Effective Interactions Between Electric Double-Layers: Jean-Pierre Hansen

Effective interactions between electric double-layers
Jean-Pierre Hansen
Department of Chemistry, Cambridge University, Lensfield Road, Cambridge CB2 1EW, United Kingdom; e-mail:
[email protected]
Hartmut Löwen
Institut für Theoretische Physik II, Heinrich-Heine-Universität, Universitätsstraße 1,
arXiv:cond-mat/0002295v1 [cond-mat.soft] 18 Feb 2000
D-40225 Düsseldorf, Germany; e-mail : [email protected]
Key Words: Density functional; charged colloids; correlations; Poisson-Boltzmann theory; Debye screening
Shortened Title: Electric double-layer interactions
Abstract
The present review summarizes and assesses recent theoretical and experimental advances, with special emphasis on
the effective interaction between charge-stabilized colloids, in the bulk or in confined geometries, and on the ambi-
guities of defining an effective charge of the colloidal particles. Some consideration is given to the often neglected
discrete solvent effects.
1
I. INTRODUCTION
Electric double layers form spontaneously whenever surfaces carrying ionizable groups are suspended in a polar
solvent, most frequently water. The high dielectric constant ǫ of the latter favours the dissociation of the functional
surface groups, so that the surface acquires a net charge per unit area, σ. Oppositely charged counterions are released
into the solvent, which generally also contains a finite concentration of microscopic anions and cations (microions),
providing thus additional coions and counterions. The electric double-layer results from the build-up of a charge
density (or ”cloud”) of opposite sign to that of the surface charge, which tends to screen the electrostatic potential
due to the latter. The width of the double-layer, which is a measure of its capacity, is determined by the competition
between the thermal motion of the microions, which tends to spread out, or homogenize their distribution, in order
to increase their entropy, and the electrostatic interactions, which attract the counterions towards the surface, while
repelling the coions.
The present review deals with electric double-layers around mesoscopic charged particles, which will be referred
to as polyions (frequently also called macroions), and more specifically with the effective interactions between electric
double-layers associated with different polyions. Such double-layers are ubiquitous in many physical, geophysical,
chemical and biological systems, including complex fluids, clay soils, polyelectrolytes (e.g. DNA) or cell membranes.
The main focus of the review will be on colloidal dispersions of spherical, rod-like or lamellar polyions, or micelles
resulting from the self-assembly of ionic surfactants [1]. Advanced experimental techniques, like surface force ma-
chines or video microscopy in combination with optical tweezers, allow a direct measurement of the forces acting
between charged surfaces or polyions. Such direct measurements contrast with indirect determinations of effective
forces via small angle X-ray or neutron scattering measurements of static structure factors [2]. For the theoretician,
the main challenge is the highly asymmetric nature of dispersions involving mesoscopic polyions, and microscopic sol-
vent molecules and ions. Clearly, at least three widely different length scales are involved, namely the characteristic
size (radius) a of the polyions, a typical width of the electric double-layer, of the order of the Debye screening length
λD , and a typical correlation length l of the solvent, of the order of a few molecular diameters. Under most physical
conditions the double inequality a ≫ λD ≫ l holds, so that a coarse-grained statistical description of the suspension
is clearly warranted. This strategy may be cast in the unifying framework of the density functional theory (DFT)
of inhomogeneous fluids [3], as explained in the following section. A number of key issues, which are presently the
object of intense experimental and theoretical scrutiny, will be addressed in subsequent sections of this review. They
include the following questions and topics :
• The limitations of the standard Poisson-Boltzmann or mean-field theory of electric double- layers and the
importance of spatial correlations between microions. A particularly important question relates to the possibility
of an attractive component of the effective interaction between polyions, induced by microion correlations.
• The notion of effective polyion charge (or charge renormalization), as controlled by dissociation equilibria and
counterion adsorption (or condensation), and of charge regulation.
• The relation between effective interactions between polyions, and the phase behaviour of their dispersions.
• The effect of confinement on the effective interactions between colloids, and more generally the influence of
electrostatic boundary conditions at interfaces.
• The introduction of solvent granularity into the statistical description of double-layers.
This review will emphasise the more recent experimental and theoretical developments in this very active field, covering
mostly work published during the past decade. Earlier work is adequately covered in some previous reviews of the
subject [4–7]. Electrostatic CGS units will be used throughout.
II. MULTI-COMPONENT VERSUS EFFECTIVE ONE-COMPONENT DESCRIPTION
Consider a suspension of N polyions, of radius a and charge Z, in a polar solvent with co and counterions, of radii
a± and charges z± e ; the radius of the solvent molecules is comparable to the microion radii a± , i.e. of the order of
0.1 to 0.2 nm. In micellar solutions, 1 nm < a < 10 nm, and 10 < |Z| < 100, while in most colloidal dispersions, 10
nm < a < 103 nm, and 102 < |Z| < 104 . A statistical description of the highly asymmetric multi-component system is
a considerable challenge, which requires some degree of coarse-graining. In most theoretical investigations of micellar
or colloidal systems, the solvent is regarded as a mere continuum, characterized by its macroscopic dielectric constant
2
ǫ. This amounts to a ”primitive model” (PM) level of description, where polyions and microions are assumed to
be charged hard spheres (non-spherical polyions will be considered later in this review) interacting via the Coulomb
potential ∼ e2 /ǫr, outside the excluded volume range of inter-particle distances r. It is implicitly assumed that the
mesoscopic polyions have the same macroscopic dielectric constant ǫ as the solvent, thus avoiding complications due
to dielectric discontinuities (e.g. image charges).
Even within the PM, the theoretician is still faced with the polyion-microion asymmetry. In micellar solutions,
where the size and charge ratios a/a± and |Z|/|z± | are roughly a factor of 10, the asymmetry may still be han-
dled within the multicomponent PM level of description, using the theoretical techniques of the theory of classical
fluids [8]. In particular, the partial pair distribution functions gαβ (r) (where α and β are species indices) may be
calculated from the usual fluid integral equations, including the hypernetted chain (HNC) [9] and mean spherical
approximations (MSA) [10], or their variants [11] and hybrid combinations [12,13]. Alternatively these correlation
functions may be obtained from Monte Carlo (MC) or Molecular Dynamics (MD) simulations of the asymmetric PM.
The multicomponent PM point of view is, in practice, limited to micellar systems with |Z| < 102 , because integral
equation closures are increasingly unreliable for asymmetric systems, and their numerical solution tends to become
unstable. Similarly simulations become more and more difficult, and are at present limited to |Z| ≤ 60 [14]. The
multi-component PM description becomes untractable in the much more asymmetric colloidal range, which requires
a coarse-grained description, based on effective interactions between polyions, to be discussed below.
Important simplifications occur within the PM in a number of limiting situations.

a) In the limit of infinitely low concentration n (number per unit volume) of polyions, only a pair of these, within
a solution of microions, need to be considered. The effective interaction between the two polyions (p) is then
exactly given by the potential of mean force, vpp (r) = −kB T log[gpp (r)] [15]. The multicomponent problem
reduces to that of an electrolyte of microions in the external field due to two fixed polyions, which leads to
considerable simplifications in simulations [16] or the numerical solution of integral equations.
b) The problem may be further simplified by taking the limit a → ∞ ; the initial dispersion then reduces to the
much simpler system of an electrolyte confined between two parallel planes carrying a, generally uniform, surface
charge σ. The classic problem of two interacting planar double-layers goes back to the work of Gouy [17], and
has been the object of considerable theoretical and experimental work, some of which will be discussed in a
subsequent section.
c) In the opposite limit of high concentration, each polyion is, at least temporarily, trapped in the ”cage” formed
by neighbouring polyions. This regime is reasonably described by a one-body model, where a single polyion is
located at the centre of a Wigner-Seitz cell, together with an inhomogeneous distribution of microions, such that
the total charge inside the cell is zero. For practical purposes the cell is chosen of a simple geometry reflecting the
shape of the polyion (e.g. a spherically symmetric cell around a spherical polyion), and the electrostatic boundary
conditions are chosen such as to mimic the average effect of the surrounding polyions [18,19]. Despite the
considerable simplifications which they imply, cell models allow valuable insight into concentrated dispersions,
and will be discussed further on.
