4012 Langmuir 1994,10, 4012-4021

Osmotic Pressure of Latex Dispersions

C. Bonnet-Gonnet,? L. Belloni,t and B. Cabane*ptJ
Equipe mixte CEA-RP, Rhbne Poulenc, 93308 Aubervilliers, France,
and Service de Chimie Moldculaire, CEN Saclay, 91191 Gif sur Yvette, France
Received April 6, 1994. In Final Form: July 15, 1994@

The resistance of aqueous latex dispersions to removal of water has been measured through osmotic
compression. For pure polystyrene particles stabilized with sulfate groups anchored at their surfaces,the
measured pressures match the predictions of models for the interactions of charged spheres in water. For
copolymerparticles (styrene-butyl acrylate)covered with acrylic acid sequenceslinkedto the core polymers,
the measured pressures are considerablyhigher than the predictions. This strongresistance to dehydration
originates from polyelectrolytes which are released by the particles or extend fiom their surfaces into the
aqueous phase. Consequentlythe behavior of such dispersions is determined by the content ofthe aqueous
phase that separates the particles.

Introduction The interactions between particles dispersed in water

Water-based dispersions of small polymeric particles
are called latex dispersions.' Such dispersions are
obtained through an emulsion polymerization process,
where hydrophobic monomers are transferred to growing
polymer chains in the particlesS2Hydrophilic species are
often added during the synthesis; they form membranes
around the particles and generate electrostatic repulsions
between them, which stabilize the dispersion against
aggregati~n.~
The constraints on this stability are severe, because
the dispersions are made and stored at high volume
fractions (4 L 0.5) where interparticle distances are short
(on the order of 10nm) and collision frequencies are quite
high. Even when electrostatic repulsions are adequate,
instability may develop after a long period of time, or in
high stress devices (pumps or filtration systems). It is
found that the stability of latex dispersions varies
considerably, according to the nature of the hydrophilic
membranes. Therefore it would be most useful to have
a method to measure this stability in the conditions of
application.
The usual way to measure the stability of colloidal
dispersions is to dilute the dispersions, depress their
stability by addingsalt, and measure the aggregation rate.*
The result is expressed as a critical coagulation concen-
tration, which is the salt concentration where the ag-
gregation of particles is no longer inhibited by electrostatic
barriers. This result measures the limit of electrostatic
stability, but not the stability in the conditions of
applications, where the volume fractions are high and
there is no added salt. In such conditions, the resistance
against aggregation is determinedby interactions between
latex particles at short distances. Thus a more general
goal would be to measure the interactions oflatex particles
in this range of distances.
T Equipe mixte CEA-RP.
* Service de Chimie MolBculaire.
@ Abstract published in Advance A C S Abstracts, September 15,
1994.
(1)(a)Vanderhof,J.W.; Bradford, E. B.; Carrington,W. K J.Polym.
Sci. Symp. 1873,41,166. (b) Kast, H. Makromol. Chem. Suppl. 1886,
10-2 1,447.(c)Zozel, A,;Heckmann,W.; Ley, G.; Mlchtle,W. Macromol.
Symp. ISSO,35-36,423.
(2)Van den Hul, H. J.,Vanderhoff, J. W. Br. Polym. J.1970,2,121.
(3)(a) Greene, B. W., Sheetz, D. P., Filer, T. D. J.Colloid Interface
Sci. 1870,32,90.(b) Greene, B. W. J. Colloid Interface Sci. 1873,43,
449.(c) Greene, B. W.: Nelson, A. R.: Keskev, _ .W. H. J. Phrs. Chem.
1980,84,1616.
(4)Vervey, E. J. W.; Overbeek, J. Th. G. Theory of the Stability of
Lyophobic Colloids; Elsevier: Amsterdam, 1948.
can be measured through three different methods. (1)
Use two macroscopic surfaces of the same nature as the
particle surfaces, and measure the force-to-distance rela-
tion with a surface forces apparatuss or an atomic force
microscope.6 At present this is not practical, because it
is difficult to stick common latex particles to the surfaces
of the SFA or AFM. (2) Measure the spontaneous
fluctuations of particle volume fraction, using the scat-
tering of incident radiation (light, X-rays, or neutrons).
However, latex dispersions are not transparent to light
and do not provide good contrast for X-rays; neutrons are
not readily available. (3) Use membrane exchanges to
measure the relation of osmotic pressure to volume fraction
in the dispersions. This method was pioneered by
Hachisu,' Rand and Parsegian,6Nagy and H ~ r k a yand ,~
Parentich and Ottewill.lo It is always applicable and
requires very little equipmentbut a fair amount of physical
chemistry work; it measures a behavior which is very close
to that observed in applications,i.e. the resistance to forced
deswelling.
In this paper we report the use of the osmotic stress
method to measure interactions between particles in latex
dispersions. Osmoticpressures and volume fractionshave
been measured for different latex dispersions, and they
have been compared with predictions based on models for
electrostatic interactions of charged spheres. If the
measured pressures do match the predictions, we may
conclude that the dispersion is well described as a collection
of hard spheres bearing electrical charges on their surfaces.
If the measured pressures are considerably higher than
the predictions, then we may conclude that the resistance
to deswelling results from interactions between charges
which are not localized on the particle surfaces.
(6)(a) Israelachvili, J. N. J.Colloid Interface Sci. 1979,44,269.
(b)
Israelachvili, J.N.;Adams, G. E. J.Chem.Soc., Faraday Trans.1 1978,
74,976.(c) de Costello,B. A,; Kim,1. T.; Luckham, P. F.; Tadros, Th.
F. Colloida Surfaces 1883,77,66.
(6)Binnig,G.;Quate, C. S.;Gerber, Ch. Phys. Reu. Lett. 1886,56,
930.
(7) Hachisu, S.,Kobayashi,Y. J. Colloid Interface Sci. 1874,46,470.
(8)(a) Parsegian, V. A,; Fuller, N. L.; Rand, R. P. Proc. Natl. Acad.
Sci. U S A . 1878,76,2750.(b) Parsegian, V. A,; Rand, R. P.; Fuller, N.
L.;Rau, D. C. In Methoda in Enzymology: Bwmembranes, Protons and
Water, Structure and Translocation; Packer, L., Ed.; Academic Press:
New York, 1986;Vol. 127,p 400.
(9)(a)Nagy, M.,Horkay, F. Acta Chim. Acad. Sci. Hung. 1880,103,
387.(b)Rohrsetzer, S.;Kovacs, P.; Nagy, M. Colloid Polym. Sci. 1986,
264,812.
(10)Goodwin, J. W.; Ottewill, R. H.; Parentich, A. Colloid Polym.
Sci. 1980,268,1131.

