Anda di halaman 1dari 13

View Article Online / Journal Homepage / Table of Contents for this issue

Proteins at interfaces and in emulsions


Stability, rheology and interactions

Published on 01 January 1998. Downloaded on 08/04/2015 23:50:55.

Eric Dickinson
Procter Department of Food Science, University of L eeds, L eeds, UK L S2 9JT

Physical properties of oil-in-water emulsions stabilized by milk proteins are determined largely by the nature of the adsorbed
layer at the surface of the dispersed droplets. There are two distinct classes of protein : the disordered caseins (especially a -casein
s1
and b-casein) and the globular whey proteins (especially b-lactoglobulin). Substantial dierences exist between these two classes in
terms of adsorbed layer structure and surface rheological properties at the oil/water interface. Theoretical modelling of adsorbed
layers of a -casein and b-casein with a simple self-consistent-eld approach is useful for understanding the excellent stabilizing
s1
properties of the caseins, and in interpreting the aggregation behaviour of casein-based emulsions as a function of ionic strength
and pH. The creaming behaviour and droplet occulation are sensitive also to the concentration of non-adsorbed casein. In
systems containing milk protein and small-molecule surfactant, competitive adsorption has a strong inuence on orthokinetic
emulsion stability, and on the viscoelasticity of heat-set b-lactoglobulin-stabilized emulsion gels. Computer simulation of model
particle gel networks shows considerable promise for providing new insight into the relationship between interparticle interactions and the structure and rheology of emulsion gels.

In the welcoming address at the start of a recent dairy science


conference, the eminent head of the host institution reminded
his audience that they were witnessing the dawn of the great
age of molecular biology and genetic engineering. According
to this vision, most advances in food quality in the future will
come about, not from our improved physico-chemical understanding, nor from developments in manufacturing practice in
the factory, but rather from the genetic control of biochemical
processes within living animals, plants and micro-organisms.
Soon, apparently, it will become routine for appropriately
trained food scientists to tailor-make biopolymer and lipid
ingredients with precisely required nutritional and functional
properties. An inevitable consequence of the coming predominance of genetics-based research is that many of the physicochemical approaches carried out in the past will become
obsolete. According to this analysis, the era of making important progress by studying the properties of milk ingredients
using approaches from physical and colloid chemistry is
nearly over. As we move into the new millennium, then, the
obituary of food colloid science is already being written !
There is, of course, a substantial aw in this argument. Put
simply, it is that, in reality, we know nothing like enough
about the relationship between biomolecular structure and
ingredient functionality for even the most imaginative molecular biologist to be able to construct the ideal hypothetical
emulsier/stabilizer/gelling agent. At one extreme of scientic
activity, there is a fundamental world comprising theoretical
models of adsorption or self-assembly that apply reasonably
well to systems of, say, idealized copolymers or simple charged
amphiphiles. At the other extreme, there is the real world of
food systems containing multicomponent mixtures of heterogeneous biopolymers and particles that are all interacting
with one another. Bridging this gap remains an awesome task.
Far from being able to reduce the amount of attention
devoted to physical chemistry aspects, what we may actually
require is much more focused eort directed towards the statistical mechanics of complex biocolloidal systems. That is, as
the methods of molecular biology and genetic engineering
become more routine, the real challenge will be how to
improve our ability to predict the consequences of changes in
molecular composition for the microstructure and colloidal
stability of whole systems. Solving this problem will require

the combined application of many emerging experimental and


computational techniques to systems of much greater complexity than has hitherto been the case. An example of the
class of systems requiring such attention is to be found in this
article.
The stability and rheological properties of emulsions are
largely determined by the interactions between the droplets.
The nature and strength of the interactions are, in turn,
dependent on the structure and composition of the adsorbed
layer at the oil/water interface. In food colloids the stabilizing
layer around emulsion droplets is compositionally and structurally complex. This obviously makes it difficult to disentangle the relationship between the colloidal interactions and
the macroscopic emulsion properties.
From the technological point of view, an especially important class of food system is that containing dispersed oil or fat
droplets, in which the stabilizing layer is mainly composed of
proteins derived from dairy ingredients. Progress in understanding the properties of these real food products is being
made by studying simpler model systems containing pure milk
proteins adsorbed at macroscopic interfaces or at the surface
of emulsion droplets. This paper reports on some of the
experimental and theoretical advances that have occurred in
this eld since a previous Faraday Research Article appeared
on this topic.1
Milk proteins are widely valued for their emulsifying and
emulsion stabilizing properties.2h6 As individual molecules or
in the form of aggregates, the milk protein emulsier becomes
rapidly adsorbed at the new oil/water interface generated
during emulsication.7 The resulting steric stabilization protects the nely dispersed droplets against immediate recoalescence and, during subsequent storage, it provides long-term
stability towards creaming and occulation.5,6 The emulsion
stability is aected by changes in interdroplet interactions
caused by changes in the conformations and interactions of
the proteins during storage. The large negative free energy
change associated with protein adsorption means that the
process is eectively irreversible with respect to serial dilution
of the bulk phase.8 During or after emulsication, however,
some of the individual protein components may be partially
or wholly displaced from the interface by other more surfaceactive protein species or by small-molecule surfactants.9,10
J. Chem. Soc., Faraday T rans., 1998, 94(12), 16571669

1657

Published on 01 January 1998. Downloaded on 08/04/2015 23:50:55.

View Article Online

Hence, emulsion stability is strongly inuenced by the


dynamic aspects of protein competitive adsorption and by the
nature of the interfacial protein interactions.9h11
While dairy colloids do contain a large number of individual protein species, many of the proteins are only present in
very small amounts and so have little inuence on physicochemical properties. From the perspective of protein functionality, it is convenient to distinguish between the disordered
exible caseins and the compact globular whey proteins.
Casein in milk is strongly aggregated into polydisperse
protein particles (ca. 200 nm) called casein micelles ; in contrast, the whey proteins are completely soluble in the aqueous
serum. The casein micelles are colloidal particles of selfassociated casein held together by nanometre-sized clusters of
calcium phosphate and sterically stabilized by an outer layer
of the glycoprotein i-casein.12h14 Removing the calcium salts
from milk casein and replacing them by sodium salts leads to
the production of a widely used food ingredient called
sodium caseinate . This exists in aqueous solution at neutral
pH as a mixture of casein monomers and small casein aggregates (so-called sub-micelles ).15
The two major individual caseins, a -casein and b-casein,
s1
comprise about three-quarters of total bovine milk casein. The
globular b-lactoglobulin comprises about three-quarters of the
total whey protein. Since much relevant recent work has
involved just these three key pure proteins, they form the
main focus of this article.

Characteristics of a -casein and b-casein


s1
The molecules of a -casein and b-casein have much in
s1
common. Both are calcium-sensitive phosphoproteins containing a high proportion of proline but no cysteine along disordered polypeptide chains of ca. 2 ] 102 residues.16 Each
casein monomer carries a net negative charge at the pH of
milk (pH 6.7), and each is substantially amphiphilic in character, having extensive regions with a high proportion of hydrophobic side-chains. Both proteins have a tendency to
self-associate in solution and to be adsorbed at both hydrophilic and hydrophobic surfaces. Each gives a low interfacial
tension and a low surface shear viscosity at the oil/water interface,17 and each can be used as an eective stabilizer of oil-inwater emulsions.18
On the other hand, one should not underestimate the
obvious structural dierences between a -casein and b-casein.
s1
The hydrophilic/hydrophobic segment distribution is considerably less random in b-casein than in a -casein.10 A notes1
worthy feature of the non-uniform distribution of
hydrophobic residues in b-casein19 is the presence of a
sequence of ca. 50 segments at the N-terminus which is highly
hydrophilic and contains all ve phosphoserines, with the rest
of molecule (ca. 160 residues) being substantially hydrophobic
and carrying no net charge.16 The distinct amphiphilicity of
the b-casein molecule rather resembles that of a typical lowmolecular-weight water-soluble surfactant ; this representation
is consistent with the known reversible association of b-casein,
at concentrations above ca. 0.05 wt.%, into spherical
surfactant-type micelles.20,21 The separation of the hydrophilic and hydrophobic regions in a -casein is less distinct :
s1
the a -casein primary sequence has three main hydrophobics1
rich regions : one in the middle (at residues 90110) and two at
the ends (140 and 150199). At neutral pH, the net negative
charge on a -casein (ca. 22 e) is higher than that on b-casein
s1
(ca. 15 e). In contrast to the spherical micelles formed by bcasein, the more highly charged a -casein apparently forms
s1
chain-like aggregates through a series of consecutive association steps.20
The equilibrium free energy of the adsorbed protein at oil/
water and air/water interfaces is slightly higher for a -casein
s1
than for b-casein. This leads to preferential adsorption of b1658

J. Chem. Soc., Faraday T rans., 1998, V ol. 94

casein in mixed casein systems at low ionic strength.22h24


Available experimental evidence25h27 at a variety of surfaces
indicates that a -casein is mainly adsorbed at slightly lower
s1
surface coverage than b-casein, and that its layers are thinner.
Droplets and latex particles coated with a -casein are more
s1
highly charged than those coated with b-casein,18,28,29 and
emulsions stabilized by a -casein are considerably more suss1
ceptible to occulation by salts.18,29,30 Hence, b-casein is a
better steric stabilizer than a -casein ; this is probably because
s1
it is a more distinctly amphiphilic molecule.

