Rheology and phase behavior of dense casein micelle dispersions

A. Bouchoux, B. Debbou, G. Gésan-Guiziou, M.-H. Famelart, J.-L. Doublier, and B. Cabane

Citation: The Journal of Chemical Physics 131, 165106 (2009); doi: 10.1063/1.3245956
View online: http://dx.doi.org/10.1063/1.3245956
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/131/16?ver=pdfcov
Published by the AIP Publishing

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.235.1.118 On: Tue, 14 Jan 2014 08:37:41
 THE JOURNAL OF CHEMICAL PHYSICS 131, 165106 共2009兲

Rheology and phase behavior of dense casein micelle dispersions

A. Bouchoux,1,2,a兲 B. Debbou,1,2 G. Gésan-Guiziou,1,2 M.-H. Famelart,1,2 J.-L. Doublier,3
and B. Cabane4
1
INRA, UMR1253 Science et Technologie du Lait et de l’Oeuf, F-35042 Rennes, France
2
AGROCAMPUS OUEST, UMR1253 Science et Technologie du Lait et de l’Oeuf, F-35042 Rennes, France
3
INRA, UR1268 Biopolymères Interactions Assemblages, F-44316 Nantes, France
4
PMMH, CNRS UMR7636, ESPCI, F-75231 Paris cedex 05, France
共Received 15 July 2009; accepted 21 September 2009; published online 29 October 2009兲

Casein micelle dispersions have been concentrated through osmotic stress and examined through
rheological experiments. In conditions where the casein micelles are separated from each other, i.e.,
below random-close packing, the dispersions have exactly the flow and dynamic properties of the
polydisperse hard-sphere fluid, demonstrating that the micelles interact only through excluded
volume effects in this regime. These interactions cause the viscosity and the elastic modulus to
increase by three orders of magnitude approaching the concentration of random-close packing
estimated at Cmax ⬇ 178 g / l. Above Cmax, the dispersions progressively turn into “gels” 共i.e., soft
solids兲 as C increases, with elastic moduli G⬘ that are nearly frequency independent. In this second
regime, the micelles deform and/or deswell as C increases, and the resistance to deformation results
from the formation of bonds between micelles combined with the intrinsic mechanical resistance of
the micelles. The variation in G⬘ with C is then very similar to that observed with concentrated
emulsions where the resistance to deformation originates from a set of membranes that separate the
droplets. As in the case of emulsions, the G⬘ values at high frequency are also nearly identical to the
osmotic pressures required to compress the casein dispersions. The rheology of sodium caseinate
dispersions in which the caseins are not structured into micelles is also reported. Such dispersions
have the behavior of associative polymer solutions at all the concentrations investigated, further
confirming the importance of structure in determining the rheological properties of casein micelle
systems. © 2009 American Institute of Physics. 关doi:10.1063/1.3245956兴

I. INTRODUCTION dispersions are concentrated, for instance, in industrial op-

erations such as membrane filtration,11 evaporation, and
Caseins make up to 80% of the protein content of cow
milk. In native milk, they are associated into large aggregates drying.12
that are called casein micelles. These “micelles” contain the In a recent study, we used the osmotic stress technique to
four caseins ␣s1, ␣s2, ␤, and ␬ in the proportion of 3:1:3:1 explore interactions in model dispersions of casein micelles
and ⬃8% in mass of phosphate and calcium ions. Their main over a wide range of casein concentrations, i.e., from 20 to
biological function is the transport and delivery of proteins 500 g/l.13 Our results could be described in terms of three
and essential minerals to the young mammals.1 The structure compression regimes: 共1兲 a dilute regime in which the casein
of the casein micelles has been studied for over 40 years, and micelles are well separated and do not interact, 共2兲 a transi-
rather precise descriptions are available, although still tion regime in which the behavior of the dispersion changes
controversial.2 It is commonly accepted that they are roughly from liquidlike to solidlike behavior as a result of stronger
spherical core-shell particles with outer diameters ranging interactions between micelles, and 共3兲 a concentrated regime
from 50 to 500 nm.3–5 The core is now generally described as in which the dispersions behave as coherent solids made of
a homogeneous web of caseins in which calcium phosphate micelles that are in direct contact with their neighbors.
nanoclusters are uniformly distributed.6–8 The shell is essen- In this paper, we present measurements of the rheologi-
tially made of ␬-caseins that extend into the aqueous phase cal properties of casein micelle dispersions in these three
as a polyelectrolyte brush and in this way produce short- concentration regimes. First, steady shear viscosities were
range repulsions between micelles.9,10 measured with casein micelle dispersions that “flow,” i.e.,
Even if the structure and composition of the casein mi- dispersions that do not behave as solids 关regimes 共1兲 and
celle itself are now reasonably known, little has been done in 共2兲兴. Second, oscillatory experiments were done to measure
understanding how these objects actually interact with each the complex shear modulus of dispersions over a large range
other and how these interactions influence the properties of of concentrations, i.e., from diluted dispersions to highly
native casein dispersions. In particular, we still need to un- concentrated dispersions obtained through osmotic stress 关re-
derstand how interactions and properties evolve when casein gimes 共1兲–共3兲兴. Additionally, creep-recovery tests were per-
formed with some highly concentrated samples. For com-
a兲
Author to whom correspondence should be addressed. Electronic mail: parison, similar experiments were performed with sodium
antoine.bouchoux@rennes.inra.fr. caseinate 共SC兲 dispersions in the same range of casein con-

0021-9606/2009/131共16兲/165106/11/$25.00 131, 165106-1 © 2009 American Institute of Physics

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.235.1.118 On: Tue, 14 Jan 2014 08:37:41
 165106-2 Bouchoux et al. J. Chem. Phys. 131, 165106 共2009兲

