified.
Along with the sensory data, some type of instrumental analysis was usually conducted.
The intention was to match deformationtype and rate as closely as possible to what occurs
in the mouth. Liquids were measured in steady shear from 0.1 to 3,000 s-. There is some
debate about the correct temperature to use. Our investigations used 22 to 23OC as a
compromise and for simplicity. In most cases, the measured viscosity at a higher or lower
temperature was well correlated with the data at room temperature so that the ranking of
the samples would remain unchanged.
For solid materials, TPA was the technique used. Free-standing samples were preferred
over ones in a container (where possible). Uncontrolled effects due to the container were
then eliminated. Compression strain level were chosen carefully. Too much and the
second cycle terms of cohesiveness and springiness are poorly measured. Too little
and the primary structure of the sample may not be broken down and measured well. For
most foods, the optimal range lies between 55 and 75% when expressed as simple strain.
A problem remains with deformation rates in PA. Ideally, these would match those in
the mouth. Most instruments cannot take data at a high enough fkequency to make this
possible. Rapid compression will result in poor detection of peaks and insufficient data to
accurately calculate the TPA parameters. Improvements in instrumentation are needed.
Traditional small strain rheological tests such as oscillatory measurements and creep /
relaxation tests do not correlate with perception in the mouth. Large strains that destroy
the sample integrity are needed to correlate with eating a food. Lately, extensional
viscosity measurements show some promise of good correlation with human perception.
This is especially true of parameter such as stringiness and short or long flow.
Investigations of flavour and aroma with electronic nose technology hold promise.
Statistical analysis was the final analysis step. The panel members were checked for
consistency using ANOVA. Simple correlation tables were built and multiple regression
was tested on some likely relations. Principle Component Analysis or PCA is a very
usefit1 technique to employ for sensory studies. PCA groups the sensory and instrumental
measures into similar factors. Products with similar characteristics can also be grouped.
In this way, the underlying and most basic measures of the product are revealed. In most
cases 4 -8 factors suffice to explain more than 90% of the sample variability.
Taken together, QDA, good instrumental analysis and a thorough statistical analysis
provide a great deal of insight into food products. Some of the products that have been
examined in this way are described in the sections that follow.
3.2 Lemon flavoured beverage

The base for this product was a commercial aspartame sweetened lemonade product
containing no gums. Three gums (CMC [cellulose gum], xanthan, and PGA [propylene
glycol alginate]) were added at three levels to this product. Steady shear viscosity and
viscoelastic properties were measured at 23OC.
Thirty one parameters were used to describe the products for appearance, mouthfeel,
aroma, flavour and aftertaste. A good correlation existed with measured viscosity at
100 s- and viscosity as perceived in the mouth, This is shown in Figure 2. The
correlation of 0.966 shows excellent agreement between sensory and instrumental
measurements. The best shear rate for prediction was 100 s, which is higher than some
others have predicted.
Figure 3 shows an interesting interrelationship between the gum used and the perceived
flavour properties. In Figure 3, it can be seen that the overall flavour of all gums except
Hydrocolloids in Real Food Systems 22 1
PGA declines as the gum level increases. Lemon flavour decreases in all cases as the gum
level increases. So why did the PGA overall flavour increase as the gum level went up?
The answer is in the bitter flavour score. As more PGA was used, the bitterness of the
lemonade increased. At high gum levels, the lemonade made with the PGA was more
bitter and less lemon-like than the other products. Clearly, there was a difference between
the gums for how they influence overall flavour of a beverage.
+
- Sensory score = 2.39* Vis$sity 013~~
r*=O. 966
8
v)
a0 ___._
___.__.._.___.______..-..-.
v)
Figure 3 Sensory flavour scores for beverage samples

3.3 Citrus punch beverage
This beverage was similar in properties to a commercial product, Sunny Delight@,sold in

the US by Procter and Gamble. For many years, Sunny Delight has used PGA as a
thickener and stabilizer with the knowledge that this thickener also added a certain bite
to the beverage. This bite agrees closely with the bitterness data in the previous section.
Table 1 below shows how flavour was modified in the presence of different hydrocolloid
systems, formulated to a constant viscosity.
Gum Overall Orange Sweet Bitter
PGA 33.2 28.6 21.Ob 7.2
Xanthan 29.9 25.4b 20.gb 6.7b
Xanthanl guar 30.7 25.5b 19.gb 7.7a
Gelian/ CMC 31.7 28.7a 23.3 5.gb
The overall flavour score was similar for all of the hydrocolloids but there were
statistically significant differences between the gums with respect to orange, sweet,
and bitter flavours. None of the gum systems tested was a perfect match for the
standard of the PGA thickened sample.
3.4 Model dessert gels
45 F
0 0 +
.- 1.3 YOgelatine
40 (meks)
g 35 :
c8
n 30
6
cn t 0.4% LA gellan+
xanthan I LBG
8 25 (cohesive) 0
2o Overall Flavour = 41.91 0 + S.OO76*Hardness
: r 2 4 . 6 4 6 (rh0.875 without mariced points) 0
15 5 . 1 * * . 7 * I * 1 1 * * * a * *
Figure 4 Overall flavour of dessert gels correlated with TPA hardness value
The exceptions are interesting because they reveal something about flavour in gels. In
order to be perceived, flavour must be expressed or physically released from the gel. The
gelatine sample had more flavour that would be expected from its hardness value because
of the well-known property for gelatine to melt at body temperature. The gel made with
gellan gum, xanthan gum, and LBG was the most elastic and cohesive gel. It did not
break down in the mouth and so the flavour remained trapped in the gel.
The dessert gels show that there are many possible correlations between sensory flavour
and instrumental measurements. Some of the correlations may be quite good. However,
there will always be exceptions and these exceptions are very helpful in extending our
understanding of sensory perception.
3.5 Salad dressings
The property of creaminess is desirable in salad dressings and especially hard to obtain
when reducing the fat or oil levels. Manufacturers of these products frequently rely on
PGA to make a product seem creamier. The evidence for this practice is shown in Table 2
below. Levels of PGA as low as 0.3% made a fat free ranch dressing seem as creamy as
the control. Of course, there were other favour aspects of PGA to consider like it
accentuation of sour / tartflavours that may have undesirable consequences.
Sensory Fat free standard No oil control Control +O. 3% PGA Control +O. 6%PGA
Creminess 36.0 28.0 35.7 42.2
Table 2 Sensory creaminess for ranch salad dressing
1.o
0.95 --
%
0.9
--
$ 0.85 --
I
t
0
0.8
0.75
--
--
--
f 0.7
--
0.65
8
- a a : II. . . . . ....L
I . I I . ...
0.55
0.1 1
10 100 1000
Shear Rate (s-l)
Figure 6 Optimal shear rates for sensory viscosity prediction in ranch dressings
The shear rate found in the mouth is predicted to be lower for thicker foods such as
dressings than it is for the beverages discussed in section 3.1. In this study, the panel
members were asked to rate dressing viscosity by pouring from a spoon, stirring, and in
the mouth by pressing between the roof of the mouth and the tongue. Using data from the
mouth evaluation, plots were made of the instrumental viscosity at various shear rates. A
linear regression was calculated and the correlation coefficient, 3, was determined for
each shear rate. The resulting plot is shown in Figure 6.
This figure clearly shows that the correlation between a sensory estimate of viscosity and
the instrumentally measured value increases to shear rates of about 100 s-l and then
subsequently decreases as the shear rate is increased above 100 sel. The industry standard
measurement, a Brookfield RVT at 20 rpm gives a shear rate of roughly 10 s- and so
could not be expected to have a correlation of better than 0.75% or so. Much better results
will be obtained at higher shear rates.
3.6 Sour cream
As is the case with dressings, dairy products such as sour cream are being formulated
today to reduce the fat content. When fat is replaced, water and some material to thicken
the water must be added to replace it. Frequently, cellulose based products such as
Avicel@microcrystalline cellulose are added. In the results below, Avicel was compared
to bacterial cellulose (CLN) in a reduced fat sour cream.
Creamy
Full Fat Conb.ol
CMC/CLN 1:l
40.51
40.35
I CMCICLN 1:l
KGEUAvicel
Thick
40.75
37.87
I KGEUCLN 1:2
KTUAvicel
-ng
23.92
23.85
KGEUCLN 1:l 39.85 KGEUCLN 1:2 36.50 Cellulon 23.02
CMCICLN 1 :2 39.33 KGEUCLN 1: l 36.10 KTUCLN 1:2 22.96
Cellulon (CLN) 38.56 Cellulon 35.73 CMClCLN 1:2 22.86
KTUAvicel 38.29 CMCICLN 1:2 34.60 KGEUAvicel 22.79
KTUCLN 1:2 37.71 KTUCLN 1.2 33.79 CMCICLN 1:1 22.25
KTUCLN 1:l
KGEUAvicel
KGEUCLN 1:2
I KTUAvicel
KTUCLN 1:1
Full Fat CoWro/
33.54
27.33
23.50
KTUCLN 1: l
KGEUCLN 1:1
Full Fat Control
21.89
21.44
20.42
I
Greasy Lumpy Gritty
KTUAvicel 14.64 KGEUAvicel 33.69 KGEUAvicel 10.35
KTUCLN 1:2 13.50 KGEUCLN 1:2 19 54 KGEUCLN 1.2
KGEUAvicel 13.48 KTUAvicel 4.11 Cellulon 5.56
CMCYCLN 1 :1 13.35 KGEUCLN 1:l 3.77 KTUAvicel 4.27
KTUCLN 1:l 12.89 CMCICLN 1:l 3.u KTUCLN 1:2 4.17
CMCICLN 1 :2 12.67 KTUCLN 1 :2 3 .23 CMCICLN 1:2 3.88
KGEUCLN 1:l 11.96 Full Fat Confro1 2.90 KTUCLN 1:1 3.83
Cellulon 11.61 KTUCLN 1:l 2.73 KGEUCLN 1: l 3.25
full Fat ConlW 11.42 CMC/CLN 1:2 2.71 cMcK=LN 1:l 2.85
KGEUCLN 1:2 11.10 I Cellulon 2.62 I Full fat Confroi 250
The full fat control was the most creamy, least thick, least coating, nearly the least greasy,
smooth and not lumpy. None of the combinations had exactly the same properties as the
full fat control but the blend of bacterial cellulose with xanthan gum (KTL/CLN 1:1) came
close. When the xanthan gum was combined with a more traditional cellulose source,
Avicel, the match was not quite as good. This shows that subtle differences between
ingredients that appear insignificant can result in substantial differences in the consumers
reaction to a product.
These data also show how well panel members in a well set up QDA panel could
distinguish sample differences. They were able to distinguish between creamy and
thick. A greasy mouthfeel was judged to be different than a creamy one. The
strength of letting the panel choose their own descriptors and definitions was shown by
these results.
4 CONCLUSIONS AND FUTURE DIRECTIONS
The effect of hydrocolloids in foods is not limited to texture modification. The evidence
is clear that these ingredients play a significant and even key role in the flavour of
products ranging from wine to salad dressing to dessert gels. The mechanisms behind
these interactions are becoming known. Adsorption, diffusion, molecular interactions, and
even precipitation of certain key flavour compounds by hydrocolloids has been shown in
model systems.
Also shown is the fact that when put into real food systems, the influence of the gum on
the aroma, flavour and texture of the food is very evident. Substitution of one gum for
another is not something to be taken lightly. It requires first hand knowledge of how
various gums act individually and in concert to alter all of the sensory properties of foods.
It now seems likely that gums may be added to foods in order to deliberately change the
aroma and flavour as well as the texture.
This understanding would not have been possible were it not for improvements that have
been made in characterisation of foods. This begins with sensory techniques that go far
beyond difference testing and simple hedonic techniques. Screening of panel members,
careful language development and careful testing can yield results as good as a physical
instrument.
Instruments have their role to play in rapid product development, consistent quality
control,and adding to the understanding of what physically and chemically controls how
we perceive foods. In the fbture, development of extensional viscosity methods,
electronic nose and tongue technology along with improved traditional tests such as
compression and steady shear measurements will provide still more insight into how we
perceive our foods and how hydrocolloidsinfluence all aspects of that perception.
References
M. Brandt, E. Skinner and J. Coleman J. Food Science, 1963,28,404
' F. W. Wood, in Rheology and Texture of Foodstuffs, Society Chemishy Industry Monographs, London,.
vol27, 1968, p 40.
F. Shama and P. Sherman, J Texture Studies, 1973,4, p 1 1 1.
R. M. Pangborn, I. M. Trabue and A. S. Szcsesniak, J Texture Studies, 1973,4, p 224.
A. Boutboul, P. Giampaoli, A. Feigenbaum, and V. Ducruet, Food Chemistry, 2000,71, p 387.
E. E. Braudo, I. G. Plashchina, V. V. Kobak, R. V. Golovnya., I. L. Zhuravleva and N. I. Krikunova.,
N a h w g , 2000,44 (3), p 173.
' C. Dufour and C. L. Bayonove, J. Agric. Food Chem, 1999,47, p 671.
* S. Bayarri, I. Rivas, E. Costell, and L. DurSm, Food Hydrocolloidr,2001,15, p 67
M. Houska, H. Vaientova, P. Novotna, J. Strohalm, J. Sestak, and J. Pokorny, J. Texture Studies, 1998, 29,
poz.3Stone,J. Sidel, S . Oliver, A. Woolsey and R. C. Singleton, Food TechZogy, 1974,28 (1 I), p 24.
I' R. Clark, Food Technology, 1997,5 I( I), p 49.
THE YIELD STRESS PHENOMENON AND RELATED ISSUES -
AN INDUSTRIAL VIEW
N.W.G. Young
Dansico Cultor, Edwin Rahrs Vej 38, DK-8220 Brabrand, Denmark.
ABSTRACT
Industrially, the yield stress is an established, product related phenomenon used to describe
structure within a given system. It is used, primarily as a descriptor of product behaviour
and hence, quality under specific circumstances. This paper examines an alternative view
of stability with respect to fruit suspension. The samples have all been subjected to
different shear histories. Flow curves are presented, together with fits for the Carreau
model. The systems stability is expressed qualitatively as a function of the gravitational
stress of the particle and the stress corresponding to the onset of shear thinning.
Quantitatively, the stability is expressed as a Stokes law velocity.
INTRODUCTION
Qualitatively, the yield stress represents a product related phenomenon and is routinely
used to determine specific quality parameters and in predicting product behaviour. Usually
it is used to indicate the stability of a given system with respect to sedimentation. It is
typically defined by saying that a material will not flow until a certain threshold value of
stress, the yield stress, is exceeded (1). Such is the hold of the yield stress that it remains
an industry benchmark parameter to aim for and is generally highly regarded. Classically,
the stability of yoghurt fruit preparations have been described and governed by the yield
stress parameter (2, 3, 4, 5 ) , and was held as standard belief until about 30 years ago.
However, with continued improvements in rheometer instrumentation, (6, 7) evidence
which counters the yield stress phenomenon is appearing.
Barnes (8) recently published an extensive review on the yield stress, quoting numerous
definitions and concluded them to be confused. He simplistically suggested the yield stress
of a solid, essentially being the point where increasing applied stress the solid first
shows liquid like behaviour, i.e. continual deformation. The concept then, suggests that
no flow is possible below a certain critical shear stress and leads to the assumption that
here viscosity tends towards infinity. However, careful measurement below so-called yield
stresses have shown previously regarded yield stress materials to exhibit idealised flow
behaviour, i.e. they possess a finite zero shear viscosity. Examples of such measurements
are given (8, 9). But, when the situation of suspended particles is thought about, it can be
argued that the yield stress measurement is essentially meaningless. After all, the yield
stress represents the response of a force that is applied only for a given instant (lo), but
suspended particles exert continuous stress on the suspending matrix, often for long
periods. Hence, a different approach is required.
Stability of a given system can be calculated if careful measurements are made,

accessing the low shear Newtonian plateau and hence, zero shear viscosity. In order to
satisfactorily measure for zero shear viscosity at such low shear rates, slip must be guarded
against. The presence of slip can lead to erroneous conclusions being made, including
predictions of pseudo yield stresses. Slip can be avoided by several methods, varying from
gluing sandpaper to geometries (1 l), using specially designed serrated plates (12), or, as
used in this paper, the vane and basket geometry (13).
Zero shear viscosity values were obtained by fitting the data to the now well established
Carreau (14) model, shown below in Equation 1.
From equation 1, the zero shear viscosity, (qJ, can be obtained and also a value of the
critical shear rate term, (A). The critical shear rate term, expressed in units of reciprocal
seconds, can be converted and expressed as a stress in units of pascals. This corresponds to
the onset of shear thinning and hence, by implication the onset of the unstable region.
Thus, in order to maintain a stable h i t preparation, the gravitational stress exerted by the
fruit piece must be less than the stress corresponding to the onset of shear thinning. The
gravitational stress may be calculated according to Equation 2,
a&
GS=-
3
where GS is the gravitational stress, a is the radius of the particle, g is the acceleration due
to gravity and Ap is the difference in density between the particle and the medium.
Once it has been ascertained that the system will be stable under rest conditions, it is
advantageous if the movement of the particle can be quantified. This is achieved through
Stokes law,
where Vs is Stokes law velocity, g is the acceleration due to gravity, d is the particle
diameter, Ap is the difference in density between the particle and the medium and qois the
zero shear viscosity.
This investigation focuses on the stability of model fruit preparations by adoption of a

non yield stress approach. Effect of shear history is considered and data presented fitting
the above equations. It is argued that a fuller understanding of the system was gained than
from yield stress alone.
MATERIALS & METHODS
The pectin sample used was GRINDSTEDTMPectin YF 450, a low ester amidated
pectin. The specified degree of esterification (DE) is 21-31% and degree of amidation
(DA) is 20-23% (15).
The fruit preparation was prepared as an artificial model system where the formulation
of 40% soluble solids (ss) was chosen, as outlined below in Table 1.
GRINDSTEDTMPectin YF 450
Sugar I
Water I
Mix 2
sugar I1 37
Water I1 50
Water I11 6
Ca-lactate, 5H,O (required) 0.023
Ca-lactate, 5H,O (- strawberry) 0.076
Total 107.799
Total %ss 40
Table 1. Formulation percentages for the model fruit preparation.
Sugar I1 replaced the normal strawberry content, typically around 50% of the recipe where
10% constituted soluble solids. Calcium lactate 5H,O was the only salt included in the
model system. When simulating the calcium level in the fruit, in this case strawberry,
calcium was added to the recipe corresponding to 20 mg Cd100 g strawberry. Additional
calcium was added as required by the pectin.
Procedure:
1. MIX 1: The pectin and sugar I were dry blended and dissolved by stirring in hot water I
(85-9OoC)in a beaker.
2. MIX 2: Sugar I1 and water I1 were mixed in a pot and brought to the boiling point.
3. MIX 1 was then added to MIX 2 with agitation.
4. The Ca lactate was dissolved in water I11 (80C) and added to the boiling sugar mass
under vigorous agitation.
5. Evaporation in the pan while boiling until 1% less than the total %SS.
6. pH was adjusted to pH 3.9 k 0.1 with 50% w/v citric acid, H,O.
7. The sugar mass was then allowed to cool down to a temperature of 80C while gently
stirring and then placed in the rheometer.
All rheological measurements were carried out using a Haake (Karlsruhe, Germany) RS
150 controlled stress rheometer. The sample was placed in the cup, pre-heated to 80C of a
vane basket system and then cooled to the measurement temperature of 25C at the rate of
1C per minute. The vane had a diameter of 40 mm and the internal diameter of the basket
was 43.4 mm. This meant the gap between the blade of the vane and the basket wall was
1.7 mm. During cooling, the sample was either sheared at a constant rate of 10,30 or 50 s-
or left unsheared. When the sample had reached the appropriate measurement temperature
the shearing, as appropriate, was ceased for 1 minute before a controlled stress viscosity
experiment was performed. The stress boundaries were 0.01 to 150 Pa and the run time
had a duration of 5 minutes.
RESULTS & DISCUSSION
The recipe used throughout the paper simulated a h i t preparation, where the calcium
concentration was adjusted, as if strawberries were present. The percentage of soluble
solids and pH were held constant at 40 and 3.9 f 0.1 respectively, and again were designed
to represent a typical fruit preparation. The pre-shearing was designed to investigate the
effects of mild process conditions.
Calculation of both the gravitational stress and Stokes law velocities first required the
size of a strawberry and secondly the density of the strawberry and the model fruit
preparation matrix. A number of large strawberries were measured and an average radius
of 17.5 mm was found. The measured density of the natural strawberries, not pre-sugared,
was found to be 1.036 g/cm3, while the density of the h i t preparation matrix was 1.160
g/cm3. Thus, according to equation 2 the gravitational stress exerted by the average
strawberry is of the order of 0.7 Pa.
Figure 1 shows the viscosity versus shear rate profile for the model fruit preparation
without shearing during cooling. Evidence of the low shear apparent Newtonian plateau
may be clearly seen and the zero shear viscosity and onset of shear thinning were
determined by the fitted Carreau model. The highly shear thinning nature of the material is
also evident from the dramatic drop in viscosity at still low shear rates, a property that
suggests relative ease of pumpability of the system.
I 03 -.
lo2 :
lo1:
loo '
1'0-$ '"";'o-i ' "";'o-i ' ""l;'o-i ' "";'o-i ' I ' iI0-l Ilcf IlO' ;id
t [l/Sl
Figure 1 , Viscosity shear rate profile for model fruit preparation (symbols) with fitted
Carreau model (solid line) measured at 25C after cooling in the absence of shear.
The effect of shear during cooling was investigated whereby samples were sheared, at
three particular shear rates of 10, 30 and 50 s-', in order to mimic process conditions on the
system. The graphs shown in Figures 2 to 4 all show basically the same trend as Figure 1,
and the important results will be summarised in Table 1 .
10 3 " )
-v1
a 2-
k 10
E
1-
10
0
10
10-9
1111111
Id-5
11111111
1d-4
I1111111
Id-3
I1111111
1:-2
I1111111
4-1 I1111111
Id0
I1111111
ldl
1 I
7 0 2
i Il/sl
Figure 2. Viscosity shear rate profile for model fruit preparation (symbols) with fitted
Carreau model (solid line) measured at 25C whilst sheared during cooling at 10 8.
Hydrocolloids in Real Food Systems 23 1
104
103
-m
v)
102
k
F
10'
100
1
i [l/Sl
Carreau model (solid line) measured at 25C whilst sheared during cooling at 30 s-'.
i [l/Sl
Carreau model (solid line) measured at 25C whilst sheared during cooling at 50 s-'.
Despite all exhibiting the same general trend of apparent low shear Newtonian plateau
followed by shear thinning, the increase in shear rate during cooling plays a distinctive
role on the model h i t preparations. The calculated zero shear viscosity is seen to drop
by nearly a factor of two. This had important implications with regards to the system's
overall stability. The differences can be more clearly seen in Table 1.
Cooling Zero shear Critical Critical Fit Stokes law Stokes law
shear rate, viscosity, shear rate, shear coefficient, d s m/m
S-l PaS S-l stress, Pa r
0 10040 0.001767 10.11 0.9995 8.242~10-~ 0.021
10 6077 0.001594 7.24 0.9969 1.361~10'~ 0.035
30 5813 0.001512 6.84 0.9906 1.423~10-~ 0.036
50 5330 0.002415 6.51 0.9967 1.552~10-~ 0.040
Table 1. Showing the zero shear viscosity, critical shear rate, critical shear stress as
determined by the Carreau equation, the relative fit coefficient and Stokes law velocities
in meters per second and meters per month as a function of shear rate during cooling.
The low shear Newtonian plateau, a feature of hydrocolloid solutions (16, 17, 18, 19)
agrees well with the fitted Carreau model to give the zero shear viscosity. The values
for the zero shear viscosity are high, and they are attributed to the extensive three
dimensional intermolecular network formed between the pectin and calcium, that is
promoted throughout the system by means of the well established egg-box model (20).
The presence of shearing during cooling is seen to drastically reduce the zero shear
viscosity, by close to a factor of two, from no shearing to 50 s-'. This drop in viscosity
influenced the relative stability of the system.
Concerning the stability of the system, the gravitational stress exerted by the average
strawberry was calculated to be 0.7 Pa. Hence, we can conclude that so long as the
critical shear stress, linked to the onset of shear thinning, is greater than 0.7 Pa for the
model fruit preparation systems, then they will be stable under rest conditions. It can be
seen then, from Table 1 that the corresponding shear stress as calculated from the
critical shear rate are all considerably greater than 0.7 Pa. This leads to the conclusion
that under norrnal conditions of rest all the above systems would be capable of
suspending the fruit piece. However, one should beware that the increase of shearing
during processing reduced the stress of the onset of shear thinning, therefore lowering
the threshold of stress required to produce unstable conditions. This implies that more
vigorous process conditions are likely to influence the internal structure of the fruit
preparation matrix, making fruit suspension more difficult, and fruit identity worse.
Gentler conditions during processing, such as those found in tubular or scraped surface
heat exchangers would be advantageous (2 1,22).
The predominant problem in fruit preparations is fruit flotation ( 5 ) , since the density of
the fruit is usually less than the density of the surrounding matrix. A real feel for the
effect of shearing on the relative stability can be gained from the Stokes law velocities
which are given in either conventional meters per second, or more industry friendly
meters per month (Table 1). The latter gives a clearer indication as to the movement of
the fruit piece with respect to the average lifetime of a fruit preparation, around, 45
days. It is seen then, that the relative velocity of the strawberry roughly doubles as
shearing during cooling is increased up to the maximum of 50 s-'. A rough rule of
thumb, quoted by Dickinson, (23) says that a system can generally be regarded as stable
if the mean settling speed is less than 1 mm per day. When the above values for Stokes
law in meters per month are converted into mm per day the following values are
obtained. For increasing shear during cooling, 0, 10, 30 and 50 s-' the movement of the
strawberry per day is 0.70, 1.16, 1.20 and 1.33 respectively. This would suggest then
that only the sample where no shear was exerted would conform to the rough rule of
thumb. Two points to consider though is that the other values are close to the cut off
and it is certain that the viscosity of the system will be greater during most of the fruit
preparations lifetime, given that storage temperature is much less than 25C as is the
case here. The easiest way to decrease the Stokes law velocity is to increase the
viscosity parameter of equation 3, easily achieved by lowering the temperature. Indeed,
product stability is assured by storing the newly made fruit preparation for 10 days at a
temperature of 10C (24). The samples in this investigation were not cooled down to
10C before subsequent measurement at 25"C, but it could be expected that a higher
viscosity will form as a result of this cooling process. It is envisaged that any resultant
increase in temperature after cooling, still within ambient regions, will have little effect
on the overall stability of the system and shift the mm per day value below the 'critical'
value of one. Evidence to support this claim comes from ongoing research where
similar pectin systems were cooled from 80C to 15C and heated back to 80C again.
Despite being reheated to the original temperature, these reheated samples maintained a
far greater level of structure and viscosity than originally observed.
CONCLUSIONS
The data presented clearly shows the presence of a low shear Newtonian plateau with
respect to viscosity, indicating a finite zero shear viscosity. The effects of slip have
been minimised by the use of the vane rotor geometry. Zero shear viscosity is seen to
be dependent on the shear history of the sample with respect to shearing during cooling.
Gravitational stress calculations showed an average sized strawberry to exert 0.7 Pa on

the fruit preparation matrix. Ali systems were therefore deemed stabie under rest
conditions. Stokes law velocities indicated, as did zero shear viscosity that the effect of
shear reduced the stability or viscosity by a factor approaching 2. Only one of the
systems showed a Stokes law velocity that corresponded to less than 1 mm per day,
concurrent with a generally accepted rule of stability. However, the samples subjected
to shear were all very close to 1 mm per day. Changing the temperature by further
cooling would increase the zero shear viscosity and hence reduce the Stokes law
velocity below this 1 mm per day limit. Other factors such as process conditions and
added calcium content also play a role.
The improvement of rheometers over the recent decade made possible the accurate
measurement of viscosity at low stresses. Hence, accessing the zero shear viscosity, a
much greater understanding about the system has been demonstrated without invoking
the yield stress.
REFERENCES
1. M.A. Rao, in Rheology of Fluid and Semisolid Foods - Principles and

Applications. Aspen Publications, Gaithersburg, Maryland 1999. chl ,p.8.
2. H.S. Ramaswamy and S. Basak, J Food Sci., 1992,57,357-360.
3. H.C.A. Pedersen, FIE, 1993,51-54.
4. S. Basak and H.S. Ramaswamy, J Food Eng., 1994,21,385-393.
5. R. Kratz and K Dengler, Food Marketing and Technology, 1995, August, 12-18.
6. J.H. Watson, in press.
7. H.A. Barnes, H. Schimanski and D. Bell, Appl. Rheol. 1999,9,69-76.
8. H.A. Barnes, J. Non-Newtonian FIuidMech., 1999,81, 133-178.
9. H.A. Barnes and KWalters, Rheol. Acta, 1985,24, 323-326.
10. M.Power, in Annual Transactions of the Nordic Rheology Society, Ed. A.
Saasen, 1993,1, 11-13.
11. L.L. Navakis and E.B. Bagely, J. Rheol. 1983,27, 519-536.
12. L.H. Block and P.P. Lamy, J. Soc. Cosmet. Chem. 1970,21,645-660.
13. H.A. Barnes and J.O. Carnali, J. Rheol. 1990,6,841-866.
14. P.J. Carreau, Trans. SOC.Rheol. 1972,16,99-127.
15. Product Description - PD 10117-1e Dansico Cultor, 2000.
16. E.R. Morris, in Gums and Stabilisers for the Food Industry, Eds, P.A. Williams,
G.O. Phillips and D.J. Wedlock. Oxford: Pergamon Press. 1984,2, p 57-78.
17. J.L. Doublier, Gums and Stabilisers for the Food Industry, Eds, G.O. Phillips,
P.A. Williams and D.J. Wedlock. Oxford: IRL Press. 1994, 7, p 257-270.
18. N.W.G. Young, PhD Thesis, University of Salford, 1997.
19. H.A. Barnes, App. Rheol. 1999,9,262-266.
20. G.T. Grant, E.R. Morris, D.A. Rees, P.J.C. Smith and D. Thorn, FEBS Lett.
1973,32 195-198.
21. G.S. Tucker and U. Bolmstedt, Liquid Foods International 1999,3, (3), 15-16.
22, U. Bolmstedt, New Food, 2000,3, (3 j, 17-22.
23. E. Dickenson, Gums and Stabilisers for the Food Industry, Eds, G.O. Phillips,
P.A. Williams and D.J. Wedlock. Oxford: IRL Press. 1988,4, p 249-265.
Y . Flaissier. Revue Laitiere Francaise. 1995. 555.4-6.
Interfacial Behaviour and
Gelation of Proteins
THE ROLES OF PROTEINS AND PEPTIDES IN FORMATION AND STABILISATION
OF EMULSIONS
Pieter Walstra
Department of Agrotechnology and Food Sciences, Wageningen University, P.O. Box

8 129,6700 EV Wageningen, the Netherlands
1 INTRODUCTION
Virtually all proteins are surface active at oil-water interfaces. A protein that is well soluble
at the given conditions can be used to make oil-in-water emulsions. Because proteins are
insoluble in oil, they cannot be used to make water-in-oil emulsions, in accordance with
Bancroft's rule.' Emulsion formation and emulsion stability,should be considered separately,
since they are for the most part governed by different factors.
We will consider simple emulsions of triglyceride oil, coarsely pre-emulsified in a protein
or a peptide solution, and subsequentlytreated in a high-pressure homogenizer. Generally, no
other surfactants were present.
2 EMULSION STABILITY
From a recent review,* the following kinds of physical instability of emulsions can be
distinguished.
O&ald ripening, i.e. the growth of large drops at the expense of small ones by iso-
thermal distillation of the disperse phase through the continuous phase. Since the disperse
phase (triglyceride oil) is insoluble in protein solutions, this instability does not occur.
Creaming. The rate of creaming strongly increaseswith the droplet diameter (4,and with
aggregation of the droplets.
Aggregation. As a general rule, a simple protein-stabilised emulsion does not show
aggregation, provided that the conditions are such that the protein is well soluble, the
surface load (I)is not too low (say, > 2 mg-m-2),the ionic strength is not too high (say, <
0.1 molar), and the droplets are not very large.
Coalescence. Proteins generally provide good protection against c o a l e ~ c e n c ebecause
~.~
they can provide strong colloidal repulsion and the interfacial tension oil-water is not very
small (say, 10 mN-m-'). Conditions for stability are that the drops are small and that the
surface load has reached its plateau value (implying a densely packed monolayer). Under
some extreme conditions, however, for instance if the emulsion is highly concentrated and
then is agitated," coalescence can still occur.
Partial coalescence. This can occur if the emulsion droplets contain fat crystals.It will not
be considered here, since the role of proteins and especially peptides in preventing partial
238 Gums and Stabilisers fur the Food Industry I1
Table 1. Some characteristics ofproteins used in this study. IEP = iso-electric pH; LAH =
lipid acyl hydrolase.
Name Source MW (kDa) IEP Particulars
P-casein bovine 24 -5 fairly hydrophobic, forms micelles

P-lactoglobulin bovine 18 5.2 1 -SH group, forms dimers or polymers
lysozyme hen 14 10.7 stable conformation
a-lactalbumin bovine 14 5.1 similar to lysozyme, less stable conformation
ovalbumin hen 45 4.5 strongly denatures at surfaces
patatin potato 43 -4 has some LAH- activity
coalescence (clumping) has been insufficiently studied. By and large, thick protein layers
are effective in preventing the phenomenon, provided that the emulsion droplets are small.
It may be concluded that the most important variables governing emulsion stability are droplet
size distribution f(d), protein surface load and physicochemical conditions (especiallypH and
ionic strength). The values of f(6) and of r will for each protein depend on conditions during
emulsion formation. Consequently, this aspect will be considered in some detail.
3 EMULSION FORMATION WITH PROTEINS
Emulsions were made with the proteins given in Table 1;these were about 98% pure. Methods
for determination of f(d) (spectroturbimetry), r (depletion) and recoalescence during emul-
sification (mixing of two different oils) are described el~ewhere.~-~ Emulsions were homog-
enized in a very small homogenizer, which implied that the emulsion had to pass through the
machine several times. The oil volume fraction was 0.2. Varied were type of protein, protein
concentration(c),homogenization pressure (often 5 MPa), and physico-chemical conditions.
Emulsions were checked for coalescence (droplet size as a function of time) and aggregation
(by microscopy).
Some of the results are given in Figure 1. Leaving out patatin for the moment, the resulting
plateau values of d32are on average 1.O pm, and vary by a factor 2. Also the plateau values of
r vary by a factor 2.The rates of recoalescence did greatly increase with decreasing protein
concentration. The recoalescence was almost the same for P-casein, P-lactoglobulin and
a-lactalbumin, and was much more pronounced for lysozyme and ovalbumin. The concen-
tration of protein needed to reach a plateau value of d3,increased with increasing molar mass.
For a technical soy protein, where the effective molar mass is quite high, a far higher protein
concentration,about 25 mg per ml, is needed to obtain a plateau value." If the homogenization
pressure was increased, a higher protein concentration was needed to obtain a plateau value
of d32, which then was smaller.
Figure 2 gives some results for P-lactoglobulin at various pH. Closer to the iso-electricpH,
a much larger d32value resulted and also the value of r was larger. It is seen (Figure 2.c) that
at the lower pH the rate of recoalescence was also markedly higher. If the P-lactoglobulinwas
heat-denatured, much larger droplet sizes were obtained, presumably because the protein then
is to some extent aggregated. Denaturation of a-lactalbumin, which then does not aggregate,
had little effect on the results obtained.
The most striking result is, however, that the droplet sizes obtained with proteins are much
larger than those obtained with most small-molecule surfactants. For instance, under similar
conditions, a value of d,, = 0.25 pm was obtained with SDS (sodium dodecyl sulfate).6 Other
Interfacial Behaviour and Gelation of Proteins 239
r(mg/m2)
t 4
I
4-
,
2- -
2-
1-
-
I
PA /
I
I I 1 I I I /
I I I I
0 2
-6
c(mg/ml)
10
- 1
c/A (mg/m2)
Figure 1 Emulsificationfor variousproteins as a function ofprotein concentration c; pH =

