Europhys. Lett., 66 (1), pp. 125–131 (2004)
1 April 2004
Sedimentation and multi-phase equilibria
in mixtures of platelets and ideal polymer
H. H. Wensink and H. N. W. Lekkerkerker(∗)
Van ’t Hoff Laboratory for Physical and Colloid Chemistry, Debye Institute
Utrecht University - Padualaan 8, 3584 CH Utrecht, The Netherlands
(received 19 November 2003; accepted in final form 26 January 2004)
PACS. 82.70.Dd – Colloids.
PACS. 64.70.Md – Transitions in liquid crystals.
Abstract. – The role of gravity in the phase behaviour of mixtures of hard colloidal plates
without and with non-adsorbing ideal polymer is explored theoretically.
(macroscopic) osmotic equilibrium conditions, we show that sedimentation of the colloidal
platelets is significant on a height range of even a centimeter. Gravity enables the system
to explore a large density range within the height of a test tube which may give rise to the
simultaneous presence of multiple phases.As to plate-polymer mixtures, it is shown that
sedimentation may lead to a four-phase equilibrium involving an isotropic gas and liquid phase,
nematic and columnar phase. The phenomenon has been observed experimentally in systems
of colloidal gibbsite platelets mixed with PDMS polymer.
By analyzing the
It is well known that adding non-adsorbing polymer to a colloidal dispersion induces an
attractive depletion potential of mean force between the colloidal particles [1–3]. For colloidal
spheres, the attractive potential has been shown to give rise to a phase separation in a colloid-
poor “gas” and colloid-rich “liquid” or “solid” phase at sufficiently high concentrations of the
colloid and the polymer [4–8]. Compared to colloidal spheres, the behaviour of dispersions
of rod- and plate-like colloids mixed with polymer is richer due to their possibility to form
liquid-crystal phases, i.e. nematic (N), smectic (Sm) and columnar (C). Recent experiments
on mixtures of colloidal gibbsite platelets and non-adsorbing polymer  have uncovered the
phase behavior of plate-polymer mixtures. A manifestation of the rich phase behaviour of
these mixtures is the observation of a four-phase equilibrium involving both isotropic gas and
liquid phases along with nematic and columnar states. The appearance of multiple phases
seems to conflict with the phase rule of Gibbs which states that the number of coexisting
phases is limited to three for an athermal binary mixture. One of the possible explanations
conjectured by the authors  is that the observation might be due to the polydispersity in
particle size. The presence of many components (i.e. platelets with different diameters and
thicknesses) in principle allows for a coexistence of arbitrarily many phases.
Another possibility to reconcile the experimental results with Gibbs’ phase rule is by
accounting for an external gravitational field. Sedimentation of particles leads to a density
(∗) E-mail: email@example.com
c ? EDP Sciences
gradient which facilitates the formation of multiple phases in a vessel of sufficient height. In
this paper we scrutinize the effect of sedimentation in systems of colloidal platelets with and
without added polymer from a simple osmotic compression treatment. We will first consider
a one-component system of colloidal platelets and then study the influence of the polymer-
induced depletion attraction using a mean-field free-volume theory .
Sedimentation equilibrium: one-component system. –
colloidal particles (platelets) in osmotic equilibrium with a dispersing solvent with a chemical
potential µ0subject to a gravitational field along the z-direction of the vessel. We assume
that the concentration profile of the colloids is sufficiently smooth so that the system is locally
in a homogeneous equilibrium state between z and z + dz. This is usually the case if the
particles are not too large and heavy and if the dispersion is not too close to a critical point.
The (macroscopic) condition for sedimentation equilibrium reads
Let us consider a vessel containing
in terms of the osmotic compressibility (∂ρ/∂Π)T,µ0of the dispersion and the buoyant mass
m∗of the colloidal particle (g is the gravitational acceleration). The concentration profile ρ(z)
of the colloids can be obtained from eq. (1) if the osmotic pressure as a function of ρ, i.e. the
equation of state (EOS), is known.