0743-7463/94/2410-4012$04.50/00 1994 American Chemical Society

Osmotic Pressure of Latex Dispersions
Methods
Osmotic stress methods are based on exchanges ofwater
molecules between the sample and a reservoir. In nature,
similar processes govern the passage of water and ions
between cells and their environment. In the laboratory,
the sample is pIaced in contact with a dialysis membrane
through which it exchanges water, ions, and surfactants
with a reservoir containinglarge amounts ofthese species.
At equilibrium, the chemical potentials of all species able
to cross the membrane are identical in the sample and in
the reservoir; therefore it is possible to control the state
of the sample through manipulation of these chemical
potentials. This is particularly useful for colloidal systems
where the adsorption of ions (H+, OH-) or molecules
(surfactants) from the aqueous solution onto the particle
surfaces determines the surface charge, and therefore the
stability of the dispersions.
The same equilibria make it possible to extract water
from the dispersion by adding to the resemoir high molar
mass species which bind water. At equilibrium, the
chemical potentials of water on either side of the mem-
brane are equal, and therefore the osmotic pressure of the
sample equals that of the polymer in the reservoir.
Through this procedure, dispersions may be concentrated
at a set value of the osmotic pressure.
The stressing polymer was dextran, chosen because it
gives the same osmotic pressure regardless of ions and
temperature.8bJ1The grade of dextran was T110, obtained
from Fluka. The distribution of molar mass in this
polymer was determined through size exclusion chroma-
tography (SEC);the number average molar mass was Mn
= 70 000, and the weight average was Mw = 150 000.The
osmotic pressures of dextran solutions were determined
with a Knauer membrane osmometer for pressures up to
4000 Pa (dextran concentrations between 0.1%and 5%).
At higher pressures the values were from the intensity of
scattered light, which gives the osmotic compressibility
(dextran concentrations between 0.2% and 15%).12In the
range where both methods overlapped, they gave the same
pressures. The results were fitted by polynomial expres-
sions for pressure ll (in Pa) versus concentration c in
weight percent:
TI = 286c + 87c2 + 5c3 (1)
The linear term corresponds to a number average molar
mass Mn = 87 000 g/mol, in good agreement with the
results from SEC. The next term is related to the second
virial coefficient, which corresponds to an exclusion
diameter of 131 8 for each macromolecule; this is
consistent with the hydrodynamic diameter obtained
through quasielastic light scattering, which is 177 8.
Finally, at high pressures, the values calculated from eq
1 may be compared with those measured by Rand and
Parsegian,8b and by VBr6to~t.l~These authors used
concentrated dextran solutions of molar mass 500 000.
Their published values match those calculated from eq 1,
even though the dextran samples are different. This is
not unexpected, since the macromolecules in these con-
centrated solutions overlap t o the extent that the sizes of
individual macromolecules become irrelevant. For these
reason we conclude that expression 1represents well the
(11) (a) Vink, H.Eur. Polym. J. 1971, 7, 1411. (b) Le Neveu, D. M.;
Rand, R. P.;Parsegian,V. A.Nuture(London) 1976,259,601.(c)Prouty,
M. S.;Schechter, A. N.; Parsegian, V. A. J.Mol. Biol. 1986,184, 517.
(12) BonnebGonnet,C. Ph.D.Thesis,Universitk Pierre et Marie Curie
Paris 6, 1993.
(13) Vkrktout, F.;Delaye, M.; Tardieu, A. J. Mol. Biol. 1989,205,
713.
Langmuir, Vol. 10,No. 11, 1994 4013
osmotic pressures of our dextran solutions in the whole
range of concentrations, which was 0.1 to 40% in weight
percent.
The membranes used for osmotic stress experiments
were dialysis bags which allowed exchange of water and
ions but not of polymers or particles; this was obtained
with a cutoff at a molar mass of 15 000 (Visking 8/32).
Prior to the experiments, these bags were conditioned in
water at the appropriate ionic strength and pH. Then
latex dispersions were placed in the bags and immersed
in the stressing solution of dextran. During the initial
deswelling of the dispersions it was necessary to refill the
bags with more latex dispersion (in order to obtain a
sufficient amount of concentrated dispersion) and to
exchange the stressing solutions (in order to avoid dilution
with water from the dispersion);equilibrium was reached
after 3 weeks.
Then the content of each bag was extracted and the
latex volume fraction was determined through drying at
120 "C. The reservoir was also analyzed, and its dextran
concentration was measured through SEC or through total
organic carbon analysis (TOC); from this concentration
the osmotic pressure was calculated according to expres-
sion 1. Consequently, the osmotic pressure of the sample
was known, and its latex volume fraction was measured;
from these measurements one point of the equation of
state of the dispersion was determined.
Materials
Three types of latex dispersions have been studied, one
containing pure polystyrene particles, and the other two con-
taining copolymer particles.
PS Latex. This dispersion was obtained from IDC (ref
815226). The particle cores are made of pure polystyrene, with
glass transition temperature 110 "C; the Hamaker constant of
pure polystyrene is 2kT.14 The particle diameters are mono-
disperse a t 61 nm. The core surfaces are largely hydrophobic,
but they carry a few charged groups: sulfate groups which are
the ends of the polystyrene chains, and sulfonate groups from
styrene sulfonate monomers which are copolymerized with the
styrene monomers. The total surface density of charged groups
has been measured through potentiometric titration;ls it is 2 =
0.14 e/nm2. The aqueous phase ofthese dispersions is pure water.
L1 Latex. This dispersion was synthetized through emulsion
copolymerization of a mixture of hydrophobic and hydrophilic
monomers.16 The particle cores are made of styrene-butyl
acrylate copolymers, with glass transition temperature 25 "C;
their Hamaker constant may be estimated as a linear combination
of the Hamaker constants of polystyrene and poly(methy1-
methacrylate), which yields 2.5kT.14The particle diameters are
nearly monodisperse a t 115 nm. Each particle core is protected
from water by a membrane made of a copolymer of acrylic acid
(AA,42%), butyl acrylate (BA, 35%),and styrene (S, 32%). The
membrane polymers are copolymerized with the core polymers.
The total amount of acrylic acid in the surface layer is 2%of the
total latex mass. The total surface density of charged groups
has been measured through potentiometric titration;16 it is 2 =
3.5 e/nm2a t pH 9 and 2 = 0.4 e/nm2at pH 3. The aqueous phase
of these dispersions contains water, loose membrane polymers,
and salt.
L2 Latex. This dispersion is similar to L1,but the particle
cores are made of styrene-butadiene copolymers, with glass
transition temperature 0 "C; their Hamaker constant is 2.3kT.14
The particle diameters are nearly monodisperse at 170 nm. Each
particle core is protected from water by a membrane made of
monomers with carboxylic acid groups (37%)copolymerizedwith
(14)Russel, W. B.; Saville, D. A.; Schowalter, W. R. Colloidal
Dispersions; Cambridge University Press: Cambridge, U.K., 1989.
(15) Van den Hul, H. J.; Vanderhoff, J. W. J. Electroanal. Chem.
1972,37, 161.
(16) (a)Chevalier,Y.;Pichot, C.;Graillat, C.;Joanicot, M.;Wong, K.;
Maquet, J.; Lindner, P.;Cabane, B. Colloid Polym. Sci. 1992,270,806.
(b) Joanicot, M.; Wong, IC; Richard, J.; Maquet, J.; Cabane, B.
Macromolecules 1993,26, 3168.
4014 Langmuir, Vol. 10, No. 11, 1994
hydrophobic monomers (63%). The total amount of acid mono-
mers in the surface layer is 1.4% of the total latex mass. The
total surface density of charged groups is 2 = 4.3 dnm2a t pH
8.6 and 2 = 2.7 dnm2 at pH 5. The aqueous phase of these
dispersions contains water, loose membrane polymers, and salt.
All these dispersions were washed (Le. the aqueous phase was
exchanged with pure water) through tangential ultrafiltration
and then exchange of ions on ion exchange resins (Amberlite
IRN 77 and IRN 78). Subsequently they were brought to the
desired pH and ionic strengthlthrough dialysis against solutions
of sodium hydroxide or sodium chloride.
PolyacrylicAcid (PAA). Pure sodium polyacrylate solutions
were also made in order to estimate the resistance of the particle
membranes to osmotic compression. These solutions were made
with sodium polyacrylate bought from Fluka (ref 81138). The
distribution of molar mass in this polymer was determined
through SEC;the number average was Mn = 110 000 g/mol and
the weight average was Mw = 330 000 g/mol.