6.7. Proteins were ovalbumin (Ov), lysozyme (L y), p-casein (BC), a-lactalbumin (AL), p-
lactoglobulin (BL), and patatin (PA). a: Volume surface average drop diameter (d,J. b:
Protein surface load (r) as a function of c/A, where A is specijk oil surface area; the broken
line gives the value that would results ifall protein becomes adsorbed. Results after Ref 8 and
Ref 9 batatin).
differences are shown in Figure 3. It is seen that the protein, in this case p-casein, is much
more surface active than SDS. On the other hand SDS gives a smaller plateau value for the
interfacial tension at the oil-water interface: y equals 2.3 for SDS versus 11 mN-rn-' for the
protein. This means that to obtain small droplets with protein as the surfactant, a much higher
homogenization pressure is needed than for SDS and similar surfactants.
It is also seen in Figure 3 that the relation between surface load and equilibrium protein
concentration differs greatlybetween quiescentadsorptionat a macroscopic oil-water interface
and oil droplets as obtained by homogenization. In other words, the latter result is not an
adsorption isotherm for a protein, hence does not represent an equilibrium situation (which it
0 5 10 0 5 10 0 25 50
c (mg/ml) c/A (mg/m2) passes
Figure 2 Making emulsions with P-lactoglobulin at two pH values (indicated).' a: Average
drop size. b;Surface load. c: Recoalescence;the extent of recoalescenceduring emulsification
is larger ifthe relative turbidity (z/zo) decreases stronger with the number ofpasses through
the homogenizer.
does for a small-molecule surfactant). The protein adsorption curve gives an increase over
the plateau value for ceq> 0.03%, and this indicates multilayer absorption; the excess over the
plateau value is readily washed away by dilution. This is not the case for the high r values of
the emulsion curve: washing by diluting a few times did not alter r.At low values of I,
addition of some protein to the emulsion did lead to a slow increase of r.
To explain the differences in results between SDS and proteins we need to consider what
happens during emulsification. This varies somewhat with the emulsification regime, which
will be in the present case bounded laminar elongational flow. The events are
Strong deformation of a drop, often followed by its break-up into smaller ones. The drop
size obtained will be about inversely proportional to the effective y value during break-up.
Surface active material is transported to the drop, leading to a lowering of y.
Adsorption of protein is followed by a change in its conformation; this can take a relatively
long time and may not lead to equilibrium within the time scale involved.
Droplets frequently encounter each other (collide),which may cause their recoalescence.
All these processes have their own time scale and for the present case it can roughly be
calculated6that deformation takes about 1 ps and absorption 10 ps. It is more difficult to
estimatethe encountertime, but it will be less than 10 ps. Presumably,conformationalchanges
may take milliseconds to hours, depending on protein and external conditions.
Break-up thus depends on yew Since all the processes occur several times, a drop will soon
attain a y value close to local equilibrium. However, break up is preceded by deformation,
which will roughly double the surface area, thereby halving r.Relations between I7 (i.e. the
decrease in y due to adsorption) and r a r e given in Figure 4. Admittedly, it applies to air-water
surfaces, but the trend will roughly be the same for oil-water. It is seen that halving Iwill in
the case of SDS still leave a substantial I7-value, but for lysozyme a very small I7 may result.
This is because of the typical surface equation of state for proteins. At very small r,the
0
-6 -4
- -2 0
log (c,,/wt.%)
Figure 3 Surface loadr ofp-casein as obtained by quiescent adsorption onto a macroscopic
oil-water interface, and as resulting_fi.omemulsiJcation.a ceq is the protein concentration
in the aqueousphase after adsorptionor emulsi9cation. The equilibrium adsorption isotherm
for sodium dodecyl sulfate (SDS) is also given.
Inte$acial Behaviour and Gelation of Proteins 24 1
relation is for all surfactants 17= RTT, where Tis expressed in moles per unit surface area; the
broken line in the figure gives this for lysozyme, and it is seen that value of 17 would be almost
negligible. However, a correction for crowding in the interface is needed, and for the case of
hard spheres (as for lysozyme molecules) this results in'*
where 0 is the surface fraction covered by the protein and r is its molecular radius. This rela-
tion well describesthe experimental results.l 3 The main conclusion is, that the high molar mass
of proteins gives rise to a situation where a considerable value of r,as expressed in mg.m-2,
is needed to obtain a substantial lowering of y. We come back to the curve for P-casein.
It has been reasoned before' and it has been experimentally confirmed for SDS as the
surfactanf, that recoalescence during emulsification is not prevented by colloidal repulsion
between the droplets, since the forces involved are generally by at least a factor 100 too small.
What is involved is the formation of interfacial tension gradients at the drop surfaces in the gap
between droplets coming close. This greatly slows down the approach of the droplets, since
it will prevent slip of the liquid flowing out from the gap; before coalescence can occur, the
drops may again move away from each other. The strength of this effect will depend on the
magnitude of the surface dilational modulus.' In the present situation,its value would be given
by
ESD= dl7/dlnT PI
The slope of the curves in Figure 4 would thus directly give the value of ESD,and it is seen that
it would be far higher for SDS than for lysozyme. In other words, there would be more
recoalescence in the latter case. This is borne out by experiments.6.8
In Figure 1 .a, the curve for patatin is rather different from the others, in that a very small
value of d3zis obtained that keeps decreasing with increasing protein concentration. Patatin
I
20
10
0 I I I
-1
- 0 1
In ( r / m g m-2)
Figure 4 The relation between surface pressure 17,i.e. the decrease in surface tension due to
adsorption, at the air-water surface, as a function of the surface excess r for three
adsorbates. l 2 Further see text.
is said to have a weak LAH-activity or, in other words, it would be a weak lipolytic enzyme.
It turned out, however, that patatin is strongly lipolytic at the oil-water interface, rapidly
producing copious amounts of free fatty acids and monoglycerides from a triglyceride
Especially the presence of monoglycerides would greatly decrease the effective y value and
increase the value of PD. In other words, it will lead to far smaller droplets than can be
obtained with a similar but non-lipolytic protein.
Can we now explain differences among proteins in their capacity to produce small droplets
by using the same arguments? One observation fits: there is a clear correlation between ave-
rage droplet size and effective molar mass of the protein. The latter value may also be high
because the protein is aggregated (e.g. due to heat denaturation) or poorly soluble (e.g. near
its iso-electric pH). The explanation would be that for a higher molar mass (i) more protein is
needed to obtain plateau values for r as well as d32;(ii) at the same protein concentration yeR
is higher, thereby causing droplet break up to be less effective; and (iii) at the same concen-
tration the value of PDis smaller, causing more extensive recoalescence of newly formed
droplets during homogenization. It also appears that the protein conformation need not have
a large effect, since denaturation of a-lactalbumin did not affect the values of d32and of r at
the same protein concentration. On the other hand, lysozyme,which is very similar in structure
to a-lactalbumin, gave a somewhat larger d32value, presumably due to faster recoalescence.
To interpret further differences among proteins, we would need to have curves like those
in Figure 4 for various proteins and at the oil-water interface, but these are not available (except
for a few proteins, which shows that the relations do depend on the nature of the interface).
Moreover, the data would be difficult to interpret because of differences in time scale.
Determining a I7 - r curve like in Figure 4 takes at least several minutes and in that time the
adsorbed proteins will have changed their conformation. Lysozyme is a protein with a high
conformational stability and it will change its effective diameter only a little after ad~orption~,
which presumablymeans that also I7 will hardly change. j3-Casein, however, is a highly flexible
protein that strongly expands its effective diameter upon adsorption, which will cause an
increase in I7.The time scale needed for such changes is quite uncertain and widely variable
estimateshave been reported. The lowest value found for P-casein appears to be 0.1 s,I4 which
is very much longer than the characteristic times for drop deformation or protein adsorption
during homogenization. This would mean that the values for 17 and for PD during emul-
sification are significantly smaller than the equilibrium values at the same r.
The latter may explain why P-casein tends to give larger droplets than, say, P-lactoglobulin.
Also the latter protein will change conformation upon adsorption, and this even takes a far
longer time,I4but the value of I7 obtained after a very short time will be at least equal to that
calculated from Eq. [l], taking for the radius r that of the molecule in solution. A flexible
protein like p-casein, however, can become adsorbed with only a very small part of each
molecule in the interface, thereby giving quite a small I7 value. This is to some extent borne
out by the high plateau value of r obtained by emulsification (Figures 1 and 4), which is not
due to multilayer or micelle adsorption,8 as mentioned; the value can only be explained by
assuming that the adsorbed layer is to some extent like a brush.
Altogether, a detailed explanation of the variation in emulsifying efficiency of proteins
would need far more study. It may even be questioned whether such a study would be worth-
while, since the unexplained differences are rather small. It may also be questioned if the
results obtained with a very small homogenizer are representative for what happens in a big
one. However, for mixed proteins, e.g. Na-caseinate, no significant differences between
homogenizers were observed.
W5.
c\I
ll(Fl
0.5
0
2Ei 5
c (mg/ml)
10
4
0 5
c/A (mg/m*)
10
s
1
0
0 .
25
passes
5
50
B
Figure 5 Making of emulsions with p-casein (BC) and with amphiphilic peptides (AP). The
peptides consisted for the most part of the amino acid sequences 29-1 Om07 of p-casein.
Further see text,
4 PEPTIDES AS EMULSIFIERS
Since it can be concluded that proteins of smaller molar mass are more effective in producing
small droplets, it is to be expected that the still smaller peptides that can be derived from
proteins are even more effective. To test this, peptides were obtained from p-casein and
0-lactoglobulin by selective enzymatic hydrolysis, followed by chromatographic and other
separation and purification steps. In this manner, almost pure and well-characterized peptides
were ~btained.'~.'~ These were used to make emulsion^.^.'^
Peptides that are not very small molecules, and have at least a somewhat amphiphilic
character, and are well soluble (some hydrophobic peptides were not), could be used to make
emulsions. An example is given in Figure 5 . It is seen that the peptide derived form p-casein
indeed gave smaller droplets, a lower plateau value of I-', and less recoalescence during
emulsification, as predicted. The same trends were found for most other peptides. However,
the emulsions generally showed some coalescence, d32doubling over a few days, for instance.
For some peptides the emulsions were stable if the pH was far away from the iso-electric point
and the ionic strength was low. It should further be realized that especially the smaller peptides
can desorb from the droplets upon dilution; in other words, the adsorption is closer to equi-
librium than for proteins.
None of the emulsions made with proteins showed coalescence, unless r was clearly
smaller than the plateau value. Then, some coalescence can occur during a limited time, pre-
sumably until it has led to the plateau value of I-'.
Acknowledgements
Most of the results given here were obtained by a number of then Ph.D. students: Petra
Caessens, Gerrit van Koningsveld, Andrk de Roos and Laurent Taisne, and especially Ine
Smulders. Part of the work was sponsored by the Innovation-oriented Research Programme
on Industrial Proteins (IOPie) of the Dutch Ministries of Economic affairs and of Agriculture,
and by DMV International, Veghel, the Netherlands.
References
P. Walstra and P.E.A. Smulders, in Modern Aspects of Emulsion Science, ed. B.P. Binks,
Roy. SOC.Chem., Cambridge 1998, ch. 2, p. 56.
P. Walstra, in Encyclopedia ofEmulsion Science, Vol. 4, ed. P. Becher, Dekker, New York
1996, ch. 1, p. 1.
E. Dickinson, B.S. Murray and G. Stainsby, J. Chem. SOC.Faraday Trans. I , 1988,84,871.
G.A. van Aken and F.D. Zoet, Langmuir, 2000,16,713 1.
P. Walstra and P.E.A. Smulders, in Food Colloids: Proteins, Lipids and Polysaccharides,
eds. E. Dickinson and B. Bergenstihl, Roy. SOC.Chem., Cambridge, 1997, p. 367.
L. Taisne, P. Walstra and B. Cabane, J. Colloid Inter- Sci.,1996, 184, 378.
P.E.A. Smulders, P.W.J.R. Caessens and P. Walstra, in Food Emulsions and Foams:
Interfaces, Interactions and Stability, eds. E. Dickinson and J.M. Rodriguez Patino, Roy.
SOC.Chem., Cambridge, 1999, p.61.
P E A . Smulders, Formation and Stability of Emulsions made with Proteins and Peptides,
Ph.D. Thesis, Wageningen University, 2000.
G. van Koningsveld, Physico-chemical and Functional Properties of Potato Proteins, Ph.D.
Thesis, Wageningen University, 2001.
10 B. Mertens,-The Functional Properties of Proteins in the Formation and Stabilization of
O/W Emulsions, Ph.D. Thesis, State University of Ghent, Belgium, 1989.
11 D.E. Graham and M.C. Phillips, J. Colloid Inter- Sci., 1979,70,415.
12 J.A. de Feyter and J. Benjamins, J. Colloid Inter- Sci., 1982,90,289.
13 A.L. de Roos and P. Walstra, Colloids Surfaces B, 1996,6,201.
14 F.J.G. Boerboom, Proteins and ProteidSurfactant Mixtures at Interfaces in Motion, Ph.D.
Thesis, Wageningen University, 2000.
15 P.W.J.R. Caessens, H. Gruppen, S. Visser, G.A. van Aken and A.G.J. Voragen, J. Agr.
Food Chem., 1997,45,2935.
16 P.W.J.R. Caessens, Enzymatic Hydrolysis ofp-Casein andp-Lactoglobulin, Ph.D. Thesis,
Wageningen University, 1999.
EFFECT OF VARIOUS FACTORS ON EMULSION STABILISING PROPERTIES
OF CHITOSAN IN A MODEL SYSTEM CONTAINING WHEY PROTEIN
ISOLATE
S. Laplante, S. L. Turgeon and P. Paquin
Dairy Reseach Center (Centre STELA), Facultk des Sciences de 1Agricultureet de

1Alimentation, Universite Laval, QuCbec (Qukbec), Canada, G1K 7P4
ABSTRACT
Chitosan (CN) is a cationic polysaccharide making it distinctive among other anionic or

neutral stabilising gums currently used in food emulsion industry. This structural
difference would bring new interesting properties on emulsion stability, considering the
impact it could bring on interactions with another important food macromolecular
component in those systems: the proteins. So, the objective of this study was to evaluate
the effect of factors (pH, p, %CN, %WPI) on the stabilising properties of CN in an acidic
model emulsion containing whey protein isolate (WI), recognised for his good
emulsifying properties. Under a central composite experimental design, we compared
various combinations of pH (4.0-6.0), %WPI (0-1%), and %CN (0-0.3%), either at low
(unadjusted) or high p (0.3 M), with a CN having those characteristics (78.1%DD, 1490
m a ) , in emulsions containing 10% canola oil. Results from phase separation evolution
measurements in tubes revealed that pH and p are the most important factors affecting
emulsion stability. At pH 4.0-5.0, we observed rapid wheying off (WO) with flocculation
between droplets, and at pH 5.5-6.0, we mainly observed small gradient creaming
phenomena with fine droplet dispersions, whereas with p=0.3 M or low CN:WPI ratio
(1 :lo), some creaming conditions at pH=5.5 destabilised towards WO. WO conditions that
occurred at pH<pI of proteins ( ~ 5 . 1 )are explained by the incompatibilitybetween CN and
adsorbed proteins, favoring depletion flocculation and wheying off. The more stable
gradient creaming conditions at pH 5.5-6.0 are due to CN-protein compatibility, that
permit stabilising effect of CN by coadsorption with protein, as shown by protein load
experiments at pH 5.5.
1 INTRODUCTION
Proteins and polysaccharides play important roles in various food systems. For instance,
under suitable conditions and processes, mixtures of protein-polysaccharide (P-PS) can
control and improve the texture and stability of lipid dispersions in various food emulsions
(homogenised milks, ice creams, whipping creams, dressings, mayonnaises, etc.). In such
oiVwater emulsions, the major functional role of proteins (ex.: casein, whey proteins) is to
emulsify the lipid phase, by forming an interfacial protein film with elastic and steric-
repulsive properties, which prevents flocculation and coalescence mechanisms leading to
creaming2 and wheyng off3. In the case of PS, their main stabilising role is by gelation or
viscosification of the continuous phase, which prevents destabilising mechanisms leading
to ~ r e a m i n g ~However,
.~. in most food systems, we cannot consider P and PS as
independant components, particularly at low PS concentrations (SO,1 % (w/w)) where, the
overall stability is less dependant on the specific role of each biopolymer, but becomes
rather more dependant on interactions between themselves, which could have important
consequences on emulsion ~ t a b i l i t y ~ * ~ , ~ * ~ ~ ~ ~ .
According to the structure of biopolymers and environmental conditions in a particular
food emulsion (pH, p, PS/P ratio, concentrations of PS and P), P-PS interactions depend
on many attractive (electrostatic > H bonding) and repulsive (electrostatic > steric,
excluded volume effect) forces. When P-PS attractive interactions predominate over P-P
or PS-PS, both biopolymers are compatible, can coadsorb at the oiVwater interface, and
the resulting emulsion stability could be increased by steric and charge repulsion between
droplets. When P-PS interactions are repulsive or unfavorable, there is incompatibility and
destabilisation towards phase separation.
Until now, PS already studied as stabilisers in low viscosity model emulsions containing
protein emulsifiers were either neutral (hydroxyethylcellulose, dextran) or anionic
(xanthan, carrageenan, guar gum, alginate, CMC)791,12~13914 . However, to date, a
distinctive cationic PS, namely chitosan (linear co-polymer of D-glucosamine and N-
acetyl-D-glucosamine units connected through 13-( 1-4) linkages), has never been studied
for its stabilising role in model emulsions containin protein emulsifier, in spite of the
numerous patented food applications that contain it! Food applications of chitosan are
motivated by the fact that it is readily available (from crustaceous shells), and many
studies revealed that hypocholesterolemic and hypolipidaemic properties of chitosan
would make it of great interest for food application^^^^'^^'^^'* . Moreover, emulsifying
properties of chitosan were reported9y20,as well as co-adsorption properties on lipid
droplets covered with anionic lipid emulsifier^'^^^^^^^'^^ . Other studies have shown that
emulsifying roperties of chitosan would be influenced by its own characteristics such as
ash content24pand the degree of deacetylation2*.
By comparing the effect of various environmental factors (pH, p, concentrations of P
and PS) on emulsion stability with a model emulsion containing whey rotein isolate
Y
(WPI), known for its good emulsifying properties in many food systems , it would be
possible to show that, as opposed to anionic PS, the cationic nature of chitosan would
favor P-PS interactions and stability by interfacial co-adsorption at pH > protein PI.
Results of this study would provide a better evaluation of potential applications of
chitosan in food emulsions containing a protein emulsifier, and extend our current
knowledge on P-PS emulsion stability to the unique case of a cationic PS.
Intelfacial Behaviour and Gelation of Proteins 241
2 MATERIAL AND METHODS
2.1 Material
Chitosan produced from shrimp chitin (Pandalus borealis) was kindly donated by Les
PCcheries Marinard Ltte. (Rivikre-au-Renard, Qc, Canada). This preparation (named CNI)
has a deacetylation degree of 78.1%, Brookfield viscosity of 4413 mPa.s (1%
chitosan/l% AcOH, 25"C, spindle #4, 60 rpm,), a viscosity average MW=1494 kDa,
0.244% ashes and 0.254% proteins (dry weight basis). Whey protein isolate (WPI) was
supplied by Davisco International (Le Sueur, MN). The composition was (moist weight
basis): 94.85% of proteins (3.2% bovine serum albumin, 13.0% a-lactalbumin, 66.1% 13-
lactoglobulin, 17.7% other proteins), 1.95% ash, 6.15% moisture. The canola oil was
purchased from a local supermarket. A lipophilic dye (Red Oil 0, Sigma Co., St. Louis,
MO) was added (0,06% (w/v)) to the oil. Reagents were purchased from Fisher
(Pittsburgh, PA) and Sigma . Distilled water was used in all preparations.
2.2 Experimental design and statistical analysis
The selected experimental design was a central composite rotatable design made up of 3
factors x 5 levels: (%(w/w) WPI= 0-0.25-0.5-0.75-1.O; %(w/w) chitosan= 0-0.075-0.15-
0.225-0.3; pH= 4.0-4.5-5.0-5.5-6.0), making a total of 19 experimental units (14
treatments and 5 repeats of the central point treatment). In a first design, we did not adjust
ionic strength (p), whereas in the second one, through the addition of NaCl, we adjusted p
at a constant value, equivalent to 0.3 M NaCl=([Na']+[Cl-])/2. Statistical analysis was
realised with SAS software.
2.3 Emulsion preparation
Each emulsion was made from 180 ml of solution containing required concentrations of
CNI and WPI in 0.2 M AcOH and 0.02% (w/v) sodium azide, with 20 ml of canola oil, in
order to get the complete emulsion containing 10% (v/v) oil. The pH adjustments were
made with NaOH. Once the oil was added, a pre-mixing was done with an Ultra-Turrax
(Janke-Kunkel, GmbH) during 30 seconds, with an electrical input of 30 Volts. The
homogenisation was done with an Emulsiflex-C5 homogeniser (Avestin Co., Canada) in 2
passes (6000 and 3000 psi).
2.4 Particle size determination
Immediately after emulsification, the volume-weighted average diameter of droplets in

emulsions was determined by Photon Correlation Spectroscopy (PCS), using a Nicomp
370 Submicron Particle Sizer (Pacific Scientific, Hiac-Royce Instruments Division, CA,
USA).
2.5 Stability of emulsions during storage
Immediately after emulsification, two glass tubes were filled with 15 ml of emulsion,
capped, and kept at room temperature (23C). Comparison of the evolution of stability
between each treatment was done over a period of 20 days by daily measurements of the
thickness (centimeter units) of serum layer (lower distinct clear or semi-transparent phase)
or creamed phase (red layer gradient at the upper phase). In the latter, the presence of red
lipophilic dye in the oil enabled us to distinguish and follow thickness evolution of the
creamed phase. For each experimental design, statistical analysis for comparison of
stability profiles between treatments, where serum layer appeared (wheying off
treatments), was established from their maximal speed of phase separation (Smax),
evaluated by the second derivative of non-linear fitting curves for each evolution profile of
thickness of phase separation vs. time. Treatments performing gradient creaming were
compared by LSD multiple comparison test between their average thickness of phase
separation over a period of 20 days,
2.6 Protein load in lipid phase
A weighed amount of each emulsion was centrifuged at 40000g for 1 hour (23OC). Protein
assay (BCA assay method, Pierce Co., Rockford, IL) was done with serum, by taking
AcOH buffer with the same concentration of chitosan as in the formulation for the blank.
By subtracting the amount of protein in the serum from the total protein amount in the
formulation, protein concentration in the lipid phase, and protein load (in mg protein/m2
interfacial area units, from the average droplet diameter determinations) were calculated.
2.7 Microscopic observations
From 1:lO dilution of each emulsion in water, 2-3 drops of 1% (wh) Methylene Blue
(Sigma Co.) were added in order to enhance the contrast around lipid droplets.
Microscopic observations and recordings were made with a Zeiss optical microscope
connected to a digital camera (Sony).
3.1 Observations of phase separation phenomena
Figure la) shows the assembly of tubes corresponding to each condition of the central
composite design for CNI, after 40 days at room temperature. Mainly two types of
creaming phenomena occurred: a) "wheyng off' as shown by a lipid-rich upper phase and
a clear lower serum phase, and b) "gradient creaming", as shown by a gradient on top of
the emulsion due to higher concentrations in the oil phase marked with red lipophilic dye.
Comparing the effect of treatments (Fig. la), reveals that most conditions at pH25.5 give
the most homogeneous and stable emulsions, even if little gradient-creamed phase
occurred on top. The only condition at pH 5.5 where there is wheying off (tube#13),
corresponds to the minimal ratio CNYWPI=l :10. At pH15.0, all treatments containing
CNI-WPI mixtures gives wheying off, including the central condition (tube#8), and the
control without chitosan (tube#7). The exception at pH 5.0 that gives gradient creaming, is
the control without W I (tube#@, where we see a thick creamed oil layer on top and a
light lipid droplet suspension remaining in the lower phase, without flocculation.
a) b)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 11 12 13 14
Tube ## 1 2 3 4 5 6 7 8 9 10 11 1213 1415

%(w/w)CNI 0.15 0.075 0.225 0.075 0.225 0.15 0 0.15 0.3 0.15 0.075 0.225 0.075 0.225 0.15
%(w/w)WPI 0.5 0.25 0.75 0 0.5 1.0 0.25 0.75 0.5
pH I 4.0 I 4.5 5.0 5.5 6.0
Figure 1 Stability tube test with CNI after 40 days (at room temperature) a) CNI and
b) CNI with ji=O.3M (the chart describes the treatment for each tube number)
Figure 2 Microscopic visualisations of typical emulsions (at 400x) with two types of
phenomena a) jloccuiarea aroplet network resulting in wheying offand b)
individual droplets of various sizes resulting in gradient creaming.
Figures 2a-b show both phenomena at microscopic scale (400x), where in the case of
wheying off (a), there is a flocculated droplet network expelling the serum phase. In the
case of gradient creaming (b), there is a fine dispersion of individual droplets with
heterogeneity in sizes.
These results indicate that of all factors and levels selected, pH is the most important,
governing stability in this emulsion system. According to the literature680,this could be
explained by the predominance of electro-attractive forces (compatibility) occurring
between adsorbed WPI proteins (mostly P-lactoglobulin, mainly negatively charged) and
chitosan (mainly positively charged) at pH values above the isoelectric point of WPI
( ~ 1 ~ 5 . 1 )This
. condition could lead to interfacial co-adsorption of both kinds of
biopolymers, which would prevent such a system fkom destabilisation (i.e., through
flocculation, coalescence or creaming) by creating steric and electrostatic repulsion
between lipid droplets. However, at pH 5.5, when CN1:WPI ratio is too low (l:lO),
wheying off with flocculation is observed. This type of flocculation associated with
compatibilit at low polysaccharide concentrations is reported as bridging
flocculations0. At pHC5.5, due to a net electro-repulsive effect between adsorbed WPI
emulsifying proteins (mainly positively charged) and chitosan in the aqueous phase, the
incompatibility between both types of biopolymers would favor adsorbed protein-protein
associations between droplets, resulting in lipid flocculation and wheyng off. As reported
in the literature6s0,this type of flocculation associated with incompatibility is a depletion
flocculation. Protein-protein associations between droplets would also be favored with
control at pH 5.0 without CNI, due to the effect of pH near isoelectric point of WPI
proteins, combined with the presence of the AcOH buffer (at p=O. 136 M), which indicates
instability in the dispersing effect of WPI proteins alone. However, in spite of a low
stability against gradient creaming with the chitosan control without protein, we could see
residual stable lipid dispersion, which confirms the reported emulsifying property of
chito~an~~~.
Figure l b shows the assembly of tubes at conditions of pH 5.5 in the central composite
design for each type of chitosan with salt added to p=0.3 M, all other pH conditions (not
shown) remaining visually unchanged in comparison to Fig. la). The increase in p has
reduced stability of some conditions at pH 5.5, as shown by transitions from previous
gradient creaming phenomena towards flocculation and wheying off (appearance and
progression of clearer lower phases).
The explanation of this phenomenon is known as the loss of compatibility between
adsorbed protein and polysaccharide due to a charge screening effect, which enhances
adsorbed protein-protein attractions between resulting in the formation of a
flocculated upper lipid phase rich in adsorbed proteins. However, in the lower turbid
phase, due to the known emulsifying properties of chitosan, we would find lipid droplets
rich in adsorbed chitosan that would resist flocculation because of higher surface charge
density and hydration properties than proteins, so they would tend to remain in suspension
along a gradient of size distribution. Consequently, after the pH effect on stability, results
showed that p is another important factor affecting stability in our model system.
3.2 Comparison of the phase separation profiles for wheying off treatments by
response-surface analysis
Until now, we have just observed the final product of destabilisation after 40 days, and
showed the predominating effect of pH on stability. However, a more accurate comparison
of stability between treatments must be based on relative evolution of destabilising
phenomena, in order to better discriminate the effect of factors other than pH and p, if
possible. In Figure 3, we see profiles of evolution of phase separation by wheying off
(continuous lines) and gradient creaming (dashed lines) with CNI, the latter phenomenon
showing slow and close evolutions relative to wheying off. If we just put our attention on
the wheying off profiles, we mainly observe rapid increases towards similar maximum
values during the first 7 days. As for corresponding treatments, we see a general tendency
of slower separation speeds at pH 4.0-4.5 in comparison to pH 5.0, due to the most
unstable pH condition at pH 5.0 (gWPI PI), as explained earlier. Moreover, among
treatments at pH 4.5, the slower evolutions are performed by compositions with the
highest CNI concentrations (0.225%), indicative of the stabilising viscosifying role of
chitosan when the concentration is increased. For experimental design at p=0.3 M (not
shown), most of wheying off evolutions were faster but similar between themselves.
Inte$acial Behuviour and Gelation of Proteins 25 1
7 , 1 +0.1510.514.0
--.-t-0.07510.2514.5
+0.22510.2514.5
-
+0.07510.7514.5
0.22510.7514.5
- - A - - 0.151015.0
+010.515.0
+0.1510.5/5.0*
+0.310.515.0
+0.1511.015.0
- - f - - 0.07510.2515.5
- - 0 - - 0.22510.2515.5
I0.07510.7515.5
Q) - - i3 - - 0.22510.7515.5
0
0
- - 0 - - 0.1510.516.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1516 17 18 1920
Time (Days)
Figure 3 Evolution of phase separation in emulsions with CNI.
However, in order to validate any possible relationship between factor effects, we

experimented with a quantitative approach for easier validation of our conclusions,
through response-surface analysis between wheying off profiles. In order to do so, we
performed a non-linear regression analysis for each wheying off evolution profile by
fitting the following mathematical relating thickness in centimeters of serum
phase (cms) vs. days of phase separation (t):
Cms = a(1- e(-k.t))P
Where a , p , and k represent fitting constants for each curve fitting, at p=0.05.
From each fitted wheying off profile from the experimental design, a new parameter of
comparison of instability, Smax (maximum speed of serum phase increase, in c d d a y
units) was calculated by setting the second derivative 62cms/(6t)2=0.The response-surface
analysis was achieved by combining those Smax as dependent variable, and setting
Smax=O for treatments performing gradient creaming because of their independancy from
wheying off.
.Figure 4 Response-surface curve for interaction effect between %CNI-%WPI on Smax.

252 Gums and Stabilisers for the Food Industry 1 1
As expected, results of wheying off response-surface analysis revealed that quadratic

effect of pH on Smax is the most significant factor (result not shown), where the maximal
instability (Smax) is reached at pH 5.0 confirming our previous observations and
discussion. However, another significant effect is the %CNI-%WPI interaction (Figure 4),
where we see a general tendency towards maximal instability when CN/WPI ratio is
minimised. As reported with other protein-polysaccharide model systems680, the
conditions of compatibility between WPI proteins and CN (pH25.5), with low CN/WPI
ratio (low [CN]) would favor emulsion destabilisation by bridging flocculation, whereas in
conditions of incompatibility (pH55 .O), it would destabilise the emulsion by the excluded
volume effect of CN, with its lower ability to adsorb to neutral lipids (vegetable oil) when
pH is decreased due to loss of hydrophobicity (higher charge density)*. Since WPI alone
in emulsion formulations does not resist wheying off, Fig. 4 indicates that an increase in
[CN] slowing down wheying off, due to the viscosifying effect, as reported for other gums
in other ~ystems~~. Finally, about the experimental design at p=0.3 M, no significant
factor effect was observed, due to comparable rapid destabilisation in most wheying off
conditions.
If we now specifically compare the effect of p on Smax for each treatment (Fig. 5 ) , we
see a general tendency of increase in Smax when p=0.3 M, which is explained by the
charge screening e f f e ~ t ~ that
~ reduces
~ ~ the electrorepulsive property of the lipid coating
(containing WPI proteins and/or chitosan). However, the p effect is negligible at pH 5.0,
because of this already lowest stable pH condition against wheying off, as discussed
earlier. However, at pH25.5, the p destabilising effect is the most effective because, as
reported in the literature, increases of p could convert compatibility into incompatibility
between adsorbed protein and polysaccharide, that converts a gradient creaming towards a
wheyng off.
n
h 10
m
-0
x 4
n o
Conditions (%CNl/O/oWPl/pH)
Figure 5 Comparison of Smax values for CNI (a),and CNI with ,u=0.3 M (0).
Error bars represents standard error of estimate at p=O.O5
Interfacial Behaviour and Gelationof Proteins 253
3.3 Comparison of the gradient creaming profiles
21
of
gradient 0.4
creaming 0
(cm) s
I 2
\ 2
I 2
\ 2
\
0
I
v)
T: 2-
v) v)
2ln
\
v)
2
I
2
\
v)
0 v) v) 7
b. N N 0
8 2 2
Conditions (O/&Nl/%Wl/pH)
Figure 6 Effect of conditions on mean thickness of gradient creamingphase for CNI
(U),and CNI with p =0.3M (0). Error bars represents the least signijkant
difference (LSD) atp=0.05
In order to better discriminate the generally close and slow gradient creaming profiles
observed (see Fig. 3), multiple comparison tests (LSD test) between average creaming
thickness for each treatment during 20 days were done. Graphical results are presented in
Figure 6 with or without p=0.3M. If we compare the effect of composition on the stability
against gradient creaming, we see mostly the same creaming thickness, with a small
tendency towards lower values at higher %CN (0.225%), as a result of the viscosifying
stabilising effect. However, with p=0.3M, we see that some value bars representing
creaming treatments are absent due to the destabilisation towards a wheying off.
Furthermore, among the remaining treatments, condition at pH 6.0 was the only one that
was significantly unaffected by p. Therefore, we conclude it is the most stable condition
from our experimental conditions. This result could be explained by this more favorable
pH condition for interaction and co-adsorption of chitosan with proteins at the interface,
associated with an increase of electrostatic and steric repulsion between droplets due to a
more cationic and hydrophilic state of the lipid coating, but also because p without salt
added at pH 6.0 (=0,19M) is the nearest to 0.3 M. Consequently, as in the case of wheying
off there is predominant effect of pH and p on the stability against gradient creaming.
3.4 Results of protein load determinations
Results of protein load for gradient creaming conditions are presented in Table 1, along
with corresponding values of mean droplet diameters. If we keep our attention at pH 5.5,
we mainly observe a decrease in protein load when %CN is increased ( % W I constant),
whereas it increases when %WPI is increased (%CN constant). These results are
indicative of interfacial co-adsorption of chitosan with protein that would be responsible
for the stabilising properties of chitosan in conditions of compatibility (pH>pI WPI).
However, more repeats and treatments would be necessary in order to evaluate the
significance of those tendancies.
Table 1 Results of protein load for gradient creaming conditions with corresponding
values of mean droplet diameter and mean thickness of creamedphase.
4 CONCLUSION
In a model system containing WPI as emulsifier and low concentrations of chitosan as

stabiliser, we observe two kinds of phase separation phenomena (wheyng off, mainly at
pH15.0 and gradient creaming at pH25.5). Among the various factors studied, pH and p
are the ones that affected stability the most, demonstrating the predominance of
electrostatic interactions between WPI proteins and chitosan on stability. In fact, due to its
cationic charge, chitosan is distinctive among other anionic gum stabilisers already used in
industrial food emulsions by the way that it could perform interfacial stabilisation by
coadsorption with WPI proteins at pH>pI, provided a CN:WPI ratio>l:10 and p<0.3 M.
These results suggest suitable applications of chitosan in low acidity milk products
(whipping and ice creams, milk chocolates, etc.)
Acknowledgements
The financial support from Conseil de Recherche en Peche et Agroalimentaire du Qutbec

and Les Pecheries Marinard Ltee. is gratefully acknowledged.
References
Y. Cao, E. Dickinson and D.J. Wedlock, Food Hydrocolloids, 1990,4(3), 185.

J. Leman and J.E. Kinsella, Crit. Rev. in Food Sci. and Nutr., 1989,28(2), 115.
P.S. Prajapati, S.K. Gupta, A.A. Pate1 and G.R. Patil, in Brief Communications of
the Z I I I International Dairy Congress, Montreal, October 8-12, vol. 11,
International Dairy Federation, Brussels, Belgium, 1990, p. 53 1.
E. Dickinson, in Food Structure-Its Creation and Evaluation, eds. J.M.V. Blanshard
and J.R. Mitchell, Butterworths Publ., London, 1988, p. 41.
E. Dickinson, in Gums and Stabilisers for the Food Industry, 4'hEdn., eds. G.O.
Phillips, D.J. Wedlock and P.A. Williams, IRL Press, Oxford, 1988, p. 249.
E. Dickinson, in Food Polysaccharides and their Applications, ed. A.M. Stephen,
Marcel Dekker, N.Y., 1995, ch.15, p. 501.
Interfacial Behviour and Gelation of Proteins 255
7 E. Dickinson, in Macromolecular Interactions in Food Technology, ACS series (No.