In the present study we will encounter phase-separated samples containing a number of
distinct phases. Since these phases are generally described by different equations of state,
it is convenient to treat each daughter phase i separately and assign a phase height Hi to
each of them. Recasting eq. (1) in dimensionless form by introducing the height parameter
ζ = z/H(i)(with 0 < ζ < 1) and dimensionless plate concentration ci= ρiD3(with D the
plate diameter) corresponding to phase i yields
with β = 1/kBT and˜Hi= Hi/ξ the height of phase i rendered dimensionless by relating it
to the gravitational length ξ = kBT/m∗g which is on the order of 10−3m for the colloidal
dispersions of gibbsite platelets we consider in this paper. The average concentration c0,iin
phase i follows from
where ct,i and cb,i denote the concentrations at the top and the bottom of the phase, re-
spectively. The average concentration c0 of the sample then follows from a simple linear
a given average sample concentration c0. In order to solve eq. (2) for colloidal platelets, we
must know the EOS Πi(ci) for the different liquid-crystal states (viz. isotropic (I), nematic
(N) and columnar (C)) encountered upon densifying these systems. As a quantitative input
we use fits to the EOS obtained from Monte Carlo simulations of hard platelets performed
by Zhang et al. . A polynomial of the K-th order was used as a fitting function so that
the concentration profile from eq. (2) in the different phases.
ic0,i˜Hi/˜H with˜H =?
i˜Hithe dimensionless sample height.
Note that in an experimental situation these concentrations are to be determined from
i, with i = I, N, C. The coefficients an,ipertaining to state i can be
found in ref. . The polynomial form of the EOS allows for a simple analytic solution of
H. H. Wensink et al.: Sedimentation in plate-polymer mixtures
Fig. 1 – a) Phase diagram for colloidal platelets with L/D = 0.05 in a gravitational field. Plotted is
the reduced sample height˜ H = H/ξ vs. the overall plate volume fraction φ0. The three-phase region
opens up at H = 11.15ξ. b) Concentration profile of a sample with overall volume fraction φ0 = 0.157
and vessel height H = 30ξ (corresponding to the open dot in a). Plotted is the relative height z/H
vs. φ. The I-N and N-C phase boundaries are indicated by the horizontal dotted lines.
The effect of gravity on the phase behaviour of the colloidal platelets is presented in fig. 1.
The curves represent so-called cloud curves which indicate the minimum sample height (and
associated overall volume fraction) needed to induce the formation of an infinitesimal amount
of a new phase at the top and/or the bottom of the vessel due to sedimentation of the particles.
On the horizontal axis we find the coexistence densities for the I-N and N-C transitions at zero
gravity, which would correspond to a vessel with potentially zero height. Figure 1a) shows
that a vessel height of about 10 gravitational lengths already leads to significant changes in the
phase diagram. A large three-phase isotropic-nematic-columnar region is encountered which
opens up at the state point indicated by the black dot. At the associated volume fraction
(φ0= 0.291) the system is fully nematic at short sample heights but as soon as the height
exceeds 11.15 gravitational lengths, two additional fractions of an isotropic and columnar
phase are split off simultaneously at the top and bottom of the sample, respectively. To
compare with actual sample heights we use the following expression for the gravitation length,
ξ = kBT/(gvplateρ∗
the colloidal gibbsite platelets dispersed in toluene (plate dimensions D = 180nm, L = 12nm
and buoyant density ρ∗
phase isotropic-nematic-columnar equilibria in fig. 1 may be expected in samples larger than
a centimeter, which is comparable to the typical height of a test tube. Figure 1b) shows an
example of a concentration profile one may encounter experimentally in a sample with overall
plate volume fraction of 15.7% and height of 2.7cm. The scenario is that the system initially
phase-separates into equal portions of an isotropic and nematic phase. At a later stage, a
columnar fraction will be formed at the bottom of the vessel due to slow sedimentation of the
platelets. At sedimentation equilibrium, the I, N and C phases, respectively, occupy 58, 39
and 3% of the system volume.