Structure of adsorbed layers of b-casein


The structure of adsorbed layers of b-casein at various solid
and liquid surfaces has been investigated experimentally by
the techniques of ellipsometry,31h34 small-angle X-ray scattering,35 dynamic light scattering,36,37 and neutron reectivity.38h41 From these studies, there has emerged a general
consensus on two main points. First, b-casein is adsorbed as a
monolayer over a wide range of bulk protein concentrations.
Secondly, the protein distribution in the monolayer appears
highly non-uniform, with the hydrophobic region(s) of the
molecule remaining very close to the surface in a densely
packed inner layer, and the more highly charged region(s)
residing well away from the surface in a much thinner outer
layer. The question arises as to what are the detailed average
congurations of individual b-casein molecules adsorbed at
such hydrophobic surfaces. Of particular interest is the specic
nature of the steric stabilizing moiety which gives b-casein its
excellent emulsion stabilizing ability.
Measurements of hydrodynamic layer thickness by dynamic
light scattering have suggested36 that parts of the adsorbed
b-casein molecules on polystyrene latices extend from 10 to 15
nm from the hydrophobic surface. An analysis of the breakdown products of adsorbed b-casein due to proteolysis by
trypsin, combined with complementary data on changes in
layer thicknesses on latices and emulsion droplets, led Dalgleish and Leaver to speculate19,42,43 that the most likely conguration of adsorbed b-casein is one with a large loop of ca.
50 residues at the N-terminus, with the remaining segments of
the molecule residing mainly very close to the interface, in
trains, as illustrated in Fig. 1(a). For reasons associated with
the interpretation of experimentally observed dierences in
rates of enzymic digestibility of various parts of the adsorbed
protein, it was suggested42,43 that this looptrain model [Fig.
1(a)] is more plausible than an alternative conguration in

Fig. 1 Suggested representations of b-casein adsorbed at a polystyrene latex surface.37 (a) The looptrain model with a boundary
between the two regions at around residue number 50. (b) The bloband-spring model in which the blob represents the cluster of negatively charged residues around residue number 20 : (i) native protein at
low ionic strength, (ii) dephosphorylated protein at low ionic strength
or native protein at high ionic strength, and (iii) native protein with
bound calcium ions.

Published on 01 January 1998. Downloaded on 08/04/2015 23:50:55.

View Article Online

which the entire sequence of 50 N-terminal segments forms a


single dangling tail. This was despite the fact that it was also
recognized42,43 that the dangling tail is probably more consistent with the large hydrodynamic layer thickness. Arguments
favouring the looptrain model were developed further by
Brookbank et al.37 in discussing the eect of electrostatic
interactions on the adsorbed layer structure of b-casein on
latices. These same authors also introduced37 an appealing
alternative representation of adsorbed b-casein which they
called the blob-and-spring model [see Fig. 1 (b)]. In this
rather quaint phenomenological representation, the cluster of
negatively charged residues at positions 1525 is regarded as a
negatively charged blob that is repelled from the surface by
electrostatic forces but restrained from moving into solution
by a stretched spring section corresponding to the rest of the
loop [state (i) in Fig. 1(b)]. When the eective charge on the
blob becomes reduced (by dephosphorylation of the protein)
or screened (by addition of salt), the spring relaxes back to
give state (ii) corresponding to a thinner layer. Specic binding
of calcium ions to phosphoserine residues can also cause the
blob to move towards the surface [state (iii)], again implying a
thinning of the adsorbed layer.
To advance beyond the sort of intuitive model described
above, a more rigorous statistical description is required that
can take proper account of the balancing entropic and enthalpic contributions to the equilibrium polymer adsorption.44
For a exible linear homopolymer composed of identical hard
spherical segments in an athermal solvent, the train/loop/tail
distribution depends only on the surfacesegment adsorption
energy and the bulk polymer concentration.45 The situation is
potentially much more complicated for a complex heteropolymer like b-casein, however, since there are some twenty or
so dierent kinds of segments along the chain. In order to
keep the degree of chemical complexity within sensible
bounds, it seems reasonable46,47 to approximate the number
of types of segment in the model to just three : non-polar
(hydrophobic), charged (or potentially charged) and polar
(hydrophilic, uncharged). To make worthwhile predictions of
changes in adsorbed layer properties as a function of aqueous
solution conditions, an essential feature that clearly has to be
included is the pH dependence of the charge distribution
along the polyelectrolyte chain.
Statistical modelling of the adsorption of a b-casein-like
polymer at a hydrophobic plane surface has been carried
out48,49 using the self-consistent-eld (SCF) theory of Scheutjens and Fleer.44,50 In this lattice-based model of the polyelectrolyte solution, the segment potential is made up from a
combination of an excluded volume interaction, a non-bonded
nearest neighbour interaction of the FloryHuggins sparameter type, and an electrostatic interaction which
depends on segment density and the local electrical potential.
The values of the s-parameters are assigned so as to give a
strong surface binding affinity for the non-polar segments and
a net attraction between small mobile ions (Na` and Cl~) and
solvent molecules. Account is taken of the pK values of the
a
weakly charged groups by allowing the ionizing segments to
assume more than one internal state.48
The SCF theory predicts that adsorption of the b-caseinlike polymer produces a monolayer at bulk protein concentrations below a certain condensation limit.48,51 The condensation threshold is reduced to lower protein concentrations
with increasing ionic strength or decreasing pH. Fig. 2 shows
the predicted eect of pH on the segment density prole o(z)
at ionic strength 0.01 M as a function of distance z from the
surface for a bulk concentration (5 ] 10~6) below the condensation threshold. The prole indicates a dense inner layer
(z \ 2 nm) and a dilute extended outer region with a gradually
decreasing segment density reaching o B 1% at z B 10 nm.
Reduction in pH leads to increased surface coverage and
segment density in the range z B 25 nm. The shoulder on o(z)

Fig. 2 Inuence of pH on total segment density prole o(z) for bcasein-like polymer predicted from SCF theory at ionic strength 0.01
M and bulk concentration 5 ] 10~6.51

at pH 5.5 is indicative of the closeness to the condensation


limit.48 Average positions of dierent parts of the b-casein
molecule with respect to the surface are plotted in Fig. 3. As
expected, the region lying furthest away from the surface is the
sequence of 4050 segments at the N-terminus. Hence, the
picture of the adsorbed b-casein suggested by Fig. 3 is one
with about three-quarters of the molecule very close to the
surface in trains and small loops, and the rest in a highly
charged tail dangling well into the bulk aqueous phase [see
Fig. 4(a)]. Segment 1 at the extreme N-terminus resides at an
average distance of ca. 5 nm from the surface at pH 7, but this
distance is reduced to ca. 31 nm at pH 5.5. Repeating the
2
calculation for a hypothetical variant of the b-casein-like
polymer having the ve phosphoserines replaced by polar
(uncharged) residues gives a larger adsorbed amount but a
smaller full (hydrodynamic) layer thickness. This can be

Fig. 3 Mean positions of dierent parts of b-casein-like polymer


from surface as predicted by SCF theory at ionic strength 0.01 M
and bulk polymer concentration 2 ] 10~11 ; pH : (- - - -) 5.5, () 6.0,
( ), 7.0. One lattice unit is ca. 3 . Also shown are the charges
on the residues at pH 6.0 and the smoothed (negative) hydrophobicity
along the chain.49

J. Chem. Soc., Faraday T rans., 1998, V ol. 94

1659

View Article Online

Published on 01 January 1998. Downloaded on 08/04/2015 23:50:55.

Fig. 4 Sketches of representative adsorbed congurations of (a) bcasein and (b) a -casein
s1

understood48,51 in terms of the poorer solubility of the dephosphorylated polymer, and it conrms the importance of
the phosphoserine residues in b-casein in maintaining a thick
steric stabilizing layer whilst avoiding interfacial protein precipitation (i.e. multilayer formation).
Without attempting to optimize the choice of the model
parameters, there is generally good agreement between the
structure of the adsorbed b-casein-like polymer predicted from
the SCF theory and the information obtained experimentally
from the various experimental techniques. A powerful emerging technique for the study of interfacial structure is specular
neutron reection.38 Sets of reectivity data obtained for bcasein adsorbed at liquid39 and solid40 surfaces have been
analysed in terms of a two-layer model consisting of a thin
dense inner layer and a lower density outer layer.1 At pH
values in the range 5.57.0 at the air/water interface, the ts to
the data indicate41 a dense protein-rich inner layer (o B 95%)
and a more diuse outer layer (o B 15%) extending into the
bulk phase. Model-independent Guinier analysis shows52 an
increase in adsorbed amount and a thickening of the layer as
the pH is lowered from 7 to 5.5. Independent experimental
conrmation of this eect of pH on the protein layer coverage
and structure has been obtained in a recent surface plasmon
resonance study53 of b-casein adsorption onto a selfassembled monolayer of octadecyl mercaptan on gold.54,55
On tting the kinetic data to a simple double-exponential
function,56 the surface plasmon resonance technique has
demonstrated53 that the rate of adsorption of b-casein at the
hydrophobic solid surface at pH 5.5 is less than half that at
pH 7.0. The very dilute tail region, predicted from the SCF
theory for the adsorbed b-casein-like polymer, is consistent
with the large hydrodynamic layer thickness (1015 nm)
derived from dynamic light-scattering.25,36,37 Mobility of the
phosphoserine-rich tail region in adsorbed b-casein in emulsions has been conrmed by 31P NMR.57 Dephosphorylation
of b-casein has been found37,58 to lead to a reduction in
hydrodynamic layer thickness accompanied by a marginal
increase in adsorbed amount ; this again agrees qualitatively
with predictions from SCF theory.48

Interactions between casein layers : relationship to


stability
The same set of SCF parameters as used to model the structure of adsorbed b-casein has also been used to describe
adsorbed a -casein.49 Whereas, for b-casein, the N-terminus
s1
forms a long tail which dangles away from the surface [Fig.
4(a)], the mean distance prole predicted for a -casein indis1
cates a loop-like structure whose end is anchored to the
surface by a strongly hydrophobic patch at around residue
number 25. Hence, the inferred adsorbed a -casein structure,
s1
which is sketched in Fig. 4(b), actually rather resembles the
loop-like conformation suggested previously19,42 for b-casein
[Fig. 1(a)]. This conformation is consistent with the inaccessibility of peptide groups in a -casein at positions 2125 to
s1
proteolytic enzymes in emulsions.59 The maximum average
extension from the surface of the loop-like region at around
residue number 60 is below 3 nm at neutral pH. This implies
that the adsorbed a -casein-like polymer extends less far from
s1
the surface, on average, than the adsorbed b-casein-like
1660