centrations. Even if they contain the same relative amount of and Muller-Buschbaum et al.26 for recent examples of its
caseins, SC dispersions are indeed quite different from use兲. In contrast, SC powder contains the same relative
casein micelle dispersions with respect to structure14–16 and amounts of caseins as NPC but does not lead to dispersions
interactions, because they do not contain the calcium phos- of stable “casein micelles.” Instead, SC dispersions contain
phate nanoclusters.13,17 caseins that are present as individual molecules or as revers-
To our knowledge, the rheology of casein micelle disper- ible self-assembled aggregates with radius of ⬃10 nm.14–16
sions has never been investigated in this way. First, the dis- The NPC powder was prepared in our laboratory accord-
persions were made from native phosphocaseinate powder ing to a protocol developed by Pierre et al.18 and Schuck et
共NPC兲 dissolved in a solvent made from ultrafiltration 共UF兲 al.27 and described in our previous work.13 The SC powder
of skimmed milk 共UF permeate兲. Casein micelles are very was produced by Armor Protéines 共Saint-Brice en Coglès,
close to their native state in such a reconstituted milk that is France兲 according to a protocol similar to that described by
depleted of serum proteins.18,19 Then, the dispersions were Segalen et al.28 In both powders, the caseins and their asso-
concentrated through osmotic stress using the same UF per- ciated minerals represent more than 90% of the total solid
meate with added hydrophilic polymers to control the chemi- content. The detailed compositions of the powders have been
cal potential of water. This equilibrium ensured that the reported.13
chemical potentials of all ions were identical to their values The UF permeate was prepared through membrane
in milk. These features are quite different from previous UF 共5000 Da cutoff兲 of fresh skimmed milk. Its approximate
works in which water was used as a solvent20 or serum pro- ionic composition is ⬃20 mM Na+, ⬃40 mM K+,
teins were still present in the dispersions.21,22 ⬃10 mM Ca2+, ⬃30 mM Cl−, ⬃10 mM phosphate,
The aim of this work is to provide some answers to the ⬃10 mM citrate, and traces of carbonate, sulfate, and mag-
following questions. nesium ions 共see Jenness and Koops29 for a full description兲.
It also contains lactose 共⬃150 mM兲 and a few other low
共i兲 Resistance to flow at concentrations below or near the
molar mass molecules such as riboflavin, a vitamin that gives
liquid-solid transition. In this regime, do the micelles
it a distinctive yellow color. Thimerosal and sodium azide,
only interact through excluded volume effects? What
both purchased from Sigma-Aldrich, were added to the UF
are the consequences of these interactions for the re-
permeate as preservatives at 0.02% and 0.1% 共w / w兲, respec-
sistance to flow of the dispersions; does it match that
tively.
of a polydisperse hard-sphere fluid? What is the effect
of the fact that casein micelles are highly hydrated
structures that can be deformed by compressive or B. Sample preparation
shear forces? These questions have already been
raised by other authors,4,21,23 but quantitative answers For casein concentration of up to 170 g/l, the NPC and
are still missing. Recent work performed by Alex- SC dispersions were prepared by thoroughly mixing the
ander et al.24 through diffusing wave spectroscopy powders in UF permeate for 15 h at 35 and 50 ° C, respec-
suggests that fat-free milk does behave like a hard- tively. Dispersions at higher casein concentrations were pre-
sphere system in a certain range of casein concentra- pared through osmotic stress, a concentration technique that
tions. is based on water exchange between the sample 共i.e., a liquid
共ii兲 Resistance to deformation at the liquid-solid transi- NPC or SC dispersion兲 and a reservoir of controlled osmotic
tion and above. In this regime, can we still describe pressure 共see our previous work13 for a full description兲. This
dense casein micelle dispersions as fluids of hard reservoir is generally a polymer solution; poly共ethylene gly-
spheres? What are the forces that cause the resistance col兲 共PEG兲 with a molar mass of 35 000 Da 共Fluka, Switzer-
to deformation? Are they the same forces that produce land兲 was used in our case. We measured the osmotic pres-
the resistance to osmotic compression?13 Is there any sures of solutions of this polymer at concentrations of up to
involvement of intermicelle adhesion? 20% 共w / w兲 at 20 ° C through membrane osmometry 共Osmo-
mat 090, Gonotec, Germany兲 and equilibrium dialysis versus
T110 Dextran solutions of known osmotic pressure.30 The
II. EXPERIMENTAL results were fitted to the following expression for the osmotic
pressure ⌸ 共Pa兲 as a function of PEG concentration 共%,
A. Materials
w / w兲:
All experiments were performed with dispersions made
log ⌸ = a + b关PEG兴c , 共1兲
from casein powders 共NPC and SC兲 dispersed in a solvent
obtained through UF of skimmed milk 共UF permeate兲. We where a = 0.49, b = 2.5, and c = 0.24.
made this choice for the following reasons. First, the use of Standard regenerated cellulose Spectra/Por 2 dialysis
the UF permeate ensures that the chemical potentials of all bags with a molecular mass cutoff of 12 000–14 000 Da and
ions are maintained to their values in the native state. Sec- a diameter of 29 mm were used 共Spectrum Laboratories,
ond, these casein powders are depleted of milk serum pro- Rancho Dominguez, CA兲. Casein dispersions of moderate
teins that could interfere with rheological measurements and concentration 共⬃100 g / l兲 were placed in the bags and im-
also with the preparation of concentrated samples through mersed in the polymer solutions kept at 20 ° C. The bags
osmotic stress. Third, NPC powder is considered as an ad- were then regularly refilled with casein dispersion until equi-
equate model for milk casein micelles 共see Huppertz et al.25 librium was reached 共about 30 days兲. This technique makes it
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.235.1.118 On: Tue, 14 Jan 2014 08:37:41
 165106-3 Rheology of casein micelle dispersions J. Chem. Phys. 131, 165106 共2009兲

possible to prepare a fair amount 共about 40 ml/bag兲 of !

homogeneous samples with casein concentrations of up to
400 g/l.
For both preparation methods, the casein concentrations !
!
in the samples were accurately determined after preparation
through drying at 105 ° C. For that purpose, the relation be-
tween total solid content and casein concentration in g/l was

η ,-.$*)
"
initially determined over a wide range of concentrations with !
model dispersions. At 20 ° C, the pH of all dispersions
matched the average pH of milk, i.e., pH = 6.7⫾ 0.1. "#
!
C. Rheology
Flow measurements were performed with a Contraves
"#
Low-Shear 30 viscometer 共Ruislip, United Kingdom兲 for dis- !
persions of ␩ ⬍ 0.01 Pa s and with a TA Instruments
AR2000 rheometer 共Guyancourt, France兲 for dispersions
" ! # #
with higher viscosities. Both instruments were operated us- ! ! ! ! !
ing a Couette geometry with inner and outer radii of 5.5 and γɺ ,*" )
6 mm 共low shear兲 and 14 and 15 mm 共AR2000兲, respectively.
During all measurements, the sample was covered with a FIG. 1. Shear rate dependence of apparent viscosity for NPC dispersions in
film of mineral oil to prevent evaporation. The viscosity of UF permeate. Casein concentrations from top to bottom: 168, 158, 146, 138,
the UF permeate was measured at 20 ° C with the Contraves 121, 92, 56, and 10 g/l. The lines are fits to Eq. 共2兲.
Low-Shear 30 viscometer and found to be ␩s = 1.2
⫻ 10−3 Pa s.
meate with casein concentrations ranging from 10 to 168 g/l.
Oscillatory shear measurements were performed with the
At concentrations below ⬃100 g / l, the dispersions behaved
AR2000 rheometer. Different geometries were used depend-
as Newtonian fluids over the range of shear rates investi-
ing on the consistency of the samples. At moderate concen-
gated. At higher concentrations, shear thinning was clearly
trations 共below 250–300 g/l for NPC dispersions兲, the
visible. Shear thinning at high shear rates is a common phe-
samples obtained through osmotic stress were soft gels so
nomenon. For the hard-sphere fluid, it is caused by a transi-
that a cone and plate geometry 共radius of 30 mm and angle
tion from a regime where the particles have pure Brownian
of 4°兲 was used. At higher concentrations 共⬎300 g / l兲, the
motions at low-shear rates to a regime where their motions
dispersions obtained through osmotic stress clearly behaved
are determined by hydrodynamics at high shear rates. This
as solids, and parallel plates with grooved surfaces were
behavior can be described through a simple model originally
used. In such cases, the sample was cut in cylindrical pieces
developed by Krieger and Dougherty,31 in which the appar-
of 2 cm in diameter and 4 mm in width. It was then quickly
ent viscosity ␩ depends solely on shear stress ␶ according to
transferred onto the lower plate, and the upper plate was
gently lowered until the sample filled the gap and a constant
normal force was reached 共about 0.2 N兲. A closed cover ␩ − ␩⬁ 1
= , 共2兲
setup was used to prevent evaporation with both geometries. ␩0 − ␩⬁ 1 + ␶/␶c
G⬘ and G⬙ were first measured as a function of strain at a set
frequency of 1 Hz. The frequency dependent response was where ␩⬁ and ␩0 are the high- and low-shear limiting vis-
then measured from 0.01 to 10 Hz at a given strain within the cosities, respectively, and ␶c is the characteristic shear stress,
region of linear response. which characterizes shear thinning.
Creep-recovery tests were performed with the AR2000 This model was successfully used by De Kruif et al.32 to
rheometer using parallel plates with grooved surfaces. Creep analyze the viscosities of silica dispersions in cyclohexane.
stress 共10 Pa兲 was applied for 2 h, and recovery 共zero stress兲 The fits shown in Fig. 1 indicate that our data are also per-
lasted for 6 h. Strain was recorded as a function of time and fectly described by this model. The reduction to a master
was converted into compliance J共t兲 = strain共t兲 / 共creep stress兲. curve, presented in Fig. 2, confirms that a single set of pa-
A closed cover setup was used to prevent evaporation. rameters ␩0, ␩⬁, and ␶c suffices to describe the shear thin-
All rheological experiments were performed at a fixed ning at all concentrations. Accordingly, casein micelle dis-
temperature of 20 ° C. Measurements were repeated two or persions made at concentrations below the liquid-solid
three times, showing very good reproducibility within transition must closely resemble hard-sphere fluids.
⫾10%. The best-fit values of the characteristic shear stress ␶c are
III. RESULTS AND DISCUSSION plotted in Fig. 3 as a function of the volume fraction ␾ oc-
cupied by the casein micelles in the dispersions. This equiva-
A. Viscous flow before the liquid-solid transition.
lent hard-sphere volume fraction was estimated using a mi-
Casein micelles as hard spheres?
celle voluminosity taken as 4.4 ml/g of dry casein.4 In
Figure 1 shows the shear rate dependence of the appar- Krieger’s model, the characteristic shear stress depends on
ent viscosity for casein micelle 共NPC兲 dispersions in UF per- the sphere radius a through
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.235.1.118 On: Tue, 14 Jan 2014 08:37:41
 165106-4 Bouchoux et al. J. Chem. Phys. 131, 165106 共2009兲