A unifying framework for the statistical description of interacting electric double-layers, or, more generally, of
inhomogeneous fluids and interfaces, is provided by density functional theory (DFT) [3].
A. DFT and effective interactions
In view of the considerable polyion-microion asymmetry, it seems natural to combine a discrete representation of
the former with a field description of the latter. Let {Ri }(1 ≤ i ≤ N ) denote the positions of the N polyions,
assumed here to be spherical, and let VNdir {Ri } be their direct interaction energy for a given configuration; VNdir is,
to a good approximation, pairwise additive, with a pair potential vpp (R) including a short-range excluded volume
repulsion, the long- range Coulomb repulsion Z 2 e2 /ǫR and a van der Waals attraction (dispersion force), of intensity
characterized by a Hamaker constant [1]. The inhomogenous distributions of co- and counterions, in the ”external”
field of the polyions, are characterized by the local densities (or concentrations) ρα (r)(α = +, −). The equilibrium
densities satisfy the variational principle [3] :
δΩ ρ∗+ (r), ρ∗− (r)

= 0; α = +, − (1)
δρ∗α (r)
ρα =ρα
∗
3
where Ω is the grand potential functional of the trial densities , given by :
Z
X
Ω ρ∗+ , ρ∗− = F ρ∗+ , ρ∗− − Φα (r)ρ∗α (r)dr

(2)
α
In (2), F denotes the intrinsic free energy functional, while Φα (r) = µα − Φext
α (r), with µα , the chemical potential of
species α ,and Φext
α , the external potential acting on ions of this species, equal to the sum of the interactions with the
N polyions :
N
X
Φext
α (r) = vpα (r − Ri ) (3)
i=1
Note that in view of (3), the equilibrium density profiles ρα (r), and the resulting grand potential Ω, depend paramet-
rically on the polyion configuration. The chemical potentials µα are either fixed at some reservoir value (semi-grand
canonical ensemble), or determined a posteriori by the canonical ensemble constraints:
1
Z
ρα (r)dr = nα (4)
V V
where V is the total volume of the dispersion and nα is the mean (macroscopic) concentration of α-ions.
Explicit calculations of the density profiles require some approximation for the generally unknown free energy func-
tional F , which is traditionally split into ideal and excess parts, F = Fid + Fex ; Fid is known exactly :
X Z
ρα (r) log Λ3α ρα (r) − 1 dr

Fid [ρ+ , ρ− ] = kB T (5)
α=+,− V
(0)
If nα denotes the concentrations of some homogeneous reference state (i.e. in the absence of polyions), and ρα (r; ξ) =
(0)
nα + ξ∆ρα (r) (0 ≤ ξ ≤ 1) denotes a continuous set of density profiles, such that ρα (r; ξ = 1) leads back to ρα (r),
the equilibrium density profiles in the presence of polyions, then Fex is given by the formally exact expression:
X Z
(0) (0) ex
Fex [ρ+ , ρ− ] = Fex n+ , n− + µα ∆ρα (r)dr (6)
α
XXZ 1
− kB T dξ(ξ − 1)
α β 0
Z Z h i
′ ′ ′
× dr dr ∆ρα (r)cαβ {ρα (ξ)} ; r, r ∆ρβ (r )
µex
α is the excess part of the chemical potential, while the cαβ [{ρα (ξ)}] are the set of direct correlation functions for
the inhomogeneous electrolyte with local densities ρα (r; ξ). For Coulombic systems, these decay asymptotically as
−vαβ (r)/kB T = −zα zβ lB /r (where lB = e2 /ǫkB T is the Bjerrum length). It proves convenient to subtract this from
the cαβ , leaving the short- ranged part csr
αβ . The excess free energy function then splits into a mean-field Coulombic
contribution, and a correlation part :
Fex [ρ+ , ρ− ] = FCoul + Fcorr (7)

′
e2 ′ ρ(r)ρ(r )
Z Z
= dr dr + Fcorr
2ǫ |r − r′ |
where ρ(r) = z+ ρ+ (r) + z− ρ− (r) is the microion charge density. The correlation term is formally given by (7), with
csr
αβ replacing cαβ .
Starting from these exact expressions, there are basically two strategies to proceed with approximations.
a) The most common strategy focusses on the density profiles, and the resulting free energies. Fcorr is approximated
by a local density ansatz (LDA) or by some weighted density approximation (WDA) [3,20] which generally
requires the direct correlation functions in a homogeneous reference state as input. Once the variational problem
(1) has been solved for the approximate free energy functional, the effective interaction energy between polyions
is given by :
4
VN ({Ri }) = VNdir ({Ri }) + Ω ({Ri }) (8)
where the equilibrium grand potential accounts for their indirect interactions, induced by the double-layer-
forming microions, which have effectively been traced out. It is very important to realize that this contribution
depends on the thermodynamic state of the dispersion, and is generally not pair-wise additive.
If correlations are neglected altogether, Fex reduces to the mean-field Coulombic part, and (1) immediately leads
to the equilibrium profiles
ρα (r) = ζα exp − Φdir

α (r) + zα eψ(r) /kB T (9)
where ζα is the fugacity of species α (equal to its reservoir concentration in the absence of correlations), Φsr
α is
the short-range (excluded volume) part of the interaction of an ion α with the N polyions, and ψ(r) is the total
electrostatic potential at r, which satisfies Poisson’s equation :
4πe ext
∇2 ψ(r) = −

ρ (r) + ρ(r) (10)
ǫ
with ρext (r) the ”external” charge density carried by the N polyions. Eqs (9) and (10) constitute the multi-centre
version of mean-field non-linear Poisson-Boltzmann (PB) theory. Despite their apparent simplicity, numerical
solution of the PB equations in the presence of N polyions is a formidable task. In practice, eq. (10) is often
solved in the domain outside the polyions, and the contribution of the latter to the electrostatics is treated as
a boundary value problem.
b) A second, more ambitious strategy is to seek information on both the density profiles, and the pair correlations
between microions within the double-layers. This may be achieved by relating the direct correlation functions,
appearing in (7), to their functional inverses, the total correlation functions hαβ , via the Ornstein-Zernike
relations for inhomogeneous fluids [8], supplemented by an approximate closure relation between the h and c
functions. This strategy leads to coupled equations for the density profiles and the pair correlation functions, as
in the widely used inhomogeneous hypernetted chain (IHNC) theory [21]. In view of its numerical complexity,
this strategy can only be applied to very simple geometries. In practice it is limited to planar geometry and will
be discussed in a subsequent section.
B. Cell model
As stated earlier, cell models, involving a single polyion, prove useful to study concentrated dispersions, including
colloidal crystals. The paradigm is provided by the much studied case of a globular polyion of radius a, placed at
the centre of a spherical cavity of radius R determined by the polyion concentration, i.e. R = (3/4πn)1/3 , containing
counterions and salt which ensure overall charge neutrality. The density profiles ρ+ (r) and ρ− (r) depend only on
the distance r from the centre (1d problem). Gauss’s law implies the boundary conditions (b.c.) that the electric
field vanish on the surface of the cell, i.e. dψ/dr|r=R = 0. The field on the surface of the particle (i.e. at r = a)
is determined by the surface charge density σ . Finally, due to the b.c. at r = R, the osmotic pressure P of the
microions is given by [19] :
P = kB T [ρ+ (R) + ρ− (R)] (11)
PB theory, embodied in eqs. (9) and (10), reduces here to a spherically symmetric one-centre problem, giving rise to
a simple second order non-linear differential equation for the local potential ψ(r) , which is easily solved numerically
[22,23]. Improvements over the mean-field approximation are achieved by including correlations within the LDA or
the WDA [23,24]. The resulting osmotic pressures are consistently lower than those predicted by PB theory, by a
factor of three or more for highly charged polyions. This trend is confirmed by MC simulations of the microions in
the cell, subjected to the electrostatic potential of the polyion [23,24].
Cell model calculations have been extended to other geometries, e.g. rods or platelets in cylindrical cells. They
prove very useful in the determination of effective polyion charges, as discussed in the next section.