J. Chem. Soc., Faraday T rans., 1998, V ol. 94

polymer. The prediction is consistent with the experimentally


determined hydrodynamic layer thickness for a -casein on
s1
latices, which is only about half that for b-casein.25 On
reducing the pH from 7 to 5.5, the predicted a -casein
s1
adsorbed amount increases, but the extension of the loop-like
region out from the surface is reduced.
So far, we have been considering just isolated casein layers.
But colloidal stability and rheology of systems of caseincoated particles are, of course, determined by the nature of
interactions between casein layers. With this in mind, the
same SCF theory has been used60 to calculate the interaction
free energy A(D) as a function of interlayer separation D for
pairs of a -casein or b-casein layers. The calculation assumes
s1
full thermodynamic equilibrium,61 with solvent, polymer and
small ions allowed to distribute themselves inside or outside
the gap in order to minimize the total free energy of the
system. For both of the pure caseins, at pH 5.57.0 and ionic
strengths 1100 mM, there was found60 to be a large positive
interaction energy (A [ 0.01 kT per lattice site) at close
separations (D O 5 nm), corresponding to strong interlayer
steric repulsion. This is illustrated in Fig. 5 for the conditions
of pH 6.0 and ionic strength 10 mM. The stronger repulsive
interaction for b-casein than for a -casein is probably due to
s1
the greater eective thickness of the b-casein tail conguration
[Fig. 4(a)] as compared with the a -casein loop conguration
s1
[Fig. 4(b)]. Direct experimental verication of the existence of
long-range repulsive forces between hydrophobized mica surfaces coated with b-casein has recently been obtained using
the interferometric surface force apparatus.62
The range of the predicted interlayer repulsion for both
casein-like polymers is reduced as the pH is lowered towards
the isoelectric point.60 Whereas the interaction energy A(D)
for b-casein remains positive for all D, irrespective of pH or
ionic strength, that for a -casein becomes negative for D [ 5
s1
nm under certain solution conditions (lower pH, higher ionic
strength). Fig. 6 shows a set of interaction potentials A(D) at
ionic strengths 0.080.12 M for a -casein layers at a constant
s1
pH of 5.5. With increasing salt concentration, there is an
increase in the depth of the potential well, A , and a shift to
well
closer separations of the position of the bottom of the well,
D . At constant pH the predicted value of A
increases
well
well
approximately linearly with ionic strength. The strength of the
attraction is reduced with increasing pH. It is interesting to
note that, under solution conditions equivalent to the aqueous
serum of milk (pH ca. 6.8, ionic strength ca. 0.08 M), the
numerical value of A
is not negligible, thereby implying a
well
considerable net attraction between a -casein layers in this
s1
natural aqueous environment.
The presence of a signicant attractive region in A(D)
implies the equilibrium occulation of particles interacting
with such a potential. Conversely, the condition A \ 0
well
implies stability with respect to occulation. Fig. 7 shows the
theoretical stability diagram (pH vs. ionic strength) based on

Fig. 5 Interaction potentials for layers coated with b-casein and a s1


casein as predicted by SCF theory at pH 6.0, ionic strength 0.01 M and
bulk polymer concentration 2 ] 10~11. The free energy A(D) (units of
kT per lattice site) is plotted against the interlayer separation D : (L),
a -casein ; (=), b-casein.60
s1

View Article Online

explained63 using a simple statistical model based on


reversible competitive adsorption and occulation of sticky
hard spheres.

Structure and surface rheology of adsorbed


b-lactoglobulin layers

Published on 01 January 1998. Downloaded on 08/04/2015 23:50:55.

Fig. 6 Interaction potentials for layers coated with a -casein as pres1


dicted by SCF theory at pH 5.5 and bulk polymer concentration
2 ] 10~11. The free energy A(D) (units of kT per lattice site) is plotted
against the interlayer separation D for dierent ionic strengths : (a)
0.08, (b) 0.09, (c) 0.10, (d) 0.11 and (e) 0.12 M.60

this criterion for particles coated with the a -casein-like


s1
polymer. For quantitative comparison with real colloidal
systems, an additional van der Waals attractive contribution
would have to be added to A(D) ; this would shift the position
of the boundary line to somewhat lower ionic strengths.
However, such a renement is not justied here since the position of the boundary is, in any case, sensitive to the precise
choice of parameters in the SCF model. The key point is that,
whereas b-casein-coated particles are predicted to be stable at
all ionic strengths, the corresponding a -casein-coated pars1
ticles are predicted to occulate at salt concentrations above
ca. 0.1 M. Impressively, this is what is actually observed18,29,30
in model oil-in-water emulsion systems : those prepared with
b-casein as sole emulsier are stable towards NaCl addition
([2 M) whereas those prepared with a -casein become extens1
sively occulated in 0.10.2 M NaCl.
Being present in milk in approximately equal amounts, a s1
casein and b-casein are the major functional components of
the commercially important ingredient sodium caseinate.
Emulsion droplets coated with this emulsifying agent have
good stability towards salt. Hence, one may assume that the
presence of an equivalent proportion of the eective steric stabilizer b-casein overwhelms the tendency of the a -casein to
s1
induce occulation at ionic strengths above ca. 0.1 M. The
situation is somewhat complicated by the competitive adsorption of the proteins, which causes b-casein to predominate in
the adsorbed layer to dierent extents, depending on the ionic
strength. Nevertheless, the strongly protective eect of bcasein has been unambiguously demonstrated30 in salt stability experiments on model emulsions made with well dened
mixtures of a -casein ] b-casein. The key factors controlling
s1
the limits of stability in this mixed system can also be

Fig. 7 Theoretical stability diagram for the interaction of a -casein


s1 replayers. Ionic strength I is plotted against pH. The boundary
resents the transition between potentials that are entirely repulsive (as
in Fig. 5) and potentials having an attractive region (as in Fig. 6).

In solution at neutral pH, the globular protein b-lactoglobulin


exists as a dimer held together by non-covalent interactions.
The b-lactoglobulin monomer (18.4 kDa) contains two intramolecular disulde linkages and a free hidden sulfhydryl
group. On adsorption at an air/water or oil/water interface,
while much of the ordered globular structure is retained, there
is also some macromolecular rearrangement and partial
denaturation.64 The adsorbed globular protein structure lies
somewhere intermediate between the native structure and the
fully denatured stateit has been suggested10 that the structure may resemble the so-called molten globule state65 which
has a native-like secondary structure and an unfolded tertiary
structure. At its simplest, the adsorbed b-lactoglobulin monolayer can be regarded66 as a pseudo-two-dimensional system
of densely packed deformable particles interacting strongly
through a combination of ionic, hydrophobic and hydrogen
bonds. Additionally, the partial unfolding of the globular
protein structure following adsorption causes exposure of the
reactive sulfhydryl group, leading to slow polymerization of
the adsorbed protein in the aged layer via sulfhydryldisulde
interchange.67
Information on the structure of layers of b-lactoglobulin
adsorbed at the air/water interface over the pH range 5.57.0
has been obtained by neutron reectivity experiments.41,68
The results indicate that the b-lactoglobulin monolayer thickness (23 nm) is substantially less than the b-casein layer
thickness (see above) measured under similar conditions. This
also agrees with analysis of earlier reectivity data69 for
another globular milk protein, bovine serum albumin (BSA).
An important dierence between b-lactoglobulin and b-casein
revealed by the reectivity measurements41,68 is the gradually
increasing surface coverage and layer thickness with elapsed
time (ca. 25 % over 45 h) following exposure of blactoglobulin to the interface. There is evidence of a change
from a two-step structure to a one-step structure as adsorbed
protein molecules gradually rearrange themselves at the interface. The rate of lm ageing and rearrangement appears to
increase as the pH is reduced, reecting changes in intermolecular interactions and surface denaturation kinetics as the
protein approaches its isoelectric point.
The mechanical properties of protein lms are expressed in
terms of their surface rheological properties.70,71 These may
be determined under conditions of shear or dilational deformation. The large structural dierence between an adsorbed
b-lactoglobulin layer and an adsorbed casein layer is most
sensitively reected in the surface shear rheology.11,72h74 For
instance, on adsorption at a hydrocarbon/water interface from
a 10~3 wt.% protein solution (pH 7, ionic strength 0.005 M),
the apparent surface shear viscosity g reached after 24 h has
s
been found11 to be [1 ] 103 mN s m~1 for b-lactoglobulin
as compared with g \1 mN s m~1 for b-casein, i.e. a diers
ence in g of more than three orders of magnitude. Fig. 8
s
shows a set of time-dependent surface shear viscosities for a
selection of dierent food proteins.11 The high surface rheological parameters for globular proteins such as b-lactoglobulin
and lysozyme can be attributed to the high packing density
and cross-linking of the globular protein molecules in the
adsorbed layer, as compared with rather loose packing for the
disordered a - and b-caseins. The surface dilatational elass1
ticity e of adsorbed b-lactoglobulin is also higher than that of
d
b-casein, but the dierence in e for the two proteins is much
d
less (i.e. below one order of magnitude).74,75 The surface dilatational properties are also inuenced by changes in the
J. Chem. Soc., Faraday T rans., 1998, V ol. 94

1661

View Article Online

An overview of observations of protein removal from solid


or liquid surfaces by various surfactants has indicated84 two
extreme types of mechanism :
(i) The solubilization mechanism
Desorption of protein from the surface arises as a result of
protein solubilization into the aqueous phase by water-soluble
surfactant which forms a proteinsurfactant complex. The surfactant does not itself have to adsorb at the surface, but it
must interact strongly with the adsorbed protein.

Published on 01 January 1998. Downloaded on 08/04/2015 23:50:55.

(ii) The replacement mechanism

Fig. 8 Time-dependent apparent surface shear viscosity g for proapp


teins adsorbing at the n-tetradecane/water interface from a bulk
aqueous phase containing 10~3 wt.% protein at neutral pH : (A)
myosin, (B) lysozyme, (C) j-casein, (D) gelatin, (E) sodium caseinate,
(F) a -casein, (G) b-casein, (H) b-lactoglobulin, (I) a-lactalbumin.
s1
(Reproduced with permission from ref. 11.)

nature of the aqueous sub-phase, e.g. through addition of


ethanol.76
An analogy can be drawn between the unfolding of blactoglobulin caused by the presence of a surface and the
unfolding that can be induced in bulk solution (e.g. by
heating). However, it is generally considered77 that molecules
adsorbing at the oil/water interface are unfolded to a greater
extent than heat-denatured ones in solution. The analogy has
its application in the reasonable proposition that an adsorbed
protein monolayer can be regarded as a very thin layer of
concentrated protein gel. This hypothesis has been tested
recently74 for b-casein and b-lactoglobulin using a simple
model to relate bulk and surface rheological properties. The
main assumptions made were (i) no change in conformation
on adsorption and (ii) homogeneity of the adsorbed layer with
respect to segment concentration and distribution. It was
found74 that a calculation from bulk properties for b-casein
leads to a value of e at the air/water interface of the right
d
order of magnitude, whereas a similar calculation for blactoglobulin leads to substantial underestimation of e . The
d
inference is74 that the b-lactoglobulin conformation is considerably inuenced by adsorption, leading to much stronger
interactions in the adsorbed layer than in the equivalent concentrated bulk solution.