650), ed. N. Parris, American Chemical Society, Washington D.C., 1996, ch.16, p.
197.
8 E. Dickinson, Trends in Food Science & Technol., 1998,9,347.
9 E. Dickinson and S.R. Euston, in Food Polymers, Gels and Colloids, ed. E. Dickinson,
Royal Society of Chemistry, Cambridge, UK, 1991, p. 132.
10 A. Syrbe, W.J. Bauer and H. Klostermeyer, Int. Dairy Journal, 1998,8, 179.
11 Y. Cao, E. Dickinson and D.J. Wedlock, Food Hydrocolloids, 1991,5(5), 443.
12 E. Dickinson and V.B. Galazka, Food Polysaccharides, 1991,5(3), 28 1.
13 S.K. Samant, R.S. Singhal, P.R. Kulkarni and D.V. Rege, Intl. J. Food Sci. Technol.,
1993,2S, 547.
14 R.S. Ward-Smith, M.J. Hey and J.R. Mitchell, Food Hydrocolloids, 1994, S(3-4),
309.
15 J.G. Winterowd and P.A. Sandford, in Food Polysaccharides and their Applications,
ed. A.M.Stephen, Marcel Dekker inc., N.Y., 1995, ch.13, p. 441.
16 R.A.A. Muzzarelli, Carbohydr. Polym., 1996,29,309.
17 R.A.A. Muzzarelli, CLMS Cell. Mol. Life Sci., 1997,53, 131.
18 J.L. Nauss and J. Nagyvary, Lipids, 1983, 18(lo), 714.
19 P.C. Schulz et al., Colloid Polym. Sci., 1998, 276, 1159.
20 L.F. Del Blanco et al., Colloid Polym. Sci., 1999,277, 1087.
21 P. Faldt, B. Bergenstahl and P.M. Claesson, Colloids and Surfaces A: Physicochem.
and Eng. Aspects, 1993, 71, p. 187
22 S. Magdassi et al., J. Microencapsulation, 1997, 14(2), 189.
23 S. Alamelu and K. Panduranga Rao, J.Microencapsulation, 1990,7(4), 541.
24 Y.I. Cho et al., J. Agric. Food Chem, 1998,46(9), 3839.
25 G.A.F. Seber and C.J.Wild, in Nonlinear Regression, Wiley series, N.Y., 1989,
ch.7, p. 327
DEPLETION FLOCCULATION BY POLYSACCHARIDES IN WHEY PROTEIN
STABILISED EMULSIONS AT HIGH WHEY PROTEIN CONCENTRATIONS
T.B.J. Blijdenstein, E. van der Linden2,T. van Vliet and G.A. van Aken.
Wageningen Centre for Food Sciences (WCFS), P.O. Box 557, 6700 AN, Wageningen,
the Netherlands
2
Laboratory of Food Physics, Department of Agrotechnology and Foodscience,
Wageningen University, P.O. Box 8 129,6700 EV, the Netherlands
3NIZ0 food research, P.O. Box 20,6710 BA Ede, the Netherlands
1 INTRODUCTION
Whey proteins are widely used in the food industry because of their nutritional value and
their gelling and emulsiQing capacities. Oil-in-water emulsions stabilised by whey
proteins show excellent stability a ainst coalescence and aggregation if sufficient protein is
f
available to cover the oil droplets. Their stability at low volume fractions of oil, however,
is strongly reduced by means of creaming.2 In order to reduce the creaming rate one
usually enhances the viscosity of the aqueous phase by using thickening agents3 However,
thickening agents (often polysaccharides) are also found to induce droplet flocculation due
to which leads in turn to an increase of the creaming rate. If the flocculation
rate dominates over the creaming rate, the oil droplets will form a space-filling network:
strongly retarding creaming of the oil droplets.6 Understanding the interplay between these
two opposing effects will add to the formulation flexibility of oil in water emulsions.
In some recent ~ o r k surfactants
~ - ~ were used as emulsifier. Food emulsions, however,
are complex systems in which the main emulsifier is often a protein and both proteins and
polysaccharides are present in the aqueous phase. Tuinier studied the effect of an
exocellular polysaccharide in whey protein-stabilised emulsions at a protein concentration
that was just high enough to cover the oil droplets. Higher protein and polysaccharide
concentrations were studied by Reiffers-Magnani et al. in another system than Tuiniers.
After centrifigation, they observed a three-phase system consisting of a cream phase at the
top of the container, a protein-enriched phase at the bottom and a polysaccharide-enriched
phase in between. At concentrations below the demixing line for proteins and
polysaccharides in the aqueous phase they found typical creaming profiles, common for
flocculated polydisperse emulsions.2
In the present paper we describe the creaming behaviour of oil in water emulsions,
which are stabilised by an excess of whey protein isolate (WPI), and contain various
polysaccharides. The polysaccharide and protein concentrations were chosen below the
demixing line for the aqueous phase. By using diffusing wave spectroscopy (DWS) we
monitored the mobility of the oil droplets, providing information on the flocculation
process.
2.1 Materials
Whey protein isolate (WPI, Bipro) was obtained from Davisco International. It contained
71 % P-lactoglobulin, 12 % a-lactalbumin, 5 % bovine serum albumin, 5 %
immunoglobulins, 2 % salt, 1 % lactose and 4 % moisture. An exocellular polysaccharide
(EPS), excreted by Zactococcus Zactis strain NIZO B40, was produced in the NIZO pilot
plant, as described elsewhere.* A commercial galactomannan (Meyprogat 7) was obtained
from Meyhall Chemical (Kreuzlingen, Switzerland). Dextran (M, = 2 x lo6 Da), NaCl and
NaN03 were purchased from Sigma Chemicals (St. Louis, USA). Thiomersal was
purchased from Merck (Schuchardt, Germany). Sunflower oil (Reddy, Vandemoortele, the
Netherlands) was purchased from a local supermarket.
2.2 Sample Preparation
All solutions were prepared from a stock solution, which contained 0.1 M NaCl and 0.02
% (w/w) thiomersal to inhibit microbial growth. Protein solutions were prepared by adding
the protein to the required amount of the stock solution. WPI was dissolved by gentle
stirring overnight, avoiding incorporation of air. The pH was set at 6.7. WPI was not
completely soluble and gave slightly turbid solutions.
Stock emulsions contained 40 % (w/w) sunflower oil and 1 % (w/w) protein. Oil was
mixed into a protein solution, using an Ultra Turrax. This pre-mix was homogenised 10
times in a Delta lab-scale homogeniser, operating at a pressure of 50 bar. The particle size
distribution was measured using a laser diffraction apparatus (Coulter LS 230, Miami,
USA).
Dextran solutions were prepared by adding dextran to the stock solution and stirring,
while heating au bain marie for 4 hours. Galactomannan solutions were prepared by
adding the powder to boiling distilled water and stirring overnight. Because the material
was not completely soluble, the resulting dispersion was centrifuged and the supernatant
was collected for hrther use. The galactomannan concentration was determined by
measuring the weight of a dried sample of the solution. The ionic strength was adjusted to
0.10 M by adding the required amount of a NaCl solution. EPS was added to the stock salt
solution and dissolved by gently stirring overnight.
Emulsion samples, containing 10 % (w/w) oil and various amounts of protein and
polysaccharide, were prepared by gently mixing various amounts of the described
solutions. In this way the droplet size distribution of the emulsions was kept constant.
2.3 Creaming Experiments

Creaming behaviour of the emulsions as a function of time was monitored using a
Turbiscan MA 2000 apparatus (Ramonville St. Agne, France), which measures the
backscattering intensity of incident laser light along the height of an optical glass tube.
Since the backscattering intensity varies with the volume fraction of oil droplets, the
backscattering profile can be interpreted qualitatively as a creaming profile.
Emulsion samples were put into tubes using an automatic pipette and were stored at
ambient temperature (20-25C). At different times after placing the sample in the tube, the
creaming profile was measured. The development of the creaming profile with time
(Figure 1) allowed us to distinguish between creaming and enhanced creaming due to
flocculation.2
Figure 1 Typical creaming profiles of polydisperse emulsions at different times. The

arrows indicate the movement of the oil droplets. A : a non-flocculating emulsion. B: a
flocculating emulsion
For determination of the amount of oil in the serum phase, emulsions were brought into
a separating funnel and left to cream for twenty hours. AAer this time, the serum phase was
collected and its weight was determined. Subsequently, the serum was mixed by gently
shaking for determination of the average oil concentration and particle size distribution.
The oil concentration was determined by the procedure described by de Jong and
Badings". From the oil concentration in the serum and the weights of the serum phase and
the cream layer (ws and w, respectively) the volume fractions of oil ($s and q$) were
calculated.
2.4 Diffusing Wave Spectroscopy (DWS)
DWS measurements were carried out in a transmission geometry" (Figure 2), using a
Heme laser ( h = 633 nm) and a cuvette with an inner width, L, of 1.98 mm. The
transmitted and scattered light was focussed by a lens into a single-mode fibre, attached to
a photomultiplier (ALV, SO-SIPD, LangedHessen, Germany). The signal was analysed by
a correlator board (ALV flex5000, LangedHessen, Germany), which calculated the
intensity autocorrelation function g2(9 of the transmitted light, collected during 1 minute.
Within this period enough fluctuations have occurred to obtain an adequate ensemble
average of light pathways through the sample. By plotting gz(9-1 against the correlation
time the half-life time tl/2 was determined from the average correlation function. This tl/2
was defined as the time at which gr(t,-l decayed to half of its initial value (Figure 3).
'
optical
Correlator
A
Figure 2 Schematic drawing of the D WS measuring setup
Figure 3 Example of an intensity autocorrelationfunction
3.1 Creaming experiments
The creaming profiles of emulsions containing varying amounts of protein and

polysaccharides were measured. Figure 4 shows the measured stability diagrams for three
polysaccharides. At higher protein concentration, the polysaccharide concentration above
which flocculation occurred (cPf)was also higher. In the range of polysaccharide
concentrations studied, however, a turbid serum phase was observed, indicating that some
of the droplets flocculated while the rest remained in the serum as single droplets. The
intensity of the backscattered light in the serum phases was higher at higher protein
concentrations. In order to investigate a possible relation with the volume fraction of oil in
the serum phase we measured the oil concentration at a polysaccharide concentration just
above cP) Table 1 shows that wc decreased with increasing protein concentration for
emulsions containing 0.24 % dextran. The average value of 4s was larger at a higher
protein content. Clearly fewer droplets creamed at higher protein concentrations. A clear
increase of q& was found with increasing protein concentration. A high volume fraction of
oil in the cream phase is indicative for creaming of single droplets, which give a less
voluminous cream layer than aggregates.I2 This suggests that fewer flocs are formed at
higher protein concentrations.
0,12
b
0
0,04
0,02
0
0 2 4 6 0 2 4 6 0 2 4 6
%I[% Whl %PI t% W h l %PI 1% wlwl

Figure 4 Stability diagrams of three different polysaccharides plotted against the WPI-
+
concentration. creaming; A enhanced creaming. A: EPSMPI; B: galactomannanMPI;
C: dextranNPI
Table 1 Values of the weights of the serum and the cream .(ws and we resp), and the
volume fractions of oil (@s and &-) in these phases for I 0 % (w/w) oil-in-water emulsions at
different protein concentrations after 20 hours of creaming
CWPI w, Wr 4 s 4r
[% WIW] k1 [gl [-I [-I
0.23 14.1 5.89 0.0011 0.36
0.45 14.4 5.59 0.002 1 0.37
0.88 15.3 4.59 0.0026 0.45
1.76 16.5 3.48 0.0038 0.59
2.65 16.8 3.20 0.0044 0.63
3.52 16.4 3.63 0.0043 0.57
5.28 17.7 2.28 0.0056 0.89
The droplet size distributions in the serum phases show, at all WPI concentrations, that
the peak of the size distribution had shifted to a smaller droplet size compared to the
original sample (Figure 5). At increasing protein concentrations, however, the shift of the
particle size decreased with increasing protein concentration. Apparently more and larger
droplets remained in the serum phase at higher protein concentrations.
. Initial emulsion
0,Ol 0,1 1 10 100
droplet diameter [pm]
Figure 5 Droplet size distribution in the serum phases of I0 % oil-in-water emulsions with
varying amounts of WPI after creaming for 20 hours, compared to the size distribution of
the initial emulsion
3.2 Diffusing Wave Spectroscopy
To observe the effect of the protein on the mobility of the single droplets, oil-in-water
emulsions at varying concentrations of protein were analysed by DWS in the absence of
dextran. The measured values of t l appeared
~ not to change significantly with time at each
protein concentration (Figure 6). However, an increase in the protein concentration caused
an increase of tl/2, implying a reduced mobility of the droplets. This suggests that protein
enhances the viscosity of the medium.
Interfacial Behaviour and Gelation of Proteins 26 1
180
160
140
120
n
5
Y
100
c! 80
CI
60
40
20
0
0 100 200 300 400 500 600
time [s]
Figure 6 Half-life times of intensity autocorrelationfunctions plotted against time, for I0
% (w/w)oil-in-water emulsions at various WPI concentrations
DWS measurements were also performed on systems containing 0.24 % dextran. The
autocorrelation functions were measured during 30 minutes (figure 7). At 0.2 % protein, a
steady increase of the t l / 2 was found, indicating a decrease of the mobility of the emulsion
droplets due to flocculation. Increasing the protein concentration led to an increase of the
initial value of t l / z , which is probably due to a higher viscosity of the aqueous phase.
Regarding the evolution of t1/2 with time, an increase was observed which can be attributed
to aggregation of the oil dr0p1ets.I~However, the increase of t l / 2 levels off sooner at higher
protein concentration and the final increase of t1/2 is smaller. Based on these findings we
speculate that the effect of depletion flocculation, caused by polysaccharides, is reduced by
protein. This would result in a protein dependent transition from creaming to enhanced
creaming (Figure 4).
250
200
150
Y
2
2 100
50
0
0 500 1000 1500 2000
time [s]
Figure 7 Half-life times of intensity autocorrelationfunctions plotted against time, for I0
% (w/w) oil-in-water emulsions, containing 0.24 % dextran, at various WPI concentrations
4 CONCLUSION
The experimental work has shown that an excess amount of whey protein in the bulk has a
marked influence on the flocculation and creaming behaviour of 10 % oil-in-water
emulsions, caused by addition of either of the polysaccharides EPS, galactomannan and
dextran. DWS measurements, volume fractions of oil in the cream layers and droplet size
distributions in the serum phase strongly indicate that the amount of droplets that
flocculate decreases with increasing protein concentration. This is likely related to the
measured shift of the peak in the particle size distribution curve of the serum phase
towards a larger particle size.
For a full understanding of this effect, a more complete characterisation of the
rheological behaviour of the aqueous phase is required. This will be the subject of a
forthcoming publication.l4
Acknowledgements:
The authors would like to thank Franklin Zoet and Ingrid Nijssen for carrying out part of
the experimental work.
References
G.A. van Aken and F.D. Zoet, Langmuir, 2000,16, 7131.
M.M. Robins, Current Opinion Coll. Int. Sci., 2000,5, 265.
A. Syrbe, W.J. Bauer and N. Klostermeyer, Int. Dairy J., 1998,8, 179.
Y.H. Cao, E. Dickinson and D.J. Wedlock, Food Hydrocolloids, 1990,4, 185.
K . Demetriades and D.J. McClements, J. Food Sci., 1999,64, 206.
P. Manoj, A. Fillery-Travis, A.D. Watson, D. Hibberd and M.M. Robins, J. Coll. Int.
Sci., 1998,207, 283.
W.C.K. Poon and M.D. Haw, Adv. Coll. Int. Sci., 1997,73, 71.
R. Tuinier, An exocellular polysaccharide and its interactions with proteins, 1999,

Thesis, Wageningen Agricultural University, the Netherlands
C.K. Reiffers-Magnani, J.L. Cuq and H.J. Watzke, Food Hydrocolloids, 2000,14, 521.
10 C. de Jong and H.T. Badings, J. high Res. Chrom., 1990,13,2,94.
11 D.A. Weitz and D.J. Pine, In Dynamic Light Scattering, ed. W . Brown, Clarendon
Press, Cambridge, 1993, ch.16, p. 652.
12 E. H. A. de Hoog, Interfaces and Crystallization in Colloid-Polymer Suspensions,

2001, Thesis, State University of Utrecht, the Netherlands
13 E. ten Grotenhuis, M. Paques and G.A. van Aken, J. Coll. Int. Sci., 2000,227, 495.
14 T.B.J. Blijdenstein, T. van Vliet, E. van der Linden, and G.A. van Aken, Depletion
flocculation in oil in water emulsions at excess concentrations of whey proteins.
Unpublished Work
AGGREGATION AND GELATION OF GLOBULAR PROTEINS WITH AND
WITHOUT ADDITION OF POLYSACCHARIDES
D. Durand'., C. Le Bon', P. Croguennoc', T. Nicolai', A.H. Clark2

Polymkres, ColloYdes, Interfaces, UMR CNRS, UniversitC du Maine, 72085 Le Mans
Cedex 9, France
Unilever Research Colworth, Sharnbrook, Bedford, MK441LQ, U.K.
E-mail address: Dominique.Durand@univ-lemans.fr
1. Introduction
Globular proteins denature if heated which generally leads to aggregation and
eventually a gel is formed if the protein concentration is sufficient.' The aggregation
process and the structure of the gels depend on external conditions such as the pH, the
ionic strength, the protein concentration and the temperature. One of the most intensively
studied globular proteins is p-lactoglobulin (p-1 ) which is the major component of whey
and has molar mass 18600g/mol and radius 2mF4 and is widely used in the Food Industry.
Thus, it is important to understand the heat induced aggregation process either because one
wants to avoid it, e.g. in sterilisation of milk, or one wants to exploit it for applications.
We present here the main results of an extensive study of the denaturation of this
protein at pH7 in various experimental conditions: ionic strength, concentration,
temperature, and presence or not of olysaccharide. This research has been carried out in
our laboratory over the last decade!I6 Experimental and technical details of the results
shown in this paper can be found in the original papers where the reader can also find
references to the relevant literature. One aspect that has received relatively little attention
until recently, is the influence of phase separation on the gel structure. We will address this
issue here for the specific case of mixed systems of p-lg and K-carrageenan ( ~ - c a r ) . ' ~ - l ~
The latter is a polysaccharide produced by blue algae and is fully compatible with native p-
lg under the conditions of our investigation. However, K-car and p-lg aggregates are under
some conditions incompatible, which leads to (micro) phase separation with important
consequences for the gel structure. We will argue that the effect of micro phase separation
may play an important role even for pure globular protein systems.
2. Aggregation and gelation of P-lactoglobulin

p-lg aggregates are stable to cooling and dilution and can be characterized in dilute
solutions at room ternperat~re.~~l2 Figure 1 shows an example of the size distribution of p-
lg after different heating times obtained by size exclusion chromatography (SEC). With
increasing heating time the aggregates grow in size and the size distribution broadens,
while the fraction of native proteins decreases. The chromatographs show a clear
separation between a narrow peak corresponding to residual native proteins and a broad
peak that corresponds to the aggregates. This observation implies that the minimum size of
the aggregates contains already many monomers and that no or very few stable oligomers
are formed. This separation is also observed in dynamic light scattering experiment^.^
0 1000 min
0 Omin v 2000min
v 20min 0 3000min
0 40min 0 4300min
A 5700min
0 100min 104
A 600 min @ 10000min
h
-
Y
0-
103
.
102
15 20 25 30 107
Elution volume (ml) 9
Figure 1 Figure 2
Chromatograms of p l g solutions at pH7, q-dependence of the scattered light
O.1M salt and C=I8.6g/l after different intensity (Ir) by dilute P l g aggregates
heating times at 60C. formed after different heating times of the
same system as Fig. 1. The straight line
at t= I0000 min. has slope -2.
In order to characterize the growth of the aggregates we have used light ~cattering.~
Figure 2 shows the scattering wave vector (9) dependence of the intensity scattered by the
aggregates formed after different heating times. The solutions were highly diluted so that
interaction may be neglected. The scattering intensity (Ir) is normalized by that of p-lg
monomers at the same concentration, so that the value of I, extrapolated to q=O
corresponds to the weight average aggregation number (m,). Clearly m, increases with
increasing heating time, while the q-dependence becomes more important because the
average size of the aggregates increases. For the largest heating time, in the light scattering
space window, we probe the internal structure of large aggregates and we observe a power
law dependence of I, on q: I,=q-df. The exponent d ~ represents 2 the so-called fractal
dimension of the aggregates. Small angle neutron scattering experiments have revealed that
the internal cut-off of the self-similar structure with a fractal dimension d ~ is2much larger
than the native protein and corresponds to a primary aggregate of about 100 proteins
with a radius 15nm. Dynamic light scattering experiments showed that the internal
dynamics are intermediate between those of fully flexible polymer and rigid aggregates.12
Figure 3 shows micrographs of p-lg aggregates obtained using cryo-TEM. At pH7
small elongated aggregates are formed at low ionic strength while at high ionic strength we
observe larger aggregates that appear to be formed by random association of the small
aggregates. At low protein concentrations and low ionic strength the primary aggregates
are inhibited from hrther association, probably due to electrostatic repulsion. However, as
shown in Figure 6, at high protein concentrations we do observe further association
partially caused by the presence of protein counterions that increases the ionic strength.
Inte@acial Behaviour and Gelation of Proteins 265
The structure of the aggregates does not depend on the heating temperature or the protein
concentration.
Figure 3 Cryo-TEA4micrographs of F l g aggregatesformed upon heating at pH7, O.OO1M

salt (a) andpH7, O.lMsa1t (a).
It appears that the aggregation of p-lg occurs in first instance by a unidirectional

growth. This first step of the aggregation leads to particles of a small and rather well
defined size. One may speculate that the aggregation is initially a process of nucleation and
growth. This is especially clear at pH2 and low ionic strength, where immediately very
large rod-like ag egates are formed even if only a tiny fraction of the native proteins has
been consumed.But also at all other conditions investigated we never observed the
formation of oligomers. At present we can only speculate and we do not know why the
initial growth of the aggregates stops at a relatively small size at pH7. The second step of
the aggregation, which is clearly distinguished at pH7, involves the entirely different
process of random association of the primary aggregates. Figure 4 shows a schematic
drawing of the aggregation process at pH7 and 0.1M salt.
a 3 T7l
4 + o ~ o * - + q @ 3 -
native native denature
- 30 nm
primary
-D Gel
dimer monomer monomer aggregate

fractal cluster of
primary aggregates
Figure 4 Schematic representation of aggregation process of p l g at pH7 and 0.lMsalt.
In order to form a gel the aggregates have to grow sufficiently large to fill up the whole
space. If aggregates have a fractal structure there is no lower concentration limit for s ace
filling and gelation to occur, as has been demonstrated by computer simulations. Of
course, as the concentration decreases the gel becomes less dense and weaker so that in
practice there is always is lower concentration limit below which the gel modulus is no
longer measurable. For globular proteins, however, a lower concentration limit for gelation
(C,) is observed for a different reason. This is illustrated in figure 5, where we plotted the
growth of the aggregates with heating time at different concentrations. Above about 1Og/L
we observe a clear divergence of the aggregate size after which a gel is formed. Below this
concentration the growth slows down after some time. The stagnation of the growth only
leads to a delay of the gelation if it occurs when the aggregates are already close to space
filling conditions, However, if the growth stagnation occurs when aggregates are still
small, the gel is not formed. We do not know why the aggregate growth stagnates, but we
have shown that it is related to the stagnation of the consumption of native monomers.*
105
10.4
s=
With decreasing ionic strength the further association of the primary aggregates becomes
slower due to electrostatic repulsion. Consequently, C, increases with decreasing ionic
strength, see Figure 6 , and if the pH is fiu-ther from the isoelectric point.
I
I 1
I
5 gll I
3 I=O,1 M
I=0,003 M
0
o
I
E 0
103
* *
102
10 102 103 104 105 I 10 100
t (min)
Figure 5
Growth of the weight average Evolution of the z-average hydrodynamic
aggregation number (inw) of P l g radius ( R h j of P l g aggregates at the
aggregates heated at 67C at different plateau for concentrations lower than the
concentrations @H7, O.IM salt). The critical concentrations of gelation: 7 gll
dashed lines indicate the gel times. The at 0.I M and 80 g/l at 0.003 M.
solid lines are guides to the eye.
At pH7 and high ionic strength the structure of globular protein gels is traditionally
described in terms of so-called particulate protein gels typically which show dense
domains with a characteristic size that varies with the ionic strength, the pH2* and the
heating conditions.21 We speculate that the dense domains result from micro phase
separation, which is observed very clearly in mixed systems.
3. Aggregation induced phase separation

We have investigated the aggregation and gelation of p-lg at pH7 in the presence of K-
car.13-16We have established that at the conditions used in our investigation the two
biopolymers have no specific interaction. Figure 7 compares micrographs of gels with and
without K-car. Clearly the presence of K-car leads to much more heterogeneous gels
containing domains rich in p-lg with the polysaccharide situated between these domains.
Under certain conditions of high K-car concentrations and slow aggregation, a gel is not
Interfacial Behuviour and Gelation of Proteins 267
formed, but instead we observe a viscous sediment containing most of the protein and a
clear supernatant which contains most of the polysaccharide.
Figure 7 TEM micrographs of p l g gels formed after 2 hours at 78C pH7, O.IM with
60gA P L g (a) and 60g/l P L g + 4g/l K-Car (b)
Figure 8 shows the fraction of unaggregated proteins as a function of heating time at

70C for a solution containing 20g/l p-lg and varying amounts of K-car between 0 and 8
g/l. Within the experimental error there is no influence of K-car on the rate of protein
consumption, i.e. on the rate of the first aggregation step.16
Figure 9 where we have plotted m, as a function of Rg for samples of different K-car
concentrations between 0 and 1 g/l. shows that, in this regime, the aggregates have the
same structure independently of the K-car concentrations.'6
1.o 10 6
0.8 105
0.6 104
Lr, E'
0.4 103
0.2 102
0.0 10'
10 100 1000
1 10 100 1000
time (min) RBZ (nm)
Figure 8 Figure 9
Fraction of unaggregated p l g as a m, as a function of Rgz for p l g
function of heating time at 70C in a aggregates formed by heating a solution
solution containing 20g/l p l g and of 20g/l p l g and various m a r
various concentrations of K-Car indicated concentrations. Symbols are as in Fig.8.
in thefgure. The straight line has slope 2.0
268 G u m s and Stabilisers for the Food Industry 11
So, it appears that neither the rate of the native protein consumption nor the structure of
the aggregates is modified by the presence of K-car. On the other hand, K-car accelerates
the second step of the aggregation process, i.e. the growth of the aggregates, as is shown in
figure 10. When we compare the growth at different Ic-car concentrations we find that
initially the growth is the same as in the absence of K-car. But once the p-lg aggregates
have reached a certain size the growth rate increases and a gel is formed rapidly. The
influence of K-car becomes significant when its concentration approaches the overlap
concentration which may be estimated as C*=3Mw/(N,47cRg,) and is about OSg/l. The size
of the p-lg aggregates where K-car influences the growth rate decreases rapidly with
increasing K-car concentration between 0.25 and 1.O g/l, although the correlation length of
K-car varies little in this range (1 6 - 26 nm). 22
106
1000
105
104
100
-5
~
tiS (I0
103
W
102
.~
~~ -
1
1 10
10 100 1000
t(min)
Figure 10 Figure 11
Time dependence of m, and R, for P l g Dependence of the storage shear modulus
aggregates formed in a solution of 20gA G' at 1Hz on the K-car concentrations
P l g and various concentrations of K-car after heating a solution of 50gA Plg for
heated at 70C. Symbols are as in Fig.8. 2h at 75C.
If we use higher K-car concentrations we observe very rapid formation of large particles
together with small aggregates and the residual fraction of native proteins. These systems
either develop into a gel or phase separate into a turbid bottom phase and a transparent top
phase. Whether these systems gel or phase separate, depends not only on the
concentrations of p-lg and K-car but also on the temperature.16
We have investigated the structure of the gels on a larger scale using confocal
microscopy. Within the resolution of the confocal microscope, the p-lg gels formed in the
presence of no, or a small amount, of K-car appear homogeneous. However, in the presence
of more K-car one observes a micro phase separation leading to the formation of p-lg rich
domains that appear to cluster.l 6
As shown in Figure 11, the rresence of Ic-car has a profound influence on the
mechanical properties of the gels. On the one hand the gels form more quickly with
increasing K-car concentration, while on the other hand the coarsening of the gel structure
caused by the microphase separation weakens the gel. The gel modulus as a function of K-
Inter$acial Behaviour and Gelation of Proteins 269
car concentration goes through a maximum, and, as mentioned above, in some cases we do
not even observe gel formation, but rather macroscopic phase separation.
We have studied the phase separation distinct from the aggregation process by mixing
p-lg aggregates and K-car at room temperature. After mixing, one observes the formation
of spherical micro-domains rich in protein aggregates, see figure 12. The micro-domains
slowly precipitate and form a viscous sediment. The micro-domains contain the larger
aggregates while the residual native proteins and smaller aggregates remain in solution
together with the polysaccharide. As can be seen in figure 12 the micro-domains have a
tendency to stick together, but the sediment can be easily redispersed in the form of small
clusters of micro-domains. If we increase the K-car concentration, smaller and smaller
aggregates will phase separate and become part of the micro-domains.
Figure 12 Micrographs of protein rich micro-domains obtained with optical microscopy

for a mixture of 5.7 g/l p l g aggregates with Rg,=20nm and 9.2 g/l K-car.
The formation of the micro-domains is due to spinodal decomposition into two liquid
phases. If the system is diluted soon after the micro-domains are formed, we recover the
initial distribution of aggregates. However, the longer we wait the slower becomes
dissolution of the micro-domains after dilution. If we wait for more than a few days the
micro-domains can no longer be dissolved by dilution. The reason for the progressive
stabilisation of the micro-domains is very slow association of the aggregates that are
concentrated in the micro-domains. We have observed such slow association at room
temperature in more concentrated solutions of p-lg aggregates even without phase
separation.
The size of the micro-domains depends somewhat on the concentration of K-car and p-
lg, but is close to that of the dense domains seen in the gels of the heated mixed systems,
(compare figures 7b and 12). We propose that the dense protein domains in the mixed gel
are due to micro phase separation of the growing aggregates. The micro-domains formed in
the heated mixtures cannot precipitate because they are trapped in the gel. The structure of
the gels formed by the heated mixtures is controlled by the balance between aggregation
and phase separation. As long as the aggregates are small enough to be compatible with K-
car the growth and the structure of the aggregates is not influenced by the presence of the
polysaccharide. However, as soon as they become larger they will start to phase separate.
The extent of the phase separation and thus the heterogeneity of the gels depend on the rate
of phase separation compared to the rate of aggregation. The latter is strongly temperature
dependent and can be slowed down by using a lower heating temperature. If the
aggregation rate is sufficiently slow the micro-domains can precipitate and we observe a
macroscopic phase separation.
The influence of micro-phase separation is particularly clear for the mixed system
discussed here. However, gel structures very similar to that of the mixed system are
observed in pure p-lg gels at high ionic strength or at pH close to the isoelectric point. It is
likely that also in these situations the gel structure is influenced by micro-phase separation.
4. Summary
Heat-induced denaturation of p-lactoglobulin at pH7 leads to the formation of small
primary aggregates, which may associate in a second step to forin self-similar aggregates.
At low concentrations the growth of the aggregates stagnates, while above a given
concentration C, the aggregates fill up the whole space and form a gel. Under conditions of
high ionic strength, pH close to the isoelectric point or added polysaccharide very large
aggregates will phase separate, which leads to the presence of protein rich micro domains.
In such situations the morphology of the protein gels is determined by the competition
between phase separation on the one hand and aggregation and gelation on the other.
Acknowledgements: We thank Katarina Edwards and Goran Person for Cryo-TEM

microscopy experiments, Dominique AusskrrC for optical microscopy experiments.
References
1 A.H. Clark, in Functional properties of food macromolecules, ed. J.R. Mitchell,
Elsevier Applied Science, London and New York, 1998, p.77.
2 H.A. McKenzie, in Milk proteins: chemistry and molecular biology, ed. H. A.
McKenzie, Academic Press, New York and London, 1971, Vol. 11, p 257.
3 M.Z. Papiz, L. Sawye, E.E. Eliopoulos, A.C.T. North, J.B.C. Findlay, R.
Sivaprasadarao, Jones, T.A. Newcomer, M.E. Kraulis, P. J. Nature, 1986,324,383.
4 H. Pessen, T.F. Kumosinski, H.M.J. Farrell, J. Ind. Microbiol., 1988,3, 89.
5 J.C. Gimel, D. Durand, T. Nicolai, Macromolecules, 1994,27,583.
6 P. Aymard, D. Durand, T. Nicolai, Int. J. Biol. Macromol., 1996,19, 213.
7 P. Aymard, D. Durand, T. Nicolai, Int. J. Polymer Analysis & Characterization, 1996,
2, 115.
8 P. Aymard, J.C. Gimel, T. Nicolai, D. Durand, J. Chim. Phys., 1996,93,987.
9 P. Aymard, D. Durand, T. Nicolai, J.C. Gimel, Fractals, 1997, 5,23.
10 P. Aymard, T. Nicolai, D. Durand, A. Clark, Macromolecules, 1999,32,2542.
11 C. Le Bon, T. Nicolai, D. Durand, Macromolecules 1999,32, 6120.
12 C. Le Bon, T. Nicolai, D. Durand, Int. J. Food Sci. Technol., 1999,34,45 1.
13 I. Capron, T. Nicolai, D. Durand, Food Hydrocolloi'ds, 1999, 13, 1.
14 I. Capron, T. Nicolai, C. Smith, Carbohydrate Polymers, 1999,40,233.
15 P. Croguennoc, D. Durand, T. Nicolai, A. Clark, Langmuir, 2001, 17,4372.
16 P. Croguennoc, D. Durand, T. Nicolai, A. Clark, Langmuir, 2001, 17,4380.
I7 J.C. Gimel, D, Durand, T. Nicolai, Phys. Rev. B, 1995,51, 11348.
18 M. Langton, A.M. Hermansson, Food Hydrocolloids, 1992,5,523.
19 M. Verheul, F.P.M. Roefs, J. Agric. Food Chem. 1998,46,4909.
20 M. Verheul, S. P. F. M. Roefs, K. G . De Kruif, J. Agric. Food Chem., 1998,46, 896.
21 M. Stading, M. Langton, A.M. Hermansson, Food Hydrocolloids, 1992,6,455.
22 P. Croguennoc, V. Meunier, D. Durand, T. Nicolai, Macromolecules, 2000,33,7471.
KINETIC AND EQUILIBRIUM PROCESSES IN THE FORMATION OF WEAK
GELATIN GELS
David Oakenfull and Alan Scott
Food Science Australia, P.O. Box 52, North Ryde, NSW 21 13, Australia.
1 INTRODUCTION
It is now well established that when a gelatin solution forms a gel the protein chains
undergo a conformational coil-to-helix transition during which they partially recover their
initial collagen structure (three left-handed helices wrapped into a right-handed super
helix). These helices are at least in part intermolecular and link the protein chains to form
a gel network.14 The aggregation process is highly complex and seems to involve a
number of intermediate steps, as shown schematically in Figure 1. It has been extensively
7
investigated - most recently using scattering techniques 596 and atomic force microscopy.
We have investigated this process further through a series of kinetic and rheological studies
of very weak gelatin gels across the temperature range 7" to 25C. Analysis of these
results provides a series of thermodynamic parameters that add to our understanding of the
initial stages in the formation of gelatin gels.
nucleated
amolecular
le helix
two chain
random coil
junction
Ti
-J --&-
Figure 1 Some possible conformations leading to gelation of gelatin (redrawnfrom r e t 4 )

2.1 Gelatin
The same sample of gelatin was used throughout. It was a commercial preparation (Davis
Gelatine, Sydney, Australia) containing 124 g/kg moisture. Its intrinsic viscosity
(measured at 38C in 0.2 M acetate buffer, pH 4.80) was 23.6 W g . The viscosity average
8
molecular weight calculated from this value was 49 300.
2.2 Preparation of Gels
All solutions were prepared by weight in deionised water. Gels (10 g samples) were
prepared by diluting a stock gelatin solution (50 g/kg) at slightly above the transition
temperature (40C) with the appropriate weight of water in glass screw capped vials
('scintillation vials') Gelatin solutions (10 g samples) were prepared in screw-capped glass
'scintillation vials'. These were then held at 7", lo", 15", 17", 20" or 25C using a series of
constant temperature rooms in which the temperature was constant to within M.5"C. The
gels were equilibrated for 18 hours before measurements were made.
2.3 Measurement of Initial Rate of Gelation
A very simple procedure was used to measure the time required for a solution to form a gel
with a predetermined (but very small) rigidity, without disturbing the system
me~hanically.~''~ A fixed weight (10 g) of gelatin solution of appropriate concentration
was placed in each of a series of flat-bottomed cylindrical glass vials (radius 12.7 mm) at
slightly above the transition temperature (40C). These were then transferred to a
thermostatted water bath at the measurement temperature and inverted sequentially. The
time required to form a gel just strong enough to remain in position was recorded. The
reciprocal of the setting time is then proportional to rate of gelation. (This procedure is
equivalent to estimating a rate constant by the initial rate method." ) To ensure
reproducible results, the vials were always cleaned in chromic acid before use.
2.4 Measurement of Absolute Shear Modulus
Absolute shear modulus (G, equivalent to G' at zero frequency) was measured by the
method of Oakenfull, Parker and Tanner." Measurements were made of the gels in the
cylindical glass vials (radius 12.7 mm) in which they were formed. A cylindrical brass
probe (of radius 1.25 mm) was inserted step-wise into the gel with a series of 5 second
pulses at a speed of 0.0847 m d s . A period of one minute was allowed between each step
for equilibration. The equilibrium force exerted on the probe was measured at each step
and the apparent Young's modulus (Y = stresshtrain) calculated from the slope of force vs
penetration. The absolute shear modulus was calculated from the formula G = 0.0208Y).
Measurements were made in the constant temperature room at the temperature at which the
gels were formed.
2.5 Calculation of Junction Zone Characteristics From G versus Concentration
The size and thermodynamic stability of junction zones in a non-covalently cross-linked

gel can be calculated from G vs concentration data using the theory developed by
Oakenfull.12 From the theory of rubber elasticity, the following equation can be derived
which relates G to the weight concentration (c) of the polymer:
G=--xRTc M[J]-c
M MJ[J]-c
M is the number average molecular weight of the polymer, MJ the number average
molecular weight of the junction zones and the quantity [JJis effectively the 'molar
concentration' of junction zones. The derivation relies on the assumption that the gels are
very weak, with polymer concentrations close to the gel threshold, so that the polymer
chains linking junction zones are long enough to approach Gaussian behaviour. This
limiting condition also makes it reasonable to assume that the formation of junction zones
is an equilibrium process, subject to the law of mass action. It then follows that if n is the
number of cross-linking loci that form a junction zone and KJ is their association constant:
KJ = [ J ] M J " { ~ ( c - M J [ J ] ) - "
By using numerical methods, the two equations can be combined, eliminating the quantity
[A, and giving a relationship between shear modulus and concentration in terms of M , MJ,
Kj and n. Values of these can then be obtained that best fit the experimental data for G
versus c, as shown (for 7" and 20C) in Figure 1.
*Oo0
6000
1
Q
2000
0
0 5 10 15 20 25 30
c (ms)
Figure 1 Plots of shear modulus (G) against concentration for gelatin gels at 7"(upper
curve) and 20C (lower curve)
5-
4-
=
n
-c
3-
-7 4. I
0.5 0.7 0.9 1.1 1.3 1.5
In(c)
Figure 2 Logarithmic plot of gelation time (t, seconds) against Concentration (c, g/kg)
2.6 Critical gelation concentration
Plots of shear modulus (G) against concentration, as shown in Figure 1, also give the
critical gelation concentration (co),the minimum concentration at which gelation can
occur. Values of co were estimated mathematically by calculating the polynomial curve
that best fitted the experimental data and from that estimating the value of c at which G
becomes zero.
3.1 Junction zone multiplicity estimated from kinetic data
Logarithmic plots of gelation time (t, seconds) against concentration (c, g k g ) were linear,
as shown in Figure 2. The slope indicates the average number of segments of polymer (n)
9,lO
which associate to form junction zones. Values of n and standard errors were
calculated by regression analysis. The results are summarised in Table 1. Junction
multiplicity (n) decreased with increasing temperature - from close to 3.0 at 7" to about 2.4
at 25" - suggesting that at higher temperatures, where gelation proceeds more slowly, more
helices involve protein chains folded back on themselves (i.e intramolecularly).
Table 1. Junction zone multiplicity (n) estimated from kinetic data at temperatures (t)
within the range 7" to 25C.
t ("C) 7 10 15 17 20 25
n 3.1 i0.2 3.0k0.2 2 . 9 ~ 0 . 2 3 . 0 ~ 0 . 1 2.9k0.1 2.4k0.2
Interjiacial Behaviour and Gelation of Proteins 275
3.2 Junction zone characteristicsestimated from the concentrationdependence of the

static shear modulus
The concentration-dependence of the static shear modulus ( G ) gives another measure of

junction multiplicity - plus the size of the junction zones (as a number-average molecular
weight, MJ)and their association constant (KJ). The results are summarised in Table 2.
Table 2 Junction zone characteristics for gelatin gels at difSerent temperatures calculated
from the concentration dependence of shear modulus (G)
7" 10" 15" 17" 20" 25"

M 50000 49000 51 000 48000 52000 49000
MJ 3 650 4 600 12 800 14900 23 600 22200
n 3.2 3 .O 3.1 3.0 2.9 2.4
KJ 1700 2565 68500 84400 269000 8405
AGOJ (kJ/mole) -0.92 -0.76 -0.34 -0.30 -0.17 -0.18
The values for junction zone multiplicity (n)were similar to those obtained
kinetically. Junction zones appear to involve segments from three protein chains, except at
the highest temperature where n drops to 2.4.
The size of the junction zones (MJ)increased with increasing temperature.
Presumably with these very weak gels, close to the critical gelation concentration (cg),the
size of the junction zones is close to their critical minimum size for stability. Flory and
13
Weaver showed that this critical size is inversely proportional to (Tm - T), where Tm is
the melting temperature of collagen (ca 36C). As shown in Figure 3, MJ was indeed
inversely proportional to (Tm- T ) for the lower temperatures. There is thus a critical
minimum size that depends on the balance between the loss of entropy in going to the more
ordered state and the enthalpic stabilisation due to the formation of hydrogen bonds in the
helix structure.
14
The free energy of junction zone formation is given by:
AGO^ = -RT In K~
Per amino acid residue in the average junction zone, AG'J was almost independent of
temperature, but decreased slightly with increasing temperature, to a value of about -0.17
kJ/mole at 20C.
3.3 The van't Hoff enthalpy of junction zone formation
The van't Hoff equation14 gives a relationship between the association constant (KJ) and
the enthalpy of formation of junction zones (&J):
In KJ = @J /RT + constant
40000
30000
MJ
20000
10000 .
0 u u1 uI I uI u
0 0.02 0.04 0.06 0.08 0.1
l/(Tm - T)
Figure 3. Plot of junction zone size (MJ) against l/(T,,, - T).
For temperatures of 20C and below, In KJ was inversely proportional to temperature, as

shown in Figure 4. This result suggests that at the lower temperatures &J is independent
of temperature. As we saw earlier, junction zone stability depends on the balance between
enthalpic stabilisation and entropic destabilisation:
From the linear part of the plot the van't Hoff enthalpy of formation of junction zones
(&J) was -288 kJ/mole. At 15"C, A&J = 45.9 kJ/mole, from which the entropy of
junction zone formation &J =-840 JWmole a strongly negative value consistent with
formation of ordered helical structure . The collagen triple helix structure is stabilised by
hydrogen bonds that weaken with increasing temperature.' Consequently larger helices
(with more hydrogen bonded interactions) are required for stability.
3.4 The Eldridge-Ferry enthalpy of junction zone formation
The Eldridge-Ferry equationI6 relates the enthalpy of formation of junction zones (&EF)
to the temperature-dependence of the critical gelation concentration (co):
In (cg) = &&'RT + constant

Integacial Behaviour and Gelation of Proteins 277
14 -
In KJ 12 -
10 -
8-
V .
3.3 3.35 3.4 3.45 3.5 3.55 3.6

1ooorr
Figure 4. Plot of In KJagainst 1OOOr" (van't Hoflplot).
-1.5 A
Figure 5 Plot of ln(c0) against 10o0/T (Eldridge-Ferry).
Figure 5 shows ln(c0) plotted against 1/T. From the slope of this line, A g l p was -75
kJ/mole. This is much smaller than &J (-288 kJ/mole) - presumably because only
smaller, rudimentary junction zones are present at the gel threshold.
27 8 Gums and Stabilisers for the Food Industry 11
4 CONCLUSIONS
In dilute gelatin gels the development of helical junction zone structures depends on
temperature, reflecting the fine balance between enthalpic stabilisation and loss of entropy.
For temperatures of up to 20C, the junction zones increased in size with increasing
temperature and appeared to involve segments from three protein chains. The enthalpy of
formation of junction zones (-288 kJ/mole) was independent of temperature; the entropy of
junction zone formation was negative (-840 J/K/mole)Lconsistent with formation of
ordered helical structures. At temperatures above 20C (but still below the helix-coil
transition temperature), junction zone structures appear to break down. At 25" the average
number of chain segments per junction zone went down to 2.4; the size decreased and the
association constant was significantly less than predicted from a van't Hoff plot by
extrapolation from lower temperatures.
References
1. A.H. Clark and S.B. Ross-Murphy, Adv. Polymer Sci., 1987,83,57.