In aqueous solutions, polyacrylate macromolecules expand t o
large dimensions. Their average radius can be calculated from
the viscosity laws measured for sodium polyacrylate in water.17J8
Above a certain concentration, denoted 0*,the macromolecules
are forced to overlap and form a semidilute solution. The overlap
concentration can be calculated from the dimensions of the
macromolecules;the values are given below for the polyacrylate
used in this study.
0*(g/g)
p H = 9 -Z=lOmSM 0.0034
pH = 9 - Z = 0.1 M 0.0082
pH = 3 - I = M 0.0142
pH = 3 - Z = 0.1 M 0.0291
In the present work, the polyacrylate concentration was always
above the overlap concentration. Consequently, the macromol-
ecules were fully interpenetrated, and their osmotic pressure
was independent of molar mass.
Models
The osmotic pressure of colloidal dispersions is mea-
sured as a resistance to a decrease in the volume available
to the particles; this resistance originates from an increase
of energy (interparticle forces) or from a loss of entropy
(configurations) or from both. In order to calculate this
pressure, it is necessary to have some information on the
interactions and configurations of the particles; the
required information is the pair distribution functiong(r)
of interparticle distance r, and the pair interaction energy
(or “potential”)u(r).19 If these functions were known, the
pressure could be calculated from the virial equation:
where I7is the pressure, Q the number density ofparticles,
k the Boltzmann constant, T the temperature, and u’(r)
the interparticle force.
However, g ( r ) and u(r) are not known; they must be
calculated from the known quantities, which are the
number density 8 , the surface charge Z of a particle, and
the concentration of salt in the reservoir, expressed as a
number density of ions e’s. At present, there is no model
which calculates these functions accurately at all volume
fractions ofthe dispersion; however, there are some models
which are very good approximations at low volume
fractions, where the particles do not get in each other’s
(17)Encyclopedia of Polymer Science and Technology; Interscience
Publishers, 1971; Vol. 14,p 736.
(18)Van Kreveld,M. E.; Van den Hoed, N. J. Chromatogr. 1978,83,
111.
(19)Hansen, J. P.; MacDonald, I. R. Theory of Simple Liquids;
Academic Press: London, 1986.
Bonnet-Gonnet et al.
way, and others which are good at high volume fractions,
where the particle positions are fixed.
(1)Hard-SphereGas. In the limit of very low volume
fractions, the osmotic pressure of a colloidal dispersion
must follow the law of perfect gases, which is the first
term of eq 2. For dispersions which are somewhat more
concentrated, the loss of available configurations caused
by the excluded volume around each particle can be taken
into account through an expansion in powers of the volume
fraction 4. A good approximation is provided by the
Carnaham-Starling (CS) equation:
n 1+4+42-43 (3)
ijE= ( 1 - 413
In the case of charged particles which repel each other
through overlap of their ionic clouds, the volume fraction
& ~ f f used by the particles must include the volume of the
ionic clouds; good results (up to = 0.6)are obtained if
the extension of ionic clouds is chosen equal to the Debye
screening length r 1 . 1 4
(2) One-Component Model. At still higher volume
fractions, or with strongly repelling particles, it is neces-
s a r y t o describe properly the interactions and the distances
between the particles. In the “one-componentmodel”this
is done by assuming that the particles interact directly
with each other through an effective potential. This
potential is calculated numerically accordingto the DLVO
theory and the procedures described below.4 Once the
interparticle potential is known, the resulting distribution
of distances g(r) is calculated according to an integral
equation (HNC) commonly used in the theory of simple
1iq~ids.l~ Finally, the pressure is calculated from the virial
equation (2).
This model works quite well at intermediate volume
fractions but fails at high volume fractions, for three
reasons. First, the DLVO pair potential is not calculated
from the correct Poisson-Boltzmann (PB) equation but
from a linearized version. At large distances this linear-
ized equation gives the same solution as the correct
equation if it uses an effective charge 2,~ which is lower
than the true surface charge 2 of the particle; however,
this procedure fails when the particles are at short
distances. Second, the screening length used in DLVO
theory is calculated from the concentration of passive salt
in the reservoir; however, at high volume fraction, this
does not match the concentration of salt in the sample.
Third, even with a good pair potential, at high volume
fractions the integral equations may not converge to the
correct distribution of distances if the correlations are
quite strong.
These shortcomings originate from trying to describe
the dispersion as a gas of particles and assuming that the
ionic clouds are simply tied to the particles; then the
pressure is proportional to the number density ofparticles,
and the ions are included only through their effect on the
interparticle potential. This simplifies the problem, as it
reduces it to a calculation of interparticle distances and
their fluctuations. However, a t high volume fractions,
this description is inaccurate, for the reasons given above.
In fact, when the interparticle distances are short, the
resistance to a decrease in volume originates from the
strong overlap of ionic clouds; then the pressure is
proportional to the number density of ions, while the
particles come in only through their effect on the ionic
density profiles.
(3) Poisson-Boltzmann Models. The next two
models tackle this situation by calculating the pressure
from the excess concentration of ions caused by the
overlap of ionic clouds. For a dispersion containing
counterions (in our case cations) at a local concentration
Osmotic Pressure of Latex Dispersions
e+(r)and co-ions (in our case anions) at a local concentra-
tion e-(r),the osmotic pressure with respect to a reservoir
containing salt at a concentration e’s iszo
-
n = [e++ @-]E=() - 2e‘, (4)
kT
where E = 0 indicates that the concentrations of coun-
terions and co-ions are measured at a point between
particles where the electric field vanishes (closer to a
particle, the concentrations may be higher or lower because
of the nonvanishing electric field, but of course the
chemical potentials are the same).
(4) Poisson-Boltzmann Cell (PBC)Model. In this
model the distribution of ions around each particle is
calculated from the Poisson-Boltzmann (PB) equation
with boundary conditions (the field vanishes) on a spheri-
cal shell of radius b around the particle.21 All counterions
of the particle are within this boundary (the cell is
electrically neutral); in addition, there are ions from
passive salt, at a chemical potential which is determined
by the reservoir. The model calculates the reduced
electrostatic potential q(r)from the PB equation, and then
the concentration profiles of counterions, g+(r),and of co-
ions, e-(r),from the conditionthat their chemical potential
equals that of passive salt which is in the reservoir at a
concentration e‘,:
e+(r>= e’, exp[-v(r)I e-(r) = e’, exp[+q(r)l ( 5 )
Then the pressure may be calculated from eqs 4 and 5;
it can be expressed as the difference, at the cell boundary
( r = b), between the arithmetic and geometric averages
of e+ and e-:
n + e-(b)l - 2[e+(b)e-(b)lY2
- = [(e+(b> (6)
kT
This model provides an excellent description of the
electrostatic effects in the middle range of volume frac-
tions. It is also the only proper way to calculate the
concentration of salt in the dispersion. Its main short-
coming is the spherical shape of the cell, which does not
match the unit cell in a very concentrated dispersion,
where the neighboring particles come very close to the
reference particle. In particular, this model does not
predict the divergence ofthe pressure at a volume fraction
Q = 0.74,because it does not prevent overlap ofneighboring
particles.
(6) Poisson-Boltzman Model with Planar Sym-
metry. When the particle surfaces are very close to each
other, the repulsions are generated in thin layers of water
which have a nearly lamellar geometry. Then it is more
appropriateto use the PBC with planar symmetry, where
the PB equation is solved in one dimension with bound-
aries at the particle surface and at the plane located
halfway between the particles. As in the previous model,
the pressure between planes is calculated from eq 4 or 6,
but the number densities of ions are taken at the midplane
where the electric field vanishes.
This model is particularly appropriate for deformable
particles which turn into dodecahedra under strong