plate) with vplate=π
4LD2the colloid volume. Using experimental data for
plate= 1.5 103kg/m3) we obtain ξ = 0.9mm. This means that the three-
Plate-polymer mixtures. –
mixed with non-adsorbing ideal polymers (denoted by “p”) in a solvent. The gravitational
length of the polymer is much larger than that of the colloidal particles (ξp? ξ1) due to its
negligible buoyant mass. We may therefore assume that there is no external force acting on
the polymer coils and that the chemical potential of the polymer can be considered constant
throughout the system. The mixture can thus be treated as an effective one-component
We now turn to systems of colloidal platelets (component “1”)
Fig. 2 – Phase diagram of a plate-polymer mixture with L/D = 0.05 and q = 0.355 in the fugacity-
volume fraction plane, reproduced from ref. . On the vertical axis, the region of stable isotropic
gas-liquid (I1-I2) equilibria is confined between a lower critical point at zpD3= 19.233 (dotted line)
and the I1-I2-N triple line at zpD3= 24.454 (lower dashed line). The upper dashed line at zpD3= 104
represents the I1-N-C triple line.
system of colloidal platelets in a gravitational field and the osmotic pressure balance now
reads, analogously to eq. (1),
at constant µp. Similar to eq. (2), we can rewrite this equilibrium condition in dimensionless
form. Substituting the EOS for a colloid-polymer mixture from a free-volume treatment of the
Asakura-Osawa model (see the appendix) yields the following differential equation describing
the colloid density profile c1,i(ζ) in the daughter phase i:
which must be solved along with the auxiliary condition for the overall concentration, eq. (3).
Comparing with eq. (2), we see that the terms between square brackets now represent an ef-
fective (inverse) osmotic compressibility. The first contribution is the inverse compressibility
of the one-component plate system, whereas the second term accounts for the effective deple-
tion attraction between the platelets due to the presence of the polymer. The strength of the
depletion attraction can be varied by changing the fugacity zpof the polymer, related to the
chemical potential via zp= exp[βµp]/V . Note that the result for a one-component system
(eq. (2)) is recovered for zp= 0, as it should. The effective compressibility also depends on the
fraction of free volume αiavailable to the polymer in the liquid-crystal state i. Explicit expres-
sions for αiare given in the appendix. It is easily verified that d2αi/dc2
for all states i = I, N, C, implying that the effective osmotic compressibility is larger than
that of a pure system of plates due to the attractive depletion forces, as we intuitively expect.
In fig. 2 we have depicted a phase diagram for the zero-gravity case reproduced from
ref. . The values for the plate aspect-ratio and the polymer-to-plate size ratio q = 2Rg/D
(with Rgthe polymer radius of gyration) are chosen such as to match the experimental values
for the gibbsite-PDMS mixtures studied by Van der Kooij et al. . The volume fractions in the
1,iis generally positive
H. H. Wensink et al.: Sedimentation in plate-polymer mixtures
Fig. 3 – a) Phase diagram of the same mixture as in fig. 2 in a gravitational field at (constant)
fugacity zpD3= 20. Plotted is the relative sample height˜ H = H/ξ vs. the overall plate volume
fraction. The four-phase region opens up at H = 11.70ξ (black dot). b) Same diagram in a reservoir
fugacity-volume fraction representation at fixed vessel height H = 1.5cm (H/ξ = 16.67).
coexisting phases can be deduced from tie lines given by horizontal lines in this representation.
At low reservoir fugacity the phase behaviour of the mixture differs only marginally from that
of the pure system. At zpD3> 19.233, the isotropic phase becomes unstable with respect to
a demixing into an isotropic gas phase (I1) and a liquid phase (I2). The gas phase is poor in
colloid but rich in polymer, vice versa for the liquid phase. The nematic-columnar transition is
unaffected by the presence of the polymer up to the I1-N-C triple line located at zpD3= 104.