Competitive adsorption of proteins with surfactants


In addition to milk proteins, many dairy-type food colloids
such as ice-cream or whipped toppings often contain smallmolecule surfactants such as polysorbates (Tweens) or monoglycerides (e.g. glycerol monostearate). It is established78 that
these surfactants are not necessary to produce a stable emulsion ; there is generally more than enough milk protein present
to act as emulsier. The primary role of the small-molecule
surfactant in ice-cream making, for instance, is to enhance the
amount of fat destabilization during the freezingaeration
process.79h83 The mechanism of the destabilization is related
to the displacement of milk proteins from the fat droplet
surface, which induces adsorption and partial coalescence
(clumping) of fat droplets at the surface of incorporated air
cells.79,80 Hence, the competitive adsorption of milk proteins
with surfactants is an important physico-chemical phenomenon in the processing of dairy colloids.
1662

J. Chem. Soc., Faraday T rans., 1998, V ol. 94

Competitive displacement of adsorbed protein arises because


the surfactantsurface interaction is stronger than the interaction between the protein (or proteinsurfactant complex)
and the surface. The surfactant does not actually have to
interact with the protein, but it must bind to the surface.
In practice, the actual mechanism will tend to lie between
these extremes, with competitive adsorption involving ionic
surfactants mainly proceeding via a solubilization mechanism,
and that involving non-ionic surfactants mainly proceeding
via replacement.
Competitive protein displacement from the oil/water interface has been observed85,86 directly by confocal laser scanning
microscopy of the system sodium caseinate ] monoglyceride
or lecithin, with the casein stained with the selective protein
dye Fluorescein isothiocyanate. Owing to optical eects
arising from dierences in refractive indices between oil and
aqueous phases,87 the adsorbed protein appears under the
confocal microscope as a bright line of considerably greater
width than the actual physical layer thickness.85 The progressive reduction in the brightness of this line has been
exploited85 in deducing the competitive displacement of
labelled protein from the interface by unlabelled monoglyceride.
The technique of specular neutron reectance has been
used52,68 to study the competitive displacement of b-casein or
b-lactoglobulin from the air/water interface by non-ionic surfactant hexaoxyethylene n-dodecyl ether (C E ). Results for
12 6
b-lactoglobulinC E at pH 6 and ionic strength 0.02 M are
12 6
shown in Fig. 9 in the form of plots of adsorbed amount C vs.
surfactant/protein molar ratio R. Two sets of experiments
were carried out on null-reecting water (8% D O in
2
H O) :68 (a) using the normal hydrogenated surfactant (h2
C E ) and (b) using a partially deuterated surfactant (d12 6
C E ) with hydrogens of the alkyl chain replaced by deute12 6
rium atoms. The non-deuterated surfactant has a very low
scattering length density, and it contributes little to the overall
reectivity ; this means that the set of data obtained for blactoglobulinh-C E essentially shows just the behaviour of
12 6
the protein. On the other hand, the set of data obtained with
b-lactoglobulind-C E gives the total amount of reective
12 6
material at the interface (i.e. protein ] surfactant). Hence, the
combined experiments can be used to determine separately the
adsorbed amounts of both protein and surfactant. Fig. 9(a)
shows the b-lactoglobulin surface coverage C(R) as a function
of molar ratio R based on the h-C E experiments. The tted
12 6
protein displacement curve is a (complementary) error function in log R. Fig. 9(b) shows the total surface coverage based
on the d-C E experiments, as well as the individual contri12 6
butions to C(R) from protein and surfactant. The tted surfactant adsorption curve is also an error function, and the
curve representing the total adsorbed amount is a sum of the
tted curves for protein and surfactant. An interesting feature
of the plot is that, at R B 0.3, when the amount of adsorbed
surfactant has almost saturated, there is apparently still a substantial amount of b-lactoglobulin at the interface compared
with what would be predicted from a simple theory of competitive polymer displacement.88,89

Published on 01 January 1998. Downloaded on 08/04/2015 23:50:55.

View Article Online

Fig. 9 Competitive adsorption of b-lactoglobulin (0.1 wt.%) and


non-ionic surfactant C E at the air/water interface at pH 6.0 and
ionic strength 0.02 M 12as 6determined by neutron reectance.68 (a)
Experiments with hydrogenated surfactant yield protein adsorbed
amount C as a function of surfactant/protein molar ratio R. (b)
Experiments with deuterated surfactant yield the total adsorbed
amount (in units of protein mass/area). Using the tted protein displacement curve from (a), the surfactant adsorption prole is adjusted
to give the total tted adsorption in (b).

Competitive adsorption in emulsion systems is typically


based on the so-called depletion method , whereby protein
surface coverage is inferred from the concentration of
unadsorbed protein in the serum phase after centrifugation.
This method has been used90,91 to demonstrate the complete
displacement of b-casein or b-lactoglobulin from the surface of
freshly made oil-in-water emulsions at pH 7 by non-ionic
water-soluble surfactants such as octaoxyethylene n-dodecyl
ether (C E ) or polyoxyethylene sorbitan monolaurate
12 8
(Tween 20). The surfactant/protein molar ratio required for

complete protein displacement (say R [ 20) is greater in emulsions than in planar interface studies (see Fig. 9) because of the
higher surface/volume ratio. The value of R required for complete displacement has been found88,89,92 to be higher for
ionic surfactants like sodium dodecyl sulfate (SDS) or sodium
lauryl ether sulfate (SLES) which form strong interfacial complexes with the adsorbed protein. On the other hand, oilsoluble surfactants such as sorbitan monooleate (Span 80) or
monoglycerides tend to be less eective at protein displacement.88,93h95 The competitive adsorption behaviour is also
sensitive to changes in protein structure induced by physical
processing. Ageing or heating of emulsions prepared with blactoglobulin reduces the ability of the surfactant to displace
the protein from the oil/water interface.9,94,96 An additional
eect on the competitive adsorption behaviour is the nature of
the oil phase, especially its polarity (hydrocarbon oil vs. triglyceride oil)97,98 and the state of crystallization.99
Surface shear rheology is very sensitive to the structure and
dynamics of adsorbed layers.11,100h102 Hence, it can provide a
valuable probe of the competitive adsorption of proteins with
surfactants and of interfacial proteinsurfactant interactions.
The displacement of b-lactoglobulin from the oil/water interface by non-ionic surfactant is associated with a drop in
apparent surface shear viscosity by several orders of magnitude.91,94,103 Although non-ionic surfactants generally interact weakly with proteins in solution, the existence of a
strongly hydrophobic binding site on b-lactoglobulin leads to
a specic 1 : 1 complex with Tween 20 which has implications
for the composition and properties of the adsorbed
layer.104h106 Eects of proteinsurfactant interactions are
even more important for anionic or cationic surfactants, to
such an extent that the surfactantprotein complexes may
have completely dierent surface rheological properties from
those of the pure proteins.103,107,108
The orthokinetic stability of an oil-in-water emulsion containing fully liquid droplets is predominantly determined by
the structure and interactions of the adsorbed protein layer.
The disruption of the adsorbed protein layer on addition of a
water-soluble surfactant (e.g. Tween 20) has been shown109 to
lead to a substantial reduction in coalescence stability of
model b-lactoglobulin-stabilized emulsions under turbulent
ow conditions. Oil droplets containing fat crystals are especially susceptible to partial coalescence when stirred, and in
dairy-type emulsions the nature of the fat crystal morphology
is generally the most important factor controlling the orthokinetic destabilization rate.110h112 Combined aeration and
freezing during ice-cream making causes the emulsion to
undergo partial coalescence, with clumps and clusters of fat
droplets forming an internal network in the frozen
product.81,82 The rate of fat destabilization is enhanced by the
presence of surfactant. Fig. 10 shows the eect of Tween 80
addition on the percentage of fat destabilized during whipping
of ice-cream mix in a barrel freezer.78 It has been shown113
that air incorporation or shearing alone (independent of ice
crystallization) does not lead to the same high rate of fat
destabilization as when whipping and freezing occur simultaneously.

Emulsion destabilization by non-adsorbed protein

Fig. 10 Inuence of non-ionic surfactant Tween 80 on the efficiency


of orthokinetic emulsion destabilization during aeration of ice-cream
mix in a batch freezer.78 The percentage of destabilized fat
(determined from turbidity measurements on 0.002 wt.% dispersions
of the ice-cream mix) is plotted as a function of whipping time : L,
with surfactant ; , without surfactant. Crystallization of ice begins
after ca. 5 min. (Reproduced with permission from ref. 82.)

In agreement with theories of depletion occulation by nonadsorbing polymers,114 it has been found6,115 that ne
casein-stabilized oil-in-water emulsions containing low concentrations of hydrocolloids such as xanthan gum or guar
gum exhibit poor creaming stability. Recently, it has also been
found116,117 that, in casein-stabilized emulsions containing
no added hydrocolloid, the presence of excess casein itself
induces a substantial reduction in creaming stability, presumably also as a result of a depletion occulation mechanism.
J. Chem. Soc., Faraday T rans., 1998, V ol. 94

1663

Published on 01 January 1998. Downloaded on 08/04/2015 23:50:55.