5
III. CHARGE REGULATION AND RENORMALIZATION
Most published work on electric double-layers is based on the assumption of constant charge on the polyions, e.g.
in the form of a constant and uniform surface charge. This is clearly an oversimplification, as would be the other
extreme assumption of constant surface potential, for two reasons : the strong coupling between electrostatics and
chemical dissociation equilibrium at the surface (charge regulation), and the adsorption and strong physical binding
of counterions to the surface, often referred to as counterion condensation, which leads to a reduction of the apparent
polyion charge seen at larger distances (charge renormalization).
A. Charge regulation
The bare (or structural) surface charge of most polyions results from the dissociation of functional groups like
sulphate or carboxylic groups, the number of which may be measured by titration [25]. Generally the ionization
is only partial, and the dissociation equilibrium, governed by a law of mass action, depends on the local ionic
environment, e.g. on local salt concentration or pH. Any variation of this environment, linked to the relative motion
of neighbouring polyions, will lead to a fluctuation of the surface charge, but for given macroscopic conditions, one may
define some average bare charge. From a theoretical point of view, the coupling of the surface chemistry to the local
inhomogeneities induced by electrostatics poses a difficult challenge. Early attempts were based on a combination of
the law of mass action with PB theory [26]. A more microscopic point of view is adopted in the ”charge regulation
primitive model” (CRPM) [27] (see also Ref [28]), which adds a strong attraction of chemical origin to the short-range
sr
part vpα of the polyion-microion pair potential, such that :
sr 1
exp −vpα (r)/kB T = Vα δ(r − dα ); r< (a + aα ) (12)
2
1
= 1; r > (a + aα )
2
This corresponds to an infinitely deep and narrow potential well localized on a sphere of radius dα from the centre of
the polyion; in practice dα must be chosen to be smaller than the polyion radius a, to prevent the chemical binding
of the same microion to two polyions.
The pair structure of the model has been analysed by diagrammatic expansion, and by numerical solutions of the
HNC equation for small polyions corresponding to mineral oxide particles [27]. The charge regulation strongly affects
the effective interaction between spherical particles, to be discussed later.
B. Charge renormalization
The bare (or structural) charge resulting from the dissociation equilibrium is frequently very large, typically
|Z| ≥ 104 for polyion radii a ≥ 102 nm. To describe electric double-layers near such highly charged polyions,
the traditional phenomenological approach is to divide the counterions into two populations. The first includes ions
that are tightly bound (or adsorbed) to the surface by the strong electric field E = 4πσ/ǫ, thus forming a so-called
Stern layer of ”condensed” counterions. A rough estimate of the thickness ∆ of a Stern layer is obtained by balancing
the electrostatic work E × ze × ∆ against the thermal energy kB T . The resulting ∆ = ǫkB T /4πσze is of the order
of a few Å in water at room temperature for typical surface charge densities and monovalent counterions; thus ∆
is comparable to the size of microions, so that the ”condensate” may be expected to be roughly a monolayer. This
monolayer, of opposite sign to the surface charge, strongly compensates the latter, reducing the total polyion charge
to an effective charge Z ∗ significantly smaller than the bare charge Z. The remaining counterions feel a much reduced
”external” potential and form the ”diffuse” part of the double-layer, which can often be treated within linearized PB
theory (LPB).
The phenomenological approach is to consider Z ∗ as a parameter adjusted to experimental data, e.g. polyion struc-
ture factors determined by light scattering [2,29–31], but much recent theoretical effort has gone into determining the
effective charge from first principles. Most schemes are based on the observation that the asymptotic behaviour of
the potential or density profiles is correctly described by the simple exponential screening predicted by LPB theory.
6
Charge renormalization should account both for non-linearities in the mean-field PB approach and for microion cor-
relations. Early attempts focused on the single polyion problem within the cell model [22,23]. Within LPB theory
the electrostatic potential in a spherical cell is easily calculated to be of the form :
Ze
AeκD r + Be−κD r

ψ(r) = C + (13)
ǫr
where κD = 1/λD is the inverse Debye screening length :
!1/2
X
κD = 4πlB nα zα2 (14)
α
The integration constants A and B are determined by the b.c. at r = a and r = R; C is conveniently chosen such
that ψ(r = R) = 0 . The validity of the LPB approximation is expected to be reasonable on the cell surface, but to be
very poor near the polyion surface if the latter is highly charged. The bare charge Z appearing in the LPB potential
(13) is then adjusted to a (lower) effective value Z ∗ by requiring that the resulting microion charge density on the cell
surface, ρ(R), match the corresponding density obtained from a numerical integration of the PB equation under the
same b.c. [22]. In view of the latter, and of Poisson’s equation, this means that the resulting LPB potential matches
the non-linear PB potential at the surface up to the third derivative. Note that a renormalization of Z also implies a
renormalization of κD at low (or vanishing) concentration of added salt, due to the overall charge neutrality require-
ment. In the absence of salt (counterions only), the matching condition also implies equality of osmotic pressures.
Since LPB theory underestimates the charge density near the polyion surface, it overestimates ρ(r = R) , compared
to the PB charge profile, so that Z ∗ is always lower than Z. Z ∗ → Z for low polyion charge, then increases with Z,
but saturates for high values of the bare charge, as a consequence of counterion condensation; the degree of counterion
condensation is not greatly affected by salt concentration [22]. The saturation value of Z ∗ is roughly proportional to
the polyion radius a.
The effects of microion correlations on the values of Z ∗ have been investigated with LDA and WDA versions of
DFT, and by MC simulations [23,24]. Correlations tend to lead to a further decrease of Z ∗ compared to Z, and even
to a maximum value, followed by an actual decrease of Z ∗ as a function of Z [23]. If the effective charge is obtained
by matching the LPB osmotic pressure to ”exact” MC estimates, the resulting Z ∗ turns out to depend only weakly
on the ratio a/R, i.e. on the concentration of polyions.
Other semi-phenomenological criteria for determining the effective charge are discussed in ref. [32]. The Debye-
Hückel-Bjerrum theory, that has proved very successful to describe the phase behaviour of PM electrolytes [33] has
been extended to charged colloids [34]. The Bjerrum pairs of simple electrolytes are replaced by clusters involving
each a polyion and a variable number of adsorbed counterions, which are in chemical equilibrium with the free, non-
adsorbed counterions, assuming a simple form for the internal partition function of a cluster. The distribution of
cluster sizes turns out to be sharply peaked, allowing a rather clear-cut definition of Z ∗ [34].
A more fundamental approach to a definition of effective charges is based on a careful analysis of the asymptotic
decay of density profiles and correlation functions. This may be achieved within the ”dressed ion” reformulation of
linear response theory, i.e. of the polarization of an electrolyte by the average electrostatic potential [35]. Within
the PM, the point charges of the microions are replaced by a local, spread-out charge distribution incorporating
the short-range fraction of the polarization charge around each bare ion, which is thus replaced by a dressed ion.
The short-range fraction of the polarization charge is defined in terms of short-range contributions hsr αβ to the pair
correlation functions hαβ (r) = gαβ (r) − 1 ; the hsr sr
αβ are related to the short-range parts cαβ of the direct correlation
functions, introduced after eq. (7), via coupled Ornstein-Zernike (OZ) relations [36]. This leads to an exact, non-local
generalization of the LPB equations, and a residue analysis of the corresponding Fourier transform of the potential
ψi (r) around an ion i leads to the following asymptotic decay for large r [35,36]:
zi‡ e exp(−κr)
ψi (r) ∼ (15)
ǫ‡ r
This is precisely of the LPB (or Debye-Hückel) form, but with renormalized values zi‡ , κ and ǫ‡ of the valence, inverse
screening length and dielectric constant of the solution (as opposed to the pure solvent), which can all be expressed
in terms of the hsr
αβ ; the latter may be calculated by supplementing the OZ relations with some approximate closure,
e.g. HNC.