At higher fugacities the depletion attraction is strong enough to induce a transition from an
isotropic gas (I1) to a columnar solid (C) phase, without the intervention of a nematic phase.
In fig. 3a) we have depicted a phase representation, analogous to fig. 1, of the same mixture
in a gravitational field at fixed reservoir fugacity zpD3= 20. Also here, we see that sedimen-
tation leads to remarkably rich phase behaviour; several multi-phase equilibria appear that
are not present in the zero-gravity case in fig. 2. Most notably, a four-phase region opens up
at H/ξ = 11.70 which, recalling that ξ = 0.9mm, is again about a centimeter. An equilibrium
involving isotropic gas, liquid, nematic and columnar phases has been observed in the gibbsite-
PDMS mixtures . We stress that the experimental observation of four distinct phases in a
tube of a few centimeters is related in a fortuitous way to the platelets’ size (and hence their
gravitational length). If the platelets had been much larger, they would rather have formed a
dense, quasi-uniform sediment at the bottom of the tube. If they were much smaller, gravity
may not have been strong enough to enforce a four-phase sedimentation equilibrium.
From an experimental standpoint it is more appropriate to fix the total sample height
rather than the reservoir fugacity. In fig. 3b) we show a representation in terms of the fugacity
vs. the overall volume fraction at fixed sample height H = 1.5cm, which is the typical length
of the test tubes used in experiment . Unlike fig. 2, this representation does not provide
information about the composition of the phases present, it merely indicates which phases can
be expected in a sample with a fixed height and a given overall density and reservoir fugacity.
Comparing with fig. 2, we see that the four-phase region in fig. 3b) must be confined within
the range 19.233 < zpD3< 24.454 since only there both stable I1-I2 and N-C two-phase
equilibria occur at low and high densities, respectively.
We can therefore conclude that gravity enables the colloidal platelets to scan a large density
range within the range of a few centimeters. Since mixtures of plates and ideal polymer display
a number of phase transitions within a relatively small range of concentrations, sedimentation
may thus lead to the presence of multiple phases in a test tube. It is important to note that
the maximum number of phases that can appear simultaneously in the tube is governed solely
by the effective interactions between the colloids. In the present system these interactions are
tuned directly by the polymer chemical potential in the reservoir. Gravity can therefore only
induce the formation of those phases that are allowed at a particular interaction strength, as
becomes evident from comparing fig. 3b) to fig. 2.
Final remarks. –
vations in other colloidal mixtures. For instance, mixtures of hard rod- with plate-like colloids
also display rich phase behaviour with a possibility of multi-phase equilibria involving up to
five phases . The five-phase equilibrium comprises (from top to bottom) an isotropic,
rod-rich nematic (N+), X-phase (its symmetry has not been identified yet), plate-rich ne-
matic (N−) and a columnar phase. Also in these mixtures, the polydispersity in particle size
(i.e. both plate-diameter and rod-length) is appreciable. Although the phase behaviour is
certainly influenced by polydispersity, the effect of gravity should not be neglected. Given the
present results, one may question to what extent these multi-phase equilibria are induced by
an external gravitational field acting on the particles. This issue will be deferred to future
investigation. However, we can already point out that in these mixtures both components will
generally be subject to the gravitational field so that the effective one-component approach,
described in this paper, cannot be applied for binary mixtures of colloids. Consequently, the
chemical potentials of the species are coupled and therefore cannot be varied independently
(as we could do for µpin this study) . Clearly, the coupling leads to more complicated
equilibrium conditions than the ones shown here.
The theoretical results obtained thus far prompt us to reconsider obser-
Free-volume theory. –
ideal polymer within a mean-field free-volume treatment of the Asakura-Oosawa model [10,11]
is given by
with β = 1/kBT. Here, N is the number of colloidal particles in the system and F(0)is the
reference (Helmholtz) free energy of the unperturbed colloidal system. Furthermore, zp is
the fugacity of the polymer in a reservoir, separated from the system by a semi-permeable
membrane which only allows for exchange of polymer and solvent. Note that for ideal poly-
mers, zp is simply equal to the polymer concentration ρp = Np/V . Finally, α represents
the free-volume fraction averaged over all colloidal configurations of the unperturbed system.