compression, as in the case of film-forming latex disper-
sions.16 The pressure exerted on a planar face can be
multiplied by the area ofthe face to give the force between
(20) Marcus, R. A. J. Chem. Phys. 1955,23, 1057.
(21)(a) Katchalsky, A.;Alexandrowicz, 0. In Chemical Physics of
ZonicSolutions;Conway,B.E.,Barradas, R. G., Eds.;Wiley: NewYork,
1966.(b)Bell, G.M.; Dunning, A. J. Trans.Faraday SOC.1970,66,500.
(c) Belloni, L.;Drifford, M.; Turq, P. Chem. Phys. 1984,83, 147.
Langmuir, Vol. 10,No. 11, 1994 4015
two dodecahedral particles; then the pressure of the
dispersion is calculated from the virial equation (2)limited
to first neighbors at the lattice positions. For a face-
centered cubic lattice of particles, where a is the lattice
parameter and u’(r)the force between neighboring par-
ticles with a center to center distance ~ 1 4 2the , virial
equation yields
(7)
For dodecahedral particles, the force is calculated as
explained above from the pressure lIpexerted on opposing
faces separated by a distance (a142 - D):
In this formula,the coefficientwhich takes the geometry
of the particles into account is of order unity; therefore
the osmotic pressure of the dispersion is nearly equal to
the pressure between infinite planes at a separation
determinedby the volume fraction of water and the particle
size.
Alternatively,ifthe particles remain spherical, the force
between spherical surfaces can be calculated from an
integration of the pressures exerted on planes located at
different distances, followed by the use of the Derjaguin
equation.* The virial equation is then written as in eq 7,
but the force uYr) between neighboring spheres can now
be expressed according to the potential up between
opposing planes:
This model behaves well at high volume fractions; in
particular, it reproduces the divergence of the pressure at
4 = 0.74 for dispersions of spherical particles and at 4 =
1for dispersions of deformable particles. It is not as good
at low volume fractions, where the particles are far apart,
because it no longer represents properly the shape of the
unit cell.
Attractive Forces. Finally, it may be necessary to
include the effect of attractive forces. For Van der Waals
attractions the interaction energy between particles of
radius R and Hamaker constant H is4
In all the situations investigated, this attractive con-
tribution was much smaller than the repulsive contribu-
tion from electrostatics; therefore it was neglected. It
becomes significant only at high ionic strengths, compa-
rable to the critical coagulation concentration of the
dispersion, i.e. 0.08 M for the PS latex and > 1M for the
other dispersions.
Results and Discussion
PS Latex. Dispersions of the PS latex were compressed
by osmotic stress at pH 7 and ionic strength 3 x M.
The original dispersion, at a volume fraction Q = 0.06,
was fluid and white; as the pressure was raised, it became
iridescent and viscous; at a pressure of 1800Pa the volume
fraction reached 0.214 and the dispersion turned to a soft
solid. At this volume fraction the mean surface to surface
separation is 31 nm; it is comparable with the Debye
screening length (55 nm). Previous work on latex disper-
4016 Langmuir, Vol. 10, No. 11, 1994
0.1 -I
~
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Figure 1. Osmotic pressures of PS latex dispersions (particle
diameter 61 nm, surface charge density 0.14 dnma)in water
at pH 7 and ionic strength 3 x M. Horizontal scale: volume
fraction of latex in the dispersion. Vertical scale: osmotic
pressure, log 10 scale. Squares: measured osmotic pressures
and volume fractions. Full line: calculated values for a perfect
gas of particles. Dashed line: values calculated through the CS
equation for particleswithhard sphere repulsions at an effective
+
radius Res = R K - ~ .
sions indicates that in such conditions the dispersions
order as colloidalcrystals.22Thereforethe iridescent colors
were the result of Bragg diffraction on reticular planes of
particles within the crystallites.
The experimental pressure us volume fiaction curve is
shown in Figure 1. Over a large range ofvolume fractions
(0.1-0.Q the pressures are between 1000and 10 000Pa.
These pressures are 3 orders of magnitude higher than
the pressures expected for a perfect gas at the same
number concentration. This discrepancy may have two
possible causes: an error in the number of objects
contributing to the pressure or a highly nonideal behavior
caused by the interactions between particles. For this
latex, the number of particles is very well known, and we
have found no other soluble species which could also
contribute to the measured osmotic pressures. Therefore
the discrepancy is more likely to come from interactions,
which are expectedto be quite strong since they are almost
unscreened.
At very low volume fractions, the effect of such repul-
sions can be estimated through the CS formula (eq 3)
applied to effective hard spheres with a radius equal to
the particle radius plus the Debye screening length. The
pressure of this hard-sphere gas is also shown in Figure
1;it rises extremely fast with volume fraction and crosses
above the data beyond Q = 0.05;this is because these
effective hard spheres use a large volume fraction;
assuming that they cannot overlap overestimates the
actual pressure.
The pressures calculated for dispersions of charged
spheres are shown in Figure 2. For this calculation, the
parameters are the particle diameter, their surface charge,
and the ionic strength of the reservoir. All these param-
eters are determined experimentally; therefore there are
no adjustable parameters. At low volume fractions it is
necessary to take into account the distribution ofdistances
between interacting particles; this was done through the
HNC equation used in the OCM model. The calculations
(22) Monovoukas, Y.;Gast, A. 3. Colloid Interface Sci. 1989, 128,
633.
Bonnet-Gonnet et al.
100000
T 8
1
/
10000
loo0 t a
’
/
/
/
11‘’
/
/
I
I 0
100 -r 1 I I
0.1 0.3 0.4 0.5 0.7
0 0.2 0.6
Figure 2. Data as in Figure 1.Fits according to electrostatic
models for interacting charged spheres. Dashed line: OCM
model. Full line: PBC model.
converge for volume fractions up to Q = 0.2; the results
are close to the data, but below.
The validity of this procedure becomes questionable
when the particles crystallize on a lattice. According to
the iridescent colors of the samples, this occurs at Q =
0.05. Then it is more appropriate to use a lattice model
such as the PBC model (eq 6). The calculated pressures
reproduce the measured pressures accurately over the
range ofvolume fractions 0.1-0.55 (Figure 2). At higher
volume fractions the model underestimates the resistance
to osmotic compression, because it does not describe the
unit cell properly; as explained above, the calculated
pressures diverge only at Q = 1instead of Q = 0.74.
The good agreement between the measured pressures
and those calculated by the PBC model indicates that the
PS latex dispersions are well described as dispersions of
charged spheres repelling through overlap of their ionic
clouds.
L1 Latex. Dispersionsofthe L1 latex were compressed
by osmotic stress at 2 pH values: pH 3, where sulfate
groups at the particle surfaces are ionized but carboxylates
are not, and pH 9, where all acid groups are ionized. In
addition, the ionic strength was varied between and
10-1 M, generating long-range or short-range screening
of the surface charges. Unless specified otherwise, the
temperature was below the glass transition temperature
of the particle cores, and therefore the cores maintained
a rigid spherical shape even under high pressure.
The original latex dispersion at volume fraction 50%
was white and fluid. Its osmotic pressure was between
1000and 10 000 Pa, dependingon ionic strength and pH:
the pressure was higher at high pH (higher surface charge)
and low ionic strength,as expected. When this dispersion
was equilibrated at a lower pressure, it took up large
amounts of water; at low pressures (100 Pa or less) the
dispersion expanded so much that the dialysis bag was
mechanicallystretched; then the pressure inside the bag
was no longer equal to the pressure of the stressing
solution. This artefact was eliminated by mounting the
open bag at the tip of a glass capillary where the liquid
dispersion could rise; then the pressure of the dispersion
could be read directly from the level difference in an
equilibrium against an aqueous solution with or without
polymer (Figure 3).
When the dispersion was equilibrated at a pressure
higher than its original pressure, it lost water, the
dispersion became unable to flow, and the resistance to
Osmotic Pressure of Latex Dispersions Langmuir, Vol. 10,No. 11, 1994 4017
10000 T 1000000
Il (Pa)
100000