7
In the weak-coupling limit of very low ionic bulk concentration, zi‡ → zi , κ → κD and ǫ‡ → ǫ. Simple Stillinger-Lovett
sum rule [37] considerations show that proper inclusion of the finite size aα of the ions leads to an enhancement of
screening, i.e. κ > κD [38]. For sufficiently strong coupling (i.e. at high ionic concentrations), the exponential decay
(15) goes over to a damped oscillatory behaviour characterized by a pair of complex conjugate inverse decay lengths
κ and κ∗ [38–40]. The location of the cross-over from monotonous to oscillatory decay depends on the approximate
closure.
The ”dressed ion” reformulation for the bulk PM may be extended to the case where one ionic species consists
of polyions, and hence zi∗ is to be identified with the effective charge Z ∗ of the latter [36]. A similar treatment has
been applied to planar electric double layers, to determine effective surface charge densities [41]. The latter tend to
saturate for monovalent counterions, increasingly so as their concentration increases, while for concentrated solutions
of divalent ions, the effective surface charge goes through a maximum before decreasing as the bare surface charge
increases. This behaviour is reminiscent of that observed in a spherical cell [23,24].
The effective polyion charge Z ∗ is not easily accessible to experiment. Indirect determinations are based on the
assumption of the validity of some simple functional form of the effective pair interaction between polyions, generally
a screened Coulomb (or DLVO) potential. The thermodynamic properties and pair structure of a system of polyions
interacting via this potential can be determined accurately from fluid integral equations or simulations [42,43]. The
effective charge Z ∗ is then adjusted to provide the best theoretical fit to experimental data, e.g. the polyion pair
structure as measured by scattering experiments [2,29–31], or the freezing line as a function of salt concentration and
bare particle charge [44]. The latter measurements confirm the saturation predicted by the PB cell model [22,23],
but should be re-analysed since the underlying simulation data do not take into account an important ”volume”
contribution to the free energies, which strongly affects the phase diagram at low salt concentration [45,46].
Recent experiments, carried out at very low electrolyte concentration, on silica particles, with weakly dissociat-
ing silanol surface groups, and on latex particles, with strongly dissociating sulfonic acid groups, were able to measure
simultaneously the bare polyion charges by conductometric titration, and the effective charges by conductivity mea-
surements [47,48]. The bare charge was controlled by varying the amount of added NaOH. Under √ these experimental
conditions, Z ∗ was not observed to saturate, but to follow very nearly a square root law Z ∗ ∼ Z for both colloidal
systems.
IV. PLANAR GEOMETRY
The simplest, and most widely studied double-layer geometry is that of uniformly charged, infinite planes in an
ionic solution [17]. These planes may represent stacks of thin charged lamellae, like clay platelets or rigid membranes,
or correspond to the surfaces of spherical colloidal particles separated by a distance h ≪ a, so that the curvature of
the facing surfaces may be neglected in first approximation. In the former situation, one may consider the simplified
model of a single charged plane within a slab of thickness h equal to the mean spacing between lamellae in the stack;
this would be the 1d equivalent of the cell model. The two situations are shown schematically in Figure 1; σ denotes′
the surface charge density, ǫ is the macroscopic dielectric constant of the solvent (within a PM representation) and ǫ
′
is the dielectric constant of the colloidal particles. When ǫ 6= ǫ, which is the rule rather than the exception, the di-
electric discontinuity at z = ±h/2 must be properly incorporated in the boundary condition, e.g. by the introduction
′
of electrostatic image charges [49]. In most published theoretical work the assumption ǫ = ǫ is made for the sake of
simplicity.
Microion density profiles ρ+ (z) and ρ− (z) depend only on z, and the intrinsic free energy functional per unit area
perpendicular to the z axis is still given by eqs (5)-(8), with the volume integrals replaced by 1d integrals over the
interval [− h2 , h2 ] . Fixed charge (or Neumann) boundary conditions imply that the potential satisfy :
dψ 4πσ
− =± (16)
dz z=± h ǫ
2
The key physical quantity, which is in principle measurable using a surface force apparatus [1], is the force per unit
area acting between two charged planes and their associated electric double-layers. This force per unit area is the net
osmotic pressure (or disjoining pressure) ∆P = P − Pbulk , where P is the osmotic pressure exerted by the microions
8
between the charged plates, and Pbulk is the pressure of the electrolyte in the reservoir which fixes the chemical
potentials of the microions. P may be calculated via the mechanical route, by averaging the local stress tensor, or
by the thermodynamic route, by differentiation of the total free energy, or grand potential, with respect to h. Both
routes lead to a natural separation of the pressure P into kinetic, electrostatic and collisional parts [49] :
P = Pkin + Pel + Pcoll (17)
where
X
Pkin = kB T ρα (z) (18)
α
and explicit expressions for Pel and Pcoll in terms of density profiles and inhomogeneous pair correlation functions
are given in ref. [49]. Note that, while each of the terms in eq. (17) depends on the coordinate z, their sum must
be independent of z for mechanical equilibrium. In practice, simplified expressions are obtained by calculating P in
mid-plane (z = 0 in Fig. 1) or at the surface of the charged planes. In the latter case the expression for P reduces to
the simple contact theorem [5] :
X 2πσ 2
P = kB T ρα (z = ±(h/2 − aα )) − (19)
α
ǫ
This expression is not very convenient for simulation purposes, since P appears as a difference of two large numbers,
the first of which involves the contact value of the counterion density profile, affected by large numerical uncertainties.
The calculation at midplane is then preferable, since the ρα are expected to be close to their bulk values there, but
the average force between microions on both sides of this plane must be evaluated [50].
A. Poisson-Boltzmann approximation
The mean-field approximation, where correlations between microions, due to excluded volume and Coulomb inter-
actions, are neglected, is well understood since the pioneering work of Gouy [17]; density profiles and forces are known
analytically, or given by simple quadratures [51]. The pressure is given throughout by :
ǫ
[E(z)]2
X
P = kB T ρα (z) − (20)
α
8π
where E(z) = −dψ(z)/dz is the

P mean local electric field. This is most easily evaluated in the midplane, where E = 0
by symmetry, i.e. P = kB T α ρα (0) , showing that the force between equally charged plates is always repulsive
within PB theory.
In the limit of low surface charge and electrolyte concentration, PB theory may be linearized, and the disjoining
pressure reduces to :
∆P = 2σ 2 e−κD h (21)
A charge renormalization to account for non-linearities, very similar to the procedure within the cell model described
earlier, leads to a much reduced effective surface charge σ ∗ , and to a saturation for large bare charges σ [52].
B. The effect of microion correlations
While PB theory always predicts a purely repulsive interaction between planar electric double-layers, it was sug-
gested by Oosawa as early as 1968 [53] that the force might turn attractive if microion correlations were taken into
account. The first convincing evidence for double-layer attraction came from MC simulations which indicated large
deviations from the predictions of PB theory in the presence of divalent counterions [50]. This early prediction of
an attractive minimum in the effective force between planar double-layers at short separations h was confirmed by
subsequent simulations in the presence of salt (1 : 2 and 2 : 2 electrolytes) [49,54] and polyvalent counterions [55], by
inhomogeneous HNC calculations [49], and an improved version based on the reference HNC (RHNC) closure [56,57],
and by LDA and WDA versions of DFT [58,59].
9
The reduction of the double-layer repulsion and eventual correlation-induced attraction may be understood by con-
sidering the contact theorem (19), or the expression for the pressure in midplane. At contact the correlations keep
the counterions in the first layer apart, thus limiting the piling up following from the uncorrelated mean field treat-
ment, and reducing the contact value of the counterion density; according to eq. (19) this leads to a lowering of the
pressure, which can even become negative (corresponding to an effective attraction), due to the negative Coulombic
contribution. The lowering of the counterion concentration immediately at contact leads to an enhanced attraction of
the next layers of counterions to the surface, so that correlations enhance the counterion density in that region. This
enhancement in turn entails a depletion of the counterion density in the midplane, while electrostatic correlations
make a negative contribution (Pel in eq. (17)) which generally dominates the smaller, positive collisional part. Overall
these various effects lead to a much reduced ∆P compared to the repulsive PB interaction. The attraction observed
for very short separations may be explained by a 2d lattice-like structuring of the adsorbed counterions, when the
counterion patterns on opposite surfaces are shifted relative to each other to minimize the electrostatic interactions
[60].