2. F.A. Johnston-Banks, in Food Gels, ed. P. Harris, Elsevier Applied Science,
Barking, 1990, p. 233.
3. D. Oakenfull, J. Pearce and R.W. Burley, in Food Proteins and their
Application, eds. S . Damodaran and A. Paraf, Marcel Dekker, New York, 1997.
4. E.G. Finer, F. Franks, M.C. Phillips, and A. Sugget, Biopolymers, 1975,14,
1995.
5. I. Pezron, M. Djabourov and J. Leblond, Polymer, 1991,32,3201.
6. T. Herning, M. Djabourov, J. Leblond and G. Takerkart, Polymer, 1991,32,
321 1 .
7. A.R. Mackie, A.P. Gunning, M.J. Ridout and V.J. Morris, Biopolyrners, 1998,
46,245.
8. J. Pouradier and A.M. Venet, J. Chim. Phys., 1950,47,391.
9. D. Oakenfull and A. Scott, in Gums and Stubilisers for the Food Industry - 3,
eds. G.O. Phillips, D.J. Wedlock and P.A. Williams, Elsevier Applied Science,
London, 1986.
10. D. Oakenfull and V.J. Morris, Chem. and Znd., 1987 (16 March), 201.
11. D.G. Oakenfull, N.S. Parker and R.I. Tanner, J. Texture Studies, 1989,19,407.
12. D. Oakenfull, J. Food Sci., 1984,49, 1 103.
13. P.J. Flory and E.S. Weaver, J. Am. Chem. SOC., 1960,82,4518.
14. S. Glasstone, Textbook of Physical Chemistry, Van Nostrand, New York, 1947,
p.274.
15. G.E. Walrafen, in Water: A Comprehenxive Treatise, vol. 1, ed. F. Franks,
Plenum Press, New York, 1972.
16. J.E. Eldridge and J.D. Ferry, J. Phys Chem., 1954,58,992.
New Materials
USING NATURE TO TAILOR HYDROCOLLOID SYSTEMS
M.J.Gidley
Unilever Research Colworth, Colworth House, Sharnbrook, Bedford MK44 lLQ, UK
1 INTRODUCTION
Biopolymers are the most abundant form of carbon in the world, yet only a small subset of
this renewable resource is utilised in hydrocolloid applications. This is primarily due to the
relatively small number of sources available for economic extraction and refinement of
individual polymer types. The two key requirements for attractive hydrocolloid production
systems are a sufficient concentration of biopolymer and its ease of extraction and
purification. Most of the worlds biopolymers are present as plant cell walls where they
serve many of the functions that are prized in hydrocolloid uses. Three areas of knowledge
that impact on exploitability of (plant cell wall) biopolymers are: (1) genetic control of
biosynthesis, (2) structure and interactions in the native biological assembly, and (3)
potential uses for Natures pre-formed structural assemblies without refinement of
components. Recent advances in both physical and biological sciences are shedding light
on the first two areas, and growing consumer interest in natural foods is providing a
driver for the third.
2 NATURES ASSEMBLIES
Many hydrocolloids perform important biological roles as structuring polymers in land and
marine plants, microbes and animals. It is likely that many of the well-known polymeric
functionalities exhibited in vitru are also exploited in viva Examples include the network-
forming abilities of e.g. pectins, alginates, carrageenans, and agars, and the viscosity/
suspensiodadhesion properties of microbial polysaccharides and plant exudates. However,
in addition, many hydrocolloid polymers are arranged by Nature into higher order
structural assemblies. Examples include the cell walls of plants, fungi and bacteria, and the
exoskeleton and connective tissue of animals. In essence, hydrocolloid extraction involves
breaking down these assemblies, and application in products involves the formation of
other polymer-containing assemblies. In principle, there are considerable advantages
possible in avoiding the extraction and refinement steps by using pre-formed natural
assemblies. This is of course exploited in many food processes where structuring is
achieved through the use of either whole tissues (e.g. b i t , vegetables, meat) or
comminuted cellular material (e.g. pastes, purees, sauces). Use of such materials in
processed foods brings benefits of authenticity/naturalness/cleanlabel etc, and is often

associated with a level of natural heterogeneity that would usually be considered
unacceptable in formulated products structured by hydrocolloids. Further potential benefits
of using pre-formed assemblies, particularly from crop plants, arise from the reduced
energy costs associated with the lower process inputs to isolate the structuring system. This
results not only in cost efficiency, but also in environmental benefits that can enhance
consumer appeal. If cereals, fruits or vegetables are used as the source of pre-formed
assemblies, then the well-known dietary advice to eat more of these food types can be
addressed in the context of formulated foods.
There are many challenges involved in the greater exploitation of pre-formed assemblies
in processed food. These can be classified under the headings of raw material selection,
process tolerance, and biological variability. Appropriate raw materials need to be available
in sufficient quantities with reliable properties after isolation of the structuring system. An
acceptable level of colour and flavour in the isolated material will also be a major concern.
This may involve informed selection of varieties, specified growth conditions, and/or
controlled harvest and post-harvest regimes. The process tolerance of natural assemblies to
e.g. heat, cold, or dehydration is often limited, and will be an important criterion for
selecting potential solutions for specific products.
Biological variability is apparent at all levels of natural materials. For example, in
plants, cell wall composition varies not only between species, but also between organs of
the same plant, between different cells within an organ, and even around the wall of a
single cell. There is also significant variation at the agricultural scale between individual
plants within one field and between fields. This places strain on making harvest decisions
in an informed manner. The holy grail in this area would be the ability to specify pre-
formed assemblies with at least the same precision and reliability as is currently possible
for refined ingredients. In a sense this is the converse of one of the driving forces behind
the success of hydrocolloids as food ingredients i.e. they are very reliable in application and
can be specified at a sufficient level to ensure predictable performance. However, as
biological systems become better understood through the post-genomic revolution in
science, it can be anticipated that the challenges identified will be capable of being met.
3 COMPOSITES OF CELLULOSE WITH OTHER POLYSACCHARIDES
In order to gain some understanding of the assembly rules underlying the formation of
cellulose-based assemblies (as found in land and marine plants), and to establish
relationships between molecular structure, assembly architecture and mechanical
properties, we have made use of a bacterial system (Acetobacter xylinus) that secretes
cellulose. As this bacterium can be fermented, it can be used as a biological production
system depositing cellulose (and no other polymers) into an extracellular medium that can
contain any desired component. We have studied the effect of having a range of
xyloglucans, mannans? pectins; and xylans in the fermentation medium to determine
what rules apply for the spontaneous assembly of composites with cellulose extruded from
the surface of the bacterium. These polymers represent the major polysaccharide
components in non-lignified land plant cell wallsa4
3.1 Classes of polysaccharide / cellulose interaction
Fermentation of Acetobacter xylinus under standard laboratory conditions results in the

production of relatively flat thin sheets of material up to several cm across and a few mm
New Materials 283
thick. After fermentation in the presence of other polysaccharides and extensive washing to
remove any excess polymer, the formation of composite materials can be assessed in a
number of ways. Chemical analysis quantifies the ratio of cellulose to added polymer.
Mobility-resolved 13C NMR spectroscopy can be used to identify the type and level of
cellulose order (crystallinity) as well as providing information on the conformation and
chain dynamics of non-cellulosic polymer^.^ Transmission electron microscopy of
shadowed replicas after rapid freezing and deep etching provides a visualisation of
rnicro~tructure.~~ As cellulose ribbons produced by A. xylinus 35324 are typically 40-60
nm wide, they are easy to distinguish from less associated polysaccharide chains by TEM.
Using a combination of these measurement techniques, three classes of cellulose /
polysaccharide composite have been identified, as described below.
3.1.1 Binding and cross-linking. Major effects on cellulose molecular order by NMR
and TEM observation of thin cross-links between cellulose ribbons provide the evidence
for binding and cross-linking of cellulose by both xyloglucan and certain galacto- and
glucomannans. For xyloglucan, the lengths of cross-links observed are relatively uniform
(30-50 nm) and are similar to those Seen in partially extracted onion cell walls by the same
measurement technique.8 This suggests that the cross-linked architecture of the cellulose /
xyloglucan matrix, thought to be the major load-bearing network in the cell wall, is
provided by a purely physico-chemical self-assembly process.
The effect of chemical composition on the ability of galacto-and glucomannans to bind
and cross-link cellulose can be probed as variants are available from Nature. For the
galactomannan family, polymers with mannose to galactose ratios >2.8 all showed NMR
evidence for molecular association with cellulose, whereas those with ratios of 1.6 or lower
did not? Glucomannan shows evidence not only for binding but also for a combination of
cross-linking and coalescence of cellulose ribbons. This behaviour was also found for
galactomannans, with less substituted mannans (e.g. mannose to galactose ratios > 5 )
showing a greater tendency towards coalescence of cellulose, and intermediate substitution
levels (e.g. mannose to galactose ratio ca 3) leading to a predominance of cross-linking.
NMR analysis of both xyloglucan and mannan composites with cellulose show evidence
for a conformational change of the added polysaccharide, associated with the fraction of
polymer showing most rigidity. This comes from chemical shift values seen in the so-
called CPMAS and SPMAS experiments carried out sequentially on one sample, that detect
relatively rigid and mobile polymer segments re~pectively.~ Table 1 below summarises
chemical shifts for the marker resonance of the anomeric carbon (C-1) with assignments
based on comparison with crystalline polymers for which conformations have been
established by diffraction methods, and by comparing to the solution state.
C- 1 chemical shifts for polysaccharide backbone residues
xyloglucan ca 105.5 103.5 103.5 105-106 (cellulose I)
glucomannan ca 105.5 103.5 103.5 105-106 (cellulose I)

102.2 101.0 101.0 102.2 (mannan I)
galactomannan 102.2 101.0 101.0 102.2 (mannan I)
Table I . 13C NMR shifts (ppm)that diagnose polysaccharide confornational state.

The data in Table 1 show that those segments of polysaccharide with high rigidity in
composites exhibit chemical shifts characteristic of cellulose I and mannan I for glucan and
mannan backbones respectively. Conversely, relatively mobile segments of polysaccharide
in cellulosic composites have chemical shifts that match those found in solution. Together
with the microscopic data described above, this provides compelling evidence for binding
of polysaccharide segments to cellulose via complementarity of backbone conformations
(extended 2-fold geometry) with non-bound segments involved in cross-linking and
exhibiting average conformations similar to those found in solution. It is interesting to note
that the C-1 chemical shift diagnostic of the 'cellulose-like' mannan I conformation is not
found in gels of glucomannans or galactomannans." It appears that a pre-formed extended
2-fold cellulose conformation is required as a template for formation of the analogous
conformation in xyloglucans or mannans.
The backbone of mannan (1,4- linked beta-D-mannose) differs from cellulose and
xyloglucan (1,4- linked beta-D-glucose) only by the orientation of substituents at C-2 with
the hydroxyl group axial in mannan and equatorial in (xy1o)glucan. The backbone of xylan
(1,4- linked beta-D-xylose) differs from glucan only by the lack of the exocyclic -CH20H
group attached to C-5. Despite this close structural similarity, we have not yet found
convincing evidence for binding or cross-linking with cellulose for any of a number of
xylans from cereal (wheat, maize) or wood (larch) sources under the experimental
fermentation conditions employed. Based on this and the relative disrupting ability of
sidechain substituents, the propensity for binding to cellulose is in the order glucan >
mannan > xylan.
3.1.2 Inter-penetrating networks. Cellulose-based composites that survive exhaustive
washing and do not show direct binding to, or cross-linking of, cellulose are diagnosed as
inter-penetrating networks of cellulose and the added polymer. One example of such an
inter-penetrating network is found for pectins of intermediate degrees of esterification (25-
80%)present in the fermentation medium in the form of a calcium-mediated network3
In this case, cellulose is deposited directly into a pre-formed pectin network, but NMR
spectra show no evidence for perturbation of cellulose crystallinity or altered pectin
conformation. This rules out significant molecular association; TEM images show an
intimate mixing of cellulose (thick strands) and pectin (thin strands). Together this suggests
that inter-penetrating networks are resent.^ An interesting effect of the cellulose network
on the pre-formed pectin network is to cause chains in the latter to aggregate, leading to
thicker strands / larger voids (TEM) and increased segmental rigidity (NMR) compared
with the starting pectin network3 It is possible that this effect arises from a local
concentration of pectin driven by cellulose deposition, and/or that the proximity of
cellulose has a significant effect on the local solvent quality, perhaps related to the relative
hydrophobicity of cellulosic ribbons.
It is not only pectins that are capable of forming inter-penetrating networks with
bacterial cellulose. Less predictably, deposition of cellulose into solutions of guar
galactomannan2(mannose to galactose ratio 1.6) or either wheat or maize arabinoxylan has
the same effect. In these cases, the proximity of cellulose is sufficient to promote network
formation from the dissolved state for these polymers. As for pectin, the physico-chemical
mechanism by which cellulose exerts this interesting effect is not yet apparent.
3.1.3 Phase separation. For polymers that do not show affinity for cellulose and do not
form networks in the presence of cellulose, washing of fermented cellulosic material results
in loss of the second polymer. The cellulose network produced in these systems appears (by
NMR and TEM) to be unaffected by the presence of the non-interacting polymer. Examples
of this type of behaviour are pectins in the absence of calcium3, and fenugreek
galactomannan2which has a low mannose to galactose ratio (ca 1.1).
New Materials 285
There is also a second category of phase separation. This is where a pre-formed network
is too strong/dense for cellulose to deposit within it. The result is a bulk phase separation of
the two networks. Low ester ( ~ 2 5 % )pectins show this behaviour in the presence of
calcium3. There are therefore two mechanisms for phase separation and one for inter-
penetrating network formation all from the same polymer. The richness of interplay
between pectin and cellulose may well fmd expression in the biological control of local cell
wall architectures and material properties via pectin esterification and calcium levels in
vivo.
3.2 Mechanical properties of cellulose-basedcomposites
Characteristic material properties of plant cell walls all start from cellulose, but are capable
of a large degree of modulation by the presence and state of other polysaccharides. To a
first approximation, cell wall cellulose can be considered as an entangled assembly of stiff
rods. The linear conformation of cellulose chains together with their aggregation into
fibredribbons results in an exceptionally high axial ratio and consequently very weight-
efficient structuring via hydrodynamic entanglements.
Under small deformation oscillatory conditions, it appears that entanglement of stiff
ribbons is the primary mechanism of structuring for a range of cell walls and model
composites." However, under larger deformation conditions e.g. uniaxial or biaxial
tension, non-cellulosic polymers have a marked influence. Fig 1 shows the very different
results obtained on uni-axial extension of cellulose, cellulose/pectin and
cellulose/xyloglucan composites. In both cases, incorporation of a second polymer makes
cellulose more extensible and weaker. For pectin-containing composites, it can be shown
that this altered mechanical behaviour is due entirely to the cellulosic component in the
composite, as removal of the pectin (by chelation of calcium with CDTA) does not change
the uniaxial mechanical proper tie^.^
1.2
0 I0 20 30 40 50 60 70 80
Strain (%)
Figure 1 Strain/stress curvesfor composites in a uni-axial tensile
Bi-axial test conditions (bulge testing) provide a closer representation of the action of
turgor pressure on cell walls in living plants, and show significant differences from uni-
axial (extension) conditions for the same composites (Figure 2).* In addition to being
significantly less stiff than other composites, finite element simulation shows that the
cellulose/xyloglucan composite behaves in a non-linear elastic fashion, Mechanical
behaviour of the other two composites are compatible with linear elasticity. This distinction
is borne out by creep tests in which cellulose/xyloglucan composites showed si nificant
time dependence under constant pressure whereas the other two materials did not. 15
I
Pressure - -
(bar) Celulose alone
0 1 2 3 4 5
Displacement (mm)
Figure 2 Bi-axial tensile behaviour of Acetobacter cellulose composites (two examples of
each material are shown)12
In summary, based on studies of Acetobacter cellulose-based composites, the following

effects of two polymer types have been found:
1. Xyloglucan. When xyloglucan molecular weights are high enough to extensive1 cross-
link cellulose, uniaxial tensile stiffness is reduced and extensibility is increased.J Under
biaxial tensile conditions, non-linear-elastic behaviour is found with very low apparent
stiffness and pronounced creep behaviour.12Thisis interpreted as being due to cross-linking
(thin) xyloglucan strands becoming partially responsible for load bearing, resulting in
weaker and more time dependent mechanical behaviour.
2. Pectin. For pectin networks with cellulose deposited into them, uniaxial extensibility is
enhanced and stiffness is decreased. This result is maintained after removal of the pectin
component with calcium-chelating agents such as CDTA. It is suggested that the deposition
of cellulose into a pre-formed pectin network results in more local alignment of cellulose
ribbons, reducing the density of effective entanglements, and making them easier to pull
apart on stretching31n contrast, biaxial tensile results show only a modest weakening and
no increase in extensibility. This is consistent with the proposed uniaxial mechanism as the
New Materials 287
geometrical constraints of the biaxial test geometry prevent the same mechanism from
being operative.12
4 IMPACT OF POST-GENOMIC BIOLOGY
Plant biology is entering a new era following the publication in December 2000 of the
complete genome sequence for the model plant Arabidupsis thaliana.l3 Other complete
sequences (e.g. rice) will follow, but the major effort will be in determining the function of
all of the ca 25,000 genes in Arabidupsis and identifying homologues in other plants with
the equivalent function. This is going to revolutionise all (plant) biology with many
profound consequences. Undoubtedly, some of these will be unpredictable; some
foreseeable possibilities of relevance to hydrocolloids are discussed here.
4.1 New sources of enzymes
Preliminary analysis of the Arabidupsis genome shows a very large number of putative
hydrocolloid processing enzymes.l3 For example, 52 genes for polygalacturonase, 20 genes
for pectin lyase, and 79 genes for pectin methyl esterase have been tentatively identified.13
It is possible that the final numbers may differ slightly, but it is clear that Nature has
chosen to exploit very many genes for catalysis of specific reactions. It is anticipated that
some of these genes will encode enzymes of equivalent function that are expressed at
different times and/or different locations in the plant. However, it is also likely that these
unexpectedly large gene numbers reflect a diversity of enzyme action that will provide a
significant resource for novel enzymic tailoring (in vitru or in vivu) of hydrocolloids.
4.2 Polysaccharide biosynthesis
Whilst the genetics and biochemistry of biosynthesis for some polysaccharides is starting to
get unravelled (starch is a notable example), for cell wall polysaccharides there is a current
bottleneck caused by an inability to identify the genes or purify the proteins involved. At
the moment there are genes identified for only three enzyme activities: cellulose ~ynthase'~,
fucosyltransferase acting on xylogl~can'~, and galactosyltransferase acting on
galact~mannan'~.Even if there is no duplication of enzyme activities at the gene level
(highly unlikely), there is a requirement for between 50 and 100 genes for the synthesis of a
typical cell wall. A very rough estimate suggests that somewhere in the region of loo0
genes may be associated with cell wall synthesis, turnover and degradation. This is a huge
genetic resource that is likely to become available in the next few years, opening up the
possibility of selecting or tailoring at the genetic level for desired polymer properties in
plants.
4.3 Gene-level specification
Although a long way off still, it should be anticipated that it will eventually become
possible to understand the whole physiology of plants at the level of genes and their
response to environmental conditions. When (and if!) this happens, then plants will
approach the same level of predictability that is possible currently at the molecular level for
purified materials. This could become a major driver in making the processing and
application performance of relatively non-purified plant extracts more predictable and
homogeneous, in turn making their commercial potential greater.
References
1 S.E.C. Whitney, J.E. Brigham, A.H. Darke, J.S.G. Reid, M.J. Gidley, Plant Journal,
1995, 8,491.
2 S.E.C. Whitney, J.E. Brigham, A.H. Darke, J.S.G. Reid and M.J. Gidley, Carbohydrate
Research, 1998,307,299.
3 E. Chanliaud and M.J. Gidley, Plant Journal, 1999,20,25.
4 N.C. Carpita and D.M. Gibeaut, Plant Journal, 1993,3, 1.
5 T.J. Foster, S. Ablett, M.C. McCann and M.J. Gidley, Biopolymers, 1996,39,51.
6 M.C. McCann and K. Roberts, in The cytoskeletal basis for plant form and function, ed.
C.W. Lloyd, Academic Press, London, 1991, p109.
7 J.E. Brigham, M.J. Gidley, R.A. Hoffmann and C.G. Smith, Food Hydrocolloids, 1994,
8,331.
8. M.C. McCann, B.Wells and K.Roberts, J. Cell Sci., 1990,%, 323.
9. I.C.M. Dea and A. Morrison, Adv. Carbohydr. Chem. Biochem, 1975,31,241.
10. M.J. Gidley, A.J. McArthur and D.R.Underwood, Food Hydrocolloids, 1991,5, 129.
11. S.E.C. Whitney, M.G.E. Gothard, J.T. Mitchell and M.J. Gidley, Plant Physiology,
1999,121,657.
12. E.M. Chanliaud, K.M. Burrows, G. Jeronimidis and M.J. Gidley, 2001, submitted for
publication.
13. S. Kaul et al, Nature, 2000,408, 796.
14. T. Arioli et al, Science, 1998,279,717.
15. R.M. Perrin, A.E. DeRocher, M. Bar-Peled, W.Q. Zeng, L. Norambuena, A. Orellana,
N.V. Raikhel and K. Keegstra, 1999, Science, 284,1976.
16. M.E. Edwards, C.A. Dickson, S. Chengappa, C. Sidebottom, M.J. Gidley and J.S.G.
Reid, 1999, Plant Journal, 19,69 1.
FORMATION OF STRONG GELS BY ENZYMIC DEBRANCHING OF GUAR GUM IN
THE PRESENCE OF ORDERED XA"
C.E. Cronin,' P. Giannouli,' B.V. McCleary? M. Brooks3and E.R. Morris'*
'Department of Food Science, Food Technology and Nutrition, University College Cork,
Cork, Ireland
'Megazyrne International Ireland Limited, Bray Business Park, Bray, Co. Wicklow, Ireland
3QuestInternational Ireland Limited, Kilnagleary, Carrigaline, Co. Cork, Ireland
1 INTRODUCTION
The gelling interaction of xanthan with locust bean gurn (LBG) was first reported around
30 years ago.',' Since then, numerous practical applications of the resulthg 'synergistic' gels
have been developed, but the gelation mechanism remains controversial and incompletely
understood.
Xanthan has a (1 + 4)-linked backbone of P-D-glucose residues, as in cellulose,
but is solubilised by charged trisaccharides sidechains attached at O(3) of alternate residues,
giving an overall pentasaccharide repeating sequence?g4 At high temperature and low
ionic strength xanthan exists in solution as a disordered coil. On cooling and/or addition
of salt, however, it undergoes a co-operative conformational transition' to a 5-fold helix?
in which the sidechains fold down to form an integral part of the ordered structure. Solutions
of xanthan in its ordered conformation show tenuous 'weak gel' properties, which appear to
arise from weak, cation-mediated association of helical sequences: but xanthan alone does
not give 'true' (self-supporting) gels. The gels formed by interaction with LBG (or related
plant polysaccharides), however, are strong and cohesive at comparatively low
concentrations (typically around 0.2-0.5% of each constituent).
LBG is a member of the galactomannan family of plant polysaccharides.' These have
a linear backbone of P-D-mannose residues linked in the same way as the P-D-glucose
residues in cellulose (by equatorial bonds at 0(1) and O(4)). Like cellulose, unsubstituted
mannan is insoluble. The galactomannans are solubilised by single-sugar sidechains of
(1 + 6)-linked a-D-galactose, with different galactomannans differing in the extent and
pattern of substitution.
The ability to form 'synergistic' gels with xanthan decreases with increasing content of
galactose.' Mixtures of xanthan with guar gum, which has a mannose/galactose (WG)ratio
of -1.55, show some enhancement of viscosity at low temperature," but formation of
strong, self-supporting gels occurs only with galactomannans of lower galactose content,
such as LBG (M/G a 3.5), cassia gum (WG = 3.0) and tara gum (M/G = 2.8). This inverse
correlation between degree of substitution and interaction with xanthan prompted an early
suggestion' that synergistic gelation occurs by attachment of unsubstituted regions of the
mannan backbone to the surface of the xanthan 5-fold helix.
More recently, however, it was suggested1'*12v'3 that interaction occurs only when
the xanthan constituent is in the disordered state, and that intermolecular junctions are
formed by direct binding of the mannan backbone of the galactomannan to the cellulosic
backbone of xanthan, with the (1 + 4)-diequatorial linkage geometry in both polymers
providing a complementary fit. This proposal was based on two main lines of evidence:
(i) X-ray fibre diffraction patterns of oriented specimens prepared from mixtures of xanthan
with LBG or gum tara showed features that were not observed in the diffraction patterns
of the individual constituents, and (ii) xanthan-galactomannan mixtures prepared at ambient
temperature remain fluid, but convert to cohesive gels after heating to a temperature where
the xanthan is disordered, and then re-cooling.
However, the xanthan-galactomannan network is itself thermoreversible, forming
on cooling and melting again on An alternative explanation of why mixtures
prepared at low temperature, below the temperature of the sol-gel transition, do not
give cohesive gels might therefore be that the mixing process prevents formation of a
continuous n e t w ~ r k . ' ~Stirring
*'~ solutions of gelling polysaccharides such as carrageenan
while cooling through the temperature range of network formation gives a stable dispersion
of microgel particle^,'^ which converts to a continuous network only after thermal melting
and subsequent quiescent cooling.
Similar disruption of continuous network structure occurs on direct mixing of solutions
of alginate or low-methoxy pectin with the divalent cations required to promote interchain
association. For these systems, formation of a cohesive gel, rather than a coagulated
dispersion, can be engineered by mixing the polysaccharide with a salt, such as calcium
sulphate, which is insoluble at neutral pH, and using glucono-&lactone (GDL) to give a
slow reduction in pH, which releases calcium ions from the salt in a controlled way,
allowing a continuous network to develop. This procedure is known as the 'internal set'
process, and is used commercially in, for example, re-structuring of fruit pulp.'' In the
present work we have used a somewhat analogous 'slow release' process to explore the
gelling interaction of xanthan with galactomannans of low galactose content.
Incubation of guar gum with a-galactosidase (rigorously purified to eliminate any
P-mannanase activity which would cleave the polymer backbone)20can be used to remove
a proportion of the galactose sidechains and give a product with an M/G ratio similar to
that of LBG. Like LBG, the resulting 'debranched' guar gum gives strong synergistic gels
with xanthan when mixed solutions of the two polymers are prepared at high temperature
and then cooled.*' In the work reported here, however, mixed solutions of xanthan and
guar gum were prepared at a temperature well below the sol-gel transition temperature,
and also well below the temperature range of the disorder-order transition of the xanthan
component. The debranching enzyme (a-galactosidase) was then added, and the rheological
consequences of 'slow release' of sparingly-substituted galactornannan in the presence of
ordered xanthan were monitored by low-amplitude oscillatory measurements of storage and
loss moduli (GI and G") and by large-deformation compression testing.
This experimental procedure was designed explicitly to distinguish between the effect
of mixing and the effect of the conformational state of the xanthan component. If the reason
why xanthan does not give cohesive gels when mixed with LBG (or debranched guar) at
low temperature is that the mixing process disrupts the synergistic network as it is forming,
then enzymic debranching of the guar gum component in a homogeneous, non-gelling
mixture would be expected to give a cohesive network. If, however, synergistic gelation
requires the xanthan component to be disordered, then debranching of guar gum in the
presence of ordered xanthan would not be expected to result in gel formation.
New Materials 29 1
Xanthan, guar gum and locust bean gum were commercial food-grade samples provided by
Quest International; the a-galactosidase was also a food-grade preparation used by Quest
in commercial production of debranched guar gum. To satisfy the ionic-strength and pH
requirements of the enzyme, all samples were prepared in 50 mM sodium acetate-acetic
acid buffer at pH 5.0. Solutions of xanthan and guar gum in this buffer were prepared
separately at equal concentrations (0.25 or 0.50 wt %) and mixed in equal volumes at 30C,
which was chosen as a temperature low enough to allow formation of synergistic gels,
while still maintaining an acceptable rate of enzymic activity. Enzyme (a-galactosidase)
was then added, at 30 U activity per gram of guar gum, and the solutions were immediately
filled into beakers for compression testing andor loaded onto an oscillatory rheometer
at 30C.
Oscillatory measurements were made at 1% strain, using parallel plate geometry
(4 cm diameter; 0.5 mm gap) on a CarriMed CSL-100 rheometer. Changes in G' and G",
(1 Hz; 1% strain) were monitored over a period of 10 h at 30C, and a mechanical
spectrum was then recorded to show the variation of G', G" and complex dynamic viscosity
[q* = (G'2 + G"~)"/w] with frequency ( o h a d s"). Samples for compression testing were
held at 30C for 3.5 h or 24 h before measurement. The containers had an internal diameter
of 7 cm and were filled with 100 rnl of sample, giving a depth of 2.6 cm. Compression
was made at a rate of 0.2 d s on a TA-XT2 Texture Analyser &omStable Microsystems,
using a cylindrical probe of diameter 3.5 cm. Comparative measurements, by both
techniques, were made using the same concentrations of xanthan and LBG, in the same
buffer. Mixtures were prepared at 80"C,cooled to 30C at l"C/min, and then treated in
the same way as the xanthan-guar gum samples.
1000
100
G' (Pa)
G"(Pa)
10
1
30 40 50 60 70 80
Temperature ("C)
Figure 1 Changes in G' fllled symbols) and G" (open symbols) during cooling (I "Chin)
for mixtures containing equal amounts of xanthan and LBG at total concentrations of
0.25 wt % (circles) or 0.50 wt % (triangles);fiequency 1 Hz; I % strain.
The temperature-course of the conformational transition of xanthan (0.125 wt % in

50 mM sodium acetate-acetic acid buffer at pH 5.0) in the absence of galactomannan was
characterised by differential scanning calorimetry (DSC), using a scan rate of 0.3"C/min on
a Setaram DSC-3 microcalorimeter.
Figure 1 shows the changes in moduli on cooling (1"Chnin) observed for solutions
containing equal amounts of xanthan and LBG (in 50 mM sodium acetate-acetic acid
buffer at pH 5.0) at total polysaccharide concentrations of 0.50 and 0.25 wt %. At both
concentrations, gel formation, seen as a steep increase in G', began at -60C and was
essentially complete by 30"C, the temperature used in investigation of the effect of enzymic
debranching of guar gum in the presence of xanthan. No further changes in moduli were
observed on holding at 30C for 10 h (as in the debranching experiments). On subsequent
heating, reduction in moduli followed the same temperature-course as the increases
observed on cooling, with no detectable thermal hysteresis. Closely similar behaviour has
been observed p r e v i o ~ s l yfor
~ ~xanthan-LBG mixtures studied under different conditions of
polymer concentration, pH and ionic strength.
Figure 2 shows DSC traces obtained for xanthan alone (in the buffer used for the mixed
systems) at 0.125 wt %, the lower of the two concentrations used in the mixtures, on
heating and cooling at 0.3"C/min. A well-defined exotherm was obtained on cooling, with
an approximately equal and opposite endothem on heating. As would be anticipated from
finite rate of heat transfer within the calorimeter ("thermal lag"), the endotherm is centred
at slightly higher temperature than the exotherm (by -2"C), but it is evident that the true
midpoint temperature of the transition is -84C. Closely similar traces, again centred
20
10
-10
Heat flow
-20
(Pw
-30
-40
-50
-60
40 50 60 70 80 90 100 110
Temperature ("C)
Figure2 DSC traces recorded during heating (e) and cooling (0) at 0.3"C/minf o r
0.125 wt % xanthan in 50 mM sodium acetateacetic acid buffer at p H 5.0.
New Materials 293
at -84"C,were obtained at a xanthan concentration of 0.25 wt %, the higher of the two

concentrations used in the mixed systems.
Inspection of the cooling trace in Figure 2 indicates that, under the ionic conditions used
in this investigation (50 mM sodium acetate-acetic acid buffer at pH 5.0), conformational
ordering of xanthan is complete well before the onset temperature of the synergistic
gelation process shown in Figure 1, which, in itself, indicates that association with
galactomannan does not require the xanthan molecule to be disordered.
It has been argued:' however, that when xanthan is cooled in the presence of LBG
(or related plant polysaccharides with which it forms synergistic gels), the disorder-order
transition may be displaced by association of disordered xanthan with the other polymer,
and that the temperature-course of conformational ordering for xanthan alone may not,
therefore, give a true indication of the conformational state of the xanthan component in
the mixed system. The same argument cannot, however, be applied, with any degree of
credibility, to non-gelling mixtures of xanthan with guar gum, prepared at a temperature
(30C) more than 50C below the mid-point temperature (Figure 2) of the disorder-order
transition. It seems reasonable to assert, therefore, that the xanthan component in the
starting solutions for the experiments on enzymic debranching was fully ordered.
The changes in moduli during incubation of the xanthan-guar gum solutions with
a-galactosidase are shown in Figure 3. At both concentrations studied (0.125 or 0.25 wt %
of each component), there was an immediate, steep increase in GI, indicating gel formation,
followed by a reduction in G", which is consistent with progressive incorporation of
unbound polysaccharide into the crosslinked network. At the end of the 10 h period over
which changes in moduli were monitored, G' was still increasing slightly, but the region of
steepest ascent was during the first 3.5 h of incubation, which was therefore chosen as one
of the incubation periods used in preparation of samples for compression testing.
1000
100
G"(Pa)
10
1
0 2 4 6 8 10
Time (h)
Figure 3 Changes in G' lfilled symbols) and G" (open symbols) during incubation with
a-galactosidase at 30C for mixtures containing equal amounts of xanthan and guar gum
at total concentrations of 0.25 wt % (circles)or 0.50 wt % (triangles).
The mechanical spectra recorded at the end of the 10 h period of observation (Figure 3)
are shown in Figure 4, in direct comparison with the spectra obtained for xanthan-LBG
mixtures, at the same polymer concentrations, after cooling from the solution state at
80C and holding for the same length of time (10 h) at 30C. At total polysaccharide
concentrations of both 0.25 wt % (Figure 4a) and 0.50 wt % (Figure 4b), the networks
formed by enzymic debranching of guar gum in the presence of ordered xanthan show
substantially greater gel-like character than the xanthan-LBG networks formed by cooling
from high temperature: G' and q* are appreciably higher, and the separation of G' and GI' is
much greater, indicating a greater degree of crosslinking.
loo0 1
a
0.1
0.1
- 1 10 100
1
Frequency (rad s- )
Frequency (rad s-')
Figure4 Mechanical spectra ( 1 % strain), recorded after I0 h at 30C,showing the

frequency-dependence of G' (squares), G" (circles) and q * (triangles)for 50/50 mixtures
of xanthan with LBG (open symbols) or with guar gum in the presence of a-galactosidase
willed symbols) at total polymer concentrations of ( a ) 0.25 wt % or (b)0.50 wt % .
New Materials 295
The gels formed by incubation of xanthan-par gum mixtures with a-galactosidase

were also much more resistant to fracture in large-deformation compression testing than
the control samples of xanthan-LBG. After incubation for 3.5 h, when, as indicated in
Figure 3, substantial further debranching has still to occur, the force required to break the
gel obtained for xanthan-guar gum at a total polymer concentration of 0.25 wt % was
about 3 times greater (Figure 5a) than for the same concentrations of xanthan and LBG
after cooling from high temperature and holding for the same time at 30C, and indeed was
about 50% greater than for the xanthan-LBG control at twice the polymer concentration.
1.0
I
t 1
0.8
0.6
0.4
0.2
0.0
0 5 10 15 20 25
Distance (mrn)
5.0
4.0
3.0
Force (kg)
2.0
1.0
0.0
0 5 10 15 20 25
Distance (mm)
Figure5 Compression curves (0.2 mmfs) for 50150 mixtures of xanthan with LBG
(open symbols) or with guar gum in the presence of a-galactosidase filled symbols);
( a ) samples measured after 3.5 h at 30C for total polymer concentrations of 0.25 wt %
(circles) or 0.50 wt 96 (triangles); (b) samples prepared at a total polymer concentration
of 0.50 wt % and measured afler 3.5 h (triangles) or 24 h (circles) at 30C.
When the xanthan-guar gum concentration was also raised to 0.50 wt %, the force
required for rupture after incubation with a-galactosidase for 24 h was about 7 times higher
than for the xanthan-LBG control prepared at the same concentration and held for the same
time at 30C (Figure 5b), and after only 3.5 h incubation the break-strength was already
about 4 times greater than that of the same concentration of xanthan-LBG after holding
for 24 h at 30C.
4 CONCLUSIONS
The main conclusions from this investigation are: (i) that treating mixed solutions of
xanthan and guar gum with a-galactosidase, to reduce the galactose content of the
galactomannan, gives firm,cohesive gels, which are much stronger than those obtained by
cooling the same concentrations of xanthan and locust bean gum,and (ii) that gelation can
occur with the xanthan component in its ordered conformation at a temperature at least
50C below the midpoint of the disorder-order transition, demonstrating that there is no
need for the xanthan chains to be disordered when forming junctions with galactomannan.
References
1 J.K. Rocks, Food Technology, 1971,25,470.