As such α depends only upon the colloid density ρ = N/V in the system. The osmotic
pressure of the colloid-polymer mixture follows from eq. (6) using the standard derivative
Π = −(∂Ω/∂V )N,T,µp:
˜Π =˜Π(0)+ zpD3
The semi-grand canonical potential for a mixture of colloids and
?= βF(0)?N,V,T,µ0,zp= 0?− zpαV
α − ρdα
in terms of the dimensionless pressure˜Π(0)≡ βΠ(0)D3of the reference plate system. An
expression for the free-volume fraction α can be obtained from scaled particle theory [14,15].
Zhang et al.  derived expressions for cut spheres with diameter D and thickness L. Cut
spheres are plate-like objects obtained by slicing two caps off a sphere of diameter D, at two
planes parallel to the equatorial plane with equal distance L/2. The general expression for
the free volume reads
α = (1 − φ)exp?−?Ay + By2+ C˜Π(0)??
H. H. Wensink et al.: Sedimentation in plate-polymer mixtures
with y ≡ φ/(1 − φ) and φ the plate volume fraction. The expression still depends on the
pressure Π(0)of the reference cut sphere system, for which no analytical expression is available
yet. Specific expressions αifor the different liquid-crystal states i can be obtained by inserting
the corresponding EOS˜Π(0)
from the simulation fits. The coefficients are given by
q?1 + 2l − l2?+ q2?2l +?π
q2?1 + 2l − l2?2
2− arcsinl??√1 − l2
?l − l3/3?
C = πq3/6
2?l − l3/3?2
with l = L/D the aspect ratio and q = 2Rg/D the size ratio of the ideal polymer coil and the
platelet. The volume fraction follows from φ = (π/4)ρD3(l − l3/3).
∗ ∗ ∗
We thank G. J. Vroege and M. Oversteegen for a critical reading of the manuscript.
 Asakura S. and Oosawa F., J. Chem. Phys., 22 (1954) 1255.
 Asakura S. and Oosawa F., J. Polym. Sci., 33 (1958) 183.
 Vrij A., Pure Appl. Chem., 48 (1976) 471.
 Vester C. F., Kolloid-Z., 84 (1938) 63.
 de Hek H. and Vrij A., J. Colloid Interface Sci., 84 (1981) 409.
 Vincent B., Edwards J., Emmet S. and Groot R. D., Colloids Surf., 31 (1988) 267.
 Gast A. P., Russel W. B. and Hall C. K., J. Colloid Interface Sci., 109 (1986) 161.
 Illet S. M., Orrock A., Poon W. C. K. and Pusey P. N., Phys. Rev. E, 51 (1995) 1344.
 van der Kooij F. M., Vogel M. and Lekkerkerker H. N. W., Phys. Rev. E, 62 (2000)
 Lekkerkerker H. N. W., Poon W. C. K., Pusey P. N., Stroobants A. and Warren
P. B., Europhys. Lett., 20 (1992) 559.
 Zhang S. D., Reynolds P. A. and van Duijneveldt J. S., J. Chem. Phys., 117 (2002) 9947.
 van der Kooij F. M. and Lekkerkerker H. N. W., Langmuir, 16 (2000) 10144.
 Vrij A., J. Chem. Phys., 72 (1980) 3735.
 Reiss H., Frisch H. L. and Lebowitz J. L., J. Chem. Phys., 31 (1959) 369.
 Lebowitz J. L., Helfand E. and Praestgaard E., J. Chem. Phys., 43 (1965) 774.
 Zhang S. D., van Duijneveldt J. S. and Reynolds P. A., Mol. Phys., 100 (2002) 3041.