1000 --

100
.=
--.
10000

1000

100

10
I
4
* 10

I 1

1 4 i
0 0.05 0.1 0.15 0.2 0.1 I

Figure 3. Swelling of latex dispersion L1 at pH 9 and ionic 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
strength 7.5 x M. Diamonds: pressures and volume Figure 4. Osmotic pressures of L1 latex dispersions at pH 9.
fractionsmeasured for dispersions placed in closed dialysis bags Squares: measured osmoticpressuresand volume fractionsat
and immersed in dilute dextran solutions.Squares: same for ionic strength loe3M. Dashed line: values calculated through
dispersions in open dialysis bags. The measured pressure is the CS equation for particles with hard-sphere repulsions at
the osmoticpressure ofthe dextran solution,outside the dialysis +
R,B = R K - ~ . Diamonds: measured values at ionic strength
bag; it equals the osmotic pressure inside the bag minus a 10-l M. Semidashed line: corresponding values calculated
contribution from the stretching of the bag. through the CS equation. Full line: perfect gas of particles.
Arrows indicate the location of the fluid-solid transition.
compression became very high. The precise location of

.
1000000 T
the crossover from fluid to solid was determined by the Il (Pa)
range of repulsions: a t low ionic strength and high pH
the transition occurs when the counterion clouds overlap 100000 c
(separation = 2 Debye lengths); at high ionic strength
and low pH, the transition occurs at the volume fraction 10000
where a hard sphere gas turns into a hard-sphere solid
(4 = 0.49 to 4 = 0.55).19
Even with very high pressures (up to 4 atm) it was not 1000
possible to concentrate the dispersion to the maximum
volume fraction of a face-centered cubicpacking (4 = 0.74); 100
this resistance originated from the particle membranes,
which are swollen with water and resist dehydration.
Some typical compression curves are shown in Figure
4 (at pH 9) and Figure 5 (at pH 3). For all the curves there
10 i
is a wide range of volume fractions (4 = 0.1 to 4 = 0.5)
where the pressure varies smoothly and slowly; then,
beyond 4 = 0.6,the pressure rises abruptly. We discuss
1
t
these regions successively. 0.1 J 4