2 P. Kovacs, Food Technology, 1973,27,26.
3 P.E.Jansson, L. Kenne and B. Lindberg, Carbohydr.Res., 1975,45,275.
4 L.D. Melton, L. Mindt, D.A. Rees and G.R. Sanderson, Carbohydr. Res., 1976,
46,245.
5 E.R.Morris, D.A. Rees, G. Young, M.D. Walkinshaw and A. Darke, J. Mol. B i d ,
1977,110, 1.
6 R. Moorhouse, M.D. Walkinshaw and S. Arnott, ACS Symp. Ser., 1977,45,90.
7 K. Okuyama, S. Amott, R. Moorhouse, M.D. Walkinshaw, E.D.T. Atkins and
Ch. Wolf-Ullish, ACS Symp. Ser., 1980,141,411.
8 S.B. Ross-Murphy, E.R. Morris and V.J. Morris, Faraday Discuss. Chem. SOC.,1983,
18, 115.
9 I.C.M. Dea. and A. Morrison, Advan. Carbohydr. Chem.Biochem., 1975,31,241.
10 L. Lopes, C.T.Andrade, M. Milas and M. Rinaudo, Carbohydr. Polym., 1992,17, 121.
11 P. Cairns, M.J. Miles and V.J. Morris, Nature (London), 1986,322,89.
12 P. Cairns, M.J. Miles, V.J. Morris and G.J. Brownsey, Carbohydr. Res., 1987,160,411.
13 V.J. Morris, in Gums and Stabilisersfor the Food Industry 6, ed. G.O.Phillips, P.A.
Williams and D.J. Wedlock, IRL Press, Oxford, UK, 1992, p 161.
14 F.M. Goycoolea, R.K. Richardson, E.R. Morris and M.J. Gidley, Macromolecules,
1995,28,8308.
15 E.R. Morris and T.J. Foster, Carbohydr. Polym., 1994,23, 133.
16 E.R.Morris, in Biopolymer Mixtures,ed. S.E. Harding, S.E. Hill and J.R. Mitchell,
Nottingham University Press, Nottingham, UK, 1995, p. 247.
17 P. Harris and S.J. Pointer, UK Patent GB2,128,871B, 1986.
18 W.J. Sime, in Food Gels, ed. P. Harris, Elsevier, Barking, Essex, UK, 1990, p. 53.
19 B.V. McCleary, R. Amado, R. Waibel and H. Neukom, Carbohydr.Res., 1981,92,269.
20 B.V. McCleary, I.C.M. Dea, J. Windust and D. Cooke, Carbohydr. Polym., 1984,
4,253.
21 V.J. Moms, G.J. Brownsey and M.J. Ridout, Carbohydr.Polym., 1994,23, 139.
NEW TEXTURES WITH HIGH ACYL GELLAN GUM
N.A.Morrison ",G.Sworn b, R.C.Clark " ,Todd Talashek" and You-Lung Chen."
" CP Kelco ,8355 Aero Drive, San Diego, CA, USA

CP Kelco, Cleeve Court, Cleeve Rd, Leatherhead, Surrey, U.K.
1 ABSTRACT
High acyl gellan gum has been recently commercialized as a new stabilizer for the food
industry. Gellan gum is a versatile food ingredient providing a wide range of textures from
soft, elastic gels to firm, brittle gels with one label declaration. Blends of high and low acyl
gellan gum can produce intermediate gel textures
Recent studies have shown that both the levels of glycerate and acetate substituents in
gellan gum can be controlled independently. A range of gellan gums with varying structural
and functional properties has been studied. Changes in the glycerate parameters, have a
profound effect on the structural and rheological characteristics of gellan gum gels. Changes
in the acetate content of high acyl gellan gum have less effect on rheological or textural
properties. The functional and textural properties of intermediate and high acyl gellan gum are
further discussed in this paper.
2 INTRODUCTION
Gellan gum is the extracellular polysaccharide produced by the microorganism
'.
Sphingomonus elodea during aerobic fermentation The primary structure of gellan gum is
composed of a linear tetrasaccharide repeat unit 3: +3)-P-D-Glcp-( 1+4)-P-D-GlcpA-( 1+4)-
P-D-Glcp-( 1-+4)-a-L-Rhap-( 1+.Gellan gum is commercially available in both the high acyl
(HA) and the low acyl form (LA). Low acyl gellan is produced by a hot strong alkali treatment
of the finalized fermentation broth, which fully deacylates the primary molecular structure of
gellan gum '.
The native biopolymer is produced with two acyl substituents present on the 3-linked
glucose 3* namely, L-glyceryl, positioned at O(2) and an acetyl substituent at O(6). X-ray
diffraction of low acyl gellan gum has shown that the polymer exists in the solid state as a co-
axial three-fold double helix.
Computer modeling of the high acyl structure, based on the knowledge of the low acyl
form concluded that the acetate substituents would be positioned on the outside of the double
helix?
Subsequent X-ray diffraction data revealed that the glycerate substituents were
accommodated by an approximately 14" rotation of the carboxyl group about the C(5)-C(6)
bond on the adjacent glucuronate residue.8 This was accompanied by a 17+3" rotation of the
glucuronate residue itself to maintain the required spacing between the carboxyl and glycerate
groups. It is proposed that this structure results in helices stabilized by interchain associations
involving the glycerate groups, with the acetyl substituents positioned on the periphery of the
helix.'
When hot solutions of gellan gum are cooled in the presence of gel promoting cations,
gels ranging in texture, from elastic (HA) through brittle (LA), are formed, through principally
cation mediated helix-helix aggregation 21. A wide range of gel textures can be produced
through manipulation of blends of high and low acyl gellan gum. The conformational change
temperature from double helix to random coil is also dependent on the molecular structure of
gellan gum. It has been demonstrated by differential scanning calorimetry (DSC) and
rheological measurements that mixtures of the high and low acyl forms exhibit two separate
conformational transitions at temperatures coincident with the individual components l o . ',
No evidence for the formation of double helices involving both high and low acyl molecules
were found ' I . This means that the problems associated with the high gelation temperature,
(-8O"C), of the high acyl gellan gum are still present in blended systems. The advantages and
limitations of the high setting temperature of HA gellan gum have been discussed in a
previous paper lo.
It has been shown that the textural properties of gellan are dependent on its acyl
content 93 I*. Partially deacylated products studied include low acetate (cold KOH treatment),
and intermediate acyl products (hot KOH treatment). In these studies, acetyl substituents were
removed in preference to glyceryl residues. New studies have shown the ability to form
acetate rich gellan gum by chemical treatment. These new products may also be achieved
through new strain development 13. A comparison is made with these new intermediate acyl
products to simple blends of high and low acyl gellan gum products.
-
Partially deacylated prototypes A range of gellan gum products have been produced by
base treatment of high acyl gellan gum. Samples were prepared from either native broth or
reconstituted high acyl gellan gum 14. The analysis of gellan gum for acyl substituent levels
has previously been described Is.
Clarification - HA gellan gum was clarified using a PHB deficient strain, which was
subsequently clarified with a chemical and enzymatic process. Low acyl gellan gum samples
were clarified by mechanical filtration.
Texture profile analysis - Samples were cast for Texture Profile Analysis (TPA) in
cylindrical molds of 14mm height and 29 mm internal diameter. Gels were removed from the
molds after overnight storage at 5C and compressed at 0.85mm.s-I to either 30% or 15% of
their original height, i.e. 70%-85% strain, depending on soluble solids loading. Modulus, a
measure of the gels initial firmness, and brittleness, a measure of the strain at rupture, were
recorded.
Setting temperature - Setting and melting behaviour of the samples were measured using a
Rheometrics SR 200 controlled stress rheometer. Tests were performed using a 6 cm parallel
plate, 0.5 to 1 mm gap, and a coolingheating rate of 2"C/min. Strain was controlled at 1%,
and frequency at 10 rad.s-'.
New Materials 299
-
Atomic force microscopy The atomic force microscopy (AFM) was performed with a
commercial instrument (Nanoscope ma, Digital Instruments, Santa Barbara, CA) using a
silicon cantilever tip. A 1-ppm gellan gum solution was prepared in 0.1M ammonium acetate
solution (gellan hydrated and heatedcooled firstly in DI water). The 1 ppm solution was
sprayed onto a freshly cleaved mica disc(s) (-1 cm2). These mica sample disc(s) were then
placed in a heated (-60 "C) vacuum chamber for one hour to remove excess water. The dried
mica disc(s) was then scanned using the Tapping Mode of the AFM.
Reduced Molecular weight - HA gellan of varying molecular weight were produced by
homogenizing a 1% clarified native gellan gum solution at 8,500 psi through a APV Gaulin
homogenizer for successive passes 16. Materials were also reduced in molecular weight
through sonication methods.
Molecular weight determination -
Dilute aqueous solutions of clarified gellan gum (0.06%), containing 0.010M TMACl
(Tetramethyl ammonium chloride) were prepared. The hydrated HA gellan gum samples were
all ion exchanged at 90C for 2 minutes using weakly acidic ionic exchange resin, Sigma,
AMBERLlTE IRC-50. These solutions were then filtered through a 0.45 micron filter.
Molecular weight distributions were determined by GPCMALLS. The system consists of a
Waters 610E HPLC pump, a Waters 717+ autosampler, a guard column, a 30 cm Waters
Ultrahydrogel 2000 column, a 30 cm Waters Ultrahydrogel Linear column, a DawnQDSP light
scattering detector and a Waters 410 differential refractometer. 0.10 ml of sample was
injected into the eluant flow (0.01 M TMACl, at 0.50 ml/min) and is separated based on
molecular size by the SEC columns. As the sample elutes from the column the molecular
weight and concentration profiles are determined by light scattering and refractive index
detectors. A refraction index increment (6n/&) of 0.145 was used. This value was chosen
based on values reported in the literature 17.

Low acyl gellan gum gels form by cooling a hot solution through its setting temperature (40-
50C) 18. Gels form in the presence of monovalent and divalent cations I . Significant thermal
hysteresis between the setting point and the melting point is observed for LA gellan gum '.
High acyl gellan gum solution properties and gel textures are much less dependent on the
concentrations of ions in solution. Gels typically form at around 80C and show no hysteresis.
This behavior is shown below in Figure 1. It has been proposed that in low acyl gellan the
large hysteresis observed results from two thermal transitions 93' I * 14.
Figure 1 - Melt set properties of L A and H A gellan
ld
Melting
1 0.25% L A Gellan in 6 M m o l C s ?*
1.
30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0
Temp ['C]
Effect of molecular weight on gel texture
A series of clarified high acyl gellan gum prototypes were produced with ranging molecular
weight. Table 1 shows that mechanical homogenization does not influence the acyl substituent
levels but does significantly reduce the molecular weight of the high acyl gellan gum 16.
-
Table 1 The effect of high pressure homogenization on HA gellan gum molecular weight
Gellan gum sample % acetate % glycerate -

Mw (daltons)
Clarified native gellan B995 16 2.7 9.8 2,461,000
1 pass at 8000 psi 2.7 10.1 1,261.000
4 passes at 8000 psi 2.6 9.9 925,000
Kelcogel 9EX003A 0.1 0.1 270,000
It can be seen from Figure 2 that a reduction of HA gellan gum molecular weight reduces the
hot viscosity and more significantly, the gel transition temperature of the HA gellan gum
molecules. Reducing the molecular weight of HA gellan is especially advantageous in
applications having high solute concentrations where high viscosity and pre-gelation may be
problematic during processing. It can also be seen from Table 1 that the homogenized HA
gellan gum samples are still significantly higher in molecular weight than LA gellan gum. To
demonstrate that this difference is not due to molecular association of the gellan gum in the
GPC column, atomic force micrographs were prepared and molecular lengths of HA gellan
gum, sonicated HA gellan gum and LA gellan gum were measured (Figure 3). The scale of the
micrograph for the HA gellan gum is 6 pm by 6 pm, and for the low acyl gellan gum 3 pm by
3 pm. The difference in the average molecular length of the HA gellan gum compared to the
shorter LA gellan gum is apparent. Another difference between the reduced molecular weight
HA samples compared to the control is the gels can have improved organoleptic properties 16.
Compared to gels made with a control HA gellan gum, gels made with reduced molecular
weight HA gellan gum exhibit reduced cohesiveness, elasticity and hardness. A possible
consequence of this difference is improved mouthfeel characteristics 16. A textural comparison
is shown in figure 4.
F i g u r e 2 - S e t T e r n p e r a t u r e o f G e l l a n G urn
a s Influenced b y Molecular W eight
1 o5
P 8
2 id
zed
3
b
1. 00
_
50.0 55.0 60.0 65.0 70.0 75.0 80.0 85.0 90.0 95.0 100.0
0 . 5 % S a m p i e 8 In 6 m Y C n " T e m p ['C]
M e a s u r e d w i t h a Rheometric Sclentific S R 2 0 0 a n d a 5 0 m m Parallel Plate T o o l a t 1 0 radls
New Materials 30 1
Figure 3 - Molecular size of H A a n d L A G e l l a n G u m

u s i n g A t o m ic F o r c e Microscopy
H A Gellan S o n ica ted L A Gellan g u m

H A Gellan
Figure 4 - Effect of homogenization on gel texture -

15% solids
Modulus
Control native gellan

-- -8;kses homogenizer
-- 10 p a s s e s homogenizer
Effect of acetate substituents on gel texture

Previous studies have evaluated the properties of low acetate high acyl gellan gum, prepared
by various levels of cold alkali treatment at 25"to 40C for various holding times These '*.
studies concluded that a low acetyl gellan gum produced a non brittle gel with reduced setting
temperature compared to a control HA gellan gum. It was also proposed that strong
aggregation of HA gellan gum double helixes was inhibited by the presence of acetyl groups,
and that the acetyl substituents rather than a certain critical level of glyceryl, are primarily
responsible for minimizing the helix-helix aggregation that gives rise to thermal hysteresis in
LA gellan gum 9, ".
Table 2 - T h e effect of acetate substituents on HA gellan gum texture and setting temperature
HA gellan gum 9% acetate 96 glycerate Ielasticity set temp "C
LA gellan - Kelcogel79119 A 0.16 0.27 18% 41C
Native gellan gum - S-60 GPE-2 2.9 7.8 86% 85C
Low acetate HA gellan (LAG -1) 0.2 8.7 80% 86C

Figure 5 - Effect of acetste removal of HA gellan texture
Figure 6: 1% Low Acetate H A Gellan MeltinglSetting Data
I-' ' ' ' I ' ' ' ' I ' ' ' ' '' ' ' ' I ' ' ' ' '' ' ' ' ' ' ' ' '4
Setting--*
101 L
'k,
30 40 50 60 70 00 90 100
Temp ['C]
2 % strain, 5 C cooling r a t e p e r m i n , B o h l i n C V O r h e o m e t e r
This study evaluated the rheological and textural parameters of a low acetyl bacterial strain
(LAG-1) of high acyl gellan gum, as shown in Table 2. This molecule, although almost devoid
of acetate groups, contains a similar level of glycerate to the control HA gellan gum.
Interestingly, it forms an elastic gel with very similar gel texture and melt set properties to the
control HA gellan gum (Figure 5) with little thermal hysteresis (Figure 6). These results
suggest that the acetate substituents play a less significant role in the texture and rheological
properties of HA gellan gum than previously reported 99 19.
Effect of glycerate substituentson gel texture

The effect of cold alkali to produce a glycerate rich HA gellan is discussed above 9v 12. When
gellan gum at temperatures above its conformational transition is subjected to alkali treatment
both acetate and glycerate are removed at a similar rate 99 12. Morris has reported that it is the
presence of glycerate residues that principally stabilize helical formation and result in the
higher setting temperatures of HA gellan gum '. New studies have shown that if HA gellan
gum is treated with various weak alkalis when the gum is in a disordered conformation
(95"C), that glycerate residues are removed at a faster rate than acetate residues 14* *O. By
varying the level of weak base, a wide range of acetate rich, partially deacylated gellan gum
molecules can be studied. The textural and setting temperature parameters of these partially
deacylated prototypes are shown below in Table 3. It can be concluded that low levels of
glycerate deacylation have a significant effect on setting temperature reduction (7 1"C to 47C)
whilst 96 yield strain remains at a high level (76% to 77%).
New Materials 303
Table 3 - Effect of acyl substituents on the properties of 0.5%wlw HA gellan gum gels.
per repeat unit set temp Modulus Yield strain Elasticity
Glycerate Acetate (C) (Ncm-) (%) (%)
(1) 0.58 0.41 71 0.391 76.0 58.6
(2) 0.53 0.39 62 0.358 79.2 46.6
(3) 0.35 0.36 47 0.335 77.0 32.1
(4) 0.07 0.23 31 0.867 56.3 23.1
(5) 0.00 0.01 31 2.850 44.9 26.3
This effect is also displayed in Figure 7, where the setting temperature and the yield strain of
HA gellan gum are plotted against the level of added trisodium phosphate. The ability to
reduce setting temperature whilst retaining many of the elastic properties of the HA gellan
gum has many advantages especially in high solids applications. Figure 8 shows the melt-
setting properties of two different acetate rich prototypes where glycerate residues have been
removed with weak alkali treatment . Compared to the HA gellan gum control, removal of
45% of the glycerate residues and only 15% of the acetate residues (figure 8a) results in a
gellan molecule with lowered setting temperature and thermoreversible melting with little
thermal hysteresis. Further removal of acyl groups with higher levels of weak alkali, so that
88% of the glycerate residues and 44% of the acetate residues are removed compared to the
HA control (figure 8b), shows very different rheological behavior. In this case, the gellan
molecule does not show thermoreversible behavior and a high level of thermal hysteresis is
observed.
This result (figure 8b) suggests that the presence of a significant level of acetate residues,
although maintaining some elasticity in the system compared to a fully deacylated gellan gum
(Table3), does little to prevent thermal hysteresis when glycerate residues are predominately
removed.
Figure 9 shows the textural impact of glycerate substituents on a new set of partially
deacylated gellan gum prototypes where the % acetate levels are all maintained at a high level
compared to the HA gellan control. The results show that small reductions of glycerate (whilst
reducing setting temperatures significantly) have little textural impact, but further removal of
the glycerate residues (in this case 67%) leads to a texture more similar to LA gellan gum,
where a gel with low cohesiveness and relatively high modulus is formed.
5 CONCLUSIONS
Previous studies had accounted for the suppression of helix-helix aggregation in high acyl
gellan gum to one of two factors . The first was that acetyl groups on the periphery of the HA
gellan gum helix inhibited helix-helix aggregation. An alternative proposal was that the
presence of the bulky glycerate residues, embedded within the double helix altered the

stereochemical structure around the carboxyl group on the gellan gum . The results from this
study suggests that it the presence of a certain level of glycerate substituents that are
principally responsible for the enhanced helical stability in HA gellan gum and also for the
lack of thermal hysteresis in HA gellan gum.
It appears that the acetate substituents play a minor role in the enhanced helical stability in HA
gellan gum, and are not the significant factor inhibiting helical aggregation by their presence
on the outside of the gellan gum double helix.
304 Gums and Stabilisersfor the Food Industry I1
Figure 7 - Effect of weak alkali treatment at 90C on HA gellan gum
80
- -
-X-
-&
Setting temperature
Yield strain %
10
0
0 1.52 3.05 6.1 12.2
Trisodium phosphate(TSP) level (Mmol)
Figure 8 - L o w glycerate H A gellan g u m , m elt-set properties

100
n 10 dm
(d
b
b l 10
.
tJ0.1
Q1
0.01 \
10 20 30 40 50 60 70 80 10 2x) 30 40 9 60 mm
Temperature ("C) Tm-=("q
Cooling ( O ) a n d heating ( 0 ) profiles o f partially d e a c y l a t e d gellan
gum (0.5% w/w in 2 m M calcium) with acyl content of,
a ) fg = 0 . 3 2 , fa = 0 . 3 5 , b ) fg = 0 . 0 7 , fa = 0 . 2 3 .
Control K g e l L T l O O s a m p l e h a d acyl content of fg = 0 . 5 8 , fa = 0 4 1
Figure 9 - Effect of glycerate substituent on gel texture

Modulus
2.500 T
Cohesiveness
--- - -
-Control g=0.79 a=0.51
g=0.65 a d . 4 9
-g=0.26 azO.43
New Materials 305
6 REFERENCES
1. Gibson, W. and Sanderson, G.R. In Thickening and Gelling Agents for Food 2"d
edition, ed. Imeson, A.P. Blackie Academic & Professional, London (1997) 119.
2. Kang, K.S. and Veeder, G.T. U.SPatent 4,326,053 (1982).
3. Jansson, P.E., Lindberg, B. and Sandford, P., Carbohydrate research, 124, (1983),
135-139
4. Kuo, M.S., Mort, A.J and Dell, A., Carbohydrate research, 156, (1986), 173-187
5. Chandrasekaran, R., Millane, R.P., Arnott, S. and Atkins, E.D.T. Carbohydr.
Res., (1988) 175, 1-15.
6. Chandrasekaran, R., Puigjaner, L.C., Joyce, K.L. and Amott, S., Carbohydr. Rex,
(1988) 181,23-40.
7. Chandrasekaran, R.& Thialambal, V.G. Carbohydr. Polym., (1990) 12,43 1-442.
8. Chandrasekaran, R., Lee, E.J., Radha, A. and Thailambal, V.G. (1992). In
Frontiers in Carbohydrate Research-2, ed. Chandrasekaran, R., Elsivier Applied
Science, New York pp 65-84.
9. Morris, E.R., Gothard, M.G.E., Hember, M.W.N., Manning, C.E. and Robinson,
G. Carbohydrate Polymers, 30, (1996), 165.
10. Morrison, N.A., Swom, G., Chen, Y-L., Talashek, T. and Clark, R.C. Progr
Colloid Poly Sci ,114, (1999), 127-131
11. Kasapis, S., Giannouli, P., Hember, M.W.N., Evageliou, V., Poulard, C., Tort-
Bourgeois, B. and Sworn, G. Carbohydrate Polymers, 38, (1999), 145-154.
12. Baird, J.K., Talashek, T.A. and Chang, H. In Gums and Stabilisers for the Food
Industry 6, eds. Phillips, G.0, Williams, P.A and Wedlock, D.J. IRL Press,
Oxford, (1992) 479.
13. Jay, A.J., Colquhoun, I.J., Ridout, M.J., Brownsey, G.J., Moms, V.J., Fialho,
A.M., Leitao, J.H. and SA-Correia, I. Carbohydrate Polymers 35, (1998) 179.
14 Sworn,G., Cranfield Univ, PhD thesis ,Chapter 4.
15. Cheetham, N.W.H. and Punrickvong, A. Carbohydrate Polymer, 5 , (1995), 399.
16. Clark, R.C., Chen, Y-L., Talashek, T., Morrison, N.A. and Burgum, D. US Patent
6,242,035 B1
17. Paradossi, G ., and Brant, D., Macromolecules 15, (1982), 874).
18. Sworn, G., Sanderson, G.R. and Gibson, W. Gellan gum fluid gels. Food
Hydrocolloids, (1995) 9,265 - 271.
19. Gothard, M.G.E., Cranfield Univ ,UK, PhD thesis , Chapter 3.
20. Sworn, G., Chen, Y-L.,Talashek, T., Morrison, N.A. and Clark, R.C. European
Patent Application 98 870 131.4 G (1998).
21. Sworn, G., Handbook of hydrocolloids., eds. Phillips, G.O. and Williams, P.A .
CRC Press, (2000), Chapter 7, 117-135.
RHEOLOGICAL PROPERTIES AND POTENTIAL INDUSTRIAL APPLICATION OF
KONKOLI (MAESOPSIS EMINI) SEED GUM
J.T. Barminas and I.C. Eromosele
Department of Chemistry, Federal University of Technology, P.M.B. 2076, Yola, Nigeria.
1 INTRODUCTION
Polysaccharide gums are obtained from a large number of sources such as seaweed, tree
excudates, plant microbial biosynthesis and seeds. The most common seed gums otherwise
called the galactomannan gums are locust bean and guar gums. These gums have found use as
thickeners mainly in the food, pharmaceutical, textile, paper and cosmetic industries. 2,3,4*5
In Nigeria, the increased demand for food gums has encouraged significant expansion
into sourcing of other natural products as alternatives to commercial gums. Research reports
indicated that water soluble hydrocolloids could be extracted from seeds of achi (Brachystegea
eurycoma) and osbono (Irvingia gabonensis) and could be utilized as stabilizers in the
production of ice cream similar to commercial gums such as carboxymethyl cellulose, kappa-
carrageenan and sodium alginate.6 Another possible food hydrocolloid that could be
employed as a thickener or suspending agent is konkoli (Maesopsis eminii) seed gum.
Maesopsis eminii plant is a medium-sized tree that is widespread in tropical Africa.7 It
produces purplish black fruits containing very hard stone seeds between August and
September every year. Powdered konkoli seeds when dispersed in water produces a viscous
solution which is used traditionally in thickening soup and widely mixed with bean meal to
make a locally fiied cowpea paste called kosai or akara in Nigeria. This product is
consumed as a snack and breakfast food and has a break-like character similar to hush
puppies.
In this paper, the results of studies on the flow behaviour of konkoli gum in aqueous
solutions and its potential uses are presented.
2 MATERIALS AND METHOD
Seeds from konkoli were purchased from a local market in Gembu, Taraba State, Nigeria. The
very hard stone seeds were washed with distilled water and dried at 55C in an oven for 5h,
powdered in a mortar and finally ground in a manually driven attrition mill. The powder was
hrther sieved through a 100pm screen to obtain a fine powder and packed in an air-tight
bottle. This sample was used in our investigation without hrther treatment and purification.
New Materials 307
Gum solutions were prepared by dispersing konkoli gum in distilled water or sodium
chloride solutions (0.5 x lOM) at a given temperature by sprinkling the gum samples on the
sides of the vortex formed in a magnetic stirrer and stirred for lh. These solutions with
concentration between 0.5-2.0%(w/v), were allowed to equilibrate at room temperature for
24h. Viscosities o f the solutions were measured using Rheotest-2 rotation viscometer
(Germany). A constant temperature bath was used to control temperature. All measurements
were made in triplicate and were averaged. Experimental data were evaluated by analysis of
variance using the general linear model procedure of the Statistical Analysis System.8 A non-
linear regression analysis of the above program was used in determining yield stress. Also
with the evaluated yield stress, a linear regression analysis was performed using the SAS
package to determine other flow properties such as consistency and flow behaviour indices
from the Hershel-Bulkley model:
z = zo + ky
where zo is the yield stress (Nm-2), K is the consistency index (NSn/m2), n is the flow
behaviour index, z is the shear stress (Nm ) and y is the shear rate (s-l).
Konkoli gum dispersions in water consist of an insoluble but swellable fraction which
confers water-swelling properties to the gum and a water-soluble polymer fraction which gives
the gum its solubility. Therefore konkoli gum may be composed of mixtures of
polysaccharides which produce viscous aqueous solutions having a texture similar to that of a
soft gel.
Figure 1 shows that konkoli gum water solutions show a pseudoplastic behaviour
similar to other galactomannans such as guar and locust bean gums. 3,9~10~11 This flow
behaviour was exhibited at all polymer concentrations and temperatures studied. The flow
curve for 2% ( w h ) gum concentration was a continuous curve. With other concentrations
140
120
%
3 100
v
Concentration
x
.-v) so
+
0 1.5
0
v)
5 60 x 2.0
4
c
2
m 40
L1
2-20
0
0 100 200 300 400 500
Shear rate (5-)

Figure 1 Apparent viscosities of aqueous konkoli gum solutions
at different concentrations (% w/v)
below this, the flow curve had a peak at a shear rate approximately 10 s-'. Beyond this shear
rate the viscosity of gum solutions gradually decreases. This behaviour could be explained in
terms of intermolecular network formation by gum molecules 3711 leading to a system that
could be described as a soft or weak gel with a yield stress. The yield stress increases with
increase in gum concentration as indicated in Table 1.
Table 1 Effect of gum concentration oneflowproperties of konkoli gum
Concentration To (Nm-2) K N R2
% (w/v)
0.5 3.98 k0.04 1.21 * 0.01 0.52 * 0.02 0.9865
1.o *
5.84 0.02 1.35 * 0.01 0.60 + 0.01 0.9877
1.5 12.68 f 0.68 *
1.60 0.02 *
0.55 0.01 0.975 1
2.0 15.86 * 0.05 2.64 + 0.01 0.54 i 0.01 0.9950
Rheological measurements were made while increasing shear rate from 0 - 437 s-l in 5
mins (ascending curve) and subsequently decreasing from 437 5-l to 0 s-l in the next 5 mins
(descending curve). Typical upward and downward flow curves for 2% (w/v) dispersion is
shown in Figure 2. This test solution showed some textural breakdown since the upward and
the downward flow curves did not overlap indicating lack of recovery of apparent viscosity.
This aspect must be considered for handling of such solutions in technological processes and
their use in consumer products.
80 I I
o ! I I I I 1
'0 100 200 300 400 500
Shear rate (s-' )
Figure 2 Upward and downward flow curves for 2 %(w/v)
aqueous konkoli gum solution at 25OC:
From the results in Table 1, the Herschel-Bulkley model determined by linear

regression described the flow behaviour of konkoli gum satisfactorily. The flow behaviour
index (n) was not affected significantly when the concentrations of gum solution were
increased. However, increase in concentration, affected the consistency index of konkoli gum
New Materials 309
solutions. Since the consistency index defines the viscous nature of solutions, this observation
is not unexpected.
Table 2 shows the influence of sodium chloride solution on the apparent viscosity of a
2% (wh) dispersion of the gum. These viscosities were not adversely affected by dilute
solutions of sodium chloride. This property shows that the konkoli gum may find application
as a suspending and thickening agent in pharmaceutical formulations. However, the apparent
viscosity decreases when 2% (w/v) konkoli gum solutions was heated as shown in Figure 3.
This effect was caused by loss of hydration around the polymer molecule thereby increasing
the flexibility of the polymer chain and weakening the interaction of molecules in s~lution.~
Table 2
Apparent Viscosity Nsm-3 of 2% (wh) of konkoli gum dissolved in drfferent concentrations of

sodium chloride at different shear rates and 2%
Concentration of NaCl ( M)
Shear rate (s'l) ..........................................................................................
0 0.5 1.o 1.5
5.40 217.03 * 1.48 *
218.22 4.13 220.33 * 1.44 222.63 * 1.84
9.00 162.28 * 2.87 *
165.14 1.44 166.54 * 1.02 167.23 0.56
16.20 126.60 * 3.55 *
127.51 1.23 128.24 * 1.25 129.33 * 2.13
27.00 81.84 * 0.86 82.84 *
0.24 83.84 * 0.25 84.50 * 0.48
48.60 60.29 * 0.77 61.03 *
2.40 62.43 * 1.84 63.55 * 0.26
81.00 43.41 * 0.86 44.23 *
0.33 45.28 * 1.23 46.43 * 2.44
P
E
2
W 200
)I
.-
c
v)
0
;. 150 4 5OoC
> 40C
c
c
2 A 30C
g 100 x 25C
a
50
0
0 50 100
Shear rate(s-')
Figure 3 Efect of temperature on the apparent viscosity of 2%(w/v)

konkoli solution measured at different temperatures
3 10 Gums and Stabilisersfor the Food Industry 1 1
3 CONCLUSION
Konkoli gum forms viscous solutions whose apparent viscosities could be comparable to guar
gum and other galactomannans. One usefbl property exhibited by konkoli gum solutions is
pseudoplastic rheology. The viscosities of the solutions were high at low shear rates, yet low
at high shear rates. This shows that these solutions could be used as a viscosity builder,
stabilizer and may provide food processors with many advantages in a wide variety of nonfat
and low fat applications including frozen desserts, baked goods, dairy products, salad
dressings and condiments. Also, it could be used as a suspending agent in pharmaceutical
formulations. The chemical and physicochemical characteristics and the interactions of
konkoli gum with xanthan gum are under investigation.
References
1 Y. Pomeranz, Functional Properties of Food components. Academic Press Inc.

Sandiego California, 1991, Ch.3, p.79.
2 J.D. Dziezak, Food Technol, 1991, 3, 1 16.
3 A.M. Elfak, G. Pass, G.O. Phillips, JSci. FoodAgric., 1979, 30, 439.
4 H. Mair, M. Anderson, C. Karl and K. Magnuson, in Industrial Gums:
Polysacchurides and their Derivatives, 3'd Edn., ed. R.L. Whistler and J.N.
BeMiller, Academic Press, New York, 1993, ch.2, p.205.
5 I.C.M. Dea and A. Morrison, Adv. Carbohydr. Chenz. Biochem., 1975,31,241.
6 A. Uzomah and R.N. Ahiligwo, Nig. Food J., 1995,3,63.
7 R.W.J. Keay, C.F.A. Onichie and D.P. Standfield Nigerian Trees V0Z.U. Federal
Department of Forest Researc, Ibadan, Nigeria 1064, ch.3, p.226.
8 Statistical Analysis System, SAS/STAT Users Guide (Release 6.03 Edn), SAS
Institute Inc., NC 1988.
9 C.A. Hoppe and A. Goswani, Polym. Prep., 1998,39,690.
10 J.A. Casas, A.F. Mohedano, F. Garcia-Ochoa, J. Sci. Food Agric., 2000, 80, 1722.
11 F. Garcia-ochoa and J.A. Casas, J Sci. FoodAgric., 1992, 59, 97.
Emulsifying Properties of Depolymerized Citrus Pectin: Role of the
Protein Fraction
Mahmood Akhtar'., Eric Dickinson', Jacques Mazoyer' and Virginie Langendorffj

1
Procter Department of Food Science, University of Leeds, Leeds LS2 9JT, U.K.
2DegussaTexturant Systems, Centre de Recherche - 50500 Baupte, France
* To whom correspondence should be addressed
1 INTRODUCTION
Pectin is composed mainly of polymerized and partly esterified galacturonic acid

(200-1 000 units), and the percentage of esterified galacturonic acid strongly influences
the functional properties of the hydrocolloid. About 5-10% of the pectin is composed of
neutral sugars such as galactose, glucose, rhamnose, arabinose and xylose. Pectins from
apple, citrus, cherry, carrot, sugar beet, potato and cabbage have the same neutral sugar.'
Pectin is extracted industrially from citrus peel (lemon, lime and grapehit) and apple
The food industry is the most important field of application of extracted
pectin. Pectins from different sources are widely used to stabilize food emulsions and
dispersions in products such as fmit drinks, and fruit and tomato pastes? Pectin can form
gels under certain circumstances, and it is used as a gelling agent in jams, jellies and
marmalades. Pectin is also used to stabilize clouding in beverages with such stability
dependent on the nature and amount of pectin present.' For a polysaccharide to function
satisfactorily as an emulsifier of flavour oils, both the concentrated flavour oil emulsion
and the diluted beverage must remain stable for a period of several months.6
While much research has been carried out on the properties of pectin:** poarticularly
on the composition of the constituent sugars: the solution properties, and the
mechanism of gelation,' the interfacial properties have received little attention.I2 In
order to understand the physico-chemical properties of pectin-stabilized emulsions, the
relationship of function to structure is particularly of interest. This is a challenge because
pectin has the most complex molecular structure found amongst food poly~accharides.'~
It has been demonstrated recently that depolymerized pectins of molecular weight
below 80 kDa derived from citrus h i t s and apples can possess good emulsion stabilizing
characteristics even though they are low in acetyl groups (< 0.8 %).I4 This good
emulsifLing behaviour was found to be associated with a much higher surface activity of
the depolymerized pectin as compared with normal citrus or apple pectin.
We report here on the surface and emulsifying properties of depolymerized pectin, in
relation to the adsorbed pectin concentrations and the interfacial protein content, and
their respective contribution in the formulation of a stable oil-in-water emulsion. The
effects of pH, pectin concentration, and degree of esterification on the emulsion stability
are investigated. For these experiments, samples of depolymerized pectin were prepared
from citrus peels by acid hydrolysis; the samples were also enzymatically treated for
demethoxylation to obtain various degrees of methoxylation.
2.1 Materials
Depolymerized citrus pectin (molecular weight 70 kDa) was obtained from Degussa
Texturant Systems, France, and the rapeseed oil was a gift from St. Ivel (Swindon, UK).
The degree of esterification and the anhydrous galacturonic acid content of the citrus
pectin are shown in Table 1. The molecular weight was estimated by intrinsic viscometry
(calibrated against known pectin products), and the degree of methoxylation and the
galacturonic acid content was determined by titration.
Table 1 Degree of esteriJication (DE) and anhydrous galacturonic acid

Content (AG) of depolymerized citrus pectin of molecular
weight 70 kDa.
Sample DE (%)a AG (%)

30 67
48 75
55 65
71 81
a Based on standard method given in 'Pectins', pp. 87-91, Compendium of