View Article Online

Justication for this interpretation comes from theoretical


treatments118,119 describing the reversible occulation of
large spheres (equivalent to casein-coated oil droplets) in the
presence of a high density of small spheres (equivalent to
casein sub-micelles).
Fig. 11 compares creaming data117 for two ne ntetradecane-in-water emulsions (35 vol% oil, 0.05 M phosphate
buer, pH 6.8) containing (a) 2 wt.% sodium caseinate and (b)
4 wt.% sodium caseinate. Both emulsions were prepared so as
to have the same average volumesurface droplet size d \
32
0.40 ^ 0.02 lm. The emulsions dier by more than a factor of
four in the concentration of non-adsorbed sodium caseinate
present in the aqueous continuous phase : (a) 0.8 wt.% and (b)
3.4 wt.%. Time-dependent gravity creaming proles of oil
volume fraction / as a function of height were determined
non-invasively at 30 C for each of these emulsions over a
quiescent storage period of approximately one month using
the technique of ultrasound velocity scanning.120 The dispersed phase volume fractions were calculated from measured
velocities using a renormalization technique121 which makes
allowance for contributions to the signal from scattering
eects. The plots show that the creaming stability of emulsion
sample (a) is very much better than that of emulsion sample
(b). The prole after 1 day in Fig. 11(a) still shows a uniform
oil content of / B 0.35 over the entire sample height, whereas
that after the same storage time in Fig. 11(b) shows extensive
phase separation. Even after 1 month the 2 wt.% caseinate
emulsion shows only a thin cream layer near the top of the
sample (ca. 20 mm thick) and no discernible serum separation
near the bottom [Fig. 11(a)]. This behaviour is characteristic
of a stable (unocculated) polydisperse emulsion of ne droplets, in contrast to the fast development of a distinct serum

layer in the 4 wt.% caseinate emulsion [Fig. 11(b)] which is


typical of systems exhibiting depletion occulation.122,123 At
the end of the 1 month experimental period, the droplet-size
distributions of both emulsions were remeasured ; it was
established117 that there was no signicant change in d
32
(^0.03 lm), indicating the absence of any irreversible aggregation or droplet coalescence.
Conrmation of the large inuence of non-adsorbed sodium
caseinate on the state of occulation of these emulsions has
come from light microscopy116,117 and small-deformation
rheology.117,124 Whereas the 2 wt.% caseinate emulsion is a
Newtonian liquid, the 4 wt.% emulsion is distinctly shearthinning with a low-stress apparent viscosity some two orders
of magnitude larger than that for the 2 wt.% caseinate system.
Consistent with a depletion occulation mechanism is the
complete reversibility of the aggregation towards dilution.
Assuming that sodium caseinate is composed of spherical particles of diameter ca. 10 nm at a volume fraction of a few per
cent, we can estimate from simple hard sphere theory118,125 a
depletion interaction of the order of 5 kT ; this is of the correct
magnitude to produce reversible occulation of the type
observed in these casein-based emulsions. (Qualitatively
similar depletion interactions have also been reported
recently126 for bidisperse silica systems with a large particle
size ratio.) The occulation behaviour of the caseinatestabilized emulsions, and its associated inuence on creaming
and rheology, has been found to be rather insensitive to ionic
strength (up to 0.5 M NaCl), but it does exhibit sensitivity to
calcium ion concentration (P5 mM CaCl )127 and to ethanol
2
content (P20 vol%).128 These latter eects can be attributed
to changes in the state of self-assembly of the caseinate due,
respectively, to the binding of Ca2` to phosphoserine residues
on a - and b-caseins, and to the reduction in aqueous phase
s1
solvent quality as a result of alcohol addition. Further
research is required to determine the relative contributions of
the unadsorbed a - and b-caseins to the depletion occus1
lation behaviour of these caseinate-stabilized emulsions. Preliminary results at Leeds on the rheology of concentrated
emulsions made with well dened mixtures of a -caseinbs1
casein suggest that the state of occulation is very sensitive to
protein emulsier composition and to temperature in the
range 040 C.

Emulsion gels and computer-simulated particle gels

Fig. 11 Inuence of sodium caseinate emulsier concentration on


the time-dependent creaming proles of caseinate-stabilized emulsions
(35 vol.% n-tetradecane, pH 6.8, d B 0.4 lm) stored at 30 C. Local
32
oil volume fraction / is plotted against
height. (a) Emulsion containing 2 wt.% protein : (K) 1 day, (>) 9 days, (|) 17 days, (), 34 days.
(b) Emulsion containing 4 wt.% protein : (K) 6 h, (>) 16.5 h, (|) 18 h,
() 21 h, (L) 24 h, (+) 27 h, ( ) 31 days. (Reproduced with permission from ref. 117.)

1664

J. Chem. Soc., Faraday T rans., 1998, V ol. 94

A gel is properly characterized129 rheologically by (i) a


storage modulus G@ which exhibits a pronounced plateau
region extending to timescales at least of the order of seconds
and (ii) a loss modulus GA which is considerably smaller than
G@ in the plateau region. This type of rheological behaviour
usually requires the constituent particles and/or macromolecules to be connected together into some sort of semipermanent three-dimensional network structure. A gel formed
predominantly from aggregated colloidal particles is called a
particle gel. Fig. 12 shows a representation of a concentrated
particle gel generated in two dimensions by computer simulation. Such a structure may provide a useful description of a
monolayer of adsorbed globular protein molecules crosslinked at a uid interface.
A protein-stabilized emulsion gel is a particle gel composed
of aggregated protein-coated oil droplets. A stable liquid-like
emulsion may be converted into an emulsion gel by causing
the protein-coated droplets to become joined together into
some sort of network structure. A depletion occulated emulsion containing excess unadsorbed sodium caseinate (see
above) is an example of a very weak emulsion gel produced by
indirect dropletdroplet interactions. Most proper emulsion
gels have more well dened interfacial proteinprotein connections between the emulsion droplets. At the molecular level

Published on 01 January 1998. Downloaded on 08/04/2015 23:50:55.

View Article Online

Fig. 12 A network of aggregated particles in two dimensions produced by computer simulation. The particle areal packing fraction is
ca. 50%. (Assuming each particle represents a single protein molecule,
this cross-linked planar structure may be a useful model of a gel-like
adsorbed globular protein layer.)

there may be numerous contributions to the dropletdroplet


interactions arising from direct chemical cross-links (covalent
bonds) and various kinds of physical bonds (van der Waals
forces, hydrogen bonding, electrostatic forces, hydrophobic
interactions).
A moderately concentrated liquid-like emulsion, stabilized
by an adsorbed layer of a globular protein like blactoglobulin, may be simply converted into an emulsion gel
by heating130,131 or high-pressure treatment.132,133 The
association of the droplets is triggered by the denaturation
and subsequent aggregation of protein adsorbed at the surface
of the droplets and in the continuous phase between droplets.
Experiments with model systems containing added surfactants
have shown108,134h138 that the resulting rheology of the heatset whey protein emulsion gel is very sensitive to the nature of
the proteinsurfactant interactions before, during and after
gelation. For instance, complexation of added anionic surfactant DATEM (diacetyl tartaric ester of monoglyceride)
with adsorbed and non-adsorbed b-lactoglobulin leads136 to a
substantial reinforcement of the heat-set emulsion gel structure and hence to a large positive eect on the elastic
modulus. This contrasts with the more complicated
behaviour135,136 arising from addition of non-ionic Tween 20
to freshly prepared b-lactoglobulin-stabilized emulsions prior
to thermal processing. At low Tween 20 contents (up to
R B 1) there is an increase in G@ attributable to 1 : 1 protein
surfactant complexation.104 With more surfactant added (to
R B 2) there is a sharp reduction in G@ ; this is attributed to
partial displacement of protein from the oil/water interface,91,99 leading to emulsion droplets that no longer contribute to the mechanical structure of the heat-set protein gel.
Further addition of surfactant may lead to the modulus
increasing again or remaining low, depending on the protein
content.135,136 In further contrast to the above, data obtained
recently108 for anionic SDS added to b-lactoglobulin emulsions prior to heating show a distinct maximum in G@ at R B 4
as indicated in Fig. 13. The insensitivity of G@ to surfactant
addition up to R B 2 can be attributed to specic SDS
binding to a small number of high-affinity protein binding
sites.139,140 The large increase in G@ at higher SDS concentrations is due to cooperative proteinsurfactant binding, which
causes extensive unfolding of the b-lactoglobulin molecules
and, hence, the formation of a greatly enhanced number of
mechanically important cross-links in the heat-set denatured
protein gel. Further addition of SDS (up to R P 10) causes a

Fig. 13 Inuence of anionic surfactant SDS on the elasticity of heatset b-lactoglobulin-stabilized emulsion gels (38 wt.% n-tetradecane,
58 wt.% protein, pH 7.0). Storage shear modulus G@ at 1 Hz and
30 C is plotted against surfactant/protein molar ratio R for three different b-lactoglobulin concentrations : (=) 5, (K) 6 and (>) 8 wt.%.
(Reproduced with permission from ref. 108.)

drastic lowering of the modulus again. This can be attributed,140 at least in part, to protein displacement from the oil/
water interface (cf. the behaviour of Tween 20). But another
contributing factor is the incorporation of denatured protein
into non-associating mixed proteinsurfactant micelles, so
that the resulting b-lactoglobulin aggregates become highly
compartmentalized and hence unavailable for network formation.
Gel-like characteristics may be induced in a concentrated
globular protein-stabilized emulsion without heating through
the addition of a second polymer that has an attractive electrostatic interaction with the adsorbed protein surface and so
can form polymer bridges between adjacent droplets.141
Bridging occulation is strong when the added polymer is of
opposite charge to the adsorbed protein (e.g. cationic gelatin
added to b-lactoglobulin-coated droplets at neutral pH)140
and relatively weak when the added polymer is of the same
charge as the adsorbed protein (e.g. anionic dextran sulfate or
i-carrageenan added to BSA-coated droplets at neutral
pH).142,143 An emulsion gel with covalent cross-links between
protein-coated droplets may be induced by treating the emulsion with the enzyme transglutaminase.144h146 At large deformations, the modulus of the enzyme-set emulsion gel
(containing chemical cross-links) increases with strain,
whereas that for the equivalent heat-set emulsion gel (with
predominantly physical bonds) has qualitatively the opposite
behaviour.140,144
Improved understanding of the quantitative relationship
between interparticle interactions and the rheology of emulsion gels can be derived from simple models of network structures formed from idealized systems of aggregating
particles.147 Recent progress has been made by a consideration of the properties of idealized model particle gels generated by computer simulation.148h151 An important aspect of
the data treatment is the use of fractal scaling analysis150 to
describe the intermediate-range microstructure. While it is the
(semi-)permanent cross-links formed during aggregation that
are mainly responsible for holding together the network structure in a strong elastic particle gel, the structure also depends
on the nature of weak reversible forces that act between the
particles before and during the formation of the cross-links.
These combined features can be taken account of in a Brownian dynamics simulation model152 which incorporates exible
irreversible cross-links, as well as non-bonded particleparticle
interactions. Such a model was used to generate the twoJ. Chem. Soc., Faraday T rans., 1998, V ol. 94

1665

Published on 01 January 1998. Downloaded on 08/04/2015 23:50:55.