φ
! !$# !$% !$& !$'
%
!$' !
(η−η∞)/(η0−η∞)

!$& #
!

!$%
#

η*
!
!$#

!
!
"# " ! #
! ! ! ! !
τ*τ+ !
!
! %! !! %! #!!
FIG. 2. Reduced viscosity 共␩ – ␩⬁兲 / 共␩0 – ␩⬁兲 vs reduced shear stress for
NPC in UF permeate at 168 g/l 共open diamonds兲, 158 g/l 共closed inverted + ,-*/)
triangles兲, 148 g/l 共open triangles兲, 138 g/l 共closed circles兲, and 121 g/l
共open squares兲. The solid line is calculated through Eq. 共2兲. FIG. 4. Concentration dependence of the relative viscosity ␩r at low shear
共␩r,0 = ␩0 / ␩s兲 for NPC in UF permeate 共solid squares兲 and SC in UF perme-
ate 共shaded diamonds兲 and at high shear 共␩r,⬁ = ␩⬁ / ␩s兲 for NPC in UF per-
␶c = kT/ba3 , 共3兲 meate 共open triangles兲. The volume fraction occupied by the casein micelles
共upper x-axis兲 was estimated using a micelle voluminosity of 4.4 ml/g 共Ref.
4兲. The dotted line is calculated through Eq. 共4兲. The solid and dashed lines
where parameter b is usually comprised between two and are calculated through the Quemada’s equation 关Eq. 共5兲兴 with ␾max = 0.64
共Cmax = 145 g / l兲 and ␾max = 0.78 共Cmax = 178 g / l兲, respectively.
three for hard spheres.32
Taking an average micelle radius a = 100 nm, the ␶c val-
ues we obtained for NPC dispersions imply that b ⬇ 1 – 3, tially because the dimensions of the investigated objects are
which is close to the value of the hard-sphere system. Figure similar 共78 nm for the silica spheres of De Kruif et al.32兲.
3 compares our results with those obtained by De Kruif et As shown in Fig. 3, ␶c first increases with ␾. It has a
al.32 for silica spheres in cyclohexane. The two sets of data clear maximum at ␾ ⬇ 0.62 共⬃140 g / l兲 and then decreases
show very similar ␶c values ranging from 0.5 to 4 Pa essen- to vanishing values at ␾ ⬇ 0.8– 0.85. This behavior is quite
similar to that of the hard-sphere fluid. However, for mono-
$ disperse hard spheres, the maximum in ␶c is located at ␾
⬇ 0.5, and ␶c decreases to zero at ␾ ⬇ ␾rcp ⬇ 0.64, i.e., at the
random-close packing fraction of spheres33 共Fig. 3兲. The ob-
served shift in volume fractions for NPC dispersions has two
# possible causes. First, the high polydispersity of casein mi-
celle dispersions, usually measured between 30% and 40%,4
could be responsible for this effect. Indeed, polydispersity
τ+ ,-./

" would lead to higher values for the volume fraction at which
the hard-sphere system has a disorder-order transition, i.e.,
the volume fraction at which ␶c is maximum. In the same
manner, the random-close packing fraction ␾rcp is also
) known to increase with polydispersity. As an example, ␾rcp
⬇ 0.78 when the polydispersity is 40%, as opposed to ␾rcp
⬇ 0.64 for the monodisperse hard-sphere fluid.34 Second, as
reported by Russel et al.,35 a similar shift is found with soft
!" !# !$ !% !& !' !( sphere systems. Accordingly, we could reasonably imagine
that casein micelles start to deform at the onset of shear
φ thinning and thus behave as soft spheres in this hydrody-
FIG. 3. Characteristic shear stress as a function of volume fraction for NPC namic regime.
in UF permeate 共closed squares兲. The volume fraction occupied by the Our data also show a remarkable variation in the viscosi-
casein micelles was estimated using a micelle voluminosity of 4.4 ml/g 共Ref. ties with casein concentration. The relative high- and low-
4兲. For comparison, the characteristic shear stress for colloidal dispersions
shear limiting viscosities, ␩r,⬁ = ␩⬁ / ␩s and ␩r,0 = ␩0 / ␩s, re-
of 78 nm radius silica spheres in cyclohexane 共Ref. 32兲 is also reported as a
function of volume fraction 共crosses兲. The dotted and dashed lines are spectively, are plotted as a function of casein concentration
guides for the eye. and casein micelle volume fraction in Fig. 4, taking ␩s
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.235.1.118 On: Tue, 14 Jan 2014 08:37:41
 165106-5 Rheology of casein micelle dispersions J. Chem. Phys. 131, 165106 共2009兲

= 1.2⫻ 10−3 Pa s for the viscosity of UF permeate at 20 ° C. #

!
At low volume fractions 共␾ ⬍ 0.35兲, for the hard-sphere
fluid, De Kruif et al.32 proposed that ␩r,0 varies with ␾ as
!
␩r,0 = 1 + 2.5␾ + 4␾2 + 42␾3 . 共4兲
Figure 4 shows that our data correlate well with Eq. 共4兲, !
!