Moderate Volume Fractions (4 = 0.1 to 4 = 0.5). In this
regime the dispersion is expected to behave as a moder-
ately concentrated gas, because the repulsions are short
range (no more than 10 nm) and there is plenty of room
available (the separation between particles is at least 50
nm). A calculation of pressure through the CS equation
shows that the pressure from the particles is expected to
vary between 1 and 500 Pa, depending on the effective
volume fraction (Figures 4 and 5). The measured pres-
sures are 3 orders of magnitude higher at pH 9, and 2
orders of magnitude higher a t pH 3. Therefore the
pressure must originate from other species besides the
particles.
These species must be numerous, since they produce a
high pressure, proportional to their number density, at a
low volume fraction, since the slope of the compression
curve is the same as that of a perfect gas, therefore of a
small molar mass, but still large enough to be excluded
by the pores of the dialysis bags.
The molar mass of these species was determined as
follows. The particles were separated by centrifugation,
and the supernatant was examined through light scat-
tering, size exclusionchromatography,and refractometry.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Figure 6. As in Figure 4, but at pH 3.
These studies indicate the presence of macromolecules
with molar mass distribution centered at M = 3000 g/mol.
Besides molar mass, some other characteristics of these
macromolecules may be inferred from the shape of their
compression curve. Indeed, the interactions between
charged species in solution are reflected in their osmotic
pressure, which may be calculated through the OCM
model, as explained above. For this calculation, the
number density of macromolecules must be known. In a
first step, we take the number density obtained from the
measured pressures at high dilutions. With this number
density, all the compression curves can be reproduced
through the OCM if the charge per macromolecule is 2 or
3 electron charges (Figures 6 and 7). Since these
macromolecules must be made of the same monomers as
the latex particles, it would then follow that each one
would have 2 or 3 AA monomers, and about 30 BA
monomers. However, macromolecules with this composi-
tion would not be soluble. On the other hand, macro-
molecules with more charges would give pressures which
rise faster with concentration.
4018 Langmuir, Vol. 10, No. 11, 1994
1000000
I
I
100000 i macromolecules which are soluble but physically trapped
10000
t the number concentration of such soluble macromolecules
1000 t (resistance to loss of water), vicosity and interfacial
1 0
100 I I
0 0.1 0.2 0.3 0.4 0.5 0.6
Figure 6. Comparison of osmotic pressures measured at low
volume fractionsfor latex L l with pressures calculated for small
macromolecules released by the particle membranes into the
aqueous phase. Squares: measured osmotic pressures and
volume fractions of the dispersion at pH 9 and ionic strength
M. Full l i e : calculated osmotic pressures for a solution
of macromolecules with molar mass M = 3000 g/mol, charge
per macromolecule 2 = 3e, and number density 9000 times
higher than that of latex particles. Dashed line: same for a
perfect gas of uncharged macromolecules.
I interactions between charged polymers anchored on
a opposing surfaces.
10000
i I
I
8
I of copolymerization. Therefore it may be useful to describe
0 water is a poor solvent for the surface polymers, e.g. at
100
0 0.1 0.2 0.3 0.4 0.5 0.6
Figure 7. Same as Figure 6,but at pH 3 and ionic strength
10-1M. Fit (fullline) with M = 3000 g/mol,charge 2 = 2e, and
number density 300 times higher than that of latex particles.
In order to use this information, we need to figure out
how macromolecules which carry many charges (they are
water soluble) do not repel very much (the pressure rises
slowly with concentration). If these macromolecules carry
sequences of charged AA monomers and sequences of
uncharged BA monomers, then they may self-associate
as other amphiphilic molecules do. At high concentrations,
this association would be more extensive, thereby reducing
the pressure.
The origin of these soluble macromolecules can be
understood by retracing the history of the dispersion.
During the synthesis, AA monomers are copolymerized
with hydrophobic monomers (mostly BA) to form a
collection of macromolecules with different proportions of
AA and BA monomers. Because the pH is kept low during
Bonnet-Gonnet et al.
synthesis, all these macromolecules condense on the
growing particles and form their membranes. After
synthesis, the dispersion is washed, and all uncondensed
monomers or oligomers are eliminated. However, the
membranes may contain a fraction of low molar mass
by other macromolecules; these species will be slowly
released during long-term storage of the dispersion. This
release will be accelerated by a dilution or by a rise in pH;
indeed, from the compression curves (Figures 6 and 71,
appears to be much larger at pH 9 than at pH 3.
The presence of such soluble macromolecules in the
aqueous phase must have consequences for the properties
of dilute latex dispersions. In addition to osmoticpressure
properties may also be modified. Thus, for latex disper-
sions with polymeric membranes, the properties of dilute
dispersions are controlled by polymers released from the
membranes.
I High Volume Fractions (4 > 0.5). The compression
0.7 curves show a steep rise when the volume fraction reaches
4 = 0.5 (at high pH) or 4 = 0.6 (at low pH) (Figures 4 and
5). At the same point, the dispersions turn from fluid to
solid. This abrupt change cannot result from the presence
of released macromolecules, because their mass concen-
tration in the aqueous phase is still small. At this point,
however, the separations between particle surfaces are
on the order of 12 nm. Moreover, the surface polymers
are charged and thus may be extended into the aqueous
phase; therefore the aqueous separation between surface
polymers of neighboring particles may be quite short. Thus
the steep rise in pressure corresponds to short-range
It may not be possible to predict the configurations of
surface polymers extending into the aqueous phase
separating particles: the simpler problem of polyelectro-
lytes grafted to a surfacehas been described theoretically,=
but the present situation is complicated by the distribution
of AA and BA sequences, which depends on the history
two extreme situations, where an accurate prediction of
interactions and pressure can be made.
(1)Assume that the surface polymers are so strongly
anchored that all ionizable groups remain located at the
particle surfaces. In this case the pressure may be
calculated through the PB models. In conditions where
I pH 3, the calculated pressure is of the same order of
0.7
magnitude as the measured pressures (Figure 8). In
conditions where water is a good solvent for the surface
polymers, e.g. at pH 9 and low ionic strength,the measured
pressures are much above the calculated pressures (Figure
91, indicating that the ionizable groups are not localized
on the particle surfaces.
(2) Assume that the charged groups of the sueace
polymers are completely free to move in the aqueous
solution between particles. Then the distribution of
negative and positive charges in these aqueous layers is
uniform, and their pressure can be calculated as that of
a perfect gas of ions. The concentrations of ions of each
type are calculated as above, by counting negative charges
from the surface polymers, positive counterions, and
passive salt in equilibrium from the reservoir (the PBC
model is used beforehand to determine the concentration
(23)(a)Van de Steeg, H. G. M.; Cohen Stuart, M. A.; de Kreizer,A.;
Bijsterbosch,B. M. Langmuir lSS!2,8,2538. (b)Blaakmeer,J.;Btihmer,
M. R.; Cohen Stuart,M. A.; Fleer, G. J. Macromolecules 1990,23,2301.
Osmotic Pressure of Latex Dispersions Langmuir, Vol. 10, No. 11, 1994 4019
1000000 1000000
It (Pa) ll (Pa)

100000 100000

10000 10000 w .'