Food Additive Specifications,FA0 FNP No. 52, Add. 1, Rome, 1992.
Depolymerized pectin sample was prepared by treating the citrus peels with nitric acid
at pH = 2.3 for 3 hours at 85 "C in the presence of various amounts of hydrogen peroxide
(4-8 ml of 30% H202 per kg of peels). After purifying the slurries by filtration, the slurry
syrups were concentrated by ultrafiltration, and the pectin samples were recovered by
precipitation into isopropyl alcohol. The products were then dried and ground.
Pectin samples were also enzymatically treated for demethoxylation. Powders were
dissolved in water and pH was adjusted to 4.5 using diluted sodium hydroxide. Various
amounts of the pectin-methyl-esterae RheozymeTMof activity 10 PEU/ml (Novozymes
S.A., Nanterre, France) were added to the solution at 40 "C, and the pH was maintained
at 4.5 by automated addition of NaOH. To inhibit the enzyme at the end of the reaction,
HNO3 was added to pH 1.8 and the temperature raised to 70 "C. Resulting pectin
solutions were precipitated into isopropyl alcohol, dried and ground.
2.2 Emulsion preparation
The aqueous buffer was prepared using double-distilled water, citric acid and sodium
citrate, with 0.01 wt% sodium azide (Sigma Chemicals) added as antimicrobial agent.
-
The buffer solution was heated to 50 "C and pectin powder was added slowly with
gentle stirring. The pH of the resulting pectin solution was adjusted to pH 4.7 or pH 7 by
adding a few drops of freshly prepared NaOH (1 M) solution.
New Materials 313
A laboratory-scalejet homogenizer was used to make oil-in-water emulsions at room

temperature. The jet homogenizer block has two chambers, one for the oil phase and the
other for the aqueous phase, in different ratios (e.g. 11:89, 2090, 4555). The water and
oil hases were passed through the chamber of the jet homogenizer at a pressure 300
bar.7 5 These emulsions were stored at room temperature. The pH values mentioned in this
paper refer to the pH of the pectin solution before emulsification.
2.3 Droplet Size Determination
Droplet-size distributions and average droplet sizes (volume-surface mean diameter

d32 and weighted average mean diameter d43) were determined immediately after
emulsion making and as a function of storage time at 25 "C using a Malvern Mastersizer
MS2000 (static light-scattering apparatus). The sample settings for the measurement
were: pump speed 1250 rpm, stirrer speed 500 rpm, measurement time 5 second and
measurement snaps 5000. The key parameters used in the droplet size analysis were:
refractive indices of the disperse phase (e.g. RSO; 1.471) and the continuous phase
(water; 1.330), general purpose model, absorption (0.001) and the size range (0.020 to
2000 pm). The average droplet size was characterized by two mean diameters, 4 2 and
d43, defined by
where ni is the number of droplets of diameter dj.. The d32 value was used to estimate the
specific surface area of freshly made emulsions, and the d43 value was used to monitor
changes in droplet-size distribution on storage. States of droplet flocculation were
assessed qualitatively by examining emulsions by light microscopy using the Normarski
differential interference contrast technique.I6 Creaming behaviour was examined by
visually measuring the height (thickness) of cream and serum layers in emulsions stored
at 22 "C at regular time intervals.
2.4 Pectin and Protein Determination
The amount of depolymerised pectin adsorbed on the droplet surface following

emulsification was determined using a colorimetric titration method with
metahydroxydiphenyl (MHDP)." A standard calibration curve was obtained using pectin
solutions at known concentrationswith the absorbance measured at 525 nm.
The emulsions were centrifuged at 1.1 x lo4 rpm for 2-4 hours, and then the serum layer
was separated and diluted for the colorimetric titration. A standard calibration curve was
used to determine the concentration of pectin in the serum layer. Since the concentration
of pectin in the aqueous phase before emulsification was known, the difference gave the
amount of pectin adsorbed onto the droplet surface. The reliability of the analysis is
largely dependent on the effectiveness of the serum separation. Under conditions of good
separation, the percentage adsorbed is considered to be precise to f 3 %.
The Coomassie blue dye-binding method was used to determine the protein content of
pectin solutions before emulsification and of the serum layers of pectin-stabilized
emulsions after centrifugation (1.1 x 1O4 rpm for 4 h).' * A standard calibration curve was
established using bovine serum albumin in 0.15 M NaCl with absorbance measured at
595 nm. For the case of the original pectin sample, the protein content was also
determined independently by the semi-micro Kjeldahl method." The values obtained
3 14 Gums and Stabilisers for the Food Industry 11
were 0.62 f 0.03 % and 0.76 f 0.04 % for the Coomassie blue and Kjeldahl methods,
respectively.
We report results of experiments carried out on depolymerised pectin samples of similar

molecular weight (70 kDa) with varying degree of esterification (see Table 1). Rapeseed
oil (RSO) (10 ~01%)was emulsified at pH 4.7 and 7 with the pectin at an aqueous phase
concentration of 4 wt%. Emulsion stability was assessed over a period of 5-7 weeks on
the basis of droplet-size measurements, creaming behaviour, and microscopic and visual
observations.
20
0
30 48 55 71
degree of esterification (YO)
Figure 1 Effect of degree of esterification on average droplet size and creaming

behaviour of RSO emulsions (20 vol% oil, ionic strength 0.2 M) stabilized by
depolymerizedpectin of molecular weight 70 kDa at p H 4.7.
Figure 1 compares the properties of rapeseed oil-in-water emulsions stabilized by

depolymerized pectin with varying degrees of esterification (DE = 30-71%) in terms of
average droplet size at pH 4.7 over a storage period of 32 days. The initial droplet size
was large (> 8 pm) for the RSO emulsions made with the pectin with a low degree of
esterification (3048%). However, the emulsions made with a high degree of
esterification (55-7 1%) have shown better emulsifying properties in terms of the initial
average droplet size (< 4 pm). On the basis of these results, it is apparent that the
depolymerized pectin samples with a high degree of esterification have good emulsifying
capabilities for vegetable oil under these acidic conditions. Hence this highly esterified
sample was chosen to further explore the emulsifying properties of depolymerized pectin.
Figure 2 shows the effects of pectin concentration and pH on the average droplet size
after one month in emulsions made with RSO (10 ~01%)at pH 7 and 4.7. It is clear that
the hydrocolloid emulsifier is highly effective at considerably lower emulsifier/oil ratios
under acidic conditions. At pH 7 there is a steady improvement in d43 with increasing
pectin content up to 12 wt%. At pH 4.7 the optimum emulsification behaviour is already
New Materials 315
reached at ca. 4 wt% pectin. At both pH values, there is a suggestion of a slight increase
in d43 at high pectin contents.
20
n +pH 4.7
f 15 *PH 7
3
.-8In 10
x 5
e
U
0 I I I I 1
0 4 8 12 16
depolymerised pectin (wt%)
Figure 2 EfSect of pectin concentration on average droplet size of RSO emulsions (10
vol% oil, ionic strength 0.2 M) stabilized by depolymerized pectin of molecular weight 70
m a with degree of esteriJication 71% at pH 7 and 4.7.
~~ ~
+pectin adsorbed
70
+adsorbed fraction
55 n
5
40 2
25
10
0 2 4 6 8
pectin in the aqueous phase (wt%)
Figure 3 EfSect of the pectin concentration in the aqueous phase on the amount of
pectin adsorbed and the fraction of pectin adsorbed in emulsions made with 20 vol%
RSO at pH4.7.
In order the estimate the amount of pectin adsorbed on the droplet surface, the
emulsions were centrifuged and the resulting serum phase was analysed for the pectin
content. The results are summarized in Figure 3 for the case of 20 vol% RSO emulsions
made at pH 4.7 with 2-6 wt% of the depolymerized pectin of molecular weight 70 kDa
(DE 7 1%). Increasing the aqueous phase pectin emulsifier concentration increases the
amount of pectin adsorbed on the droplet surface under similar homogenization
conditions. However, independent of the concentration of emulsifier used, the fraction of
3 16 Gums and Stabilisers for the Food Industry I1
pectin adsorbed at the oil-water interface appears to remain constant at ca. 25%. The
aqueous phases of the original pectin solutions were also analysed with respect to protein
content. A small amount of protein (- 0.7%) was found to be associated with the
depolymerized pectin.
In order to investigate further the degree of importance of the pectin protein content in
relation to emulsion stabilization, a set of three 20 vol% RSO emulsions was prepared
sequentially from a solution of 4 wt% pectin emulsifier at pH 4.7. In this set of
experiments, the aqueous phases used to prepare the second and third emulsions were
taken, respectively, from the serum layers separated after centrifugation of the first and
second emulsions. The data for the pectin and protein contents before and after
emulsification, and the time-dependent emulsion droplet sizes, are compared in Table 2.
A negligible proportion of the original protein content of the pectin sample was detected
in the aqueous phase after emulsification (see Table 2). Hence the protein fraction of the
hydrocolloid emulsifier is considered to be almost entirely associated with the adsorbed
layer around the droplets. This suggests a strong correlation between the protein fraction
of the depolymerized pectin and its emulsifying capability.
It has been shown that the non-adsorbing pectin fraction arising from centrifuging the
first emulsion acts as an extremely poor emulsifier in the subsequent emulsion as
reflected by a large increase in the average droplet (see Table 2). The specific surface
area reduces dramatically from 4.3 to 0.4 m2g-' between the first and second emulsion,
and a further reduction by a factor of 2 between the second and third. The large reduction
in specific surface area clearly demonstrates the loss of interfacial functionality during
subsequent emulsificatiodcentrifugation.
The data in Table 2 also show that the amount of non-adsorbed protein remaining in
the serum layer becomes increasingly reduced with each subsequent stage of
emulsificatiodcentrifugation.Hence, the more protein that is removed from the solution
of the depolymerized pectin sample, the increasingly less effective the hydrocolloid
becomes as an emulsifier.
Table 2 Properties of emulsions made with pectin fraction in serum phase of previously
centrifuged emulsions (20 vol% RSO, pH 4.7); CO= pectin concentration in the
aqueous phase before emulsijication; CS = pectin concentration in the serum
phase after centrifugation, Fads = fraction of pectin adsorbed; A , = speciJic
surjace area; d43 = average droplet size; CPS= protein concentration in serum
layer after centrifugation.
Pectin Sample Co (wt%) Cs (wt%) Facis (%) d43 (p) ]!S$ , CPS (%)
(m g )
Initial emulsion 4 3.1 23 4 4.3 0.008
Serum 1 emulsion 3.1 2.9 7 68 0.4 0.003
Serum 2 emulsion 2.9 2.8 3 164 0.2 0.001
'Based on d32 values determined for each emulsion
4 CONCLUSIONS
It has been demonstrated that depolymerized pectin derived from citrus peel can be used
as an effective emulsifying agent for formulating food emulsions under acidic conditions.
New Materials 317
In particular, rapeseed oil-in-water emulsions made with depolymerized pectin of

molecular weight 70 kDa and degree of esterification 70% at relatively low pectidoil
ratios were found to have excellent stability in terms of average droplet size and
creaming behaviour over a two-month storage period. Emulsions made at pH 4.7 were
found to be more stable than those made under neutral pH conditions.
The pectin and protein analysis of the serum layers shows that only a modest
proportion of the pectin used as emulsifier actual adsorbs on the droplets, and that this
fraction (ca. 25%) contains most of the protein present in the aqueous phase before
emulsification. In addition, the non-adsorbed pectin in the serum layer separated by
centrifugation is found to be a very poor emulsifying agent. Hence we infer that the
presence of the protein moiety in the depolymerized pectin plays a major functional role
in enhancing the emulsification properties of pectin following depolymerization.
Presumably, as with gum arabiq6 the hydrophobic proteinaceous component generates
the surface activity of the pectin emulsifier and acts as its strong point of anchor at the
oil-water interface, whilst the attached hydrophilic polysaccharide chains provide the
thick protective layer that confers effective steric stabilization during extended storage.20
References
1. C. Rolin, and J. De Vries, Pectin. In P. Harris (Ed.), Food Gels, London: Elsevier
Applied Science, 1990, p. 401.
2. T. Sakai, T. Sakamoto, J. Hallaert and E. J. Vandamme, Adv. Appl. Microbiol.,
1993,39,213.
3. C. D. May, Pectins. In A. Imeson (Ed.), Thickening and Gelling Agentsfor Food,
Glasgow: Blackie, 1992, p. 124.
4. E . Costell, E. Carbonell and L. Duran, J. Text. Studies, 1993,24,375.
5. Z. Elshamei and M. Elzoghbi, Nahrung (Food), 1994,38, 158.
6. E. Dickinson, D. J. Elverson and B. S. Murray, Food Hydrocolloids, 1989,3, 101.
7. Y. A. Antonov, N. P. Lashko, Y. K. Glotova A. Malovikova and 0. Markovich,
Food Hydrocolloids, 1996,10,1.
8. C. R. F. Gross0 and M. A. Rao, Food Hydrocolloids, 1998, 12,357.
9. A. Kawabata and S . Sawayama, J. Japanese Soc. of Food Nutrition, 1975,28,
395.
10. A. Kawabata, S. Sawayama and T. Nagoya, J. Japanese SOC.Food Nutrition
1977,30, 149.
11. B. R. Thakur, R. K. Singh and A. K. Handa, Crit. Rev. Food Sci. Nutrition, 1997,
37,47.
12. D. R. Coffin and M. L. Fishman, J. Applied folym. Sci., 1994,54, 1311.
13. R.J. Redgwell and R.R. Selvendran, Carbohydr. Res., 1986,157, 183.
14. J. Mazoyer, J. Leroux and G. Bruneau, 1999, U.S. Patent No. 5,900,268.
15. I. Burgaud, E. Dickinson, and P. V. Nelson, Int. J. Food Sci. Technol., 1990,25,
39.
16. C. F. A. Cullings, Modern Microscopy: Elementary Theory and Practice.
London: Butterworths, 1974.
17. P. K. Kintner, and J. P. van Buren, Journal of Food Science, 1982,47,756.
18. M. M. Bradford, Analytical Biochemistry, 1976, 72, 248.
19. W. Horwitz (Ed.), Oflcial Methods of Analysis, 1980,3 edn. 858. Association
of Official Analytical Chemists.
20. E. Dickinson, An Introduction to Food Colloids. Oxford: University Press,
1992.
MODIFIED HYDROCOLLOIDS WITH ENHANCED EMULSIFYING PROPERTIES
Florian M. Ward, Ph.D.
TIC GUMS,Inc., Belcamp MD 21017, USA
1 . ABSTRACT:
Hydrocolloids or water-soluble gums, which are mainly plant polysaccharides, perform a

variety of fbnctions and are widely used in the industry as thickeners, stabilizers, gelling
agents, emulsifying agents, and a fiber source. Modification of the hydrocolloids by the
incorporation of lipophilic moieties, e.g., by esterification can yield enhanced emulsifying
properties in addition to their inherent fbnctionality. Several gum substrates for the
modification were employed, including, but not limited to non-emulsifying gum acacia,
guar gum, arabinogalactan and inulin. AAer modification, the products were tested in
beverage emulsions, spraydried flavors, and salad dressings. Emulsion stability inckes
were analysed, including particle size distribution by a Coulter counter device, beverage
ring tests, etc. Viscosities of the solutions were measured by using a programmable
Brookfield DVIII rheometer. Results show significantly higher emulsion stability at lower
gum levels as compared to the control samples.
2. INTRODUCTION
Typically, gum arabic or gum acacia (Acacia species) or modified starches are used in
stabilizing oil-in-water flavor emulsions. EmulsifLing grades of gum acacia are relatively
expensive and subject to seasonal variations. Recently the supply of gum acacia from the
Sudan was subjected to U.S. trade sanctions and embargoes. Additionally, some corn
starch producers have been unable to guarantee that their products are not derived from
genetically modified organisms (GMO), creating a marketing problem where consumers
demand products that are "GMO-free". Due to the above considerations, it would be
desirable to investigate other options by evaluating polysaccharide or hydrocolloid
emulsifiers that are not based on cornstarch or its derivatives, and do not have the supply
problem and high cost of emulsifying grade gum acacia. This study involves the
evaluation of the stability of oil-in-water emulsions using non-starch modified gum
systems.
3 . MATERIALS AND METHODS
The modification of the gums or hydrocolloids is conducted using a method that is "patent
pending". It involves the introduction of lipophilic groups to the polysaccharide by
controlled esterification procedures. After chemical treatment the modified gum system
New Materials 319
may be used as a solution or subjected to dehydration methods such as spray-drying. The

substrates used for the esterification include guar gum (Cyamopsis tetragonolobus),
arabinogalactan (Larix sp.), non-emulsifying gum acacia (Acacia seyal), and inulin, a
fructo-oligosaccharide from chicory root.
After modification, the gums are evaluated for emulsifying properties in flavor or
beverage emulsions and salad dressing base formulations. The particle size distribution of
the emulsions is determined by using a Coulter Counter device (from ElectrozoneRarticle
Data, Inc.). Typically, the smaller the particle size, the more stable is the emulsion. Other
tests include beverage ring tests where the concentrated emulsion is diluted 1:500 with
water, the pH adjusted to 3.8 by addition of citric acid or phosphoric acid, and ring
formation is observed on the finished beverage.
The beverage emulsions were prepared with and without weighting agents. The
weighting agent used is glyceryl abietate, commonly known as ester gum. A 5 % solution
of the modified gum was prepared by dispersing the gum in water and allowing it to
hydrate for 30 minutes with slow agitation while mixing. Then 10 % of the orange oil is
added to the gum solution. A coarse emulsion is prepared by mixing in a Ross mixer for
10 minutes. The oil-in-water emulsion is hrther homogenized at 3000-3500 psi to reduce
the particle size distribution of the emulsion.
The salad dressing base emulsions were prepared by using 0.75 - 1.25% of the
modified gum system at 17 - 25% oil level. The gum is dispersed in water and allowed to
hydrate with mild agitation. The vegetable oil is added gradually, vinegar and salt added,
and the mixture is passed through a colloid mill or a homogenizer. The particle size
distribution of the salad dressing base is analyzed using the Coulter Counter device.
The particle size distribution of the diluted beverage emulsions is determined by using a
Coulter Counter device on Day 1 and Day 7. A comparison of the results using various
modified gum systems is shown in Table 1. Emulsions using control samples without
esterification, which include the emulsifying gum acacia at 5 and 15% gum levels were
used for comparison.
Hydrocolloid Emulsion Particle Size After 7 Days

YOLess than 2pm I
Median Size (pm)
Modified ArabidAcacia 91.3 1.17

Modified Gum Arabic/Guar Gum 91.7 1.13
Modified Inulin 97.8 1.15
Modified Polydextrose 97.6 1.35
Emulsihing Gum Acacia 32.6 2.28

(at 5 % level)
Emulsihing Gum Acacia 63.7 1.73
(at 15 YOlevel)
Non-Emulsijjing Gum Acacia 38.1 2.14
Table 1 Particle size distribution of unweighted beverage emulsions using 5% solution of

modrfied hydrocolloids as compared to those prepared using unmodijed emulsifiing gum
acacia Control (emulsifiing type) at 5 and 15% gum level.
The analyses of the data in Table 1 reveal a significant decrease in particle size
distribution, associated with increased stability in unweighted emulsions prepared with 5 YO
solution of modified gum systems at a 10% oil level. The modified starches typically
recommend 12% gum level and for emulsifying acacia, 15 to 20% gum level is generally
recommended at 10-15% oil level, to yield a stable emulsion. Hence a reduction in gum
level using the modified gums also indicates a considerable cost savings for the
manufacturer.
In weighted emulsions where ester gum was used a weighting agent, the particle
size distributions of emulsions at 5% gum and 10% oil levels are shown in Figure 1 at day
1 and day 7, showing a median size of 1.697pm and 1.704 pm respectively. At Day 1, the
percentage distribution of particle size less than 2 pm was at a high level of 92.0% and at
Day 7 was shown to be 86.4%.
12
10
n
8 8
I5 6
8 4
G0 2
0
0.47 0.8 1.37 2.36 4.03 6.94 11.84
Particle Size Diameter (pm)
Figure I Particle size distribution of a weighted oil-in-water emulsion using 5% solution

of a mod#ed emulsifying system containing acacia and guar gums at 10% oil level, at
days I and 7.
The modified gum systems were also evaluated in salad dressing emulsions where
the modified gum is used to replace propylene glycol alginate and xanthan gum and guar
gum are added as thickeners and suspending agents. The initial results show that the
modified gums systems may be used as emulsifying agents for salad dressing emulsions,
based on particle distribution of emulsions that were prepared using a colloid mill or a
homogenizer. As expected, median particle size using the Colloid mill was larger (2.584
pm) than the homogenized emulsion (1.943 pm), see Figure 2.
5. CONCLUSIONS AND RECOMMENDATIONS

The results of this study show that chemical modification of hydrocolloids to introduce
lipophilic groups yields emulsifying properties that are significantly enhanced compared to
both traditional gums and modified starches used at higher levels. Hence these new
systems may serve as potential substitutes for gum acacia in beverage emulsions and
encapsulated flavors. In salad dressings they may be used as low cost replacements for
New Materials 321
propylene glycol alginate, now typically used in the industry. More work should be done
on verifying the work on other types of oil-in-water emulsions as well as low-fat
formulations since the threshold value of flavor components is influenced by the ratio of
hydrophobic and hydrophilic groups in the medium.
~
A: %c2 pm, Day 7 : = 49.8%

Mean Size = 1.943 rrm
~~
1.67 1.00 1.50 2.26 3.40 5.11 7.68
I Particle Size Diameter ( p m)
Figure 2 Particle size distribution of salad dressing base emulsion containing 17% oil
and 1.25% mod'lfiedgum system with acacia and guar gums after homogenization (A) or
processing through a colloid mill (B).
6 . ACKNOWLEDGMENTS
The author would like to acknowledge the financial support of Steve Andon and Chris
Andon, proprietors of TIC Gums, Inc.; the technical support of the R & D group,
particularly Jim Caulfield, Ken Kuschwara, and Dr. Mar Nieto; and the editing and
assistance of Dr. Richard Ward, IU3W Consulting.
References:
1. M. Glicksman, Food HydrocolZoia?s, 1982, 3, CRC Press, Inc., Boca Raton, Florida,
U.S.A.
2. P. Jurasek, M. Kosik, and G.O. Philips, FoodHydrocolloids, 1993, 7 , 73-85.
3 . M. Kenyon, Modfled Starch, Maltodextrin and Corn Syrup Solids as Wall Materials
for Food Encapsulation, 1995, in American Chemical Society (ACS) Symposium
Series 590.
4. G.A. Reineccius, F.M. Ward, C. Whorton, and S . Andon, Developments in Gum

Acacias for the Encapsulation of Flavors, 1995, in American Chemical Society (ACS)
Symposium Series.
5 . S.J. Risch and G.A. Reineccius, Per-mer and Flavorist, 1990, 15, pp. 55-58.
6. F. Thevenet, in Flavor Encapsulation, 1988. S.J. Risch, and G.A. Reineccius, Eds.
American Chemical Society. Washington, D.C., 37-44.
7. F.M. Ward, Guar Specialty Products, in Gums and Stabilizers for the Food Industry
2000, lo., P.A. Williams and G.O. Philips, Eds., Royal Society of Chemistry,
Cambridge, UK.
8. R.L. Whistler and J.N. BeMiller, Eds., Industrial Gums,3'd ed. 1993, Academic Press.
San Diego, CA , U.S.A.
9. P.A. Williams, G.O. Philips, and R.C. Randall, in Gums and Stabilizers for the Food
Industry, 5, 1990, G.O. Philips, P.A. Williams, and D.J. Wedlock, Eds. Oxford
University Press, Oxford, UK.
Hydrocolloids and Health
EUROPEAN LEGISLATION AND HEALTH CLAIMS
Peter Berry Ottaway
Berry Ottaway & Associates Ltd

Hereford, England
1 INTRODUCTION
The European Union (EU) currently consists of 15 member states and from the mid 1960s
the European Commission has been working to achieve harmonisation of food legislation
across the EU. As a consequence, 85 to 90% of the food law in any member state is
derived from European law.
Over the past decade, there have been a number of European laws that have increased
the complexity of the procedures for companies wishing to place new food additives or
ingredients on the European market or significantly changing the use of existing additives
or ingredients. In addition, the EU laws surrounding the marketing and claims for
bctional ingredients are very restrictive.
2 APPROVAL AS A FOOD ADDITIVE
Under European law, a substance added to a food for a technological purpose (e.g. an
emulsifier, stabiliser or thickener) is classified as a food additive and comes under the
control of the EU Directive on Food Additives other than Colours and Sweeteners
(95/2/EC).' This directive lists the permitted additives, the categories of foods in which
they can be used and the maximum level of use in each particular food category. There is
also a separate directive laying down the specifications and purity criteria for each
permitted additive.
The control of each food additive is based on its assessed Acceptable Daily Intake
(ADI) with substances with the highest ADIs being permitted in a wider range of foods
than those with low ADIs.
The situation regarding the permission to use various gums and stabilisers as food
additives is complex. Some gums, such as acacia gum, gum tragacanth and gellan gum can
be used in most foods with maximum levels based on the quantum satis GMP principle.
There are, however, some foods in which they are not allowed or where only one is
allowed, such as acacia gum in chocolate products. Four gums - guar, locust bean, xanthan
and tara - have a relatively wide permitted use, but are prohibited in dehydrated products
that are intended to rehydrate on ingestion (i.e. some fibre supplements). Karaya gum is
restricted to only eight categories of foods with maximum levels, as g/kg or g/l, given for
each category.
If an additive does not appear in the permitted list, or does not comply with the physical
or chemical parameters given in the specification, it cannot be used in foods within the EU
unless it has received official approval. The procedures for obtaining approval for new
additives are both complex and expensive. The substance has to be assessed for safety by
the Scientific Committee on Food (SCF) of the European Commission. Whilst the focus of
the assessment is on safety, there is also a requirement to provide a case of need. This has
326 Gums and Stabilisersfor the Food Industry 1I
to be based on technological need of benefit to the consumer, and applications with an

emphasis of commercial need are not looked upon favourably.
In general, applications for approval of new food additives are made to the European
Commission. However, there is written into the fiamework Directive on Food Additives
(89/107/EEC)2 a procedure whereby a member state can, on its own assessment, give
provisional authorisation for a food additive to be sold within its own territory. This
authorisation is subject to a number of conditions, including a maximum time limit of two
years; the requirement on the member state to officially monitor the use of the additive
and, in some cases, a special indication on the label of the product containing the additive.
The additive must be formally reviewed and authorised by the European Commission
before the two year period expires.
3 APPROVAL AS A FOOD INGREDIENT
From the 15 May 1997, all food ingredients that were not on the European market in a
significant quantity before that date have to be officially assessed and a p p r ~ v e d .This
~
requirement is as a result of EC Regulation No 258/97 on Novel Foods and Novel Food
Ingredients, and since it came into force in 1997 it has seriously inhibited product
innovation within the EU.
The regulation defines a novel food or ingredient as one that hasnot hitherto been used
for human consumption to a significant degree within the European Community and that
falls into one of the specified categories (Table 1). The categories are almost all embracing
in terms of ingredients and cover plant extracts, new molecular structures and ingredients
consisting of, or derived fiom, genetic modification.
This means that a new food hydrocolloid that is intended for use as a food ingredient
rather than an additive will most likely have to be subjected to official review and approval
if there is no evidence of a significant history of food use in the EU before May 1997.
Table 1 Categories of foods and ingredients defined in the European Regulation M258/97
foods and food ingredients containing or consisting of genetically modified

organisms within the meaning of Directive 90/220/EEC
foods and food ingredients produced fiom, but not containing, genetically modified
organisms
foods and food ingredients with a new or intentionally modified primary
molecular structure
foods and food ingredients consisting of or isolated fiom micro-organisms, fungi or
algae
foods and food ingredients consisting of or isolated fkom plants and food
ingredients isolated fiom animals, except for foods and food ingredients
obtained by traditional propagating or breeding practices and having a history
of safe food use
foods and food ingredients to which has been applied a production process not
currently used, where that process gives rise to significant changes in the
composition or structure of the foods or food ingredients which affect their
nutritional value, metabolism or level of undesirable substances
Hydrocolloidsand Health 327
To obtain approval for a novel food or ingredient, an application must be made to a

reviewing committee in the member state of intended first sale of the product. The
application must be accompanied by a very detailed dossier on the substance which must
include a comprehensive set of toxicity studies. A summary of the dossier has also to be
sent to the European Commission and the other 14 member states. In theory, the committee
in the reviewing country should deliver their initial assessment of the application within 90
days and there is a further 60 days allowed for comments fiom the other member states. In
practice, these time limits have not been achieved on any application filed in the last four
years.
Since the law came into force in May 1997, it has been found to have a number of
serious flaws. Although the assessment should only be carried out by one member state, it
has been common for the other member states to request the detailed dossier and
effectively carry out a full review. This means that, in reality, the assessment is being
carried out by a number of committees, often with conflicting opinions on the data. In the
case of any concerns or differences of opinion within the member states, the application is
referred to the Scientific Committee on Food (SCF) of the European Commission in
Brussels for their opinion. Unfortunately, there is no legal time limit imposed on the SCF
and in some cases this has resulted in more than a years delay in arriving at a final
decision.
Another serious problem with the law is that it places no obligation upon a member
state to accept ingredients that can be demonstrated to have been on the market in another
member state in significant quantities before 1997. There are a number of cases where
ingredients have been well established on the British market for a number of years before
1997 but are still not recognised by the French and German authorities.
4 HEALTHCLAIMS
With the increasing market for healthy foods, functional foods and food supplements,
health claims have become an important marketing tool.
The European Union directive on Food Labelling (2000/13/EC formerly 79/112/EEC)
prohibits the attribution to any foodstuff of the pro rty of preventing, treating or curing a
ge
human disease, or any reference to such properties. Human disease has been interpreted as
any ailment, injury or adverse condition, whether of body or mind. Any claim, expressed
or implied, that a product can prevent, treat or cure a human condition or disease is
regarded as a medicinal claim and the product has to be regulated as a medicine under the
requirements of directive 65/65/EEC.5
Claims for specific biological functions of recognised nutrients such as calcium aids in
the development of strong bones and teeth or vitamin Bg is important for the maintenance
of a healthy nervous system are regarded as nutrition claims and are normally permitted
under food law. Other accepted nutrition claims relate to the amount of nutrients or
components in a food such as low energy, sugar eee or a rich source of protein.
A third set of claims are those which state or imply that the consumption of a food, or
the component of a food, has a specific health benefit or avoids a specific aspect
detrimental to health. Whilst a health claim can be a nutrient h c t i o n claim such as
vitamin E protects the fat in body tissues fiom oxidation, there are other health claims
such as those made for the benefits of products containing beneficial gut bacteria (probiotic
products) which are not nutrient specific. Since the latter part of the 1990s, health claims
have become a more important aspect of food marketing and development, particularly in
the context of functional food concepts such as the antioxidant nutrients, probiotics and
prebiotics (the supply of selected nutrients to the gut bacteria). Legislation constantly lags
behind innovation and this has been the European situation with regard to health claims for
foods.
5 NUTRITION AND HEALTH CLAIMS IN THE EU
As early as 1980 the European Commission recognised that the area of food claims
required harmonisation and circulated the first proposal for a directive. After considerable
discussion, agreement could not be reached between the relatively few member states at
that time and the proposal was dropped. Over ten years later, in 1990, the Commission
attempted to revive the proposals and then in 1992 there was a working document as a
proposal for legislation on food claims fiom the Commission.6
Between June 1992 and June 1995 there were a number of revisions to the proposal as
agreement could not be reached by the member states on any of the major points. In mid
1995 it was announced within the Commission that it would no longer be working on a
directive for food claims. This meant that the individual national regulations would
continue to remain in force, resulting in a diversity of approaches across the EU. The
Commission then announced that it would try to introduce food claims principles into
existing EU legislative texts such as nutrition labelling and misleading advertising.
By early 1999, this approach had also not succeeded, and the Commission began
making further noises about developing EU legislation on health claims. The issue has
become even more important with the growth of the functional food market in Europe. In
January 2000, the Commission announced its intention to amend the Directive on Food
Labelling (79/112/EEC, now 2000/13/EC) to spec* conditions under which functional
claims and nutritional c l a i i may be made. The target date for adoption of the
amendments by the Commission was given as July 2001, and adoption by the European
Council and Parliament in July 2002.
The vacuum caused by the collapse of the EU harmonisation initiative on claims has, to
some extent, been filled by national attempts to regulate this area. This has resulted in
codes of practice on health c l a i i being developed and introduced in Sweden, Belgium,
the Netherlands and the United Kingdom. In addition, there has been a voluntary
agreement in Spain between the Ministry of Health and the Spanish Food and Drinks
Federation.
In France, the Conseil National de IAlimentation (National Dietary Council) has
produced a draft opinion and proposals on claims linking diet and health. This follows a
number of meetings of a working group fiom the end of 1997. During these meetings
evidence was considered fiom consumer associations, industry and scientists.
Whilst all six countries have taken slightly different approaches to obtain the same
objectives, they are all agreed on the major points. There appears to be a general consensus
that some provision should be made to allow health claims for foods. These would be in
addition to the nutrient h c t i o n claims (e.g. calcium for healthy bones and teeth) that are
already permitted, but would still exclude therapeutic claims.
By the beginning of 1999, there had been considerable progress made within five of the
member states. There had been a general acceptance over the previous two to three years
that the developing areas of hctional foods and dietary supplements would require a
significant re-evaluation of previously adopted positions on health claims. In addition, it
W ~ Sbeing conceded that legislation developed in the 1960s and 1970s was becoming
inappropriate in the light of recent scientific developments in nutrition. For example, the
concept of fiee radical damage to body tissues and the effect of antioxidant nutrients only
Hydrocolloids and Health 329
goes back to the late 1970s. When the Medicines Directive 65/65/EEC and the Food
Labelling Directive 79/112/EEC were developed, one of the primary distinctions between
foods and medicines was the claims for prevention, treatment or cure. At that time the
term prevention was interpreted in a prophylactic context such as prevention against
malaria. It is now believed that the intake of certain components fiom foods, such as
carotenoids and flavonoids, are the bodys protection against some forms of fkee radical
damage. Under present legislation, the preventative effects of these substances cannot be
claimed as such for the foods. Whilst progress towards a review and possible revision of
the term prevention is not as fast as the food industry would like, this area is being given
consideration at both European Commission and member state level. In the interim, a
number of countries have continued to make progress on internal policies for health claims.
In order to avoid conflict with the term prevention, a number of countries, including
individual EU member states and the Codex Alimentarius Commission, are currently
investigating the possibility of permitting claims for the reduction of a risk of disease
provided there is sound scientific substantiation that a food, or the quantity of a component
of a food, can make a direct contribution to a reduction in the risk of a specific disease or
condition.
Although it may be considered to come under the concept of prevention, and thus
illegal, a claim that a food or food component helps in reducing the risk of a disease is
different f+omthe claim that it prevents a disease. In the USA a limited number of disease-
risk reduction claim have been permitted by the Food and Drugs Administration, provided
all the conditions in the permission are met by the food. Examples of these claims are those
for:
Calcium and osteoporosis
Dietary lipids and cancer
Sodium and hypertension
Folate and neural tube defects
Fruit and vegetables and cancer
One of the problems facing the authorities is finding unequivocal evidence of the
relationship between the food or diet and the reduction in the risk of a disease and,
inevitably, more claims have been rejected by the authorities than have been accepted.
6 C U R R E STATUS OF HEALTH CLAIMS IN EU MEMBER STATES
As already stated, in the absence of an EU directive, a number of member states have