The reason why an attraction is in general observed only with divalent (e.g. Ca++ ) counterions is that the en-
tropic cost associated with their ”condensation” near the surface is smaller than for monovalent ions, since only half
as many counterions are needed to provide the same electrostatic shielding. Another related aspect of microion cor-
relations is the phenomenon of charge reversal (or inversion). The apparent charge density of the surface placed at
z = −h/2, seen at the abscissa z is [61] :
Z z X
′ ′
σ ∗ (z) = σ + e zα ρα (z )dz (22)
−h
2 α
∗
Within PB theory σ never changes sign, but in the presence of correlations this may occur at a critical z, beyond
which σ ∗ is of opposite sign to σ , due to an over-compensation of the bare surface charge by the counterions. Thus
a test-charge will be attracted, rather than repelled, by a plane carrying a surface charge σ of the same sign, beyond
a critical separation. Charge inversion can only occur when the microion density profiles have a non-monotonic be-
haviour. The over- compensation of the surface charges of two parallel plates is enhanced when the counterions are
linked together to form polyelectrolyte chains, due once more to the reduction of entropic repulsion [61,62].
Charge inversion and effective attraction between platelets may even be observed in the case of monovalent counteri-
ons, provided the coions are larger than the counterions [57].
Attractive forces between charged surfaces immersed in aqueous solutions of CaCl2 and Ca(NO3 )2 have been measured
directly with a surface force apparatus [63]. The existence of correlation-induced attraction between charged surfaces
of equal sign has important consequences on colloid stability, and provides a plausible explanation for the common
observation that the addition of multivalent ions to solutions or suspensions of polyions often leads to precipitation.
V. SPHERICAL POLYIONS
Aqueous solutions and dispersions of spherical or quasi-spherical polyions, including globular proteins and charge-
stabilized colloidal particles, like silica mineral particles or polymer latex spheres, have been thoroughly studied
experimentally and theoretically for many decades. Much of the interest stems from the realization that such colloidal
systems exhibit a phase behaviour reminiscent of that of simple molecular systems, albeit on very different scales. A
particularly attractive feature of colloidal dispersions is that the effective interactions between particles may be tuned
by the experimentalist, e.g. by varying the concentration of added electrolyte, thus providing an additional handle
on phase behaviour. The beautiful early microscopy observations by Hachisu et al. [64] unambiguously showed the
existence, at low volume fractions, of colloidal crystals and crystalline alloys, which Bragg-reflect visible light and are,
in particular, responsible for the iridescence of opals.
The coexistence of ordered crystalline and disordered fluid phases has been monitored by very careful experiments
using confocal laser scanning microscopy, which allows the direct observations of ordered and disordered configura-
tions, and ultra-small-angle X-ray scattering (USAXS), which provides statistically averaged information on ordering
of colloids via the static structure factor S(k) [47,48]. These measurements provide three-dimensional phase diagrams
by varying the colloid concentration n or packing fraction η = 4πna3 /3, the effective colloid charge Z ∗ (as determined
by conductivity measurements), and the monovalent salt concentration ns , equal to the concentration n− of coions,
10
typically in the range 10−6 − 10−5 M. The observed phase diagrams exhibit a striking re-entrant behaviour, e.g. for
fixed n and ns , the colloidal dispersion crystallizes into an ordered BCC lattice upon increasing Z ∗ , as one might
expect, but remelts into a disordered fluid phase upon further increase of Z ∗ , . This re-entrant behaviour may be
qualitatively understood in terms of an increase of the total ionic strength, linked to an increase of the counterion
concentration n+ with Z ∗ , which enhances the screening power of the microions and hence reduces the range of the
screened Coulomb repulsion between polyions. However, the standard DLVO representation of the latter [65] fails to
provide a quantitative explanation of the experimental data [66].
Another striking feature of low concentration dispersions (η < 0.05) , reported by the Kyoto School, is the ap-
pearance of strongly inhomogeneous patterns in highly deionized samples. USAXS measurements and confocal laser
scanning microscopy show clear evidence of regions of relatively high colloid concentration, which may be crystalline
[67], amorphous [68] or fluid [69] in character, coexisting with regions of extremely low concentration, or voids. Apart
from direct visual observation, the existence of voids, and the fraction of the total volume occupied by them, may be
inferred from a comparison between the mean inter-particle spacing, dn ∼ n−1/3 , calculated under the assumption of
a homogeneous dispersion, and the spacing dx deduced from the position of the main X-ray diffraction peak in the
structure factor S(k). The observation that dx is significantly less than d points to the existence of voids occupying a
fraction f = 1 − (dx /d)3 of the total sample volume. The considerable literature on the subject is summarized in ref
[70]. The existence of voids, and the related, but controversial [71], observation of a complete separation between a
high concentration colloidal ”liquid” phase, and a much more dilute ”gas” phase [72], are claimed to be evidence for a
long range attractive component of the effective pair interaction between colloidal particles carrying charges of equal
sign. The same attraction is also invoked to explain the re-entrant liquid-solid coexistence described earlier [66].
However recent direct measurements of the effective interaction between pairs of colloidal particles show no evi-
dence of an attraction, at least at low concentration. Such direct measurements are based on video microscopy. In
the simplest method, a large number of instantaneous configurations of a dilute suspension are recorded, and the
colloid-colloid pair distribution function g(r) is calculated from the measured distribution of inter-particle distances
[73]. The effective pair potential coincides with the potential of mean force in the low concentration limit [8] and is
hence given by :
v(r) = −kB T log[g(r)] (23)
An alternative method follows the relative Brownian motion of a pair of colloidal particles released from initial
positions where they were localized by optical tweezers [74]. Both sets of measurements confirm that the effective
pair potential is purely repulsive, and may be fitted to a screened Coulomb (DLVO) form by adjusting the effective
charge Z ∗ and the inverse screening length κD , compatible with estimated salt concentrations. However this same
DLVO potential, used within a harmonic approximation, seems to be unable to reproduce the measured bulk modulus
of a colloidal crystal at much higher concentrations [75,76], although the difference between experiment and theory
may be linked to the omission of the ”volume” contribution to the effective interaction energy between polyions, to
be discussed in the following section.
A. DLVO theory revisited
The effective pair potential between spherical polyions, first derived by Derjaguin and Landau, and Verwey and
Overbeek [65], is easily recovered within the framework of DFT. Upon tracing out the microion degrees of freedom, the
effective interaction energy between polyions is given by eq. (8), once the grand potential Ω has been determined from
the variational principle (1). The required intrinsic free energy functional is defined by eqs. (5) and (8). Neglecting
Fcorr in the latter amounts to mean-field theory, and the Euler-Lagrange equation associated with the variational
principle (1) leads back to the PB equation (10) for the local electrostatic potential ψ(r) . Since the external charge
associated with N polyions is a multi-centre distribution, the numerical task of solving the PB equation for N inter-
acting polyions is a formidable one, which can only be handled numerically using advanced optimization techniques,
coupling Molecular Dynamics (MD) simulations with DFT [77–79].