,*'
w
,*'*

1000 1000

100
0 0.2 0.4 0.6
Figure 8. Comparison of osmotic pressuresmeasured at high
vnliima fvootinna f,-w lotnv T . l with nvnamivna
interactingcharged particles. Squares: data tor tne dispersion
at pH 3 and ionic strength M. Full line: calculated
pressures from a model where all ionizable groupsof the particle
membranes have been released into the aqueousphase. Dashed
lines: calculated pressures from models where the ionizable
groups are localized at the particle surfaces, and only coun-
terions are into the aqueous phase; the models are PBC
(semidashedline) and PB with planar geometry (dashed line).
1000000
ll (Pa)
100000
10000 m the amount of PAA in the aqueous phase is chosen: we
w assumed that all PAA present in the membranes was
I
1000
/
I
I mass (semidilute regime).
100
0 0.2 0.4 0.6
Figure 9. As in Figure 8, but at pH 9.
ofpassive salt). The resulting pressure is shownin Figures
8 and 9; it is the highest pressure that could be generated
by the dispersion, since a reduction in volume would
suppress a maximum number of degrees of freedom. At
moderate volume fractions this pressure is much higher
than the measured pressures; the discrepancy disappears
in the limit of high volume fractions, where the confine-
ment is so strongthat the distribution of charges is uniform
anyway.
As shown in Figures 8 and 9, the measured pressures
are intermediate between these two extremes, indicating
that the ionizable groups are neither completely localized
nor completely free. As mentioned above, this is because
the AA monomers are linked as macromolecules which
**
Q
100 I
0.8 0 0.2 0.4 0.6 0.8
Figure 10. Comparison of osmotic pressures measured at high
oslonlatnnrl fnr volume fractions for latex L1 with pressures measured for
polyelectrolytes confined in the aqueous layers separatingthe
particles. Squares: data for the dispersion at pH 9 and ionic
strength M. Full line: osmotic pressuresof PAA solutions
with the same amount of aqueous phase and an amount of PAA
corresponding to the release of all the PAA from the particle
membranes;the molar mass of PAAmacromoleculeswas chosen
at 2 x lo5 g/mol. Diamonds: measured pressures and volume
fractions of the dispersion at pH 3 and ionic strength M.
Dashed line: osmotic pressuresof PAA solutionswith the same
amount of aqueous phase.
extend into the aqueous phase. Thus a better ap-
proximation would be to forget the anchoring and retain
the polymer effect, by considering free PAA macromol-
ecules confined in the aqueous phase separating the
particles.
This comparison is performed as follows. First, the
,.
*,',
concentration of passive salt in the aqueous phase is
determined through a PBC calculation, as above. Then,
released in the aqueous phase; accordingly the amount of
PA4 was 2% of the amount of latex. Finally, the osmotic
pressure of a PA4 solution at this concentration in an
aqueous solution of matching pH and ionic strength was
measured; the molar mass of PAA was chosen high enough
so that its osmotic pressure was independent of molar
Q The pressures of such PAA solutions are compared in
I Figure 10 with the pressures for latex dispersions. There
0.8 is a range of volume fractions where the match is quite
good, indicating that in this range the dominant interac-
tions are between surface polymers extending into the
aqueous phase separating the particles. At higher volume
fractions, where there is a pressure "wall" in the com-
pression curves of the latex dispersion, the curves from
the PAA solutions continue to rise smoothly. This
discrepancy is caused by the rigid shape of the latex
particles, which forces the dense parts of the membranes
to be in contact at a volume fraction where there is still
a substantial amount of water in the dispersion.
This comparison is instructive, because it determines
precisely the extension ofthe three regimes for the osmotic
pressures of such latex dispersions: moderate volume
fractions, where the resistance to compression originates
from soluble macromolecules released by the membranes;
high volume fractions, where it originates from interac-
4020 Langmuir, Vol. 10, No. 11, 1994
1000000
ll (Pa)
100000
10000
1000
100 4 : ~ 4
0 0.2 0.4 0.6 0.8 1 0
Figure 11. Comparison of osmotic pressures measured for
latex L2 with pressures calculated for membrane polymers
released or extended into the aqueousphase. Squares: data for
the dispersion at pH 8.6 and ionic strength 10-1M. Dashed
line: calculatedpressures for a solution of macromoleculeswith
molar mass M = 3000 g/mol,charge per macromolecule2 = 2e,
and number density 400 times higher than that of latex
particles. Semidashed line: pressures of PAA solutions with
an amount of PAA correspondingto the release into the aqueous
phase of all the PAA sequences of the particle membranes.
Arrow indicates the location of the fluid-solid transition.
tions of macromolecules extending from the membranes
into the aqueous phase; contact, where it results from the
resistance to dehydration of the dense regions of the
particle membranes.
It is remarkable that, in all fluid states ofthe dispersion,
the pressure ofthe dispersion is the pressure of an aqueous
phase containing a certain amount of the polymers which
make the particle surfaces.
L2Latex. Dispersions of the L2 latex were compressed
in the same conditions as those of L1 latex, but higher
volume fractions were reached, up to 4 = 0.95. This was
a consequence of the low glass transition temperature of
the core polymers, which made the particles soft enough
to deform under pressure.
The compression curves are shown in Figures 11and
12. The fluid-solid transition is indicated as an arrow
on each graph. At low ionic strength its location cor-
responds approximatelyto the volume fraction for which
the counterion clouds overlap (separation of surfaces = 2
Debye lengths). At high ionic strength it corresponds to
the fluid-solid transition of a hard-sphere gas.
At volume fractions below the fluid-solid transition,
the interactions between particles are largely screened,
yet the pressure remains high even at large dilutions. As
in the case of the L1 latex, this excess pressure originates
from soluble macromolecules which are released by the
particle membranes during storage of the dispersion. The
concentration and interactions of these soluble macro-
molecules can be determined from a fit ofthe compression
curve by the model pressure for the same type of soluble
species as in the case of the L1 latex; this fit is shown in
Figures 11 and 12.
At volume fractions above the fluid-solid transition,
the pressure rises much above the pressure from the
soluble macromolecules (Figure 11).As in the case of the
L1 latex, this rise results from repulsions between surface
polymers which are extended into the aqueous layers
separating particles. This contribution to the pressure
can be approximated as the pressure from free PAA
Bonnet-Gonnet et al.
1000000
ll (Pa)
100000
10000
I
1000
I
Q
100
0.2 0.4 0.6 0.8 1
Figure 12. Data as in Figure 11but at ionic strength M.