η ,-.$*)
giving further support to the correspondence between NPC
dispersions and hard-sphere systems. Then, above ␾
⬇ 0.4– 0.45 共⬃100– 110 g / l of casein兲, the relative low- "
!
shear limiting viscosity increases much faster. The hard-
sphere fluid also has a similar behavior upon approaching the
glass transition; the divergence of ␩r is usually modeled with !
"#

Quemada’s equation,32,35
␩r = 共1 − ␾/␾max兲−2 or ␩r = 共1 − C/Cmax兲−2 , 共5兲 "#
!
with ␾max,0 = 0.64 and ␾max,⬁ = 0.71 for ␩r,0 and ␩r,⬁, respec- " ! # #
tively. ! ! ! ! !
In our case ␩r,0 does not follow Eq. 共5兲 with ␾max = 0.64. γɺ ,*" )
As already suggested, this difference might be explained by
the polydispersity of casein micelle dispersions. Indeed, the FIG. 5. Shear rate dependence of apparent viscosity for SC dispersions in
random-close packing fraction ␾max,0 = 0.64 relates to mono- UF permeate. Casein concentrations from top to bottom are 125, 119, 98,
83, 47, and 6 g/l.
disperse dispersions of spheres, whereas for a 40% polydis-
perse hard-sphere fluid, it is ␾max,0 = 0.78.34 Figure 4 shows
that the measured viscosities are located between these two assembled aggregates of about 11 caseins and ⬃10 nm in
extreme cases of monodisperse and 40% polydisperse hard- radius that are expected to overlap and interpenetrate even at
spheres systems. low concentrations 共starting from C ⬎ 50 g / l兲.14
In contrast to ␩r,0, the relative high shear limiting vis- At low shear, the specific viscosity of SC dispersions,
cosity ␩r,⬁ increased steadily over the whole range of con- ␩sp = 共␩0 − ␩s兲 / ␩s varies with casein concentration in a man-
centrations and did not diverge at the highest concentrations ner that is typical of polymer systems37,38 共Fig. 6兲. At con-
共Fig. 4兲. On one hand, the polydispersity might explain this centrations below 80 g/l, there is a first regime in which the
result since it would also affect the value of ␾max,⬁, thus specific viscosity increases with C5/2; this exponent is in be-
shifting the divergence in ␩r,⬁ to volume fractions that were tween those expected for polymer solutions in the unen-
not reached in our experiments. On the other hand, this fea- tangled 共␩ ⬃ C5/4兲 and entangled 共␩ ⬃ C15/4兲 regimes.37,39
ture could also be interpreted as a slight deviation from the
hard-sphere fluid behavior. Indeed, high shear rates might %
cause the micelles to deform and align along the flow direc- )
tion. Such increased shear thinning has been observed in
model emulsions of soft droplets by Saiki et al.36 )
$
In order to better understand the role of structure and
interactions on the flow properties of casein micelles, we #
also measured the same properties on solutions of SC in )
which the caseins are not structured into micelles. Figure 5
presents the shear rate dependence of the apparent viscosity "
η()

)
for SC dispersions in UF permeate. For 119 and 125 g/l of
caseins, the range of shear rates was rather limited because )
the flow became unstable at high shear rates. A comparison )
with Fig. 1 indicates that the flow behavior of SC was quite
different from that of NPC. In particular, the flow was New- )
tonian over a much wider range of shear rates in SC disper-
sions. The corresponding low-shear limiting viscosities are #)
reported in Fig. 4 as a function of casein concentration. )
Clearly, the values of ␩r,0 of SC dispersions are much higher ) % )
共up to ⬎2 orders of magnitude兲 than those of NPC disper- * ,,-./
sions. All these differences in flow behavior may be ex-
plained through the loss of the peculiar structure of the FIG. 6. Concentration dependence of specific viscosity for SC in UF per-
casein micelle in SC dispersions. These dispersions then be- meate 共solid diamonds兲, SC in pure water 关open squares, pH of 6.7, T
= 20 ° C, data from Pitkowski et al. 共Ref. 43兲兴, and SC in phosphate buffer
have as an “ordinary” entangled polymer system. This as-
+0.1 M NaCl 关open triangles, pH of 6.8, T = 25 ° C, data from Farrer and
sumption is in direct agreement with recent work, which Lips 共Ref. 17兲兴. The dashed and solid lines have slopes of 5/2 and 15,
shows that SC presumably forms highly hydrated self- respectively.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.235.1.118 On: Tue, 14 Jan 2014 08:37:41
 165106-6 Bouchoux et al. J. Chem. Phys. 131, 165106 共2009兲

%
Then, at concentrations above 80 g/l, there is another regime ! ,.)
in which the data follow an extremely steep power law ␩ #'! -*/
⬃ C15. This type of dependence is usually related to a “hy- %
perentanglement” regime in which entanglements are supple- !
mented by some local attractions between the casein chains40 # ! -*/
#
as for associating polymers systems, for instance.41,42 !

234 233 ,-.)

Finally, it is important to remark that the viscosities of
SC in UF permeate are well above those reported by other #
authors17,43 for SC in different solvents 共Fig. 6兲. This dis- !
crepancy most likely originates from the presence of 10 mM '' -*/
calcium ions in UF permeate, in contrast to the calcium free ! #& -*/
solvents used by these authors. It is indeed known that cal-
cium ions produce strong attractive interactions between ! & -*/ ∼(
casein chains,44 which may explain the enhanced associative
!
#
behavior of SC in UF permeate. ∼(
"
The results presented in this first part can be summarized !
"# " !
as follows. ! ! ! !
共1兲 Flow resistance of dispersions in which caseins are as- ( ,67)
sociated into micelles (NPC, “native” form). At low-
shear rates, casein micelle dispersions of concentrations %
below 170 g/l behave qualitatively and quantitatively as ! ,G)
+ H #'! -*/
polydisperse hard-sphere fluids, suggesting that in these
%
conditions, the micelles interact through excluded vol- !
ume effects only. This is fully consistent with the view
that casein micelles interact through short-range repul- #
sions produced by the 7 nm height ␬-casein brush lo- !
234 233 ,-.)

+*::) 9:+<=:*?
cated on their surfaces.9,10 The analogy with hard- !$'
#
sphere systems suggests a glass transition at C ! !$&

D , ! -. )
⬇ 178 g / l, a concentration that agrees with our former
"
observations based on osmotic pressure "# !$%
measurements.13 At high shear rates, the limiting vis- !
cosities did not diverge at the highest concentrations !$#
!
investigated. This behavior could be caused entirely by ! !$!
polydispersity. However, it might also be interpreted as ! !! #!! #!! %!! %!!
a slight deviation from hard-sphere fluid behavior @ ,ABC)
"
caused by the deformability of the micelles. !
"% "% "# "# " !
共2兲 Flow resistance of dispersions in which caseins are not ! ! ! ! ! ! !
“organized” into such micelles (SC). These dispersions ( ,67)
behave as a polymer solution over the range of concen-
trations investigated with a hyperentanglement regime FIG. 7. 共a兲 Frequency dependence of the elastic G⬘ 共solid symbols兲 and loss
G⬙ 共open symbols兲 moduli for NPC dispersions in UF permeate at different
that starts above 80 g/l of caseins. This is fully consis- casein concentrations. The lines are guides for the eye. 共b兲 Frequency de-
tent with the latest characterizations of such pendence of the G⬘ and G⬙ for NPC in UF permeate at 370 g/l: data from
systems.14,43 oscillatory experiments 共inverted triangles兲 and data from creep-recovery
experiments 共diamonds兲. The lines are guides for the eye. The inset shows
The differences between these two types of dispersions the results of creep-recovery experiments in terms of compliance J vs time.
demonstrate the crucial importance of structure and interac-
tions in determining the flow behavior of casein dispersions. casein concentrations ranging from 136 to 370 g/l. At the
lowest concentration 共C = 136 g / l and ␾ ⬇ 0.6兲, the disper-
sion behaved as a viscoelastic fluid with G⬘ and G⬙ scaling
B. Viscoelastic properties in the liquid-solid as f 2 and f 1, respectively. The strong frequency dependence
transition region and above. Casein micelles as
persisted at the next concentration 共161 g/l and ␾ ⬇ 0.71兲.
deformable, compressible, and adhesive spheres?
Then, at 178 g/l 共␾ ⬇ 0.78兲, there was a crossover of G⬘ and
At concentrations above C ⬇ 130 g / l, the NPC disper- G⬙ at frequencies just below 0.01 Hz, and the frequency
sions began to show some elastic properties. This viscoelas- dependence was much less, indicating a transition to another
tic behavior was characterized through two types of experi- regime. At the higher concentrations 共210 and 370 g/l兲, G⬘
ments at small strains: through the response to harmonic and G⬙ varied only slightly on frequency and G⬘ clearly
stresses and through creep-recovery experiments 共Fig. 7兲. dominated G⬙, meaning that the dispersions behaved prima-
Figure 7共a兲 shows the frequency dependence of the elastic rily as viscoelastic gels. At these two high casein concentra-
modulus G⬘ and loss modulus G⬙ of NPC dispersions with tions, the apparent volume fractions that can be estimated
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.235.1.118 On: Tue, 14 Jan 2014 08:37:41
 165106-7 Rheology of casein micelle dispersions J. Chem. Phys. 131, 165106 共2009兲

!$% !$& !$' φ remain in the same ratio. We take this change in frequency
dependence at Cmax = 178 g / l as the “gel point” of the dis-
%
! persion. This point is also roughly the point of crossover
%
between G⬘ and G⬙ at both low and high frequencies 共Fig. 7兲,
! another common although not universal47 definition of the
# gel point.
! At C ⬎ 178 g / l, G⬘ and G⬙ still grow with casein con-
234 Π ,-.)