A more phenomenological approach may be adopted, whereby the bare polyion charge is reduced to a considerably
lower effective value by a Stern layer of tightly bound counterions. The coupling between the polyions and the remain-
ing microions (forming the so-called ”diffuse” double-layer) is accordingly strongly reduced, so that the corresponding
density profiles ρα (r) vary much more smoothly in the vicinity of the polyion surfaces. In that case the integrands of
11
the ideal contributions to F in eq. (8) may be expanded to second order in the deviations ∆ρα (r) = ρα (r) − nα of the
local densities from their bulk values, i.e. :
X kB T
Z
2
Fid ≈ Fid (V, T, nα ) + [∆ρα (r)] dr (24)
α=±
2nα V
where Fid (V, T, nα ) is the Helmholtz free energy of an ideal gas of density nα . The free energy functional defined
by (5) and (7), neglecting Fcorr , is now a quadratic functional of the local densities, and the resulting Euler-Lagrange
equation reduces to a linear multi-centre PB (LPB) equation for the local potential ψ(r) , in the familiar form :
4π
∇2 − κ2D ψ(r) = − ρext (r)

(25)
ǫ
This linear equation is easily solved by Fourier transformation in the bulk, with the boundary condition that the
potential and its gradient vanish at infinity. Since ρext is a linear superposition of contributions from the N polyions,
the same is true of ψ(r) and the resulting ρα (r) . Since microions cannot penetrate the spherical polyions, the
excluded volume condition, ρα (r) = 0; |r − Ri | < a(1 ≤ i ≤ N ) , must be imposed via a constraint, or by the
use of a polyion- microion pseudo potential [45,46]. Note that, strictly speaking, the linear superposition of densities
only holds provided the electric double-layer associated with a given polyion does not overlap a neighbouring polyion
(”weak overlap approximation”). This is not a significant limitation except at high polyion packing fraction. The
excluded volume condition leads to an additional polyion charge renormalization such that the effective charge Z ∗ is
multiplied by the DLVO factor eκD /(1 + κD a); the product will henceforth be designated by Z ∗ . The local potential
is finally of the form :
N N
X X Z ∗ e exp (−κD |r − Ri |)
ψ(r) = ψ (i) (r) = (26)
i=1 i=1
ǫ |r − Ri |
When (26) and the resulting ρα (r) are substituted into the quadratic free energy functional, the effective interaction
energy (8) between the N polyions reduces to :
X
VN ({Ri }) = V0 + v (|Ri − Rj |) (27)
i<j
where V0 (T, n, n+ , n− ) is a state-dependent ”volume” term [80], the detailed expression of which is given in [46] and
[81]. This term has been generally overlooked, but it strongly influences the calculated phase behaviour of charge-
stabilized colloidal dispersions, as discussed later. The main contribution to the volume term is the sum of cohesive
electrostatic free energies arising from the electrostatic attraction between each polyion and its associated double layer
of opposite charge. The physical interpretation of V0 is discussed in detail in ref. [82]. The effective pair potential is
precisely of the well known DLVO form [65], namely:
Z ∗2 e2 e−κD R
v(R) = (28)
ǫ R
Note that the pair-wise additivity in eq. (27) is a direct consequence of the quadratic nature of the approximate free
energy functional; the exact ideal contribution (5) to the free energy would lead to many-body effective interactions.
There are alternative contraction procedures to reduce the initial asymmetric, multicomponent PM to an effec-
tive one-component system, based on the OZ equations and approximate closures, like the MSA, for the partial pair
distribution functions gαβ (r) [10–13]. The simple assumption of the asymptotic behaviour cαβ (r) → vαβ (r)/kB T of
the direct correlation functions is sufficient to show that the effective pair interaction is asymptotically of the DLVO
form (28), and reduces to (28) in the appropriate limits. Corrections to the DLVO potential at short range may
be obtained from numerical solutions of the fluid integral equations, and confirm the purely repulsive nature of the
effective pair potential at low colloid concentration, at least when all microions are monovalent [13].
The phase diagram of a monodisperse system of particles interacting via the DLVO pair potential (28) has been
determined by extensive MC simulations as a function of colloid and salt concentrations, the latter determining the
inverse Debye screening length κD [42,43]. As expected from the purely repulsive nature of the effective pair po-
tential, the phase diagram exhibits a first- order transition between a single (disordered) fluid phase and (ordered)
FCC or BCC crystalline phases, BCC being the stable phase in the κD → 0 limit, which corresponds to the widely
12
studied ”one-component plasma” (OCP) model [83]. However the coarse-graining procedure leading from the initial
multicomponent polyion/microion ”mixture” to the effective one-component system also introduces the volume term
V0 into the effective interaction energy between polyions. Since V0 is a non-linear function of the polyion density n,
due to the overall charge neutrality condition nZ ∗ + n+ z+ + n− z− = 0, this term must be included in the free energy
for a proper determination of the phase behaviour [45,46,82]. At high concentrations of added salt, the variation
of V0 with n is sufficiently slow not to affect the phase behaviour significantly. At salt concentrations lower than
10−5 M, the volume term drives a van der Waals-like instability of the fluid, which separates into two phases of very
different colloid concentrations, reminiscent of the ”gas” and ”liquid” phases of ordinary molecular fluids. Depending
on the effective colloid surface charge, this phase separation is completely disconnected from the freezing transition
driven by the repulsive pair interaction (28), and exhibits upper and lower critical points (re-entrant behaviour) in
the n − ns plane (see Figure 2); or it merges with the freezing line, leading to a considerable broadening of the
fluid-solid coexistence region, and the possibility of upper and lower triple points,as seen in Figure 3. The possibility
of a fluid miscibility gap in polyelectrolytes had been conjectured already in 1938 by Langmuir, who referred to it
as ”unipolar coacervation” [84], and the complex phase scenario of references [45,46,82] has recently been confirmed
by an extension of Debye-Hückel theory to the highly asymmetric PM [85]. These calculations provide a natural
explanation for the observed phase behaviour of charge-stabilized colloids, including the formation of voids [67–72],
without the assumption of a long-range attractive component in the effective pair interaction which is frequently made
[86], but lacks a firm theoretical basis.
B. Beyond DLVO
It was shown earlier that the effective force between two uniformly charged plates may turn attractive at short
distances if correlations between microions are properly accounted for. A similar effect is expected to hold between
two spherical polyions, as first conjectured for two isolated spheres by Patey [15] on the basis of numerical solutions
of the HNC equation. When two spheres are sufficiently close, such that the shortest separation h between their sur-
faces is much less than their radius a, the problem reduces essentially to that of two charged planes discussed earlier
: a short-range correlation- induced attraction may be expected for divalent counterions. A collective polarization
mechanism of adsorbed counterions operates in the opposite limit h ≫ a, and leads to a ”doubly screened” attractive
component, proportional to e−2κD R , which is always dominated by the DLVO repulsion [87]. It is important to
stress that effective attractions between spherical polyions require a finite concentration of the latter in the absence
of salt, i.e. when only counterions are present [87,88]. No such restriction holds at finite salt concentration, where the
problem of two isolated polyions is a meaningful limit.
A similar ”doubly screened” attraction between spherical polyions occurs in the charge regulation PM introduced
earlier (c.f. eq. (12)); explicit HNC calculations have been carried out for small mineral oxide particles [27]. Note
that the effective interaction between two isolated colloids is purely attractive (for R > 2a) in the case of a symmetrical
adsorption of co and counterions (V+ = V− in eq. (12)), because the colloidal particles are then, on average, neutral.
C. Computer simulations
Density-functional molecular dynamics simulations were performed for monovalent counterions without [78,89] and
with [90] added salt. The effective interaction between the polyions was found to be repulsive, and in quantitative
agreement with DLVO theory, provided the charge was renormalized according to the cell model prescription incor-
porating the excluded volume correction [91]. This approach includes counterion correlations, but is not completely
equivalent to a full simulation of the PM since an approximate polyion-counterion pseudopotential is used and the ex-
act density functional of the electrolyte is unknown. Due to the high charge asymmetry between poly- and microions,
a full simulation of the PM requires the inclusion of many counter- and salt ions and is not feasible on present-day
computers. There are two ways to escape from this: either by considering only a small number of polyions or by
reducing the charge asymmetry significantly. By considering only two polyions within the PM, the effective pair
potential can be calculated within a simulation by averaging over the microscopic ions, while keeping the polyion
positions fixed. For monovalent microions, the simulation yields repulsive forces correctly described by DLVO theory
both with [92] and without [93] added salt. For divalent counterions, attractive forces are obtained, which may be
attributed to counterion correlations [88] and Coulomb depletion [94], resulting from a depletion zone of counterions
between two nearly touching polyions due to the strong counterion repulsion. A typical snapshot of the counterions
around two fixed polyions and the averaged force versus polyion distance are shown in Figure 4. A system of three
13
polyions was simulated in order to extract effective triplet interactions which were found, however, to be small with
respect to the pairwise part [95].