Dashed line: fit with macromoleculesof number density 12 000
times higher than that of latex particles. Semidashed line:
osmotic pressures of PAA solutions with an amount of PAA
corresponding to the release into the aqueous phase of all the
PAA sequences of the particle membranes.
macromolecules in this aqueous phase; this fit is also
shown in Figures 11 and 12. The good match between
pressures of the dispersion and pressures of the PAA
solution, up to the highest volume fraction (6 = 0.951,
indicates that the main contribution to osmotic pressure
is the resistance to dehydration of the PAA sequences at
the particle surfaces.
A closer look at the compression curves in the high
volume fraction range reveals that the pressure actually
rises in two steps (Figure 11). We have observed this
feature for all latex dispersions where the particle cores
are soft,the membranesare made of hydrophilicpolymers,
and the ionic strength is high. We have not observed it
at low ionic strengths and high pH, where the pressure
is dominatedby the soluble macromolecules released from
the particle membranes. Hence, this feature reflects the
direct interaction of soft particles with hydrophilic mem-
branes.
The first step occurs in a range of volume fractions where
the particles must deform from spheres to dodecahedra;
therefore it is likely that this step includes a mechanical
resistance to deformation, while the second step must
reflect only the resistance to dehydration of the swollen
particle membranes in a fixed geometry. At low ionic
strengths (Figure 12) this two-step behavior is harder to
recognize because the resistance to deformationis small
compared to the resistance to compression of highly
extended PAA sequences.
Conclusion
In this work, the process of removing water from an
aqueous dispersion has been studied. Initially, the
dispersion was in a fluid state, where the particles were
dispersed in a large volume of water. At the end of the
transformation, the dispersion was in a solid state, where
the particles had been forced into contact. At a macro-
scopic level, this transformation corresponds to a process
where charged surfaces are displaced with respect to each
other, which is a very general process in aqueous disper-
sions. It was expected that the behavior of different
dispersions in this process should have some common,
maybe universal, features. In order to characterize this
behavior, different dispersions were concentrated through
Osmotic Pressure of Latex Dispersions
osmotic stress, and their osmotic pressures were mea-
sured. It was found that the response of the dispersions
to osmotic stress changed with the nature of the particles
and with the content of aqueous phase in which they were
dispersed.
Two situations were examined: one where all ionizable
groups were localized on the particle surfaces, and the
aqueous phase contained only the counterions of these
groups; in the other one the aqueous phase was loaded
with polyelectrolytes which were either released from
surfaces or extended from the surfaces.
The first situation was obtained with “model” disper-
sions made from pure PS and stabilized with sulfate groups
anchored to the particle surfaces. In this case, significant
osmotic pressures (1000-10 000 Pa or0.01-0.1 atm)were
obtained only at very low ionic strengths. These osmotic
pressures were reproduced theoretically by a model which
described the overlap ofthe counterion clouds surrounding
the particles in concentrated dispersions. Since no
adjustable parameters were used in this model, the
agreement between measured and calculated pressures
indicates that the overlap of ionic clouds is indeed the
only cause of resistance to dehydration in these disper-
sions.
The other situation was obtained with “industrial”
dispersions made of S-BAcopolymers and stabilized with
PAA sequences linked to the core polymers. In this case
significant osmotic pressures were obtained even in the
presence of passive salt in the aqueous phase. At low
volume fractions of latex, these osmoticpressures originate
from soluble macromolecules released from the particle
membranes. These macromolecules are presumablyAA-
BA copolymers which are normally associated with the
membranes or with each other but dissolve in the aqueous
phase at high dilution, high pH, and low ionic strength.
At a high volume fraction of latex, other membrane
polymers which extend from the particle surfaces into the
aqueous phase become the dominant source of osmotic
pressure, because the compression forces them to inter-
penetrate. This osmotic pressure was reproduced by the
pressure of a semidilute PAA solution with a concentration
corresponding to a complete swelling of the membrane
PAA into the aqueous phase separating the particles. Since
Langmuir, Vol. 10, No. 11, 1994 4021
no adjustable parameters were used in this comparison,
the agreement between PAA pressures and dispersion
pressures indicates that the presence of polyelectrolyte
sequencesin the aqueous layers is indeed the cause of the
very strong resistance to dehydration of these dispersions.
From these results, we can also draw two conclusions
which have implications beyond the scope of this study:
one that is specific to latex dispersions and another one
which concerns the resistance to dehydration of aqueous
dispersions.
(1)Through osmotic pressure measurements, we have
observed that latex particles are rarely “passive”objects.
They can carry adsorbed amphiphilic species, including
macromolecules, surfactants, and other molecules which
are marginally water soluble; these molecules will be
bound or released according to the solvent quality of the
aqueous phase. They also carry grafted macromolecules,
and again,depending on ionic conditions in the aqueous
phase, these macromolecules will adsorb (collapse)on the
particle surfaces or desorb and extend into the aqueous
phase. Consequently, by modifying their environment,
latex particles may yield dispersions that function dif-
ferently from other particle dispersions.
(2) For all dispersions, the osmotic pressure is basically
proportional to the number of particles. With particles
that are at least 100 nm in diameter, we have found that
the osmotic pressure is never going to be very high, even
with strong repulsions between them. A high resistance
to dehydration can only be obtained if the dispersion has
many more degrees of freedom that would be lost upon
dehydration. This can be achieved if the aqueous phase
contains large molecules or macromolecules which resist
dehydration; we have seen that these molecules can be
either dissolved or grafted. Because the number density
of these macromolecules may be orders of magnitude
higher than that of particles, the resistance to dehydra-
tion which they confer to the dispersion is also many
orders of magnitude higher. Similarly, the ability of a
dispersion to rehydrate spontaneously will be greatly
increased if the continuous phase contains molecular or
macromolecular species which have a strong hydration
pressure.