# centration, but this increase is much less pronounced. At

! !
!

"
these concentrations, the magnitude of the high-frequency
!

23) ,1CB@* <( σ**)

! !
"#
elastic modulus is then very close to that of the osmotic
pressure 共Fig. 8兲. This correspondence has already been ob-
!
!
"#
!
served in other systems that were concentrated through os-
"%

"
! motic stress, most remarkably by Mason et al.48,49 The com-
! !
"%
parison of osmotic and mechanical results shows that for
"# !$% !$& !$' !$' !$+ $! NPC dispersions, there is a transition at Cmax = 178 g / l be-
! φ
tween a low concentration “liquid” regime where G⬘ Ⰶ ⌸
!! #!! #!! %!! and a high concentration “gel” regime where G⬘ ⬇ ⌸.
We first consider the region where C ⬍ Cmax. In this re-
+ ,-*/) gime, the strong frequency dependence of the viscoelastic
moduli indicates that the resistance to hydrodynamic flow is
FIG. 8. The low- 共f = 0.01 Hz, triangles兲 and high-frequency 共f = 10 Hz, still the dominant contribution to G⬘ and G⬙. Calculations of
squares兲 elastic moduli G⬘ of NPC in UF permeate as a function of casein
concentration. The open circles are the osmotic pressures ⌸ of the disper- the high-frequency elastic modulus of a hard-sphere fluid50
sions as measured by equilibrium dialysis 共Ref. 13兲. The upper x-axis is the indicate a frequency dependence G⬘ ⬃ f 1/2 at high volume
volume fraction occupied by the casein micelles in the dispersions estimated fractions, close enough to the experimental dependence,
using a micelle voluminosity of 4.4 ml/g 共Ref. 4兲. The filled area represents
the ␾ = 0.64– 0.78 region. The inset shows the scaled plateau elastic modulus
which is G⬘ ⬃ f 2/3 at C = 161 g / l. Moreover, the magnitude
G⬘p 共crosses兲 reported by Mason et al. 共Refs. 48 and 49兲 for monodisperse of G⬘ increases by three decades 共at high frequency兲 at the
emulsions having radii from 0.25 to 0.74 ␮m and plotted as a function of ␾. end of this concentration regime. This is also the behavior
The filled area represents the ␾ = 0.58– 0.64 region. expected from a hard-sphere fluid upon approaching the
random-close packing fraction where the concentration is so
from the voluminosity of an uncompressed micelle4 high that particles can hardly move. Indeed, for systems such
共4.4 ml g−1兲 are ␾ ⬇ 0.9 and 1.6, respectively, indicating that as silica spheres and emulsions, numerous experiments have
the applied osmotic stress forced the casein micelles to de- shown that the high-frequency modulus G⬘ increases steeply
form 共␾ ⬎ ␾rcp兲 and deswell 共␾ ⬎ 1兲. near the random-close packing fraction.48–54 The inset of Fig.
In addition to oscillatory measurements, creep-recovery 8 reproduces the plateau values of the elastic modulus G⬘p
tests were performed with the NPC in UF permeate at 370 obtained by Mason et al.48,49 with emulsions. In the range of
g/l. A creep stress of 10 Pa was applied to the sample for 2 h. volume fractions ␾ = 0.58– 0.64, where the droplets are ex-
Stress was then set to zero, and the recovery period lasted for pected to behave as hard spheres, G⬘p increases in exactly
6 h. During the entire test, strain was recorded as a function the same way as observed here. There are only two differ-
of time and was transformed into compliance J共t兲 ences between our results on casein micelles and Mason’s
= strain共t兲 / 共creep stress兲 关inset of Fig. 7共b兲兴. We converted results on monodisperse emulsions. First, in casein disper-
the J共t兲 results to the frequency domain using a procedure sions, the steep increase in the G⬘ occurs at higher volume
described by different authors.45,46 The G⬘ and G⬙ obtained fractions 共␾ = 0.64– 0.78; Fig. 8兲 because NPC dispersions
in this way cover frequencies from 10−5 to 0.1 Hz and are in are polydisperse 共about 40%兲 and can pack more efficiently
very good agreement with those obtained at higher frequen- up to ␾rcp ⬇ 0.78.34 Second, experiments on casein micelles
cies from oscillatory experiments 关Fig. 7共b兲兴. However, they did not exactly reach the “plateau” limit G⬘p as defined by
have a crossover point at very low frequency 共f ⬇ 3 Mason et al. most probably because casein micelles are
⫻ 10−5 Hz兲. This crossover shows that the NPC dispersion smaller than emulsion droplets and consequently have faster
even at this extreme casein concentration does not behave as motions. Once these differences are accounted for, both sys-
a true elastic gel with permanent connections. This is also tems can be analyzed as concentrated hard-sphere systems
demonstrated by the fact that the dispersion did not com- close to ␾rcp, and their mechanical response can be calcu-
pletely recover after stress was removed 关inset of Fig. 7共b兲兴, lated using classical hydrodynamics of hard-sphere liquids.
indicating plastic flow at very long time scales. Figure 9 compares G⬘ measured in NPC dispersions with the
Figure 8 gives the concentration dependence of the elas- predictions of Lionberger and Russel for concentrated disper-
tic moduli obtained at low 共f = 0.01 Hz兲 and high 共f sions of free draining hard spheres.50,55 This model recog-
= 10 Hz兲 frequencies. In both cases, G⬘ dramatically in- nizes that hydrodynamic interactions reduce the relative mo-
creases over a narrow range of casein concentrations from bility of the spheres at high ␾ but neglects lubrication forces,
⬃145 to ⬃178 g / l. At the end of this concentration region, which become significant at extremely high frequencies only.
the values of G⬘ measured at both frequencies have come Since the measurements in NPC dispersions were performed
much closer to each other, and at higher concentrations they at intermediate frequencies, the free draining model is ex-
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.235.1.118 On: Tue, 14 Jan 2014 08:37:41
 165106-8 Bouchoux et al. J. Chem. Phys. 131, 165106 共2009兲

%
! %
,.)
!
#
!
#
!
%
!

23 ,-.)
23 ,-.)