View Article Online

dimensional particle gel illustrated in Fig. 12. This type of


simulation in both two and three dimensions has
demonstrated147,150 how the degree of heterogeneity of the
particle gel network depends on the balance between the rates
of particle cross-linking and local phase separation under the
inuence of attractive non-bonded interactions. In agreement
with general experimental observations, the frequencydependent small-deformation rheology has been shown153,154
to be mainly inuenced by the overall density of particle
particle cross-links, and to be rather insensitive to the large-scale
microstructure. On extension to large-strain deformations, however, the rheology does become quite sensitive
to the detailed microstructure. When allowance is made for
bond breakage as well as bond formation, the interparticle
stress response reveals a distinctive stress overshoot characteristic of viscoelastic materials.155
Fig. 14 shows a set of snapshots of particle gel structures
simulated from a three-dimensional model of soft spherical
monodisperse particles aggregating via exible surface-tosurface bonds.153 In the version of the model used for these
large-strain Brownian dynamics simulations,155 any bond
once formed is allowed to break again if the particleparticle
force exceeds a certain limiting value. The basic simulation
cell contains N \ 1000 particles of diameter p at a volume
fraction of / \ 0.05. Values of the other parameters and a
detailed description of the simulation methodology are given
elsewhere.153,155 The conguration in Fig. 14(a) is the particle
gel structure formed by Brownian aggregation of particles in
the absence of any applied stress. No bond breaking occurs
under these conditions. The eective fractal dimensionality of
the resulting network is d \ 1.9 ^ 0.05 over the distance
f
range 25 p, based on the slope of a plot of log n(r) against log
r, where n(r) is the average number of particle centres lying
within a distance r from a given particle.150 Fig. 14(b) shows
the same particle gel, initially formed by Brownian aggregation, but then subjected to a constant shear stress of magnitude S \ 0.2 (in reduced units).155 The network structure is
now slightly dierent in detail from that in Fig. 14(a), but it
still contains a single interconnected aggregate and has the
same value of the fractal dimensionality. In contrast, however,
the snapshot in Fig. 14(c) is rather dierent. This simulated
network was formed156 with the aggregation/gelation process
itself taking place under conditions of constant shear stress (of
magnitude S \ 0.2). The network has a coarser microstructure, larger pores and no signicant fractal scaling region.
These results156 illustrate the eect that shear ow during
gelation can have in causing local aggregate rearrangements
which can then have a large inuence on the nal particle gel
microstructure.
A slight elaboration of the simulation approach described
above is to consider the formation of a particle gel from a
mixture of two dierent kinds of aggregating particles. A wide
variety of network structures can be formed in two dimensions,157 depending on the relative values of the dierent
simulation parameters, and it has been suggested158 that this
type of model may be useful in understanding properties of
mixed protein lms at interfaces. In three-dimensional simulations of binary mixtures of hard particles of dierent sizes,154
this type of model can provide a starting point for predicting
the rheology of lled protein gels such as heat-set blactoglobulin-stabilized emulsion gels. The mixed particle gel
simulations indicate154 that small ller particles give a larger
increase in elastic modulus than do larger ller particles ; the
same general trend has also been found experimentally with
protein-based emulsion gels.134,159 Another prediction in reasonable agreement with experiment is that systems containing
ller particles that do not interact with the network have little
eect on the small-deformation rheology or on the kinetics of
gelation. A deciency of this type of model, though, is that it
treats all the particles as hard spheres. Therefore it cannot
1666

J. Chem. Soc., Faraday T rans., 1998, V ol. 94

Fig. 14 Congurations of particles in an aggregated network formed


by Brownian dynamics simulation of a three-dimensional model
system of exibly bonded soft-sphere particles.153 The snapshots show
the basic simulation cell (containing N \ 103 particles) and one half of
the adjacent image boxes (i.e. a total of 4N particles). (a) Particle gel
formed by Brownian diusion with fractal dimensionality d B 1.9. (b)
f
As (a), except particle gel is subjected to moderate shear stress
after
formation. (c) As (a), except system is subjected to moderate shear
stress during gelation.

aspire to describe the known inuences of particle rigidity or


droplet deformability on the mixed gel rheology.160
An issue of current interest150,161 is the extent to which a
single parameter such as the fractal dimensionality can be
useful for describing the structure and rheology of milk
protein gels formed made from caseinate162 or blactoglobulin.163 In the opinion of this author, based on
recent computer simulations of particle gelation, the existence
of a fractal scaling regime has clearly been established in
certain aggregated network structures,150 but other gels
formed with dierent interaction parameters also appear
essentially non-fractal. To be able to describe fully the

View Article Online

complex gel microstructure, as well as properties such as


porosity and viscoelasticity, it would seem reasonable to
assume that a more detailed description than just a single
fractal parameter is needed. Further simulations using dierent particle gelling models should contribute towards a
resolution of this issue.

Published on 01 January 1998. Downloaded on 08/04/2015 23:50:55.

Conclusions
This article has described recent progress in understanding the
relationship between the properties of adsorbed protein layers
and the stability and rheology of protein-based oil-in-water
emulsion systems. Emphasis here on the surface properties of
the major milk proteins (a -casein, b-casein and bs1
lactoglobulin) seems justied for a number of reasons : (i) the
interesting molecular contrast between the disordered caseins
and the globular whey proteins, (ii) the emergence of new
experimental data on adsorbed layer structure and surface
rheology for these pure proteins, (iii) the challenge posed by
the dierent theoretical predictions of interlayer interaction
potentials for the two dierent caseins, and (iv) the importance
of these proteins in relation to the processing of dairy-based
food colloids.
Based on recent progress, there is good reason to expect
that research in this eld will continue to benet from
advances in instrumental techniques for the study of macromolecular structure and dynamics at wet surfaces, and from
developments in the numerical modelling of complex
assemblies of surfactants and biopolymers. Further steps
towards the much greater complexity of real food systems can
be explored by extending these powerful experimental and
computational techniques to mixed proteinsurfactant
systems.
Milk protein-stabilized emulsions are susceptible to changes
in physical properties on quiescent storage or when subjected
to stirring. They can be converted into emulsion gels by
various processing methods such as heating, high-pressure
treatment, addition of polymers, or enzymic cross-linking. An
important challenge for us now is to be able to relate the
structural and rheological characteristics of these colloidal
systems to the nature of the interdroplet interactions
occurring during and after gelation. One powerful approach
towards establishing the complex link between interactions
and rheology is through computer simulation of particle gel
networks. The success of recent comparisons between simulation and experiment indicates that some encouraging progress
is beginning to be made.
Nevertheless, important issues do remain unresolved. First,
despite some exciting new advances, we still know little about
the molecular congurations of adsorbed proteins. Much
more detailed work is required using a wide range of spectroscopic and scattering techniques in order to establish the
structure of adsorbed proteins at oil/water and air/water interfaces to the same level of detail as has been established for the
structure of native proteins in solution. Secondly, we need to
know a lot more about the congurational structure of individual proteins in mixed lms containing surfactants and
other proteins. Thirdly, there is still disappointingly scant
understanding of the relationship (if any) between surface
rheological properties of adsorbed protein layers and stability
of protein-stabilized emulsions. This is not because of any
shortage of data in the literature on the static and dynamic
properties of pure protein layers or mixed proteinlipid lms,
far from it. What is lacking in most cases is any clear indication of useful quantitative links between the surface lm measurements and the colloidal stability properties. Without the
establishment of such links, a sceptical food scientist would be
justied in questioning the practical value of such data.
Fourthly, there is the question of the structural nature of the

cross-over spatial region between surface layer and bulk


phase. In attempting to interpret the stability and rheological
behaviour of emulsions containing high concentrations of proteins and surfactants, we are often impeded by the absence of
reliable information on the detailed self-assembly properties of
the species in the bulk aqueous phase. Specically in connection with dairy emulsion stability, it would seem to this
author that much more experimental and computational
eort should be directed towards understanding the selfassembly properties of concentrated solutions of mixed
caseins.
These problems cannot be solved quickly. Most protein
surface properties are sensitive to the age of the adsorbed
layer, and no two proteins behave in the same way even under
identical solution conditions. There is no reason to suppose
that the functional properties of new proteins, as produced by
methods of genetic or enzymic modication, will be any easier
to explain than those of the existing milk proteins. Computer
simulation will certainly oer exciting opportunities for modelling the behaviour of hypothetical proteins and peptides
before actually making them, but it will still be necessary to
test such predictions against careful experimental measurements using a wide range of physico-chemical techniques. For
these reasons, surely the gloomy prophesy of the imminent
demise of physico-chemical research at the expense of genetic
engineering will fail to be realized ?
The author acknowledges BBSRC for continued support to
his research group over the past several years.
References
1 E. Dickinson, J. Chem. Soc., Faraday T rans., 1992, 88, 2973.
2 C. V. Morr, in Developments in Dairy Chemistry1, ed. P. F.
Fox, Applied Science, London, 1982, p. 375.
3 D. M. Mulvihill and P. F. Fox, in Developments in Dairy
Chemistry4, ed. P. F. Fox, Elsevier Applied Science, London,
1989, p. 131.
4 J. F. Zayas, Functionality of Proteins in Food, Springer, Berlin,
1997, ch. 3.
5 D. G. Dalgleish, T rends Food Sci. T echnol., 1997, 8, 1.
6 E. Dickinson, J. Dairy Sci., 1997, 80, 2607.
7 P. Walstra and I. Smulders, in Food Colloids : Proteins, L ipids
and Polysaccharides, ed. E. Dickinson and B. Bergensta- hl, Royal
Society of Chemistry, Cambridge, 1997, p. 367.
8 E. Dickinson and D. J. McClements, Advances in Food Colloids,
Blackie, Glasgow, 1995, ch. 2.
9 J. Chen, E. Dickinson and G. Iveson, Food Struct., 1993, 12, 135.
10 E. Dickinson and Y. Matsumura, Colloids Surf. B, 1994, 3, 1.
11 B. S. Murray and E. Dickinson, Food Sci. T echnol. Int. (Jpn.),
1996, 2, 131.
12 D. G. Schmidt and T. A. J. Payens, in Surface Colloid Science,
ed. E. Matijevic, Wiley, New York, 1976, vol. 9, p. 165.
13 P. Walstra, J. Dairy Sci., 1990, 73, 1965.
14 C. Holt and D. S. Horne, Neth. Milk Dairy J., 1996, 50, 85.
15 L. K. Creamer and G. P. Berry, J. Dairy Res., 1975, 42, 169.
16 H. E. Swaisgood, in ref. 2, p. 1.
17 E. Dickinson, J. Dairy Res., 1989, 56, 471.
18 E. Dickinson, R. H. Whyman and D. G. Dalgleish, in Food
Emulsions and Foams, ed. E. Dickinson, Royal Society of Chemistry, London, 1987, p. 40.
19 D. G. Dalgleish and J. Leaver, J. Colloid Interface Sci., 1991,
141, 288.
20 D. G. Schmidt, in ref. 2, p. 61.
21 E. Leclerc and P. Calmettes, Phys. Rev. L ett., 1997, 78, 150.
22 E. Dickinson, S. E. Rolfe and D. G. Dalgleish, Food Hydrocoll.,
1988, 2, 397.
23 E. Dickinson, J. Food Eng., 1994, 22, 59.
24 K. Anand and S. Damodaran, J. Agric. Food Chem., 1996, 44,
1022.
25 D. G. Dalgleish, Colloids Surf. B, 1993, 1, 1.
26 D. V. Brooksbank, J. Leaver and D. S. Horne, J. Colloid Interface Sci., 1993, 161, 38.
27 D. G. Dalgleish and Y. Fang, in Food Proteins : Structure and
Functionality, ed. K. D. Schwenke and R. Mothes, VCH, Weinheim, 1993, p. 255.