undertaken their own initiatives.
6.1 Belgium
In Belgium, there is a Code of Conduct on Health Claims based on work undertaken in the
mid-1990s by Fevia, the Belgian food industry federation? This code defines the various
forms of health claims that can be made and accepts that disease risk reduction claim can
be made if the effect of the food, or the effect of the foods nutrients or other components
have already been established on the basis of recognized and accepted scientific
knowledge, or can be proven to help reduce the risk of contracting a particular illness.
Such a claim must not imply prevention. The code states that the relationship between the
food and the disease or condition should not relate directly to the prevention of illness but
rather about extolling the importance of a healthy diet in maintaining better health. It is an
essential requirement of the Belgian code that the health claim must be scientifically
justifiable and there is a section dealing with the data required for scientific substantiation.
6.2 France
During a plenary session held in September 1997, the French Conseil nationd de
lalimentation (CNA - National Dietary Council) undertook the task of considering claims
linking diet and health, including functional claims.* The deliberations of the Council have
resulted in an opinion following a wide consultation which included consumer
associations, the food and drink federations and food scientists. The Council has also heard
evidence fiom people involved in advertising and from market researchers studying dietary
behaviour.
The opinion of the Council was that the prohibition on therapeutic claims must remain
but, provided certain principles were maintained, health claims would be acceptable.
Claims which could be permitted with suitable controls were:
A claim for the food product as aiding a reduction of a risk of a disease
A claim for a positive contribution to health when presented as having an
innuence on the modification of a physiological state or biological parameter
A functional nutritional claim which describes the positive role of the nutrient in
the normal hctioning of the body.
The Council felt that such claims would require prior authorisation before marketing
unless they appeared on an official approved list.
For new claims, and possibly for all claims, the validation of the scientific data which
forms the basis of the claims should be undertaken by independent organisations such as
the Human Nutrition Research Centres. The Council also proposed that a brief trial period
should be introduced for the recommended system before it became permanent. The
Council was concerned that the quality of the scientific justification should be of the
highest and that there should be more emphasis on punishing those making ffaudulent
claims.
Progress on the introduction of a formal procedure has been slower than anticipated,
partly due to the formation of the French Food Agency and a change in Committee
responsibilities.
6.3 Netherlands
After a considerable amount of discussion, a Code of Practice for assessing the scientific
evidence for health benefits fiom food and the health claims made for such foods and
drinks was drawn up on the initiative of the Voedingscentrum (the Netherlands Nutrition
Centre). This code, which was completed in April 1998, contains a procedure for the
assessment of the scientific evidence for health benefits stated in health
The code was developed under the auspices of the Dutch Food and Drink Industry
Associations, the Consumers Associations, the Association of Dutch Advertisers, Retail
Associations and the Netherlands Nutrition Centre. The code was presented to
representatives of the Ministry of Public Health, Welfare and Sport, the Ministry of
Agriculture, Nature Management and Fisheries, and the Dutch Association of Branded
Product Manufacturers at a meeting on 21 April 1998. The government representatives
welcomed the procedures laid down in the code and declared official support for it.
However, this code of practice has not been officially adopted by the government.
Hydrocolloids and Health 33 1
Under Dutch law, a health claim is defined as A direct, indirect or implied claim that a
food carries special qualities which improve or maintain the users health. This is
interpreted as a claim made by any promotional means and by any means of
communication. Products imported into the Netherlands have to meet the same criteria.
The Dutch code of practice does not cover medical claims nor does it apply to nutrient
content claims or product safety, as these aspects are already covered by law.
The code requires that health benefit claims are assessed by an independent panel of
experts appointed by the Netherlands Nutrition Centre. This assessment covers the
scientific content of the supporting evidence for the claim. There is a primary requirement
that the evidence must be based on relevant data from human subjects. It must be
demonstrated that any effectiveness of the substance or substances determined fiom
research is not reduced when included in a commercial product as presented to the
consumer. The data used must be relevant to normal use of the product by the target
population and the health benefit claimed must also be relevant to the target population. It
is also necessary to show that the scientific evidence is reproducible. The code also has a
requirement that the claimed health benefit does not clash with any dietary guidelines laid
down by any of the authoritative bodies in the country.
6.4 Spain
An agreement on health claims for foods was signed on 20 March 1998 between the
Ministerio de Sonidad y Consumo (Spanish Ministry of Health) and the Federacibn de
Industrias de Alimentacibn y Bebidas (FIAB - the Spanish Federation of Food and Drinks
Manufacturers).lo This agreement was voluntary and clarifies the situation relating to
health claims within the legislation covering the labelling and advertising of foods. Within
the agreement, health claims are defined as any claim relating to:
0 the fhction of one or more nutrients in the human body
the effects of the consumption of one or more food products on human health
the consumption of certain foods as part of a healthy diet
The primary purpose of the agreement is to lay down the guidelines which permit health
claims in the categories listed above to legally be made and without S i n g i n g the law.
The agreement is intended to apply to the labelling and advertising of all foods and
drinks with the exception of products which come under the classification of foods for
particular nutritional uses (PARNUTS) and mineral waters. Nutrition content claims or
claims such as contains vitamin A or contains iron and calcium are also excluded from
the scope of the agreement as they are already controlled by specific legislation.
There is a general requirement that all health claims must be truthful and be able to be
clearly substantiated by scientific evidence. Where a claim is made for the beneficial
properties of an ingredient rather than the food itself, the nutrient or component which is
claimed to have the beneficial properties must be present in the food in sufficient quantities
to produce the claimed effect. The same applies to claims for the absence of a specific
component or nutrient (e.g. saturated fat).
There is a requirement that whenever a health claim is made in the labelling or in
promotional materials for a food it should be accompanied by a statement about the
importance of maintaining a healthy and balanced diet. Health claims must also, where
appropriate, comply with the rules laid down for the nutritional labelling of foods. It is not
permitted to claim a particular beneficial effect for a specific brand when all foods in the
same category have the same properties.
332 Gums and Stabilisers for the Food Industry 1 1
The agreement requires that, as a general principle, all health claims must avoid
categorical statements guaranteeing the beneficial effects claimed for the food. Instead,
words such as aids, facilitates and eases should be used to indicate that the
consumption of the food helps to achieve the beneficial effects but that other factors such
as a healthy diet, lifestyle and physical exercise must also be taken into consideration.
As part of the agreement, a joint committee composed of officials fiom the ministry and
representatives of the food industry has been set up to review proposed claims which can
be submitted on a voluntary basis before marketing. The committee is required to give an
opinion within ten days of submission. Failure to respond to an application within the
prescribed time means a tacit favourable opinion on the use of the claim.
6.6 Sweden
Sweden was the first of the six EU countries developing guidelines on health claims to
agree and adopt a self-regulating procedure. The first programme to regulate health claims
was agreed by the food industry and came into effect in August 1990. This was monitored
for three years until July 1993 and revised self regulating rules were agreed in August 1996
and came into effect fiom 1 January 1997. This revised programme was developed and
agreed by representatives of all sectors of the Swedish food industry including the
Federation of Swedish Food Industries, the Swedish Food Retail Association, the
Federation of Swedish Farmers and the Grocery Manufacturers of Sweden.
In the explanatory document, a health claim is defined as an assessment of the positive
health effects of a foodstuff, i.e. a claim that the nutritional composition of the product can
be connected with prophylactic effects or the reduced risk of a diet related disease. This
definition is qualified by the requirement that the health claim must be based on the
importance of the product in a balanced diet and must be in line with the official Swedish
dietary recommendations. Table 2 gives diet related risk factors which may form the basis
of health claims in Sweden.
A health claim must consist of two parts. The first must provide information on the diet
and the health relationship of the food. The second part is to consist of information on the
composition of the product. Two examples given as acceptable health claims are:
Part 1. Iron deficiency is common among women but can be prevented by good
dietary habits.
Part 2. Product X is an important source of the type of iron that is readily
absorbed by the body.
Part 1. Omega-3 fatty acids have a positive effect on blood lipid and can
therefore help protect against cardiovascular disease.
Part 2. Fish product X is rich in omega-3 fatty acids.
Six other examples of approved health claims are given in the document. These include
claims for constipation and dietary fibre, osteoporosis and calcium rich foods,
atherosclerosis and low saturated fatty acids and / or low salt content, and dental caries and
sugar fiee products. With each approved claim there is an example of an unapproved
statement or reference regarding the product composition and effect. For example it is
permissible for the second part of a claim relating to cholesterol and cardiovascular disease
to say Brand X contains a low amount of saturated fat and total fat but it is not
permissible to say Brand X provides excellent protection against cardiovascular disease
through its low content of saturated fatty acids or Brand X will help you reduce your
blood cholesterol levels.
Hydrocolloidr and Health 333
Table 2 Diet related riskfactors which may form the basis of health claims in Sweden
1. Obesity 5. Constipation
2. Cholesterol level in blood 6. Osteoporosis
3. Blood pressure 7. Dentalcaries
4. Atherosclerosis 8. Iron deficiency
Provision has been made as part of the initiative for the organisations and companies to
attend seminars in which the Swedish Nutrition Foundation will be involved. The seminars
will be for those involved in the marketing of food to understand the relationship between
diet and health and the legal issues involved in making claims, particularly health claims.
The organisations who are signatories to the rules are required to continuously provide
relevant information on the above issues to companies in the food industry. Each
organisation has appointed a contact who is responsible for maintaining communications
with the authorities and the Swedish Nutrition Foundation and to disseminate the
information to their member companies.
6.7 United Kingdom
In December 1996 the Food Advisory Committee (FAC) completed its review of the
British market for hctional foods and the control of health claims. As art of this review,
the FAC published draft guidelines on health claims for foodstufs."These guidelines
were intended to set out the conditions which food manufacturers or retailers were required
to follow when malung health claims for their products. The scope of these guidelines was
that they applied to all foods and drinks, including food supplements, but with the
exception of those products controlled under the EC directive on foods for particular
nutritional uses (PARNUTS).
A health claim is defined in the draft as any statement, suggestion or implication in food
labelling or advertising that a food is beneficial to health. Nutrient content claims and
medical claims are excluded fiom this definition. Health claims could be subdivided into
three categories:
claims that refer to possible disease risk factors (e.g. can help lower blood
cholesterol)
nutrient h c t i o n claims (e.g. calcium is needed to build strong bones and teeth)
recommended dietary practice (e.g. eat more oily fish for a healthy life style)
The draft guidelines devoted a section to the principles underlying health claims and
one to the scientific substantiation of claims.
Following a period of consultation, which in general met with a positive response, the
Joint Health Claims Initiative (JHCI) was established in June 1997. This was a joint
venture between consumer organisations, enforcement authorities and industry bodies with
the objective of establishing a code of practice for the use of health claims for foods.
The code of practice which was developed by the JHCI further separates health claims
into generic claims and new claims.13 Generic health claims are defined as those based on
well-established and generally accepted knowledge, evidence in scientific literature and /
or recommendations fiom national or international public health bodies. A regularly
reviewed and updated list of generic health claims was to be maintained as part of the
administration of the code. As part of the preparation for such a list, the Medical Research
Councils Human Nutrition Research Group was asked in 1999 to consider the scientific
evidence substantiating 14 putative generic claims. Their report, which was published in
early 2000, could not find good evidence for some of the claims.14
A new health claim must be based on scientific evidence as applied to either existing or
new foods. The scientific evidence must be substantiated in accordance with detailed rules
laid down in the code. Companies wishing to market a food with a new health claim must
show that the claim is likely to be true and that the evidence in its support outweighs any
opposing evidence or opinions. Companies are also considered to be responsible for
ensuring that the validation of a new health claim is carried out alongside all other checks
necessary when assessing the suitability of the food for marketing.
There are a number of criteria that must be met when using a health claim. It is essential
to demonstrate that the food, or its components, will cause or contribute to a significant
and positive physiological benefit when consumed by the target population as part of their
normal diet. The effect must be achieved by the consumption of a reasonable amount of
food on a regular basis or by the food making a reasonable contribution to the diet. The
claimed effect must be maintained over a reasonable period of time and it should not be a
short term response to which the body adjusts, unless the claim is relevant for only a short
or medium term benefit. An example of the latter would be the use of folic acid just before
and during the early stages of pregnancy in the case of neural tube defects.
The substantiating evidence is required to demonstrate the minimum or maximum
amount and the frequency of consumption of the food required to achieve the effect.
Alternatively, it must be shown that the food can provide a reasonable dietary contribution
to that required to achieve the effect. The substantiation should also explain how the effect
is brought about. The code accepts that the exact biological mechanism need not be fully
understood or explicable.
The scientific data needed to support a claim is described in some detail in the code.
The claim must be based on a systematic review of all the available evidence, including
published scientific literature, relating to the validity of the health claim. The conclusions
drawn fiom this review should be based on the totality of the evidence and not just on the
data that supports the claim.
The conclusions should be based on human studies which are the most
methodologically sound available or other human evidence and not just on biochemical,
cellular or animal studies. Whilst, ideally, the conclusions should be based on experimental
studies in humans, it is accepted that observational studies may be acceptable in some
circumstances. The research should assess the effect of foods on the health status of human
subjects. Where obtaining full clinical evidence is dficult, lengthy or expensive, it would
be acceptable under the code to provide evidence of the effect of the foods on bio-markers,
provided there is a strong correlation between the bio-markers and a relevant indicator of
well being, disease or risk factor.
Companies wishing to make a health claim should submit the details of the clam and
the substantiating scientific evidence to the Code Administration Body (CAB). The CAI3
will seek the opinion of a panel of experts which is described in the code as the Expert
Authority. The CAB consists of a Council and a Secretariat. The Council is conceived as a
tripartite body, with representation fiom enforcement, consumer and industry, and has the
responsibility for monitoring the code, for the selection of members of the Expert
Authority and for considering any amendments to the code in the light of experience. The
Secretariat services both the requirements of the Council and the Expert Authority and acts
as the primary point of contact for all interests. The code came into effect in December
2000.
7 SUBSTANTIATION OF HEALTH CLAIMS
Common to all the European national initiatives on health claims is the need for good
scientific substantiation of the claim All the initiatives require that the scientific evidence
is of good quality and it is either expressed or implied that the science should sustain peer-
review. The British code of practice details the type of data that should be presented. Both
the British and the Dutch codes have a primary requirement that the evidence must be
based on relevant data fiom humans. There is also a need to demonstrate that the food,
when consumed in reasonable quantities, can achieve or contribute to the claimed
beneficial effect. The Dutch code has a requirement that it must also be demonstrated that
the effectiveness of any substance or substances as determined by research is not reduced
when incorporated into a commercial product. When considered collectively, it is probable
that the requirements for scientific substantiation will have a limiting effect on the number
of non-generic health claims that will receive approval.
8 THEFUTURE
The development of fimctional foods in Europe has been increasing rapidly since 1997
with a number of multinational food companies taking an active part in the growth of this
sector. In addition, the food supplements market is still showing growth, particularly in
countries on mainland Europe. Paramount to the growth of both areas is the ability to make
claims in the marketing of the products. As discussed, this has already been appreciated
and addressed by six countries in the EU who have some form of agreement or code of
practice acceptable to the industry, the enforcement authorities and consumer interests. The
other countries in the EU, such as Germany, who have not already embarked on the
establishment of procedures for handling health claims, are in most cases in early internal
discussions on the subject.
The inconsistencies between the existing codes could lead to barriers to trade in the
products on a pan-European basis. For example, the Belgian code of practice requires only
that a dossier containing the scientific evidence be retained for examination on request,
whereas in the Netherlands and United Kingdom an independent review of the data will be
the norm. There has been so much national activity in the area of health claims that the
European Commission is forced to re-activate its work on the proposed directive. It may
now be even more difficult to achieve h o n i s a t i o n as a number of the member states will
fight to retain their recently developed national positions.
A further, and more complex aspect, is that some products are being developed in such
a way as to blur the border between foods and medicines.
The Medicines Directive 65/65/EEC contains the defmition of a medicine which is in
two parts. The first part relates to products which are intended to prevent, cure or treat a
human condition or disease. The prohibition of claims or implications of prevention,
treatment or cure has been carried over into European, and hence national, food law.
What is more difficult to interpret is the second part of the definition which states: Any
substance or combination of substances which may be administered to human beings or
animals with a view to making a medicinal diagnosis or to restoring, correcting or
m0-g physiological functions in human beings or in animals is likewise considered a
medicinal product
The appearance on the market of fat spreads (e.g. margarine) containing increased
levels of phytosterol esters for the reduction of cholesterol and products aimed at women
containing high levels of added isoflavones is making it more difficult to clearly identrfL
the legal status of a product. In the past the rule of thumb has been that products which
maintain the healthy functions of the bodys organs, tissues or systems are regarded as
foods, whilst those that are developed or promoted to induce changes to the normal
physiological functions of the body may be medicinal. Deliberately inducing diuresis, for
example with a product designed for slimmers, could be considered to be mod- the
normal functions of the bodys renal system. In such a case the classification would pivot
on the intent of the product, as diuresis can also be a side effect fkom the consumption of
foods or drinks containing substances such as caffeine or alcohol.
The positive developments on health claims by some of the member states of the EU
and the discussions on hctional foods that have been initiated at European Commission
and national government level should hopehlly lead to a modernisation of the legislation
but, unfortunately, not in the short term.
References
1. European Parliament and Council Directive 95/2/EC on Food Additives other than
Colours andsweeteners. O.J. 0fE.C. L61/1 of 18 March 1995.
2. European Council Directive 89/107/EEC on Food Additivesfor use in foods for

human consumption. O.J. of E.C. L40/27 of 2 1 September 1998.
3. European Parliament and Council Regulation (EC) No258/97 concerning novel

foods and novel food ingredients. O.J. of E.C. L43/1 of 27 January 1997.
4. European Council Directive 79/112/EEC relating to the labelling, presentation

and advertising of foodstufs. O.J. of E.C. L33/1 of 8 February 1979 (replaced
by Directive 2000/13/EC O.J. of E.C. L109/29 of 6 May 2000).
5. European Council Directive 65165fEEC relating to proprietary medicinal

products. O.J. of E.C. 22 February 1965.
6. European Commission. Draft proposal for a Council Directive on the use of

claims relating to foodstufs. SPN62 ORG: Fr of 26 June 1992.
7. Federatie Voedingsindustrie/ Federation de IIndustrie Alimentaire (Belgium).

Health Claims Code of Conduct, draft, 21 Oct 1998.
8. Conseil National de 1Alimentation (France). Claims linking Diet and Health.

Draft opinion version N 8, 1998.
9. Voedingscentrum(Netherlands), Code of practice assessing the scientifjc

evidence for health benefits stated in health claims on food and drinkproducts,
April 1998.
1 0. Joint Ministerio de Sonidad y Consumo (Ministry of Health, Spain) and

Federacibn de Industrias de Alimentacih y Bebidas (Spanish Federation of
Food and Drink Manufacturers) agreement on health claims on foods of 20

March 1998.
11. Federation of Swedish Food Industries et al. Health Claims in the Labelling and
Marketing of Food Products. Food Industrys Rules (Self Regulating
Programme) Revised programme of 28 Aug 1996.
12. Ministry of Agriculture, Fisheries and Food (United Kingdom), Food Advisory
Co&ttee Review of Functional Foods and Health Claims, 16 Dec 1996.
13. Joint Health Claims Initiative (United Kingdom). Code of Practice on Health Claims
on Foods. Final text, 9 Nov 1998.
14. Medical Research Council (2000) Validationof generic health claims. Briefing
paper by MRC Human Nutrition Research, Cambridge, England.
MANAGING HEALTHY LEVELS OF BLOOD GLUCOSE AND CHOLESTEROL
WITH KONJAC FLOUR
Guy Crosby
Opta Food Ingredients, Inc., 25 Wiggins Avenue, Bedford, Massachusetts, 01730, USA
1 INTRODUCTION
Konjac flour is a partially acetylated glucomannan polysaccharide used as an ingredient in

food both for its unique water-binding properties and well-documented nutritional benefits.
In the Far East it has been prepared from the tubers of the Amorphophallus konjac plant for
well over a thousand years. Konjac flour, which has been freshly prepared from tubers that
have been grown for three years, has a molecular weight in the range of 1-2 million
Daltons. The high molecular weight results in very high viscosity at very low
concentrations. The viscosity of a dispersion made with 1% commercial konjac flour can
vary from 30,000-100,000 mPa.s, depending on the preparation.
In addition to its function as a powerful water-binding agent in food, konjac flour has
been proven in numerous published clinical studies to significant1 lower glucose and
cholesterol levels in the blood, when consumed in modest amounts?>'Studies have shown
that the reduction in blood glucose levels caused by various hydrocolloids is proportional
to their vi~cosity.~Thus, increasing the viscosity of the intestinal fluid limits mass transport
of glucose to the bloodstream.
The mechanism of action of konjac flour in lowering blood cholesterol levels is less
well understood, but it is generally assumed that the reduction is related to increased bile
acid ~ecretion,~ Konjac flour therefore appears to have significant value when formulated
in foods for diabetics, or for those individuals who want to maintain healthy levels of
glucose or cholesterol in the blood.
Because of konjac flour's high viscosity, it has largely been formulated in solid foods such
as meat, surimi, baked goods, and nutrition bars, with very little effort to develop liquid
formulations. The requirement for high viscosity to modulate blood glucose and cholesterol
levels appears to be at odds with the requirements for formulating nutritional beverages
which will appeal to consumers. New patent-pending technology developed at Opta offers
a solution to the dilema. Hvdrolvzed corn starch (maltodextrins) with specific molecular
weight
- ranges
- have been shown to dramatically reduce the viscositv of koniac sols. The
effect is dependent on the concentration and molecular weight of the maltodextrin.
The graph in Figure 1 illustrates the effect of maltodextrin concentration on the viscosity
of konjac dispersions in water. The molecular weight of the maltodextrin, expressed as
Dextrose Equivalents (DE), was held constant. Konjac flour was added to increasing levels
of DE 5 maltodextrin in solution. The final concentration of konjac flour was 2% by
weight. Viscosity was determined using a Brookfield type viscometer at 2 rpm. The order
of addition of ingredients had no effect on the final viscosity.
400 I 3
0
0.000 5.000 10.000 15.000 20.000
O
h Maltodextrin
Figure 1 Effect of Maltodextrin on Viscosity of Konjac Flour Dispersions
The graph in figure 2 shows the effect of the molecular weight of the maltodextrins,
expressed as DE, on the viscosity of aqueous dispersions of konjac flour. The
maltodextrins ranged from DE values of 18 to 1, corresponding to average degrees of
polymerization (DP) from about 6 to greater than 50. The lower DE maltodextrins (DE 1
to 10) had the greatest effect on reducing the viscosity of konjac flour dispersions. These
results are most surprising because it is well documented in the literature that common corn
starch is synergistic with konjac flour in building viscosity.
Microscopic examination of the konjac flour-maltodextrin dispersion in water indicates
the formation of a phase-separated biopolymer system, in which the konjac phase is
dispersed as fine droplets within a continuous maltodextrin phase! Thus the entire system
exhibits a viscosity similar to a solution of maltodextrin. Depending on the method of
preparation the konjac flour-maltodextrin dispersion may be stable from 24 hours to an
indefinite period of time before separation occurs.
As an example of this effect, a milk-based beverage was prepared with 1% konjac flour
and 14.5% DE 5 maltodextrin (DP of 22). The viscosity of the beverage was 190 mPa.s.
The same beverage made without the maltodextrin had a viscosity of more than 8700
mPa.s.
The rapid hydrolysis of maltodextrin to glucose by enzymes present in the GI tract
offers a mechanism for regenerating the beneficial viscosity of konjac flour following
ingestion. This effect has been confirmed with in vitro experiments. Digestion with
amylase. as would occur in the small intestine. rapidlv restores the viscositv of a mixture of
1% konjac flour and 15% maltodextrin within one hour. Results are shown in Table 1.
1000 I I
+3% RS + 10% MD
800
'
0
*)
%
600
Y
E
5
0
400
0
u)
' 200
.I
0
0 5 10 15 20
Maltodextrin DE
Figure 2 Effect of Maltodextrin DE on Viscosity of Konjac Flour Dispersions
For this experiment a dispersion was prepared by adding 8 g of konjac flour to 392 g of
0.02 M phosphate buffer solution (pH 6.9) at room temperature for 30 mins. A 400 g
solution of maltodextrin was prepared by dissolving 120 g of DE 5 maltodextrin in 280 g of
buffer by heating to 70" C under agitation until the solution turned clear. The maltodextrin
solution was then combined with the previously prepared dispersion of konjac flour using
an overhead mixer. The combined dispersion was allowed to cool to 37" C in a water bath.
The dispersion was divided into two 400 g portions. To one of the two portions was added
4 mL of porcine a-amylase solution, containing 3000 Units of activity per mL. To the other
solution was added 4 mL of phosphate buffer as a control. The two samples were incubated
in a 37" C water bath for 24 hrs. Viscosity was measured at time 0 (prior to enzyme
addition) and at 1 , 2 and 24 hrs using a Brookfield type viscometer at 60 rpm.
Table 1 Viscosity Recovery Upon Amylase Digestion in a Konjac Flour-Maltodextrin

Mixture
Time Viscosity (cP)

(hours) No Enzyme With Enzyme
0 58 58
1 73 4,150
2 80 5,140
24 88 6,8 17
CONCLUSION
A technology is now available to formulate low viscosity nutritional beverages with

physiologically significant levels of konjac flour capable of regenerating the beneficial
high viscosity upon ingestion.
Hydrocolloidsand Health 34 1
References
1 K. Nishinari, P.A. Williams, and G.O. Phillips, Food Hydrocolloids. 1992, 6, 199.
2 V. Vuksan, D.J.A. Jenkins, P. Spadafora, J.L. Sievenpiper, R. Owen, E. Vidgen, F.
Brighenti, R. Josse, L.A. Leiter, and C. Bruce-Thompson, Diabetes Care, 1999,22,
913.
3 K. Doi, Eur. J. of Clin. Nutr., 1995,49, S190.
4 A. Arvil and L. Bodin, Am. J. Clin. Nub-., 1995,61,585.
5 C.A. Edwards, N.A. Blackburn, L. Craigen, P. Davison, J. Tomlin, K. Sugden, I.T.
Johnson, and N.W. Read, Am. J. Clin. Nutr., 1987,46,72.
6 S . Kasapis, E.R. Morris, I.T. Norton and M.J. Gidley, Carbohydrate Polymers, 1993,
21,249.
THE ROLE OF POLYSACCHARIDES IN THE GASTRIC PROCESSING OF FOODS
A.Fillery-Travis', L.Marciani2, P.A.Gowland*, and R.C.Spiller3
'Institute of Food Research, Norwich Research Park, Norwich NR4 7UA

2MagneticResonance Centre, School of Physics and Astronomy, University of
Nottingham.
3
Gastroenterology, Queens Medical Centre, University of Nottingham.
1 INTRODUCTION
Within Europe and Northern America mortality and morbidity fiom diet-related chronic
disease exceeds that fiom other causes. Until recently the response fiom the health sector
has been a pharmaceutical strategy. This is highly expensive and concentrates on the
relief fiom disease. A nutritional approach offers a preventative approach concentrating
on the maintenance of health. The consumer awareness of this debate is growing rapidly
and food manufacturers are responding with fortified and hnctional products.
Unfortunately it is rarely possible to define the proportion of the (micro-) nutrient or
fortificant available for absorption by the body due to the paucity of information available
on the gastrointestinal processing of foods and the impact on (micro-) nutrient release.
The digestion of real diets has received little attention in the past, mainly because of a
lack of direct methods to investigate it. In vivo studies are invasive and difficult to
control whereas in vitro methods have been over simplistic and rarely validated against
an in vivo model. Magnetic Resonance Imaging and, in particular, the fast sequencing
methods of echo-planar magnetic resonance imaging (EPI) provides a non-invasive
method for investigation of gastric function and the intragastric processing of food in real
time. We review here the results of two major studies with human volunteers in which
the influence of polysaccharides on the gastric processing of simple meals was
investigated. The influence of meal viscosity on meal hydration, mixing, gastric
emptying and particle break-up is reported together with the results of parallel studies on
the sense of satiety and fullness experienced by the volunteers during the studies.
2 METHODS AND RESULTS
2.1 Subjects and test meals.
For each of the studies between eight and sixteen volunteers were recruited. They
were healthy, with no history of gastrointestinal disorders and within *25% of ideal BMI.
They were asked to fast overnight and the experiments were performed more than 1 day
and less than 7 days apart. A Latin square randomised study design was used in all cases.
The subjects were scanned once every 20 minutes over a period of several hours after
ingestion of the meal. During scanning the subjects lay supine in the scanning bed but
between scans they sat upright. All procedures were approved by the University Medical
School ethics Committee and the volunteers gave informed written consent before the
experiments.
2.2 Scanning
All experiments were performed with a whole-body OST purpose-built EPI scanner
equipped with actively shielded gradient coils and a 50 cm diameter linear, bird, cage,
RF coil. Single shot MBEST EPI images were acquired with an in-built resolution of
3.5mm x 2.5mm and a slice thickness of 1 cm. Each image was acquired in 130ms using
a 128 x 128 matrix with an effective echo time of 40ms. Images were processed using in
house software packages and Analyze (Biomedical Imaging Resource, Mayo Foundation,
Rochester, MN)'.
3.0 Study A. Influence of meal viscosity on gastric processing of a model meal and
self-assessed satiety.
3.1 ExperirnentaZ Design: Each of 12 volunteers ingested four meals of 500mls total
volume: low viscosity non-nutrient (LVC), high viscosity non-nutrient (HVC), low
viscosity nutrient (LVN) and high viscosity nutrient (HVN). The non-nutrient meals
consisted of a locust bean gum (LBG) solution with 16g of dextrose to equalise
osmolality with the nutrient meals. The nutrient meals contained the LBG solution and an
olive oil emulsion stabilised with sorbitan monostearate, Crillet 3 VEG and Vykamol
83G, (Croda Food Services, Oldham, UK) such that the total energy of the meal was
1,350k.J of which 37% was supplied as lipid. Concentrated banana flavouring was used to
obtain a palatable solution. Each subject was trained to self-assess their satiety by
questionnaire and every 12 minutes after ingestion of the meal they reported on their
feelings of fullness, hunger and appetite.
The LBG solutions were prepared by adding the powdered gum to hot water which
was then maintained at 90C for 1 hours before cooling to 37C . The steady-state
viscosities of the prepared solutions were measured in a Bohlin controlled stress
rheometer within a double gap geometry. The LVC and LVN meals were prepared with
0.25%w/w LBG (qo=O.O16Pa.s) and the HVC and HVN meals with a 1.5%w/w LBG
(qo=29.5Pa.s) solution. The nuclear magnetic relaxation of the LBG solutions depends
principally on the LBG concentration therefore it was possible to calibrate the dilution of
the LBG in vitro with the inverse of the T2 relaxation time (transverse relaxation rate).
This was validated in vivo by aspirating samples of the gastric contents (after
measurement of the T2) and measuring the viscosity of the
3.2 Results: All meals provided good intrinsic contrast with the surrounding tissue and
it was possible to monitor the dilution and gastric emptying of the meal in real time. It
was found that the high viscosity meals were diluted rapidly by the stomach thus there
was no significant effect of viscosity on gastric emptying of the nutrient-containing
meals and the form of the emptying was linear. For the non-nutrient meals the emptying
was exponential in form and the high viscosity meal was slower to empty fiom the
stomach. The presence of nutrients in the meal triggered an increase in secretion volume
by the stomach (95ml&18 for LVC cf 101mle7 for LVN) but this was not as great as
that produced by an increase in viscosity (135mls*lO for HVC cf. 157mlst12 for HVN).
There was no change in the frequency or velocity of the antral contractions (the main
muscular movement by the stomach for mixing and emptying of the meal) with meal
type, which suggests that the force of contraction had increased within increasing
viscosity.
An example of the hydration map that was obtained as a function of time for a high
viscosity meal is shown in figure 1.
Figure 1: A colour-coded map acquired after ingestion of a high viscosity meal. The
conventional EPI image is shown on the left as an anatomic road map (L, left: R, right).
The T2 maps of the stomach contents are shown in the three pictures on the right
progressing in time after meal ingestion. The calibration is shown on the far right as a
colour bar indicating degree of dilution of the meal.
It can be seen that contrary to previous suggestions high viscosity meals are not mixed
homogeneously within the stomach but dilution occurs fiom the outside of the bolus of
the meal, and penetrates to the centre, with time. The more dilute portion of the meal
then empties preferentially fiom the stomach. This has implications not only for meal
components but also for oral drug delivery systems and in vitro models of digestion.
Subjects who ingested a high viscosity meal also experienced an increased sense of
fullness for the same gastric volumes (figure 2). We suggest this is due to the response of
the stretch receptors in the antrum (lower portion of stomach adjunct to the pylorus).
4.0 Study B. The effect of particulates within a meal.
Our common diet includes solid as well as liquid components. Initial communition
occurs through the action of chewing within the mouth and the resulting size distribution
of the particulates can vary between meals and between individuals. Ultrasound and MRI
imaging have demonstrated that the maximum shear is experienced within the antrum
where both forward and backward flow has been observed. There is a suggestion that at
this point selective emptying occurs of the smaller particles leaving the larger particulates
within the stomach. Little quantitative information is available due to the complexity of
flow pattern and the difficulty of tracking particulates with these techniques. We have
developed a series of agar gel beads of varying fracture strength as a probe of intragastrk
forces.
LVN
Fullness (scores 7
9min)
6
1
:i
3
21
10
I
20 30
I I
40
1
50
Volume (ml)
9
8
1 =Lvc
Fullness (scores 7
min) 6
5
4
3
I I I 1
10 20 30 40 50
Volume (ml)
Figure 2: Self-assessment scores for the sense of fullness as a function of total gastric
volumefor thefour LBG meals.
4.1 Preparation and characterisation of the beah. The beads were of sufficiently
small diameter (1.27cm) that they could be swallowed whole of the volunteers. Trace
amounts of barium sulphate (e-Z-H powder) and ferrous fbmerate (fersamal syrup;
Forley, Dublin, Ireland) were added to increase the density and MRI contrast,
respectively. The agar gel was autoclaved at 121C for 1 hour before injection into an
aluminium mould, which was then left to cool slowly overnight. The rheological
behaviour of the agar gel was assessed using a Bohlin VO rheometer with a cone and
plate geometry. The contrast agents were found to have no effect on the gel properties of
the agar. Compression tests were carried out using a TAXT2 texture analyser (Stable
Microsystem) with a 5kg load cell. The force-displacement data were recorded for eight
batches of beads for each of six different gel concentrations from 0.75% and 3% ( d w t ) .
The maximum force corresponding to failure of the bead and the initial slope of the
force-displacement curve, which is proportional to the stiffness of the bead were
calculated fiom this data.
4.2 Stu& design and results. Each of sixteen subjects ingested (without chewing)
fifteen of the beads from one of the seven strength batches as part of a 50Oml test meal.
The meals were of the same composition as the LVN and HVN meals prepared in Study
A. Multislice data sets were acquired every 15 minutes and the number of intact beads
remaining in the stomach was calculated. The half-residence time (RTs) of the intact
beads in the stomach was averaged for each bead strength. The results for both meal
types are shown in Figure 3.
80
70
W
Q 40
30
20
10
0
0.15 0.28 0.4 0.53 0.65 0.78 0.9
Beads fracture force (N)
Figure 3: Half-residence time (RTvJ of intact agar gel beads in the gastric lumen for
each of the seven bead breaMownforces and for both L W (open bars) and HVN @led
bars) meals. *P<O.002 vs. each of the strengths of the lower five beads.
The average decline in the number of intact beads within the stomach with time was
exponential in shape for both meals (?=0.79). With the LVN meal the RTs of the intact
beads in the gastric lumen doubled when the fiacture threshold of the beads was >0.65N.
For the HVN meal an increase in RTs of -40% was observed at bead fiacture thresholds
of >0.53Nbut this was not significant (P<0.45). There was, however, an increase in the
emptying of the total gastric contents with increasing bead strength (LVN P<0.03: HVN
Pc0.0 1) whereas no significant effect on antral contraction frequency or propagation
speed was observed. Thus softer beads were rapidly broken up with the stomach and
emptied at the same rate as a liquid meal. The harder beads persisted within the stomach
before breaking up producing longer gastric emptying times.
The subjects were asked to assess their sense of fullness, appetite and hunger as
before and the results are given below in Figure 4.
The lowest bead strength gave a linear response to gastric volume whereas the highest
bead strength deviated fiom linearity significantly showing a higher sense of bllness for
a given gastric volume. For the HVN meal the sense of fdlness was also non-linear with
respect to gastric volume but there was no significant difference between the extremes of
bead strength. We suggest that the hard beads increased satiety by triggering a stretch
mechanism that involves the antrum.
T

,
,
,AI
3 7 -
20
5:
W
36-
U
n
E 5 -
4-
-- 1
3 I I I 7-
I - - -
100 200 300 400 500 600

Volume (ml)
Figure 4: Self-assessment scores for the sense of fullness plotted against total gastric
volumes grouped by corresponding time points for the lowest. (0.15N)and highest.
(0.9N) bead strengthfor the L VN meal.
5 DISSCUSSION
We have shown EPI imaging to be an effective tool for the investigation of gastric
processing of foods within the stomach. Our studies have identified the gastric response
to meals of varying viscosity with and without particulates. In particular the stomach
responded to ingestion of a viscous fluid by rapid secretion resulting in a significant
reduction in viscosity. Thus gastric emptying times of low and high viscosity meals were
comparable and little effect of the initial viscosity of the meal was found. The non-
uniform dilution of viscous meals was also of note. In particular the dilution was seen to
start at the edge of the meal bolus adjacent to the stomach wall and gradually penetrate
the bolus with time. This pattern of dilution was unexpected and does not agree with the
previous assumed mode of homogeneous dilution and mixing of the bolus within the
stomach. The diluted meal fiaction was then preferentially emptied fiom the stomach.
This result has significant impact on the design of in vitro models of digestion. A viscous
meal will only experience the low pH values of the gastric secretion for a relatively short
period before emptying and this will have implications for the predicted nutrient release
(and chemical breakdown) fiom the food structure.
The presence of particulates within the meal has been shown to influence both the
gastric emptying of the meal and the sense of fullness experienced by the subject. The
analysis of the beads residence time has allowed a semi-quantitative measure to be made
of the maximum grinding force exerted by the antrum5.
So far the studies have concentrated on relatively simple meals to achieve an
understanding of the relative importance of factors such as viscosity and particulate
nature. The results so far have already contributed to our understanding of gastric
processing of foods and the breakup of food structure during digestion.
References.
1. M.K.Stehling, R.Turner and P-Mansfield, Science, 1991,254,43
2. L.Marciani, P.A.Gowland, R.C.Spiller, P.Manoj, R.J.Moore, P.Young, S.Al-
Sahab, D.Bush, J. Wright and A.J.Fillery-Traivs, J Nu&, 2000, 130, 122
3. L.Marciani, P.A.Gowland, R.C.Spiller, P.Manoj, R.J.Moore, P.Young and
A.Fillery-Travis, A m J Physiol. Gastrointest Liver Physiol, 2001,280, G1227
4. L.Marciani, P.A.Gowland, R.C.Spiller, P.Manoj, R.J.Moore, P.Young and
A-Fillery-Travis,A m J Physiol. Gastrointest Liver Physiol, 200 1,280, G844
CELL SIGNALLING POTENTIAL OF HYDROCOLLOIDS: POSSIBLE NEW
APPLICATIONS?
A.O. Phillips
Institute of Nephrology, University of Wales College of Medicine, Heath Park, Cardiff

CF14 4XN, UK.
Introduction
The aim of this review is to bridge the boundary between the image of
hydrocolloids held by the food industry and that of the biomedical research community.
Traditionally, hydrocolloids have been considered as emulsifiers, stabilisers and gelling
agents in foods and in medicine as scaffold components of the extracellular matrix.
Increasing evidence now suggests that in vivo hydrocolloids have a much wider
functional importance in that they may modify cell function and therefore impact on both
normal homeostatic processes and also on the response to disease and injury. In this
article I will use as examples a human hydrocolloid (hyaluronan), a seaweed hydrocolloid
(alginate) and plant hydrocolloids, to illustrate how our perception of the biological
functionality of hydrocolloids is changing.
Hyaluronan - the natural human hydrocolloid.
Hyaluronan (HA) is a ubiquitous connective tissue polysaccharide which in vivo is

present as a key component of the extracellular matrix. Under normal conditions, it exists
as an extremely high molecular weight molecule, which is involved in regulation of the
normal homeostatic state. It has only recently been recognised that it can perform more
subtle functions than just serving as a structural scaffold and shock absorber. For
example under pathological conditions HA may be generated in higher concentrations,
frequently in a lower molecular weight form. Several workers have reported that HA-
oligosaccharides may stimulate gene expression and protein synthesis of chemokines
and interstitial collagens2,thus initiating a disruption of normal homeostasis. Therefore it
is likely that in both high and low molecular weight forms, HA interacts with cells via
receptors on their surface and thus function as cellular signalling molecules. The main
cell-surface receptor for HA is CD44,which has been implicated both as a signalling
350 Gums and Stubilisers for the Food Industry I1
receptor and also the principal mediator of HA internalisation . HA has therefore been
implicated in a number of normal and pathological biological processes, including
embryonic development, tumour growth, chronic inflammation and wound healing .
One of our major interests has been the relationship between HA and pathological
changes in the kidney, and I intent to use examples from the world of nephrology to
illustrate the emerging role now ascribed to HA in disease. It is now clear that the rate of
progression of renal disease whatever its cause is closely correlated to the degree of renal
interstitial fibrosis. Our studies with proximal tubular epithelial cells (PTC) have focused
on their potential pro-fibrotic influence on the renal interstitium. In the normal kidney
HA and its major receptor CD44, are expressed predominantly in the interstitium of the
renal papilla. Numerous studies have now demonstrated that the expression of both HA
and its receptor CD44 may be altered during renal disease. This may occur during
reversible ischaemic renal injury 5, and associated with models of acute renal
inflammation 6* . In humans alteration in H N C D 4 4 have been identified in the acute
phase of renal transplant rejection * and also during the progressive renal fibrosis which is
associated with chronic transplant rejection (Figure 1). This raises the question as to the
source and the function of HA at sites in these models of renal injury? We have
demonstrated increased HA synthesis by human renal proximal tubular cells in response
to either IL-113, to mimic the inflammatory state, or elevated 25 mM D-glucose, to mimic
the diabetic state.
Figure 1: Increased exmession of hvaluronan during Drogressive

renal fibrosis associated with chronic renal transdant reiection
(adaoted form reference 9)
Hydrocol loids and Health 351
Furthermore, this increased generation of HA is associated with a change in the molecular

weight distribution of the HA generated. Stimulation with either IL-18 or 25 mM D-
glucose also resulted in increased expression of the functional-HA binding form of the
HA receptor CD44, the net result being increased binding of low-molecular weight HA to
the cell surface (Figure 2).
An interstitial macrophage influx has been implicated in the pathogenesis of interstitial
fibrosis related to progressive renal disease of diverse aetiology. HA induces monocyte
chemoattractant protein- 1 expression l o and up-regulation of ICAM- 1 and VCAM- 1 I in
renal tubular epithelial cells. Interestingly up-regulation of adhesion molecule expression
in PTC was also dependent on activation of the transcription factors NF-KB and AP-1 .
Significantly both adhesion molecule up-regulation and induction of chemokine synthesis
in PTC were stimulated by low molecular weight HA polysaccharides such as those
stimulated by exposure to elevated D-glucose concentrations. This suggests that there
may be a positive feedback loop by which inflammation stimulates HA generation which
may then amplify the inflammatory response by increasing recruitment of inflammatory
cells to the site of injury. Alteration in the synthesis of this human hydrocolloid may
therefore directly influence the response of the kidney corticointerstitum to injury. The
relevant question in this conference is: do other hydrocolloids of non-human origin
induce biological effects and can these be utilised to treat or ameliorate human disease
states?
5 800 1 5 m M D Glucose 5 m M D Glucose T
Internalisation Cell surface Internalisation Cell surface

of FITC binding of FITC of FITC binding of FITC
labelled HA labelled HA labelled HA labelled HA
Figure 2: Stimulation of HA binding and internalisation by pro-

inflammatory stimuli and a high concentration of D-glucose
Alginates
Alginates are polysaccharides with gel-forming properties composed of 1,4-1inked

beta-D-mannuronic acid (M), alpha-L-guluronic acid (G) and alternating (MG) blocks.
Outside the food industry, alginates have been used as a matrix for implanted cells, and as
a major component in wound dressings. Microencapsulation of hormone producing cells
has been used for the treatment of diabetes, liver disease and parathyroid disease in
experimental animal models. One of the problems with such an approach is stimulation
of a fibrotic response by the alginate capsules. It is clear that poly (M) binds to the
monocyte cell surface receptor CD14 13. Furthermore this interaction leads to
monocyte stimulation and induction of pro-inflammatory cytokine generation. In contrast
alginates with high G content do not exhibit this monocyte stimulatory activity 14. This
research has obvious implications for the design of alginate delivery vehicles. It
nevertheless establishes the cellular activity of alginate.
In contrast to these undesirable pro-inflammatory effects of alginates there are

also potential beneficial effects of alginates on cell function, which may be exploited.
Studies on angiogenesis using alginates as a delivery vehicle for the angiogenic growth
factor Vascular Endothelial Growth Factor (VEGF), suggest that the alginate itself may
be a cofactor for VEGF-dependent stimulation of endothelial cell growth 15. In addition,
the pro-inflammatory effect of the poly-M alginate may be of benefit in certain
circumstances, as mannuronan has been demonstrated to enhance survival of lethally
irradiated mice by stimulation of haematopoiesis 16. This has obvious potential clinical
therapeutic value in treating bone marrow suppression in humans.
Plant Hydrocolloids
Gum arabic is a tree gum exudate, which has had wide range of commercial uses
since ancient times being used for embalming by the Egyptians. Returning to the renal
field of renal medicine, it is well established that supplementation of the diet with gum
arabic soluble fibre lowers serum urea nitrogen in rodents l7 and in patients with chronic
renal failure I*. This effect is the result of increased colonic bacterial fermentation of gum
arabic thus providing them with energy for growth and nitrogen incorporation, and in turn
increasing faecal bacterial mass and nitrogen excretion. More recently a pilot study
performed in collaboration with the Dialysis Centre in Khartoum suggests that dietary
gum arabic supplementation also reduces serum creatinine (Table 1). This cannot be
related to alterations in colonic bacterial fermentation, but rather suggest that there may
be a direct beneficial effect on renal function. The mechanism of these potentially
important therapeutic observations is currently the subject of ongoing research.
Studies dating back over 20years have shown that locust bean gum, a non-digestible
polymer of mannose and galactose derived form the seeds of the ceratunia siliqua tree,
was an efficient sorbent which bound to many of the potential toxic substances found in
patients with chronic renal failure 19. Subsequently, in a study of only two patients, it was
suggested that dietary supplementation with locust bean gum improved both blood
pressure control and also reduced serum creatinine 2o (Figure 2). These observations,
which have not been investigated further, if confirmed would suggest a mechanism other
than that related to its sorbent properties by which locust bean gum may influence
progressive renal disease.
In addition to the work with gum arabic and locust bean gum, a single study using dietary
supplementation with the hemicellulosic ispaghula husk also suggested a beneficial effect
Number of patients treated 19