!
!
!
#
" !
!
23 ∼ +,+"+A.4)
"#
! 23 ∼ ,+"+A.4)
C

#
"# !
! ! !!
!$' !$' !$+ $!
+"+A.4 ,-*/)
+*+A.4
FIG. 9. The “hard-sphere liquid” contribution to the elastic modulus of NPC
dispersions. Experimental values of the storage modulus G⬘共f兲 at frequen- !!
,G)
cies f = 0.01 Hz 共open triangles兲 and f = 10 Hz 共filled triangles兲 as a func-
tion of C / Cmax with Cmax = 178 g / l. The lines are calculated through the
expressions given by Lionberger and Russel 共Ref. 50兲 in the case of the
hard-sphere liquid 共free draining model兲. Those calculations were made tak-
2:4 5*.+@B<C ,<)
ing an average micelle radius of ⬃100 nm and a micelle voluminosity of
4.4 ml/g 共Ref. 4兲.
* H !$%

pected to be sufficient. In fact, considering that the data were !

not obtained in the high-frequency limit required by the
model and that no parameters have been adjusted, the agree-
ment with the model is quite satisfactory.
*H
We will now consider the region where C ⬎ Cmax
= 178 g / l. In this regime, the much weaker frequency depen-
dence of G⬘ and G⬙ indicates that elastic forces are the domi- *
2:4 (*.+@B<C ∼ ,+"+A.4)
nant contribution to G⬘ and G⬙. There are two possible
sources of elastic resistance to deformation in this regime. ! !!
The first is the resistance to deformation of the micelles
themselves. Indeed, at C ⬎ Cmax, the estimated volume frac-
+"+A.4 ,-*/)
tion that is occupied by the micelles is ⬎0.78, which is the
FIG. 10. 共a兲 The high- 共f = 10 Hz, solid squares兲 and low-frequency 共f
random-close packing volume fraction for a 40% polydis- = 0.01 Hz, solid triangles兲 elastic moduli G⬘ of NPC dispersions in UF
perse fluid of spheres.34 Thus, we can reasonably assume that permeate as a function of C – Cmax with Cmax = 178 g / l 共␾ ⬇ 0.78兲. The solid
the micelles are in contact at the onset of this regime and and dotted lines are fits to the models discussed in the text. 共b兲 The remain-
ing gel fraction in casein dispersions after reswelling in UF permeate and
then progressively deform and/or deswell as concentration plotted vs C – Cmax, with C as the casein concentration before reswelling and
increases. The assembly of compressed micelles may then Cmax = 178 g / l 共␾ ⬇ 0.78兲 关experiments are detailed in our previous work
behave as the compressed emulsions of Mason et al.,48,49 共Ref. 13兲兴. The solid lines are fits to the model discussed in the text.
which do have an elastic resistance to shear stresses due to
the elasticity of their interfaces. Mason et al. analyzed this
resistance through a “network spring model” in which the shear stress. In this case, the storage modulus would scale as
number of springs between neighboring emulsion droplets ␴ / r, where ␴ would be the tension of the membranes and r
grows as C – Cmax, while the number of droplets per unit would be the radius of the micelles. Using the exact same
volume grows as C. The elastic modulus is then expected to expressions as those used by Mason et al.,48,49 we evaluate
grow as C共C – Cmax兲 as the system is compressed beyond ␴ = 0.01 N / m, which is not unreasonable for a membrane
Cmax.48 Surprisingly, this emulsion model matches the results that would surround the micelle and serve as an anchor to the
obtained with casein micelles over two decades in G⬘, both ␬-casein brush.
at high and at low frequencies 关solid lines in Fig. 10共a兲兴. It is Alternatively, one may reject any such precise structural
then interesting to examine whether the elastic resistance of features and try to analyze the behavior of compressed casein
compressed casein micelles could have a similar origin as dispersions at C ⬎ Cmax according to a general percolation
that of compressed emulsions, i.e., a network of membranes model. In this case the elastic properties would be limited by
that would be deformed by the compression and by applied the resistance of bonds between micelles. If the number of
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.235.1.118 On: Tue, 14 Jan 2014 08:37:41
 165106-9 Rheology of casein micelle dispersions J. Chem. Phys. 131, 165106 共2009兲

such bonds grows as the number of contact points per mi- ,.)
celles, assumed to be linear in 共C – Cmax兲, then percolation %
theory predicts that the storage modulus should vary as !
共C – Cmax兲n.56 This type of power law also fits the variation in #%# -*/
the storage modulus G⬘ with casein concentration 关Fig. #
!

234 233 ,-.)

10共a兲兴. However, the exponent is n = 1.3, whereas most ex- ##! -*/
&& -*/
periments on physical gels yield exponents n in the range of
two to three.56 This lower value of the exponent for NPC # '& -*/
dispersions may indicate that bonds are not created at ran- !
dom locations through the casein gel, which is not surprising
considering this gel is made of casein micelles. On the other
#% -*/
hand, recent numerical simulations of random packings of !
“jamming” spherical particles also led to such low values of
exponents n 关between 0.5 and 1 共Refs. 57 and 58兲兴. !
An indication of this transformation to a gel phase !
"# " !
through the creation of bonds between the micelles is the ! ! ! !
observation of the gel fraction after reswelling, i.e., the frac- ( ,67)
tion of gel that did not redisperse after ⬃15 h in UF perme-
ate under strong agitation. Indeed, the gel fraction increases %
continuously with casein concentration13 关Fig. 10共b兲兴, further !
,G)
confirming that more bonds are formed upon compression.
We find that the gel fraction also varies as 共C – Cmax兲s with !
%

s ⬇ 0.5– 1, s = 1 corresponding to the “ideal” case where the

gel fraction increases linearly with the number of bonds per #
!
234 233 ,-.)
micelle. Even if reswelling experiments indicate that intermi-
celle bonds are still present after ⬃15 h, it is important to
note that these junctions are probably not “permanent” as the #
!
NPC dispersions still flow at very long time scales
关Fig. 7共b兲兴.
In the same manner as in part A, it is interesting to !
compare the viscoelastic properties of NPC dispersions to
SC dispersions where the caseins are not structured into mi- !
!
celles. Figure 11共a兲 shows the frequency dependence of G⬘
and G⬙ of such SC dispersions at casein concentrations that
are similar to those reported for NPC in Fig. 7共a兲, i.e., from !! #!! #!! %!! %!!
134 to 343 g/l. Clearly, the viscoelastic behavior of SC is + ,-*/)
quite different from that of NPC 共see Figs. 7 and 11兲. Most
remarkably, SC dispersions gave moduli that were strongly FIG. 11. 共a兲 Frequency dependence of the elastic G⬘ 共solid symbols兲 and
loss G⬙ 共open symbols兲 moduli for SC dispersions in UF permeate at casein
dependent on frequency over the whole range of concentra-
concentrations similar to those in Fig. 7. The lines are guides for the eye. 共b兲
tions, whereas G⬘ and G⬙ were almost frequency indepen- The high-frequency 共f = 10 Hz兲 elastic G⬘ 共solid symbols兲 and loss G⬙ 共open
dent for NPC dispersions at C ⱖ 210 g / l 共Fig. 7兲. Moreover, symbols兲 moduli of SC 共triangles兲 and NPC 共small squares兲 dispersions in
SC dispersions show the typical behavior of entangled poly- UF permeate as a function of casein concentration. The line has a slope
of 11.
mer systems at all concentrations with a clear crossover of
G⬘ and G⬙ at a frequency that shifts gradually to lower val-
ues as concentration increases. All these observations match dispersions at C ⬍ 150 g / l, the high-frequency value of G⬘
those from authors who worked on comparable SC follows a much steeper power law where G⬘ ⬃ C11 关Fig.
systems.17,43 11共b兲兴. Besides the fact that the G⬘ values reported here are
Figure 11共b兲 gives the concentration dependence of the not strictly identifiable with plateau values 共a limit that we
elastic and loss moduli obtained at high frequency 共f did not reach in our experiments兲, this faster variation might
= 10 Hz兲 for both SC and NPC dispersions. As previously originate from local attractions between casein chains that
discussed, the NPC dispersion changes abruptly from a low- increase the resistance against deformation. The existence of
viscosity liquid to a viscoelastic gel in a very narrow range such a hyperentanglement regime was already suggested in
of casein concentrations 共⬃145– 178 g / l兲. Conversely, the the first part of this work when discussing the viscosity re-
high-frequency G⬘ and G⬙ of SC dispersions increase much sults of SC dispersions at C ⬎ 80 g / l 共Fig. 6兲. To some ex-
more slowly with concentration. For entangled polymer so- tent, these results are similar to those of English et al.,63
lutions in good solvent, the usual variation in the plateau which also showed a stronger concentration dependence
modulus G⬘p is a power law Cn with an exponent n 共n = 3.2兲 of G⬘p in associating polymer solutions compared to
⬇ 2 – 2.3 共Refs. 59 and 60兲 that is in agreement with the “standard” polymers. Then, starting from C ⬇ 150 g / l, G⬘
theoretical predictions of de Gennes61,62 共n = 9 / 4兲. For SC starts to deviate from this C11 power law and increases more
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.235.1.118 On: Tue, 14 Jan 2014 08:37:41
 165106-10 Bouchoux et al. J. Chem. Phys. 131, 165106 共2009兲