On the other hand, there are quite a number of full simulations of the PM with a reduced charge asymmetry corre-
sponding to micellar rather than colloidal polyions. Most of the work is summarized in a review by Vlachy [96]. As
for more recent work, the role of salt valency has been investigated in some detail [97]. Moreover Linse and Lobaskin
pushed the charge asymmetry to Z : z+ = 60 : 1, 60 : 2, 60 : 3 using efficient cluster move algorithms [14,98] and
treating 80 polyions simultaneously [99]. For the 60 : 3 asymmetry clear indications for a phase separation were
reported which can be attributed to an effective attraction between polyions.
VI. POLYIONS IN CONFINED GEOMETRIES
The bulk behaviour of polyionic dispersions is expected to change significantly in the presence of neutral or charged
interfaces confining the dispersion to a restricted volume. Such confinement occurs naturally in the vicinity of the
sample container, which may be approximated locally by an infinite plane restricting the suspension to a half space.
Many experiments are carried out in a slit geometry, with the suspension confined between parallel planes (e.g. glass
plates); if the interplanar spacing is comparable to the diameter 2a of the polyions, the suspension behaves like a
quasi-two- dimensional (2d) many-body system exhibiting interesting phase behaviour [100].
The electrostatic effects of confinement are essentially threefold. (i) The interface between two different dielectric
media, e.g. the suspension and the glass of the container wall, must be characterized by appropriate boundary condi-
tions; in particular the dielectric discontinuity implies the presence of electrostatic image charges. (ii). The confining
surfaces lead to a reduction in screening power of the electrolyte, which ceases to be exponential in directions parallel
to the surface [101]. (iii) If the surfaces are charged, they attract or repel the polyions and microions, and release
additional counterions into the suspension.
These effects will modify the electrostatic interactions between double-layers associated with polyions close to the con-
fining walls. This has been clearly demonstrated by a number of direct measurements of the effective forces between
spherical polyions confined near a glass wall, or in narrow slits, using digital video microscopy techniques similar to
those mentioned earlier for measurements in the bulk. Measurements of the colloid pair distribution function g(r)
carried out at sufficiently low concentration allow the effective pair potential to be extracted directly from eq. (23)
[102,103], while data collected from more concentrated samples, where the effective potential v(r) is expected to differ
significantly from the potential of mean force defined by eq. (23), require an elaborate inversion procedure based on
fluid integral equations (properly adapted to a 2d geometry) [104,105], or on an iterative method involving computer
simulations [102]. These experiments leave little doubt for the existence of an attractive component of the effective
polyion pair potential at low ionic strength (i.e. for salt concentrations estimated to be below about 10−5 M), when
the polyions are highly confined, as achieved in narrow slits [102,104], or by localizing them close to a single plane
by optical tweezers [103,106,107]. The observed attractive well in v(r) is relatively shallow (depth of the order of
kB T ), and long-ranged; typically v(r) becomes negative for interparticle distances considerably larger than the par-
ticle diameter 2a. The experiments carried out for two particles close to a charged plate with the centre-to-centre
vector parallel to the plate, clearly show that the attractive well disappears when the particles are moved away from
the plate; the measured v(r) returns to its bulk DLVO form (28) when the distance to the plate is several particle
diameters [103,106,107]. An additional twist for suspensions of colloidal particles confined to a slit is provided by the
introduction of quenched obstacles, in the form of larger, charged colloids, which form a disordered porous 2d matrix.
As the concentration of such obstacles is increased, the effective interaction potential between the smaller colloids is
observed to develop a second attractive component extending to larger inter-particle separations [108]. Closely related
observations of metastable colloidal crystallites suggest that facets of such crystals behave very much like charged
plates, inducing effective attractions between nearby polyions [106] due to many-body interactions.
The DFT formulation of DLVO theory in the bulk has been extended to the case of confined particles. The electro- ′
static potential due to a point polyion near a planar surface separating two media of dielectric constants ǫ and ǫ was
determined within LPB theory by Stillinger as early as 1961 [109]. The case of two spherical polyions of finite radius
a near an uncharged planar surface was examined within LPB theory in [110]. Since microions cannot leak beyond
the dividing surface, the spherical symmetry of the electric double layers around each of the polyions is broken. The
effective repulsion between the polyions is enhanced, and decays like 1/R3 , where R is the separation parallel to the
surface, as expected from the considerations in [101].
14
If the confining surface carries an electric charge density σ , as would be the case for a glass plate in contact
with an aqueous dispersion, the planar electric double-layer building up near the surface will modify the local co and
counterion concentrations in the vicinity of nearby polyions, and hence influence their mutual interaction. Considering
the charged wall as an additional colloidal particle of infinite radius, it is clear from the inherent pair-wise additivity
that the quadratic free energy functional, defined by eqs (24) and (8) (with Fcorr = 0), which underlies LPB theory,
will not affect the effective interaction between two polyions near the wall. Clearly the intrinsic three-body nature
of the problem must be taken into account. The minimal theory to achieve this is to carry the expansion (24) one
order further, allowing for a direct coupling between the three double-layers associated with the polyions and surface.
Application of DFT perturbation theory [78] leads to the prediction of a long-range attraction between polyions,
decaying like 1/R3 along the surface, and exponentially normal to the surface, for finite concentrations of polyions,
to allow for an imbalance between co and counterion concentrations [111]. Both the location and depth (typically of
the order of kB T ) of the predicted potential well are in semi-quantitative agreement with direct experimental mea-
surements [103,106,107]. The attraction predicted within mean-field theory occurs when the surfaces of the spherical
polyions are several Debye lengths apart, and has nothing to do with the correlation-induced-attraction at very short
range predicted in the bulk, and which is significant only for multivalent counterion [87,88,94,99,97].
The generalization of DLVO theory may also be extended to slit [112,113] and pore [114] geometries. The quasi-
1d confinement in the latter case leads to an even slower decay of the long- range attraction. The existence of an
effective attraction between two like-charged polyions confined by a charged cylindrical surface was also predicted on
the basis of numerical solutions of the non-linear PB equation [115], but it has since been proved rigorously that PB
theory can only lead to purely repulsive effective interaction between two identical polyions confined to a cylinder
of arbitrary cross-section [116,117], whatever the boundary conditions at the charged surfaces. This proof does not
invalidate the results of the perturbation theory [111,113,114], since charged renormalization (which accounts for
short-range correlations), followed by an LPB perturbation treatment of the non-adsorbed counterions is not a mere
approximation to non-linear PB theory, which is, moreover, unphysical near highly charged surfaces.
A direct simulation of the PM for confined colloids is only feasible for low surface charges and small particle sizes [118].
In this regime, the attraction between colloids is strongly enhanced by the presence of a charged wall with respect to
that in the bulk [94]. Also the wall-particle interaction was found [118] to be attractive for divalent counterions.
VII. ROD-LIKE POLYIONS
There is a large variety of rod-like polyions ranging from synthetic suspensions to biological macromolecules.
These include colloidal β-FeOOH [119], imogolite [120], boehmite [121–123], polytetrafluoroethylene [124], ellipsoidal
polystyrene latex particles [125], cylindrical micellar aggregates [126,127] as well as virus solutions from tobacco-
mosaic (TMV) [128,129], or bacterial fd [130,131] and Pf1 viruses [132]. Another motivation to study rod-like colloids
comes from the self-assembly of charged stiff biopolymers such as DNA strands [133,134], F-actin fibers [135] and
microtubules [135] which constitute rod-like polyions on a supramolecular rather than on a colloidal length scale.
The simplest model system is the electric double layer around a single infinitely long cylindrical polyion of ra-
dius a which is homogeneously charged with line charge density λ. LPB theory leads to the electric potential
Ψ(r) = λK0 (κD r)/aK1 (κD a)ǫ where r is the distance from the rod axis and K0 (x), K1 (x) are Bessel functions of
imaginary argument. Consequently, LPB theory predicts the effective interaction v(r) between two parallel rods per
unit length to be of the following form:
λ2
v(r) = K0 (κD r) (29)
(κD aK1 (κD a))2 ǫ
where the factor 1/[κD aK1 (κD a)] is the excluded volume correction. Contrarily to the three-dimension al case, PB
theory can be solved analytically within a cylindrical cell around the charged rod for vanishing salt concentration
[136,18]. The analytical solution for the counterion density field around the colloidal rod exhibits a strong peak
at contact [137] providing a theoretical framework of Manning counterion condensation [138,139]. As for spherical
macroions, one cannot extract an effective interaction from the PB solution for a single double-layer directly. One
possibility is to proceed as for the spherical PB cell model. Matching the LPB theory at the cell boundary [140], one
derives the effective interaction potential v(r) = λ∗2 /ǫK0 (κr), where κ and λ∗ are renormalized according to the cell
model prescription.