Mean decrease in serum creatinine 3 1.7%
Mean decrease in blood urea nitrogen 43.9%
I Mean decrease in serum phosphate I 6.2% I
Table 1: Effect of dietary supplementation with Gum arabic (50g/day for 2 months) in
patients with renal impairment (prior to onset of end stage renal failure)
354
--
Methyldopa(75hg/day)
Gums and Stabilisers for the Food Industry I 1
Locust bean gum (25gJday)
2 4 6 8 10 12 14 16 18 20
Time (weeks)
Figure 2: The effect of locust bean gum on renal function and blood pressure (Adapted
from reference 20).
on renal function beyond increased faecal bacterial loss *I. Again these observations
received little attention and the potential of the observation has largely been ignored.
In summary it is clear that our perception of the hydrocolloids is changing as we

identify more ways in which cellular function may be modified by these polysaccharides.
Furthermore the literature is littered with clinical observation, of which I have focused on
the renal field, which suggest that there are many as yet unexplored avenues of potential
therapeutic applications of these compounds.
References:
1. C. M. McKee, M. B. Penno, M. Cowman, M. D. Burdick, R. M. Strieter, C. Bao and
P. W. Noble, J. Clin. Invest., 1996,98,2403
2. P. Rooney, M. Wang, P. Kumar and S. Kumar, J Cell Sci, 1993,105,213
3. Q. Hua, C. B. Knudson and W. Knudson, J Cell Sci, 1993,106,365
4. T. C. Laurent and R. E. Fraser, FASEB, 1992,6,2397
5. A. J. P. Lewington, B. J. Padanilam, D. R. Martin and M. R. Hammeramn, Am. J.
Physiol., 2000,278, R247
6. P. S. Benz, X. Fan and R. P. Wuthrich, Kidney Znt, 1996,50, 156
7. V. Sibalic, X. Fan, J. Loffing and R. P. Wiirthrich, Nephrol. Dial. Transplant., 1997,

12,1344
8 . A. Wells, E. Larsson, E. Hanas, T. Laurent, R. Hallgren and G. Tufvesson,
Transplantation, 1993,556; 1346
9. A. F. Wells, E. Larsson, A. Tengblad, B. Fellstrom, G. Tufvesson, L. Klareskog and T.
C. Laurent, Transplantation, 1990,50,2; 240
10. B. Beck-Schimmer, B. Oertli, T. Pasch and R. P. Wiithrich, J. Am. SOC.Nephrol.,
1998,9,2283
1 1 . B. Oertli, B. Beck-Schimmer, X. Fan and R. P. Wiithrich, J. Zmmunol., 1998, 161,
343 1
12. T. G. Jahr, L. Ryan, A. Sundan, H. S. Lichenstein, G. Skjak-Braek and T. Espevik,
Infect. Zmmun., 1997,65, 1; 89
13. T. Espervik, M. Otterlei, G. Skjak-Braek, L. Ryan, S. D. Wright and A. Sudan, Eur J
Zmmunol, 1993,23,255
14. M. Otterlei, K. Ostgaard, G. Skjak-Braek, 0. Smidsrod, P. Soon-Shiong and T.
Espevik, J Immunother, 1991, 10,286
15. A. Kawada, N. Hiura, S. Tajima and H. Takahara, Arch Dermatol Res, 1999,291,
542
16. 0. Halaas, W. M. Olsen, 0. P. Veiby, D. Lovhaug, G. Skjak-Braek, R. Vik and T.
Espervik, S c a d J Immunol, 1997,46,358
17. S . A. Assimon and T. P. Stein, Nutrition, 1994,10,6; 544
18. D. Z. Bliss, T. P. Stein, C. R. Schleifer and R. G. Settle, Am J Clin Nutr, 1996, 63,
392
19. H. Yatzidis, Kidney Int, 1977, 12, Suppl8; 152A
20. H. Yatzidis, D. Koutsicos and P. Digenis, Clin Nephrol, 1979, 11, 2; 105
21. D. S. Rampton, S. L. Cohen, V. B. Crammond, J. Goibbons, M. F. Lilburn, J. Y.
Rabet, A. J. Vince, J. D. Wager and 0. M. Wrong, Clin Nephrol, 1984,21,3;159
ALGINATES IN RELATION TO HUMAN HEALTH;
FROM COMMODITY TO HIGH-COST PRODUCTS?
K.I. Draget and G. Skj&-Braek
NOBIPOL, Dept. of Biotechnology, Norwegian University of Science and Technology,

N-7491 Trondheim, Norway
1 INTRODUCTION
Alginates are glycuronans extracted from seaweed or produced by some bacteria. The
molecules are linear chains of (1+4)-linked residues of p-D-mannuronic acid (M) and a-
L-guluronic acid (G) in different proportion and sequential arrangements. The most
common arrangement is that of a block co-polymer, in which long homo-polymeric
sequences of ManA residues ("MM-blocks") and similar sequences of GulA residues
("GG-blocks") are interspersed between sequences of mixed composition ("MG-blocks");
Figure 1. They form gels with cations and the gel-forming capacity correlates with the
content and length of the G-blocks.'
coo-
P-D-mannuronate (M) a-L-guluronate (G)
QOC
OH
G G M M G
--
M M M MGM GGGGGMGM GGGGGGGGMM GM GM GGM
Mblock
*
-
G-block G-block MG-bIock
Figure 1 Structural characteristics of alginates: (a) alginate monomers, (b) chain

conformation, (c) block distribution
Alginates are among the most versatile polysaccharides with respect to the number
of applications.*These range from industrial (as thickeners in textile printing, paper and
board treatment, welding rod production, water treatment and can sealing), through foods
and pet-food (e.g. the restructuring of foods, in baking creams, as stabilisers and thickeners
in ice cream, emulsions, salad dressings and beers) to pharmaceutical and dental
applications. The latter include dental impression materials, non-woven wound dressings
and pharmaceutical recipes against oesophageal reflux.
Over the last decade, a considerable amount of new knowledge has emerged
following the enormous popularity that alginate has gained as an immobilisation matrix for
living cells.3 Perhaps the most significant finding in this context is the immune stimulating
properties of mannuronate-rich alginates. This was first discovered in the animal
transplantation trials of encapsulated Langerhans islets for diabetes control. Here,
overgrowth of alginate capsules by pha ocytes and fibroblasts, resembling a foreign
body/inflammatory reaction, was reported.% Induction of tumour necrosis factor (TNF) and
several interleukins (cytokines having pro-inflammatory activities) in bioassays showed
that the inducibility depended upon the content of mannuronate in the alginate ample.^
Similar effects are found for other polyuronides containing p- 1+4 di-equatorial linked
sequences. High M-alginate and lipopolysaccharide (LPS) were found to stimulate human
monocytes by a similar mechanism, which involved the CD14 LPSLBP receptor.
The scope of this paper is to summarise this new structure/function knowledge on
alginates in molecular biology that points towards an added value development of the
alginate molecule.
2. ALGINATE AS IMMOBILISATION MATRIX
Entrapment of cells within Ca-alginate spheres has become the most widely used technique
for the immobilisation of living cells.3 This immobilisation procedure can be carried out in
a single step process under very mild conditions and is therefore compatible with most
cells. The cell suspension is mixed with a sodium alginate solution and the mixture dripped
into a solution containing multivalent cations (usually Ca2'). The droplets then
instantaneously form gel-spheres entrapping the cells in a three-dimensional lattice of
ionically cross-linked alginate. The possible uses for such systems in industry, medicine,
and agriculture are numerous, ranging from production of ethanol by yeast, monoclonal
antibodies by hybridoma cells to mass production of artificial seed by entrapment of plant
embryos?
The perhaps most exciting prospect for alginate gel immobilised cells is their
potential use in cell transplantation. Here, the main purpose of the gel is to act as a barrier
between the transplant and the immune system of the host. Different cells have been
suggested for gel immobilisation including parathyroid cells for treatment of hypocalcemia
and dopamine producing adrenal chromaffin cells for treatment of Parkinson's disease.6
However, major interest has been focused on insulin producing cells for the treatment of
Type I diabetes. Alginate/poly-Llysine capsules containing pancreatic Langerhans islets
have been shown to reverse diabetes in large animals and are currently being clinically
tested in Table 1 lists some biomedical application of alginate encapsulated
cells. As a result of this development, one alginate producer (Pronova Biomedical) now
commercially manufactures ultrapure alginates highly compatible with mammalian
biological systems. These qualities are low in pyrogens, and facilitate sterilisation of the
alginate solution by filtration due to a low content of aggregates.
Table 1 Some potential biomedical applications of alginate encapsulated cells
Cell type Treatment of
Adrenal chromaffi cells Parkinson' disease'

Hepatocyte Liver f aiIure'
Paratyroid cells Hypocalcemia'
Langerhan islets Diabetes537
(P-cells)
Genetically altered cells ~ance?~
3 ALGINATE AS AN IMMUNE STIMULATING AGENT

One of the main initial problems with the injection of alginate capsules for treating
diabetes was overgrowth of the ca sules with phagocytes and fibroblasts resembling a
Y
foreign bodyhnflammatory reaction. When alginates were tested for cytokine induction on
human monocytes, it became evident that the ability of alginates to induce TNF, IL-1 and
IL-6 correlated with the ManA-content of the alginate, as well as on the molecular size.*
This result also directly explains the observed capsule overgrowth; mannuronate rich
fragments, which do not take part in the gel network, will leach out of the capsules and
directly trigger an immune response? This is illustrated in Figure 2a, where alginates with
different contents of mannuronic acid were tested for induction of TNF from human
monocytes. The highest potency is found for polymers containing more than 95 %
mannuronic acid residues and with a molecular size above 50 OOO Dalton. This material
was isolated from Pseudomonas aeruginosa and is designated poly-M. An analogue
structure, a di-equatorially p( 1+4) linked D-glucuronic acid, prepared by a selective
oxidation at C-6 in cellulose (CGOXY), also stimulates monocytes to produce TNF
although with less potency compared to poly-M (Figure 2b).
The 3D structure of C6OXY (94 9% D-GlcA) is similar to that of poly-M except that
the consecutive uronic acid residues in C6OXY are broken up with D-Glc. The stimulatory
effect of CGOXY on TNF production depends on the amount of D-GlcA in the polymer,
suggesting that the TNF induction from monocytes may occur with different types of
p( 1+4)-linked uronic polymers. The TNF-inducing capacity increases with molecular
weight of the poly-M up to 5oooO Dalton. By degrading it into oligomers with a DP, =20,
either with a controlled acid hydrolysis or by treatment with an M-specific lyase, the TNF
inducing capacity is lost. Immune stimulation can, however, be regained and even strongly
potentiated by linking the oligomer to microparticles.''
Figure 2 Stimulation of tumour necrosis factor (TNF)production as effect of A: M-content

in alginate, and B: GlcA-content in C6OXY
3.1 Effects in vivo
The potent cytokine inducing ability of p( 1+4)-linked uronic acid polymers on monocytes
in vitru points towards possible effects also in different in vivo models. Until recently,
limited knowledge has been available on the immune stimulating effects of alginates in
animal models. Table 2 summarises the most important data on the biological effect of
poly-M and C6OXY.
Table 2 Effects of poly uronic acids on immune cell functions; -biological effects
Induction of TNF, IL- 1, IG6, GM-CSF, and IL-12 p40 in human monocytes
Induction of TNF and I L 6 in mice
Protection against lethal gram positive and gram negative infections
Protection against lethal effects of irradiation
Increases the generation of myeloid progenitor cells
Increases the amount of antibody producing cells
Increases non-specific immunity in turbots
These data demonstrate that p( 1+4)-linked uronic acid polymers are also active
inducers of cytokine in vivo. Of particular interest is that poly-M can protect mice against
lethal infection with E. coli or S. aureus." In line with these observations, poly-M also
gives a marked protection of mice against lethal irradiation.12 Subjecting animals to
irradiation leads to loss in the ability of the bone marrow to generate white blood cells.
Irradiated animals thus may die from infections caused by bacteria, which normally are
well tolerated (Figure 3). In combination with sub-optimal concentration of colony
stimulating factors, poly-M enhances the formation of GM-CSF colonies suggesting that
poly-M can increase the production of myeloid blood ce11.12 This could be one mechanism
behind the radio-protective effect of poly-M. The stimulating effect of poly-M on
hematopoietic cells may be clinically important.
Alginate rich in mannuronic acid also has immune stimulating properties in fish.
Juvenile turbots fed with an alginate rich in mannuronic acid obtained an increased
360 Gums and Stabilisersfor the Food Industry I 1
protection against one type of pathogenic ba~teria.'~Evidently, these data show that
p( 1+4)-linked uronic acid polymers have potent biological effects in several biological
systems and that these effects most probably are caused by stimulation of the
monoc ytes/macrophages.
\'4 * * * * * *
"x-x-x-x -x-x -x -x -xx -x-x -xx -x -xx -x -x *
3 60
5
U
0:NaCl
3 0: 0.5 m d k b bw
v)
0 : 1.0 mg/kg bw
40
8 X: 2.0 m a g bw
20
0
0 5 10 15 20 25 30 35 40
DAYS AFTER X-RAY
Figure 3 EfSect of prophylactic (24 h; intraperitoneal) administration of mnnuronan on

the survival of lethally irradiated (7.3 Gy) C57B1/6 mice
3.2 Molecular Mechanisms
The immune response of both lipopolysaccharide (LPS)from gram negative bacteria and
poly-M involve CD14 on the monocyte membrane14, and poly-M binds to CD14 on
monocytes in the presence of serum.I5 The binding of both poly-M and LPS to monocytes
can be inhibited by addition of G-blocks suggesting a common binding site for these
apparently different polysaccharides.' After this initial observation, several reports have
now implicated a role for CD14 in response to a variety of different compounds such as
soluble peptidoglycan fragments and protein free phenol extracts from S. a ~ r e u s ' ~ ' ~ ~ ,
rhamnose-glucose polymers from Streptococcus mutans", chitosan from arthropods", and
insoluble cell walls from different gram positive bacteria.20These data implicate that CD14
has a broad specificity for compounds containing different types of sugar residues.
LPS can also stimulate cells that do not express membrane CD14 indicating a wide
variety of different cell types affected by LPS.In contrast, poly-M is not able to stimulate
cell types lacking membrane CD14.I5 These data suggest that LPS can interact with cells
via several (different) modes of action. This broad stimulatory pattern of LPS is likely to
be important for its lethal effect in vivo.
The apparent specific effect of uronic acid polymers on CD14-positive cells may
result in low systemic toxicity and suggests potential applications as an immuno-
modulator. Thus, poly-M may activate the non-specific immune system resulting in
increased protection against various types of infections.
Hydrocolloids and Health 36 1
4 TAILORING OF ALGINATES BY IN VZTRO MODIFICATION
One element of development suggesting a rapid movement of alginate into speciality

polymers rather than staying a commodity, is the discovery of the mannuronan C5
epimerases. Alginate with a high content of guluronic acid can be prepared from special
algal tissues, by chemical fractionation or by in vitro enzymatic modification of the
alginate using these mannuronan C-5 epimerases from A. vinelandii.2-24These epimerases,
which convert M to G in the polymer chain, has recently allowed for the production of
highly programmed alginates with respect to chemical composition and sequence. A.
vinelandii encodes a family of 7 exo-cellular iso-enzymes with the capacity to epimerise
all sorts of alginates and other mannuronate-containing polymers as shown in Figure 4,
where the mode of action of AlgW (giving alternating introduction of G) is presented.
/cz%xr-W=h
HOOCCH ma4
p-D-ManpA
M M M
C-5 Epimerasl AlgE4
%wo
%wt
- a 4
a4 ma4
M G M OH 4 wL-GuIPA
G
Figure 4 Mode of actionfor the mnnuronun CS-epimerase AlgE4
Although the genes have a high degree of homology, the enzymes they encode
exhibit different specificities. Different epimerases may give alginates with different
distribution of M and G and thus, alginates with tailored physical and chemical properties
can be made. None of the enzymatically modified polymers are, however, commercially
available at present. Table 3 lists the modular structure of the mannuronan C-5 epimerase
family and their specific action.
Table 3 The seven AlgE epimerasesfrom A. vinelandii

Type [kDa] Modular structure Products
Bi-functional
AlgE 1 147.2 mklk1mmllR4) G-blocks
+ MG-blocks
AIgE2 103.1 ~ ~ l ~ m
G-blocks (Short) ~ ]
AIgE3 191 I f l ~ ~ ~ ( A 2 ~ ~ l k 1 1 R 6 1Bi-functional
( R 7 1G-blocks
+ MG-blocks
AlgE4 57.7 [AF] MG-blocks
AlgE5 103.7 FImrdmFl G-blocks (medium)
AIgE6 90.2 mmmk1 G-blocks (long)
AIgE7 90.4 (AllmmkR3] Lyase activity

+ G-blocks
+ MG-blocks
-
A - 385 amino acids. R 155 amino acids
5 CONCLUSIONS
It is expected that future growth in the alginate market most likely will be of a qualitative
rather than a quantitative nature. General predictions suggest manufacturers to move away
from commodity alginate production, partially due to an increased competition from low-
cost bulk polymers, into more refined products for e.g. the pharmaceutical industry.
Highly purified alginates for immobilisation of human cells and as immune
stimulating polymers offer a possibility for alginate to move in to the pharmaceutical area
in addition to the "commodity" type of applications.
6 ACKNOWLEDGEMENTS
The authors would like to emphasise that this paper presents a brief summary of the most
important discoveries within a highly interdisciplinary research area, which for 10 years
has been co-ordinated by the Norwegian Biopolymer Laboratory, where all NOBIPOL's
scientific partners have taken part. A special thanks to Wenche Strand, Anne Bremnes and
Ingrid Aune for helping to prepare the manuscript.
References
1 0. Smidsrad, J. Chem. SOC.Farad. Trans, 1974,57, 263.
2 E. Onsgyen, Carbohydr. Eur., 1996,14,26.
3 0.Smidsrad and G. SkjAk-Brak, Trends in Biotechnology, 1990,8,71.
4 P. Soon-Shiong, M. Otterlei, G. SkjAk-Braek, 0. Smidsrgd, R. Heintz, R.P. Lanza
and T. Espevik, Transplant. Proc., 1991,23,758.
5 P. Soon-Shiong, E. Feldman, R. Nelson, R. Heintz, 2. Yao, Q. Yao, T. Zheng, N.
Merideth, G. SkjAk-Brak, T. Espevik, 0. Smidsrad and P. Sandford, Proc. Nat.
Acad. Sci. USA, 1993,90,5843
6 P. Aebisher, M. Goddard and P.A. Tresco, Animal Encapsulation and
Immobilization,CRC Press, Boca Raton, 1993, 197.
7 P. Soon-Shiong, R.E. Heintz, N. Merideth, Q.X. Yao, Z.W. Yao, T.L. Zheng, M.
Murphy, M.K.Moloney, M. Schmehl, M. Hams, R. Mendez, R. Mendez, R. and
P.A. Sandford, Lancet, 1994,343,950.
8 M. Otterlei, A. Sundan, G. SkjAk-Braek, L. Ryan, 0. Smidsrgd and T. Espevik,
Infect. Immun., 1993,61, 1917.
9 B.T. Stokke, 0. Smidsrprd, F. Zanetti, W. Strand and G. Skjflk-Braek, Carbohydr.

Polym., 1993,21,39.
10 G. Berntsen, L. Kilaas, T.H. Flo, A. Medvedev, G. Skjflk-Braek A. Sundan and T.
Espevik, 1998 J. Clinical and Diagnostic Laboratory Immunology, 1998,5,355.
11 T. Espevik and G. Skjflk-Braek, Carbohydr. Eur., 1996,14,19.
12 0.Halaas, W.M. Olsen, O.P. Veiby, D. Lgvhaug, G.Skj&-Braek and T. Espevik,
Scand. J. of Immunol , 1997,46,358.
13 J. Skjermo, T. Defoort, M. Delasque, T. Espevik, Y. Olsen and G. Skjflk-Braek,
Fish & Shel&sh Immunol., 1995,5,531.
14 M. Otterlei, T. Espevik, K. Ostgaard, G. Skjflk-Brak, P. Soon-Shiong and 0.
Smidsrprd, 1991, 10,286.
15 T. Espevik, M. Otterlei, G. Skj&-Braek, L. Ryan, S.D. Wrigth and A. Sundan, Eur.
J. Immunology, 1993,23,255.
16 B. Weidemann, H. Brade, E.T. Rietschel, R. Dziarski, V. Bazil, S. Kusomoto, H.D.
Flad and A.J. Ulmer, Infect. Immun., 1994,62,4709.
17 T. Kusunoki, E. Hailman, T.S.-C. Juan, H.S.Lichenstein and S.D. Wright, J. Exp.
Med., 1995, 182, 1673.
18 M. Soell, E. Lett, F. Holveck, M. Scholler, D. Wachsmann and J. Klein, J.
Immunol., 1995,154, 85 1.
19 M. Otterlei, K.M. Vhum, L. Ryan and T. Espevik, Vaccine, 1994,12,825.
20 J. Pugin, D. Haumann, A. Tomasz, V.V. Kravchenko, M.P. Glauser, P.S. Tobias
and R.J. Ulevich, Immunity, 1994, 1,509.
21 H. Ertesvdg, B. Larsen, G. Skjflk-Brak and S. Valla, J. of Bacteriol., 1994, 176,
2846.
22 H. Ertesvd , H.K. Hpridal, I.K. Hals, A. Rian, B. Doseth and S Valla, Mol.
Microbiol, f995,16,719
23 H. Ertesvdg, H. Hgydal, G. Skjflk-Braek, and S. Valla, J. of Biol. Cem., 1998, 273,
30927.
24 H. Hprydal, H. Ertesvig, B.T. Stokke, G. Skjflk-Braek and S. Valla, J. of Biol.
Chem., 1999,274,12316.
25 T.-A. Read, D.R. Sorensen, R. Mahesparan, P.0. Enger, R. Timpl, B.R.Olsen,
M.H.B. Hjelstuen, 0. Haraldseth, R. Bjerkevig, Nature Biotechnology, 2000. 19,
29.
Subject Index
Absorption spectra 55 Biosynthesis, of polysaccharides 287

Acacia gum 318-321 Blood glucose 338-341
Acceptable daily intake 325 Blood urea nitrogen 353
Acetobacter xylinius 282,283 Borate ions 61-63
Acidified milk 209 Bovine serum albumin 257-262
Activation energy 41, 107-1 1 1 Bridging flocculants 203
Adsorbed pectin 315
Adsorption, protein 65-71 C-5 epimerase 361
Agar 4,188-198 Calcium aids 327
fluid gels 96-103 Calcium chloride 148,150,151
Agarose 10&111 Capillary electropherogram, of
glass transition 47 carrageenan 202
Aggregation 237 Capillary electrophoresis 13-24
Aggregation of lactoglobulin 263-270 Carboxymethylcellulose4,5
Algal polysaccharide gels 104-11 1 Carrageenan4,5,104-111,188-198
Alginates 4,5 , 188-198,352,356-362 and gelatin 172-184
Amazonas (soft wheat) 137-144 gelation of 158-164
Amylase 339 gels 93
Amylose 146 hybrid 202-209
Amylose content 138 in dairy dessert gels 201-210
Amylose-lipid dissociation 140 mixed systems 112-1 19,172-184
Andrade equation 40,41 weak gels 165-171
Anion exchange chromatography 16- Carrageenan-pro tein interactions 205,
24 206
Arabinogalactans 218 Carreau model 227-230
Arrhenius model 107 Casein 238-243
Aspergillus niger 16 Caseinate 205,213-216
Atomic force microscopy 166-171, Cassia gum 4,5
299-303 CA-type starches 145
A-type starches 145 CD44349-351
Cell signalling 349-354
Beverage emulsions 3 19 Cellulose, composites of 282-287
Beverage, citrus punch 222 Chain conformation 181
Beverage, lemon flavoured 220-221 Chemical shift 29,283
Bi-axial test 286 Chitosan, emulsion stabilisation 245-
Binodal 195 254
Biopolymer mixtures 1 12-119,129- Chocolate milk 201
136 Cholesterol 338-341
phase separated, 134 Circular dichroism 85-94
Citrus pectin 3 11-3 17 Dynamic light scattering 264

Citrus punch beverage 222 Dynamic mechnical analysis 105-1 11
Coalescence 237 Dynamic viscoelastic properties, of
Composites, of cellulose 282-287 carrageenan 166-1 71
Compression curves, for xanthan-LBG
295 Elastic modulus 90
Compression measurements 173-1 76 Eldridge-Ferry enthalpy 276,277
Confocal micrograph 193, 197 Electron micrographs, of gellan 191
Connectivity 35 Electrophoresis, capillary 13-24
Controlled release 197-1 98 Emulsifyng properties,
Coomassie blue 3 13 enhanced 3 18-32 1
Corn starch 104-1 11,155 of citrus pectin 3 11-3 17
Correlation length 61 Emulsion complex modulus 129
COSY 31-38 Emulsion stabilising properties, of
CPMAS 283 chitosan 245-254
Cream layer 75 Emulsions 2 11-2 16
Creaming 237,251-253 microscopic visualisations 249
Creaming behaviour 3 14 of lysozyme 75
Creaming effects 133 preparation 247,3 12
Creaming experiments 257-26 1 protein and peptide 237-244
Critical gelation concentration 274, stabilised 256-262
276 water-in-water 128-136
Critical yield stress 121 whey proteins 256-262
Crosslinking 61-63 Endo polygalacturonase 16
Crosslinks 283 Endothermic enthalpy 142
Cryo-TEM micrographs 265 Enthalpies, transition 142
Cylindrical gel particles 113 Enthalpy,
Eldridge-Ferry 276,277
Dairy dessert gels 20 1-2 10 vant Hoff 275-277
Deacylated gellan gum 191 Enzymes, new sources 287
Deformation, of gelled particles 116 Enzymic debranching 289-296
Depletion flocculation 203,256-262 ESR spectroscopy
Depolymerised citrus pectin 3 11-3 17 Euchema cottonii 158, 159,202
Dextran 259 Euchema denticulatum 202
Dichroic ratio analysis 65 Euchema spinosum 158,202
Differential scanning calorimetry 78- Evanescent wave 66
80,87-94
of carrageenan 159-164 Fabricated food products 187
of starch 138-144 Fenugreek 36
of Sugar starch 145-156 First-order kinetics 107
of xanthan 292 Fish gelatin 172-1 84
Diffusing wave spectroscopy 258-262 Fish muscle 79
Disorder-rder transition 159-1 64 Flavour compounds 2 18
Dispersed particles 120 Flavour intensity 198
Dispersed phase morphology 112 Flavour release 2 17-225
Drop diameter 239 Flocculation, bridging 203
Droplet rise experiments 120-1 27 Flocculation, depletion 203,256-262
Droplet size 196 Flow plasticity 121
determination 3 13 Fluid gels 95-103
distribution 130 Fluorescein-labelling 54
Subject Index 367
Fluorescence anisotropy 54-64 sheared 95-1 03

Fluorescence polarisation spectroscopy thennoreversible 85-94
54-64 xanthan-galactomman 289-296
Fluoroscence spectroscopy 79 Gel-sol transition 88
Food ingredient approval 326 Gene-level specification 287
Food products, fluid-like 120 Glass transition 47,48
Free volume theory 40 Glassy state 39,41, 51
Frequency dependence 167 Globular proteins 262-270
Friction coefficients 39,41,44,45 Glucomannan 338-341
chemical shifts 283
Galactomannan, chemical shifts 283 Gravitational stress 227
Galactomannans 188-1 98 Guar gum 4 , 6
Galactosidase 290 debranching of 289-296
Galacturonic acid 18, 19 Guluronate 356
Gastric processing, 342-348 Gum acacia 3 18-32 1
Gel component 114,115 Gum arabic 4 3 , 352,352
Gel particles 112, 192
Gelatin 4 , 6 Health claims 325-336
and maltodextrin 128-136 Helices, carrageenan 159, 165
droplet size 196 Helix-coil transition 85-94, 180
gels 271-278 Hen egg white lysozyme 67-71
fish172-184 Hershel Bulkley model 307
free volume 46 Hertz model 101,102
friction coeffcients 44,45 High solids systems 39-51, 190
mixed systems 172-1 84 Hofmeister series 151
shift factors for 44,45,46 Hyaluronan 349-3 5 1
Gelatin<arrageenan interactions 172- Hybid carrageenan gels 203-209
184,177-1 84 Hydrocolloids,
Gelatinisation, cell signalling potential 349-354
enthalpy 146-157 infbence on flavour release 2 17-
of sago starch 145-156 225
starch 138-144 in real food systems 187-1 99
Gelation, Hydrophobic groups 74
critical concentration 274 Hydrophobic regions 2 19
of carrageenan 158-1 64 Hydroxypropyl p a r 54-64
rate of 272 Hydroxypropylmethylcellulose 4 , s
Gellan gum 4,6,85-94,188-198
mixed gels 112-1 19 Immune response 360
molecular weight 299 Immunoglobulins 257-262
new textures with 297-305 Interface, silicdsolution 65-7 1
Gellan, flavour modification 222 Interfacial composition 2 12
Gellan, glass transition 47-49 Interfacial tension 113, 128, 131
Gels, Interpenetratingnetworks 284
algal polysaccharide 104-1 11 Intrinsic viscosities 138
carrageenan 93,165-171 Isoelectric point 238
dairy dessert 201-210 Junction zone characteristics 273-278
fluid 95-1 03 Junction zones 85,88-94, 174
gelatin 271-278
hybrid carrageenan 203-209 Kappaphycus elvarezii 202
mechanical relaxation of 107 Karaya 4 , 6
Kinetic processes 271-278 Microcrystalline cellulose 4, 5

Kinetics, Microgel suspensions 97
first-order 107 Microstructure 95-1 03
of film formation 65-71 Milk reactivity 203
Konjac flour 338-341 Milk, acidified 209
Konkoli seed gum 306-3 10 Mixed hydrocolloid systems 193-1 98
Mixed system, gelatidcarrageenan
Lactalbumin 257-262 177-1 84
Lactoglobulin 238-243,25 7-270 Mixtures, biopolymer 112-1 19
Legislation, European 32 5-3 36 Molecular weight determination, of
Lemon, beverage 220-222 gellan gum 299
Linear viscoelastic emulsion model Morphology 196
128-136 Morphology control 112-1 19
Linear viscoelastic measurements 132- Morphology, particle 114
136 Muscle protein 79
Linear viscoelastic strain region 168
Lipid acyl hydrolase 238 Network, carrageenan 170
Lipid, amylose 140 NIR-FT-Raman spectroscopy 74
Lipid-protein interactions 73-80 NMR spectroscopy,
Lithium chloride 148, 149, 151 molecular order by 283
Locust bean gum 4,6,189-198,353, principles 28
354 two dimensional 27-3 8
Locust bean gum, gels with xanthan NOESY 35-3 8
289-296 Nutrition 328
Lysozyrne 65-71,75-80,238-243
Oil-in-water emulsions 21 1-216,256
Maesopis eminii plant 306 Oil-water interfaces 237
Magnesium chloride 151 Ostwald ripening 237
Magnetic resonance imaging 342-348 Ovalbumin 238-243
Maltodextran-gelatin mixtures 128-
136,194-196 Paliernes model 129
Maltodextrins 338-34 1 Particles,
Mannuronate 356 deformation 116
Market overview 3-9 gel 192
Maxwell model 129 morphology 114
Mechanical characteristics of gels, size determination 247
104-1 11 Patatin 238-243
Mechanical relaxation 107 Pea starch 154,155
Mechanical spectra, Pectin 4, 6, 188-198,218
of carrageenan 169 adsorbed 3 15
of starch 141-144 citrus 3 11-3 17
Meromyosine 78 composites 285,286
Metahydroxyldiphenyl3 13 fine structure 13-24
Methyl cellulose 4, 5 glass transition 47
Methyl linoleate 78, 80 low ester amidated 228-233
Methylene blue 18 1-1 84 Pentosans 137-144
Methylesterification, Peptides, role of in emulsions 237-244
degree of 13 Phase diagrams 179,194-196
distribution of 13, 14 Phase separation 178, 193-195,248-
Micelles 205-2 10 254,266-270,284
Subject Index 369
Phenylalanine 67 Sensory-instrumental correlations 2 17-

Polygalacturonic acid 19 225
Polyglycol alginate 222-224 Serum creatinine 353
Polysaccharides, in gastric processing Serum phosphate 353
342-3 48 Setting temperature 298,300
Potato starch 155 Shear compliance 42
Process effects 191 Shear field 196
Protein film 65-71 Shear modulus, measurement of 272
Protein load 253,254 Shear rheology, of xanthan 123-126
Protein surface load 239-243 Shear-cooling process 113
Protein-carrageenan interactions 205, Sheared gels 95-103
206 Shift factor 40,44,45,49
Protein-lipid interactions 73-80 Silicdsolution interface 65-7 1
Proteins, Size exclusion chromatography 263,
globular 263-270 264
role of in emulsions 237-244 Small amplitude oscillatory rheology
water soluble 137-144 131-136
Small deformation oscillatory
Quantitative Descriptive Analysis 219 measurements 159
Sorraia (hard wheat) 137-1 44
Raman spectroscopy 73-80 Sour cream 224
Reel-chain model 88, 89 Specific ellipticity 85-94
Relaxation spectra 43,44 Spin-spin coupling 29
Relaxations, mechanical 39-5 1 Splingomonas elodea 297
Renal fibrosis 350 SPMAS 283
Retrogradation 139-144 Stability, of emulsions 247-250
Rhamnogalacturonan 36 Starch 4 , 6
Rheological behaviour, of carrageenan corn 104-111
165,171,203-209 gelatinisation 138-144
Rheological properties, granules 104-1 11
of gelatidcarrageenan 177-1 84 oxidised 196
of starch 137-144 rheological properties 137-144
Rheology, rice 104-1 11
of fluid gels 95-103 sago 145-156
of suspensions 118 soluble 104-1 11
of water in water emulsions 128- tapioca 155
136 wheat 104-111,137-144
Rice starch 104-1 11 Steady flow properties, of carrageenan
Rotational correlation times 60 166-171
Rotational difhsion constant 59 Stokes law 227
Rouse theory 44 Storage and loss moduli,
Rubber elasticity, theory of 273 of agarose 98,99-102,106-111
Rubbery region 42 of carrageenan 106-1 11,159-164
Storage modulus, of starch 140-144
Sago starch 145-156 Strain dependence 168
Salad dressing 223,224,3 19 Strawberry content 228
Salting-in ions 145-157 Stress sweeps, of xanthan 123-126
Salting-out ions 145-1 57 Structural analysis 27-38
Sedimentation 133 Structure/fhction relationships 188-
198
Sugar, high 39-51 Weak gel systems 120

Surface pressure 241-243 Weak gels, carrageenan 165-1 71
Suspension rheology 118 Wheat starch 104-1 11,137-144
Wheat, NMR 33,34,37
Tapioca starch 155 Whey protein emulsions 256-262
TEM micrographs 265,267,269 Whey protein isolate 205,2 14-2 16,
Texture 2 17 257
control of 2 16 WLF theory 40,107
Texture profile analysis 298
Thermoreversible gels 85-94, 165-1 7 1 Xanthan,
Thixotropy 126 flavour modification 222
Time resolved fluorescence 56-64 gels with guar gum 289-296
Time resolved polarised attenuated gum, market potential 4 , 7
total reflection spectroscopy 65- yield stress of 120-127
71 Xanthomonas campestris 120
TOCSY 32-3 8 Xyloglucan,
Tragacanth 4 , 6 chemical shifts 283
TRATR spectroscopy 65-71 composites 285,286
Tryptophan 67
Turbidity 160-1 64 Yellow mustard gum 36
measurements 179 Yield stress 220-233,307
Tyrosine 67,76 of xanthan 120-127
Yoghurt fruit preparations 226-233
vant Hoff enthalpy 275 Youngs modulus 8694,160-164
Viscoelastic behaviour, of starch 138- of carrageenan 174
144 of mixed gels 175
Water-in-water emulsions 128-1 36 Zero shear viscosity 130,227-233

Water-absorbing capacity 143 Zipper model 89

Gums and Stabilisers For The Food Industry

Uploaded by

Copyright:

Available Formats

Gums and Stabilisers For The Food Industry

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Gums and Stabilisers For The Food Industry

Uploaded by

Copyright:

Available Formats

t

for the Food Industry 11

Special Publication No. 278

ISBN 0-85404-83 6-7

0 The Royal Society of Chemistry 2002

All rights reserved.

Published by The Royal Society of Chemistry,

For further information see our web site at www.rsc.org

Structure, Characterisationand Interactions

Capillary Electrophoresis as a Tool for the Characterisation of Pectin Fine

Application of Two Dimensional (2D) NMR Spectroscopy in the Structural

a and p Mechanical Relaxations in High Sugar Biopolymer Mixtures 39

Fluorescence Anisotropy Measurements of the Dynamic Behaviour of

Time Resolved Polarised Attenuated Total Reflection Spectroscopy of a

Elucidation of Protein-Lipid Interactions 73

Helix-Coil Transition in ThermoreversibleGels 85

Microstructural Origins of the Rheology of Fluid Gels 95

Dynamic Mechanical Characteristics of Red Algal Polysaccharide Gels

Morphology Control in Disperse Biopolymer Mixtures with at Least One

Determining the Yield Stress of Xanthan Solutions via Droplet Rise

Predicting the Rheology of Water-in-water Emulsions 128

Influence of Pentosans on the Rheological and Thermal Properties of

Gelatinisation Properties of Sago Starch in the Presence of Salts 145

Effect of K+ and Ca2+Cations on Gelation of K-Carrageenan 158

Rheological Properties of K-Carrageenan Weak Gels 165

Mixed Systems of Fish Gelatin and Ic-Carrageenan 172

Effect of Gelatidcarrageenan Interactions on the Structure and the

Hydrocolloids in Real Food Systems

Hydrocolloids in Real Food Systems 187

Interaction of Carrageenan with Other Ingredients in Dairy Dessert Gels 20 1

Emulsions as Ingredients in Foods - Surface Compositions and Their

Influence of Hydrocolloids on Flavour Release and Sensory-Instrumental

Interfacial Behaviour and Gelation of Proteins

The Roles of Proteins and Peptides in Formation and Stabilisation of

Effect of Various Factors on Emulsion Stabilising Properties of Chitosan

Depletion Flocculation by Polysaccharides in Whey Protein Stabilised

Aggregation and Gelation of Globular Proteins with and without Addition

Kinetic and Equilibrium Processes in the Formation of Weak Gelatin Gels 27 1

Using Nature to Tailor Hydrocolloid Systems 28 1

Formation of Strong Gels by Enzymic Debranching of Guar Gum in the

New Textures with High Acyl Gellan Gum 297

Rheological Properties and Potential Industrial Application of Kongoli

Emulsifying Properties of Depolymerised Citrus Pectin: Role of the

Modified Hydrocolloids with Enhanced Emulsifying Properties 318

Hydrocolloids and Health

European Legislation and Health Claims 325

Managing Healthy Levels of Blood Glucose and Cholesterol with Konjac

The Role of Polysaccharides in the Gastric Processing of Foods 342

Cell Signalling Potential of Hydrocolloids: Possible New Applications? 349

Alginates in Relation to Human Health; from Commodity to High Cost

Subject Index 365

Members of the Organising Committee

Mr P Cowburn Danisco Ingredients Ltd

Realistic Market Potential

Source : IMR estimates from The Quarterly Review of Food Hydrocolloids

Arabic Notoriously unstable in supply. Is expected to have a steady decline in

Carrageenan Multi-functional hydrocolloid, good long term potential. The use of

CMC Low profitability hydrocolloid. Fairly mature in USfood applications with no

MCC Micro-crystalline cellulose is a unique hydrocolloid with some niche markets.