slowly with C 关Fig. 11共b兲兴, suggesting a crossover to another ⬃170 g / l兲 and at low-shear rates, the casein micelle
regime. This change in dynamic behavior is probably related dispersions behave qualitatively and quantitatively as
to the progressive “gelation” of the system. Indeed, simple the polydisperse hard-sphere fluid. At high shear rates,
manipulation of the most concentrated SC samples 共C a slight deviation from the hard-sphere fluid behavior
⬎ 300 g / l兲 produced by osmotic stress shows that they be- was observed, presumably caused by deformations of
have as elastic solids.13 Such a solidlike behavior at short the micelles in the flow. In all cases, the flow proper-
time scales is also visible in Fig. 11共a兲, where the SC disper- ties indicate that the micelles are well separated from
sion at 343 g/l shows a predominant elastic character at high each other and interact through excluded volume ef-
frequencies. On the other hand and in contrast to NPC, the fects only.
compression of SC dispersions is always fully reversible 共ii兲 Resistance to deformation at the liquid-solid transi-
共there is no remaining gel fraction after reswelling13兲, sug- tion and above. Starting from C ⬇ 130 g / l, the NPC
gesting that “gels” of SC are held together by weak connec- dispersions begin to show elastic properties. Below
tions with short lifetimes. All these observations further con- ⬃178 g / l, they behave as viscoelastic liquids with
firm that SC dispersions have features similar to those of properties similar to those of polydisperse hard-sphere
reversibly associating polymer solutions; such solutions in- systems. Accordingly, excluded volume effects are the
deed show a reversible gelation transition with concentration cause of the resistance to deformation at these con-
and/or temperature but still can easily evolve and transform centrations. At Cmax ⬇ 178 g / l, a phase transition
through the breaking and rebuilding of connections.64,65 takes place, and the NPC dispersions start to behave
The results presented in this second part can be summa- as elastic gels with a storage and a loss modulus G⬘
rized as follows. and G⬙ that are much less dependent on frequency
and concentration. Since Cmax corresponds to the
共1兲 Mechanical resistance of dispersions in which caseins
random-close packing of the micelles, we assume that
are associated into micelles (NPC). Up to a casein con-
the micelles are then forced to deform and deswell as
centration that matches the random-close packing of a
in the case of the concentrated emulsions studied by
40% polydisperse hard-sphere fluid 共␾rcp = 0.78 and C
Mason et al.48,49 In this concentration regime, the
⬇ 178 g / l兲, NPC dispersions still behave qualitatively
shear resistance must result from the resistance to de-
and quantitatively as polydisperse hard-sphere fluids;
formation of individual micelles, combined with the
this is in accordance with the viscosity results of part A.
formation of bonds between the micelles. The in-
At higher concentrations, there is another regime in
volvement of such intermicelle adhesions is consistent
which casein micelles deform and deswell as concen-
with our former observations of a remaining gel frac-
tration increases. In this regime, the NPC dispersions
tion after reswelling of highly dense NPC
behave as elastic gels and their resistance to shear re-
dispersions.13 In this last concentration regime, the
sults from the formation of bonds between micelles
magnitude of the elastic modulus is very close to that
combined with the elastic properties of the micelles. At
of the osmotic pressure. This correspondence could
very long times, however, dense NPC dispersions creep
indicate that the same forces produce the resistance of
under constant stress, indicating that these bonds be-
the NPC dispersions to both osmotic compression and
tween micelles may be nonpermanent.
shear deformation. The exact nature of these forces is
共2兲 Mechanical resistance of dispersions in which caseins
not known at present and is the aim of future work.
are not associated into micelles (SC). These dispersions
behave as associating polymer solutions over the whole Additionally, the results obtained with native casein mi-
range of concentrations. Starting from 150 g/l, they un- celle dispersions have been compared to those obtained with
dergo a slow and reversible gelation transition where SC dispersions in which the caseins are not structured into
weak connections are involved. These connections are micelles. Interestingly, the rheological properties of the SC
easily disrupted and do not withstand strong stress. dispersions were quite different from those of the native
form and quite close to those of common “hyperentangled”
or associative polymer solutions. This is further confirmation
IV. CONCLUSIONS
of the importance of structure in determining the rheological
The rheological properties of native casein micelle dis- behavior of casein micelle systems. It also gives further sup-
persions 共NPC兲 have been examined over a wide range of port to the idea that the forces involved in intermicelle cohe-
concentrations from dilute liquidlike dispersions 共⬃10 g / l兲 sion are structure-dependent forces that are disrupted when
to dense solidlike dispersions obtained through osmotic the micelles are dissociated.
stress 共up to ⬃400 g / l兲. Permanent flow and oscillatory
shear measurements were conducted; the results were ana- ACKNOWLEDGMENTS
lyzed in details and compared with previous results of os-
We thank J. Lefebvre for the help in the treatment of the
motic stress experiments.13 Accordingly, we can now pro-
creep-recovery data as well as R. Buscall and W. B. Russel
pose the following answers to the questions raised in the
for interesting discussions. We are also grateful to J. Fau-
introduction.
quant, F. Garnier-Lambrouin, and P. Schuck from UMR1253
共i兲 Resistance to flow at concentrations below or near the for preparing and providing us with the NPC powders and
liquid-solid transition. In this regime 共below the UF permeate of skimmed milk.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.235.1.118 On: Tue, 14 Jan 2014 08:37:41
 165106-11 Rheology of casein micelle dispersions J. Chem. Phys. 131, 165106 共2009兲