J. Chem. Soc., Faraday T rans., 1998, V ol. 94

1667

View Article Online

28
29
30
31
32
33
34
35
36
37

Published on 01 January 1998. Downloaded on 08/04/2015 23:50:55.

38

39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67

1668

D. G. Dalgleish, E. Dickinson and R. H. Whyman, J. Colloid


Interface Sci., 1985, 108, 174.
E. Dickinson, M. G. Semenova and A. S. Antipova, Food
Hydrocoll., accepted.
H. Casanova and E. Dickinson, J. Agric. Food Chem., 1998, 46,
72.
J. Benjamins, J. A. de Feijter, M. J. A. Evans, D. E. Graham and
M. C. Phillips, Faraday Discuss. Chem. Soc., 1975, 59, 218.
B. W. Morrissey and C. C. Han, J. Colloid Interface Sci., 1978,
65, 423.
T. Nylander and M. Wahlgren, J. Colloid Interface Sci., 1994,
162, 151.
T. Kull, T. Nylander, F. Tiberg and N. M. Wahlgren, L angmuir,
1997, 13, 5141.
A. R. Mackie, J. Mingins and A. N. North, J. Chem. Soc.,
Faraday T rans., 1991, 87, 3043.
D. G. Dalgleish, Colloids Surf., 1990, 46, 141.
D. V. Brooksbank, C. M. Davidson, D. S. Horne and J. Leaver,
J. Chem. Soc., Faraday T rans., 1993, 89, 3419.
J. Penfold, R. M. Richardson, A. Zarbakhsh, J. R. P. Webster,
D. G. Bucknall, A. R. Rennie, R. A. L. Jones, T. Cosgrove, R. K.
Thomas, J. S. Higgins, P. D. I. Fletcher, E. Dickinson, S. J.
Roser, I. A. McLure, A. R. Hillman, R. W. Richards, E. J.
Staples, A. N. Burgess, E. A. Simister and J. W. White, J. Chem.
Soc., Faraday T rans., 1997, 93, 2847.
E. Dickinson, D. S. Horne, J. S. Phipps and R. M. Richardson,
L angmuir, 1993, 9, 242.
G. Fragneto, R. K. Thomas, A. R. Rennie and J. Penfold,
Science, 1995, 267, 657.
P. J. Atkinson, E. Dickinson, D. S. Horne and R. M. Richardson, J. Chem. Soc., Faraday T rans., 1995, 91, 2847.
J. Leaver and D. G. Dalgleish, Biochim. Biophys. Acta, 1999,
1041, 217.
D. G. Dalgleish and J. Leaver, in Food Polymers, Gels and Colloids, ed. E. Dickinson, Royal Society of Chemistry, Cambridge,
1991, p. 113.
G. J. Fleer, M. A. Cohen Stuart, J. M. H. M. Scheutjens, T.
Cosgrove and B. Vincent, Polymers at Interfaces, Chapman and
Hall, London, 1993.
M. C. Bujan-Nun8 ez and E. Dickinson, J. Chem. Soc., Faraday
T rans., 1996, 92, 2275.
D. E. Graham and M. C. Phillips, J. Colloid Interface Sci., 1979,
70, 427.
E. Dickinson and S. R. Euston, Adv. Colloid Interface Sci., 1992,
42, 89.
F. A. M. Leermakers, P. J. Atkinson, E. Dickinson and D. S.
Horne, J. Colloid Interface Sci., 1996, 178, 681.
E. Dickinson, D. S. Horne, V. J. Pineld and F. A. M. Leermakers, J. Chem. Soc., Faraday T rans., 1997, 93, 425.
M. A. Cohen Stuart, G. J. Fleer, J. Lyklema, W. Norde and
J. M. H. M. Scheutjens, Adv. Colloid Interface Sci., 1991, 34, 477.
P. J. Atkinson, E. Dickinson, D. S. Horne, J. Leaver, F. A. M.
Leermakers and R. M. Richardson, in ref. 7, p. 217.
E. Dickinson, D. S. Horne and R. M. Richardson, Food Hydrocoll., 1993, 7, 497.
T. M. Flynn, Ph.D. thesis, University of Leeds, 1997.
C. D. Bain and S. D. Evans, Chem. Br., 1995, 31, 46.
S. D. Evans, T. M. Flynn and A. Ulman, L angmuir, 1995, 11,
3811.
T. A. Morton, D. G. Myszka and I. M. Chaiken, Anal. Biochem.,
1995, 227, 176.
L. C. ter Beek, M. Ketelaars, D. C. McCain, P. E. A. Smulders,
P. Walstra and M. A. Hemminga, Biophys. J., 1996, 70, 2396.
F. A. Husband, P. J. Wilde, A. R. Mackie and M. J. Garrood, J.
Colloid Interface Sci., 1997, 195, 77.
M. Shimizu, A. Ametani, S. Kaminogawa and K. Yamauchi,
Biochim. Biophys. Acta, 1986, 869, 259.
E. Dickinson, V. J. Pineld, D. S. Horne and F. A. M. Leermakers, J. Chem. Soc., Faraday T rans., 1997, 93, 1785.
O. A. Evers, J. M. H. M. Scheutjens and G. J. Fleer, Macromolecules, 1991, 24, 5558.
T. Nylander and N. M. Wahlgren, L angmuir, 1997, 13, 6219.
E. Dickinson, J. Chem. Soc., Faraday T rans., 1997, 93, 2297.
E. Dickinson, in Interactions of Surfactants with Polymers and
Proteins, ed. E. D. Goddard and Ananthapadmanabhan, CRC
Press, Boca Raton, Florida, 1993, p. 295.
H. Hirose, T rends Food Sci. T echnol., 1993, 4, 48.
J. A. de Feijter and J. Benjamins, J. Colloid Interface Sci., 1982,
90, 289.
E. Dickinson and Y. Matsumura, Int. J. Biol. Macromol., 1991,
13, 26.