1
C. Holt, Adv. Protein Chem. 43, 63 共1992兲. 35
W. B. Russel, D. A. Saville, and W. R. Schowalter, Colloidal Dispersions
2
D. S. Horne, Curr. Opin. Colloid Interface Sci. 11, 148 共2006兲. 共Cambridge University Press, Cambridge, 1989兲, p. 466.
3 36
D. G. Dalgleish, P. A. Spagnuolo, and H. D. Goff, Int. Dairy J. 14, 1025 Y. Saiki, C. A. Prestidge, and R. G. Horn, Colloids Surf., A 299, 65
共2004兲. 共2007兲.
4
C. G. De Kruif, J. Dairy Sci. 81, 3019 共1998兲. 37
D. C. Boris and R. H. Colby, Macromolecules 31, 5746 共1998兲.
5
D. J. McMahon and W. R. McManus, J. Dairy Sci. 81, 2985 共1998兲. 38
S. Dou and R. H. Colby, Macromolecules 41, 6505 共2008兲.
6
D. S. Horne, Curr. Opin. Colloid Interface Sci. 7, 456 共2002兲. 39
M. Rubinstein, R. H. Colby, and A. V. Dobrynin, Phys. Rev. Lett. 73,
7
S. Marchin, J. L. Putaux, F. Pignon, and J. Léonil, J. Chem. Phys. 126, 2776 共1994兲.
045101 共2007兲. 40
J. D. Ferry, Viscoelastic Properties of Polymers 共Wiley, New York, 1980兲,
8
D. J. McMahon and B. S. Oommen, J. Dairy Sci. 91, 1709 共2008兲. p. 509.
9
C. G. De Kruif and E. B. Zhulina, Colloids Surf., A 117, 151 共1996兲. 41
T. Annable, R. Buscall, R. Ettelaie, and D. Whittlestone, J. Rheol. 37,
10
R. Tuinier and C. G. De Kruif, J. Chem. Phys. 117, 1290 共2002兲. 695 共1993兲.
11
G. Brans, C. G. P. H. Schroen, R. G. M. Van Der Sman, and R. M. Boom, 42
T. Annable, R. Buscall, and R. Ettelaie, Colloids Surf., A 112, 97 共1996兲.
J. Membr. Sci. 243, 263 共2004兲. 43
A. Pitkowski, D. Durand, and T. Nicolai, J. Colloid Interface Sci. 326, 96
12
P. Schuck, Lait 82, 375 共2002兲. 共2008兲.
13
A. Bouchoux, P. E. Cayemitte, J. Jardin, G. Gésan-Guiziou, and B. Ca- 44
J. M. Manski, L. E. van Riemsdijk, A. J. van der Goot, and R. M. Boom,
bane, Biophys. J. 96, 693 共2009兲. Biomacromolecules 8, 3540 共2007兲.
14
A. Hadj-Sadok, A. Pitkowski, T. Nicolai, L. Benyahia, and N. Moulai- 45
J. Lefebvre, Rheol. Acta 45, 525 共2006兲.
Mostefa, Food Hydrocolloids 22, 1460 共2008兲. 46
W. Wolthers, D. van den Ende, V. Breedveld, M. H. G. Duits, A. A.
15
J. A. Lucey, M. Srinivasan, H. Singh, and P. A. Munro, J. Agric. Food Potanin, R. H. W. Wientjes, and J. Mellema, Phys. Rev. E 56, 5726
Chem. 48, 1610 共2000兲. 共1997兲.
16
S. J. Radford and E. Dickinson, Colloids Surf., A 238, 71 共2004兲. 47
17 S. Mortimer, A. J. Ryan, and J. L. Stanford, Macromolecules 34, 2973
D. Farrer and A. Lips, Int. Dairy J. 9, 281 共1999兲.
18 共2001兲.
A. Pierre, J. Fauquant, Y. Le Graët, M. Piot, and J. L. Maubois, Lait 72, 48
T. G. Mason, J. Bibette, and D. A. Weitz, Phys. Rev. Lett. 75, 2051
461 共1992兲.
19 共1995兲.
M. H. Famelart, F. Lepesant, F. Gaucheron, Y. Le Graët, and P. Schuck, 49
T. G. Mason, M. D. Lacasse, G. S. Grest, D. Levine, J. Bibette, and D. A.
Lait 76, 445 共1996兲.
20 Weitz, Phys. Rev. E 56, 3150 共1997兲.
C. David, F. Pignon, T. Narayanan, M. Sztucki, G. Gesan-Guiziou, and 50
R. A. Lionberger and W. B. Russel, J. Rheol. 38, 1885 共1994兲.
A. Magnin, Langmuir 24, 4523 共2008兲. 51
21
A. O. Karlsson, R. Ipsen, K. Schrader, and Y. Ardo, J. Dairy Sci. 88, F. M. Horn, W. Richtering, J. Bergenholtz, N. Willenbacher, and N. J.
3784 共2005兲. Wagner, J. Colloid Interface Sci. 225, 166 共2000兲.
52
22
J. Kromkamp, S. Rijnsent, R. Huttenhuis, K. Schroen, and R. Boom, J. I. T. Kim and P. F. Luckham, J. Colloid Interface Sci. 144, 174 共1991兲.
53
Food Eng. 80, 257 共2007兲. T. G. Mason and D. A. Weitz, Phys. Rev. Lett. 75, 2770 共1995兲.
54
23
D. S. Horne, Colloids Surf., A 213, 255 共2003兲. A. Weiss, M. Ballauff, and N. Willenbacher, J. Colloid Interface Sci.
24
M. Alexander, L. F. Rojas-Ochoa, M. Leser, and P. Schurtenberger, J. 216, 185 共1999兲.
55
Colloid Interface Sci. 253, 35 共2002兲. G. Fritz, B. J. Maranzano, N. J. Wagner, and N. Willenbacher, J. Non-
25
T. Huppertz, A. L. Kelly, and C. G. De Kruif, J. Dairy Res. 73, 294 Newtonian Fluid Mech. 102, 149 共2002兲.
56
共2006兲. L. Guo, R. H. Colby, C. P. Lusignan, and A. M. Howe, Macromolecules
26
P. Muller-Buschbaum, R. Gebhardt, S. V. Roth, E. Metwalli, and W. 36, 10009 共2003兲.
57
Doster, Biophys. J. 93, 960 共2007兲. C. S. O’Hern, S. A. Langer, A. J. Liu, and S. R. Nagel, Phys. Rev. Lett.
27
P. Schuck, M. Piot, S. Méjean, Y. Le Graët, J. Fauquant, G. Brulé, and J. 88, 075507 共2002兲.
58
L. Maubois, Lait 74, 375 共1994兲. C. S. O’Hern, L. E. Silbert, A. J. Liu, and S. R. Nagel, Phys. Rev. E 68,
28
P. Segalen, M. Boulle, and G. Gwozdz, in Laits et Produits Laitiers. 2. 011306 共2003兲.
59
Les Produits Laitiers, Transformations et Technologies, edited by F. M. Y. H. Lin and J. H. Juang, Macromolecules 32, 181 共1999兲.
60
Luquet 共Tec et Doc - Lavoisier, Paris, 1985兲, p. 147. Juliani and L. A. Archer, Macromolecules 35, 6953 共2002兲.
61
29
R. Jenness and J. Koops, Neth. Milk Dairy J. 16, 153 共1962兲. R. H. Colby and M. Rubinstein, Macromolecules 23, 2753 共1990兲.
62
30
C. Bonnet-Gonnet, L. Belloni, and B. Cabane, Langmuir 10, 4012 J. D. Ferry, Viscoelastic Properties of Polymers 共Wiley, New York, 1980兲,
共1994兲. p. 501.
31 63
I. M. Krieger and T. J. Dougherty, J. Rheol. 3, 137 共1959兲. R. J. English, J. H. Laurer, R. J. Spontak, and S. A. Khan, Ind. Eng.
32
C. G. De Kruif, E. M. F. van Iersel, A. Vrij, and W. B. Russel, J. Chem. Chem. Res. 41, 6425 共2002兲.
64
Phys. 83, 4717 共1985兲. S. K. Kumar and A. Z. Panagiotopoulos, Phys. Rev. Lett. 82, 5060
33
J. G. Berryman, Phys. Rev. A 27, 1053 共1983兲. 共1999兲.
34
W. Schaertl and H. Sillescu, J. Stat. Phys. 77, 1007 共1994兲. 65
S. K. Kumar and J. F. Douglas, Phys. Rev. Lett. 87, 188301 共2001兲.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.235.1.118 On: Tue, 14 Jan 2014 08:37:41