J. Chem. Soc., Faraday T rans., 1998, V ol. 94

68

69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106

107

D. S. Horne, P. J. Atkinson, E. Dickinson, V. J. Pineld and


R. M. Richardson, in Proceedings of 5th Conference of Food Engineering, ed. G. V. Barbosa-Canovas, S. Lombardo, G. Narsimhan
and M. R. Okos, American Institute of Chemical Engineers,
New York, 1997, p. 111.
A. Eaglesham, T. M. Herrington and J. Penfold, Colloids Surf.,
1992, 65, 9.
E. H. Lucassen-Reynders, Food Struct., 1993, 12, 1.
B. Warburton, in T echniques in Rheological Measurement, ed.
A. A. Collyer, Chapman & Hall, London, p. 55.
J. Castle, E. Dickinson, B. S. Murray and G. Stainsby, ACS
Symp. Ser., 1987, 343, 118.
E. Dickinson, ACS Symp. Ser., 1991, 448, 114.
F. J. G. Boerboom, A. E. A. de Groot-Mostert, A. Prins and T.
van Vliet, Neth. Milk Dairy J., 1996, 50, 183.
A. Williams and A. Prins, Colloids Surf. A, 1996, 114, 267.
M. R. Rodr guez Nin8 o, P. J. Wilde, D. C. Clark, F. A. Husband
and J. M. Rodr guez Patino, J. Agric. Food Chem., 1997, 45,
3010.
J. Lefebvre and P. Relkin, in Surface Activity of Proteins, ed. S.
Magdassi, Marcel Dekker, New York, 1996, p. 181.
H. D. Go, M. Libo, W. K. Jordan and J. E. Kinsella, Food
Microstruct., 1987, 6, 193.
H. D. Go and Jordan, J. Dairy Sci., 1989, 72, 18.
N. M. Barfod, N. Krog, G. Larsen and W. Buchheim, Fat Sci.
T echnol., 1991, 93, 24.
J-L. Gelin, L. Poyen, J-L. Courthaudon, M. Le Meste and D.
Lorient, Food Hydrocoll., 1994, 8, 299.
H. D. Go, J. Dairy Sci., 1997, 80, 2620.
B. M. C. Pelan, K. M. Watts, I. J. Campbell and A. Lips, J.
Dairy Sci., 1997, 80, 2631.
M. Bos, T. Nylander, T. Arnebrant and D. C. Clark, in Food
Emulsiers and T heir Applications, ed. G. L. Hasenhuettl and
R. W. Hartel, Chapman & Hall, New York, 1997, p. 95.
I. Heertje, J. Nederlof, H. A. C. M. Hendrickx and E. H.
Lucassen-Reynders, Food Struct., 1990, 9, 305.
I. Heertje, H. van Aalst, J. C. G. Blonk, A. Don, J. Nederlof and
E. H. Lucassen-Reynders, L ebensm. W iss. T echnol., 1996, 29,
217.
B. E. Brooker, in New Physico-chemical T echniques for the
Characterization of Complex Food Systems, ed. E. Dickinson,
Blackie, Glasgow, 1995, p. 53.
J. A. de Feijter, J. Benjamins and M. Tamboer, Colloids Surf.,
1987, 27, 243.
E. Dickinson and C. M. Woskett, in Food Colloids, ed. R. D.
Bee, P. Richmond and J. Mingins, Royal Society of Chemistry,
London, 1989, p. 74.
J-L. Courthaudon, E. Dickinson and D. G. Dalgleish, J. Colloid
Interface Sci., 1991, 145, 390.
J-L. Courthaudon, E. Dickinson, Y. Matsumura and D. C.
Clark, Colloids Surf., 1991, 56, 293.
J. Chen and E. Dickinson, Colloids Surf. A, 1995, 101, 77.
E. Dickinson, R. K. Owusu, S. Tan and A. Williams, J. Food
Sci., 1993, 58, 295.
E. Dickinson and S-T. Hong, J. Agric. Food Chem., 1994, 42,
1602.
M. Cornec, A. R. Mackie, P. J. Wilde and D. C. Clark, Colloids
Surf. A, 1996, 114, 237.
J. Chen, J. Evison and E. Dickinson, in Food Macromolecules
and Colloids, ed. E. Dickinson and D. Lorient, Royal Society of
Chemistry, Cambridge, 1995, p. 256.
E. Dickinson and S. Tanai, J. Agric. Food Chem., 1992, 40, 179.
E. M. Stevenson, D. S. Horne and J. Leaver, Food Hydrocoll.,
1997, 11, 3.
J. Chen and E. Dickinson, J. Sci. Food Agric., 1993, 62, 283.
R. Miller, R. Wustneck, J. Kragel and G. Kretzschmar, Colloids
Surf. A, 1996, 111, 75.
B. Warburton, Curr. Opin. Colloid Interface Sci., 1996, 1, 481.
R. Miller, V. B. Fainerman, J. Kragel and G. Loglio, Curr. Opin.
Colloids Interface Sci., 1997, 2, 578.
J. Chen and E. Dickinson, Food Hydrocoll., 1995, 9, 35.
M. Coke, P. J. Wilde, E. J. Russell and D. C. Clark, J. Colloid
Interface Sci., 1990, 138, 489.
P. J. Wilde and D. C. Clark, J. Colloid Interface Sci., 1993, 155,
48.
D. R. Wilson, P. J. Wilde and D. C. Clark, in Food Colloids and
Polymers : Stability and Mechanical Properties, ed. E. Dickinson
and P. Walstra, Royal Society of Chemistry, Cambridge, 1993,
p. 415.
R. Wustneck, J. Kragel, R. Miller, P. J. Wilde and D. C. Clark,
Colloids Surf. A, 1996, 114, 255.

Published on 01 January 1998. Downloaded on 08/04/2015 23:50:55.

View Article Online

108 E. Dickinson and S-T. Hong, Colloids Surf. A, 1997, 127, 1.


109 E. Dickinson, R. K. Owusu and A. Williams, J. Chem. Soc.,
Faraday T rans., 1993, 89, 865.
110 K. Boode and P. Walstra, Colloids Surf. A, 1993, 81, 121.
111 K. Boode, P. Walstra and A. E. A. de Groot-Mostert, Colloids
Surf. A, 1993, 81, 139.
112 K. Boode and P. Walstra, in ref. 106, p. 23.
113 H. D. Go, B. Freslon, M. E. Sahagian, T. D. Hauber, A. P.
Stone and D. W. Stanley, J. T exture Stud., 1995, 26, 517.
114 B. Vincent, J. Edwards, S. Emmett and A. Jones, Colloids Surf.,
1986, 18, 61.
115 Y. Cao, E. Dickinson and D. J. Wedlock, Food Hydrocoll., 1990,
4, 185.
116 E. Dickinson, M. Golding and M. J. W. Povey, J. Colloid Interface Sci., 1997, 185, 515.
117 E. Dickinson and M. Golding, Food Hydrocoll., 1997, 11, 13.
118 S. Asakura and F. Oosawa, J. Chem. Phys., 1954, 22, 1255.
119 Y. Mao, M. E. Cates and H. N. W. Lekkerkerker, Physica A,
1995, 222, 10.
120 M. J. W. Povey, in ref. 87, p. 196.
121 V. J. Pineld, M. J. W. Povey and E. Dickinson, Ultrasonics,
1995, 33, 246.
122 Y. Cao, E. Dickinson and D. J. Wedlock, Food Hydrocoll., 1991,
5, 443.
123 E. Dickinson, J. Ma and M. J. W. Povey, Food Hydrocoll., 1994,
8, 481.
124 E. Dickinson and M. Golding, J. Colloid Interface Sci., 1997,
191, 166.
125 M. P. Aronson, Colloids Surf., 1991, 58, 195.
126 W. Xu, A. D. Nikolov and D. T. Wasan, AIChE J., 1997, 43,
3215.
127 M. Golding, Ph.D. thesis, University of Leeds, 1997.
128 E. Dickinson and M. Golding, J. Colloid Interface Sci., 1998,
197, 133.
129 K. Almdal, J. Dyre, S. Hvidt and O. Kramer, Polym. Gels
Network, 1993, 1, 5.
130 R. Jost, R. Baechler and G. Masson, J. Food Sci., 1986, 51, 440.
131 E. Dickinson, S-T. Hong and Y. Yamamoto, Neth. Milk Dairy
J., 1996, 50, 199.
132 E. Dumay, C. Lambert, S. Funtenberger and J-C. Cheftel,
L ebensm.-W iss. T echnol., 1996, 29, 606.
133 E. Dickinson and J. D. James, in Proceedings of 35th European
High-Pressure Research Conference, 711th September, 1997,
Reading, U.K.
134 D. J. McClements, F. J. Monahan and J. E. Kinsella, J. T exture
Stud., 1993, 24, 411.
135 E. Dickinson and S-T. Hong, J. Agric. Food Chem., 1995, 43,
2560.
136 E. Dickinson and S-T. Hong, in Gums and Stabilisers for the
Food Industry, ed. G. O. Phillips, P. A. Williams and D. J.
Wedlock, Oxford University Press, Oxford, 1996, vol. 8. p. 319.

137 E. Dickinson and Y. Yamamoto, Food Hydrocoll., 1996, 10, 301.


138 J. Chen and E. Dickinson, J. Agric. Food Chem., 1998, 46, 91.
139 M. N. Jones, Chem. Soc. Rev., 1992, 21, 127.
140 E. Dickinson, in Proceedings of 1st International Symposium on
Food Rheology and Structure, ed. E. J. Windhab and B. Wolf,
Vincentz Verlag, Hannover, 1997, p. 50.
141 E. Dickinson, J. Chem. Soc., Faraday T rans., 1995, 91, 4413.
142 E. Dickinson and K. Pawlowsky, in ref. 136, p. 181.
143 E. Dickinson and K. Pawlowsky, J. Agric. Food Chem., 1997,
45, 3799.
144 E. Dickinson and Y. Yamamoto, J. Agric. Food Chem., 1996, 44,
1371.
145 E. Dickinson, T rends Food Sci. T echnol., 1997, 8, 334.
146 Y. Chanyongvorakul, Y. Matsumura, A. Sawa, N. Nio and T.
Mori, Food Hydrocoll., 1997, 11, 449.
147 E. Dickinson, in ref. 7, p. 107.
148 M. D. Haw, M. Sievwright, W. C. K. Poon and P. N. Pusey,
Physica A, 1995, 217, 231.
149 H. F. van Garderen, W. H. Dokter, T. P. N. Beelen, R. A. van
Santen, E. Pantos, M. A. J. Michels and P. A. J. Hilbers, J.
Chem. Phys., 1995, 102, 480.
150 B. H. Bijsterbosch, M. T. A. Bos, E. Dickinson, J. H. J. van
Opheusden and P. Walstra, Faraday Discuss., 1995, 101, 51.
151 J. F. M. Lodge and D. M. Heyes, J. Chem. Soc., Faraday T rans.,
1997, 93, 437.
152 E. Dickinson, J. Chem. Soc., Faraday T rans., 1994, 90, 173.
153 M. Whittle and E. Dickinson, Mol. Phys., 1997, 90, 739.
154 C. M. Wijmans and E. Dickinson, J. Chem. Soc., Faraday
T rans., 1998, 94, 129.
155 M. Whittle and E. Dickinson, J. Chem. Phys., 1997, 107, 10191.
156 M. Whittle and E. Dickinson, to be published.
157 E. Dickinson, J. Chem. Soc., Faraday T rans., 1995, 91, 51.
158 E. Dickinson, in Biopolymer Mixtures, ed. S. E. Harding, S. E.
Hill and J. R. Mitchell, Nottingham University Press, Nottingham, 1996, p. 349.
159 Y. Matsumura, I-J. Kang, H. Sakamoto, M. Motoki and T.
Mori, Food Hydrocoll., 1993, 7, 227.
160 G. J. Brownsey, H. S. Ellis, M. J. Ridout and S. G. Ring, J.
Rheol., 1987, 31, 635.
161 T. van Vliet and W. Kloek, in ref. 140, p. 101.
162 L. G. B. Bremer, T. van Vliet and P. Walstra, J. Chem. Soc.,
Faraday T rans., 1989, 85, 3359.
163 T. Hagiwara, H. Kumagai and T. Matsunaga, J. Agric. Food
Chem., 1997, 45, 3807.

Paper 8/01167B ; Received 10th February, 1998

J. Chem. Soc., Faraday T rans., 1998, V ol. 94

1669

Anda mungkin juga menyukai