15
A more direct theory, which requires a larger numerical effort, is to solve the two-dimensional PB theory for two
parallel rods in a slit geometry and extract the effective force from there. In a recent numerical study [141] no attrac-
tion was found, consistent with the exact result [116,117] in three dimensions.
Beyond the PB level of description, computer simulations for the effective interaction between two parallel, homoge-
neously charged rods have been performed for divalent counterions and no added salt [142]. An effective attraction
between the rods was found. The question is whether the attraction is due to correlations [143] or to fluctuations
[144,145]. If the former is true, the attraction should increase with decreasing temperature and may even persist for
zero temperature, while fluctuation- induced attraction should increase with temperature. On the basis of a simple
model [146], Levin and coworkers gave significant support for a correlation picture. This can be intuitively under-
stood as the gain in electrostatic energy upon bringing together staggered arrays of adsorbed counterions [60,147].
Fluctuation [148,149] and polarization effects [150] may still play an important additional role. For parallel rods on
a triangular lattice, a negative pressure was found by computer simulation with divalent counterions [151], which can
be attributed to some attractive component in the effective interactions.
Qualitatively, the situation closely resembles the case of spherical macroions. But there are also differences. First,
Manning condensation theory applies only to rod-like polyions. A possible mechanism of attraction of nearly touching
rods by sharing Manning-condensed counterions was proposed, which is only possible for rods [152]. Secondly, the
Coulomb depletion mechanism for attraction [94] was found to be irrelevant for parallel rods.
For finite rod lengths and arbitray orientations of the rods, the effective interaction between two rods will depend both
on the rod orientations and on their center-of-mass separation. Starting from point charges distributed along the rods,
LPB theory results in a Yukawa-segment model [153] which was confirmed by density-functional Molecular-Dynamics
simulations for monovalent counterions [140], provided the charge and the screening constant κ are renormalized ac-
cording to the cell model prescription. The associated phase diagram for parameters suitable for the TMV suspension
was calculated, involving different liquid crystalline phases [154].
A full simulation of the PM was performed for stiff polyelectrolytes [155], resulting in bundle formation, which is
a possible sign of an effective attraction. Clearly, the effective pair potential picture is insufficient for bundles [156]
and many-body interactions play a significant role [148,149].
VIII. LAMELLAR POLYIONS
Lamellar polyions may be schematically looked upon as rigid or flexible charged surfaces or platelets, providing a 2d
counterpart of charged rods or polyelectrolytes. Examples are the geologically and technologically important smectite
clays [157] and self-assembled bilayers of ionic surfactants, which constitute the proto-type of biological membranes
[51].
Consideration will first be given to electric double-layers around infinitely thin, uniformly charged circular or square
platelets, which constitute a reasonable model of smectite clay particles, and in particular of the widely studied syn-
thetic Laponite mineral particles. To a good approximation the latter are rigid, thin discs, of thickness d ≈ 1nm,
radius a ≈ 15nm, carrying a structural surface charge σ ≈ −e/(nm)2 . Natural montmorrilonite clays are silicate
mineral platelets of similar chemical composition and crystal structure, but of irregular and polydisperse shape, and
of much larger lateral dimension, implying some degree of bending flexibility.
Dry powders of clay will swell upon the addition of water, releasing counterions into the interlamellar volume, and
building up interacting electric double-layers; the swelling is essentially driven by the double-layer repulsion of mostly
entropic origin. During the initial stages of swelling, the spacing h between platelets remains small compared to
their lateral extension, so that a moderately swollen lamellar phase may be reasonably modelled by a stack of infinite
charged planes [158]. The results for planar geometry, discussed earlier, apply then directly to the swollen phase. In
particular, limited swelling, often observed in the presence of divalent counterions, may be related to the cohesive
behaviour (effective attraction between planes, or negative disjoining pressure) due to correlation effects within the
PM [159]. An important issue in the understanding of the swelling behaviour is the competitive condensation of
counterions of different valence and/or size [159,160]. The PM model is, however, expected to be inadequate when the
interlamellar spacing h is only of the order of a few microion diameters. Under those conditions the molecular nature
16
of the solvent (water) must be explicitly taken into account. Constant pressure MC simulations clearly show the
importance of counterion hydration in determining the swelling behaviour, and point to the role of K+ ions as a clay
swelling inhibitor [161]. When swelling proceeds until the interlamellar spacing h becomes comparable to the platelet
radius a, finite size (edge) effects become important. The force acting on a platelet follows from the integration of
the stress tensor Π over the two faces Σ+ and Σ− of the platelet:
Z
F=− Πds (30)
Σ+ ,Σ−
where ds = n̂ds is the surface element oriented along the outward normal n̂, and the stress tensor has the standard
form :
h ǫ i ǫ
Π = P (r) + |E|2 I − E(r) ⊗ E(r) (31)
8π 4π
which generalizes the uniaxial expression (20). Within LPB theory, eq. (30) leads to the following expression for the
force between two coaxial discs separated by h, immersed in an ionic solution of inverse Debye length κD [162]:
4πσ 2 ∞ 2

1 h
Z q
Fz (h) = (πa2 ) × J1 (x) exp − x2 + κ2D a2 dx (32)
ǫ 0 x a
Numerical solutions of the non-linear PB equation, for the same coaxial geometry, show that the LPB expression (32)
strongly overestimates the force [163], as in the spherical case, thus requiring a proper renormalization of the surface
charge density σ. Recent MC simulations show that the force may become attractive for divalent counterions, as in
the case of infinite charged planes, but that the finite size of the discs leads to significant differences in the relative
weight of electrostatic and contact contributions [164].
Finite concentration effects in highly swollen stacks of parallel platelets of finite size may be examined within a
cell model, compatible with the shape of the platelet. A coaxial cylindrical cell is chosen for a disc-shaped platelet,
while a parallelepipedic cell is better adapted to square platelets [162,165]. Note that the platelet concentration
determines the cell volume, but not the aspect ratio of the cell, e.g. the ratio R/h in the case of a cylindrical cell of
radius R and height h equal to the inter-lamellar spacing; the optimum aspect ratio for given volume and electrolyte
concentration is determined by minimizing the free energy of the microions in the cell. The latter has been calculated
within LPB theory which can be solved analytically [162,165], and within non-linear PB theory, which requires nu-
merical solution; the latter is greatly simplified within a Green’s function formulation [163,166].
The osmotic pressures calculated as a function of Laponite concentration within PB theory agree reasonably well
with experimental data [167], for reservoir salt concentrations of the order of 10−3 M or larger, but differ dramatically
from the predictions of 1d PB theory for stacks of infinite platelets, pointing to the importance of edge effects in
highly swollen clays [166].
In very dilute dispersions of clay platelets, such that the distance between the centres of neighbouring polyions
is significantly larger than the particle radius, the platelets can rotate more or less freely (sol phase). As the concen-
tration increases, a gel phase is formed, depending on salt concentration. Several recent experiments have attempted
to establish a link between the mesoscopic fractal structure of the gel, and its rheological properties [167–169], but
no clear-cut scenario has yet emerged, due to metastability and ageing of the dispersions [170].
A theoretical description of the sol-gel transition of clay dispersions hinges on a knowledge of the highly anisotropic
effective interaction between charged platelets in an electrolyte. The charged segment, or site-site model, introduced
earlier for charged rods has been generalized to charged circular platelets, and used in MD simulations of Laponite
dispersions [171].
For well separated platelets, the effective pair potential reduces to a sum of screened interactions between the electro-
static multipoles associated with the anisotropic electric double layers around each platelet [172]. A simplified version
of such an effective interaction, involving infinitely thin discs carrying an unscreened quadrupole moment, has been
used in MC simulations which predict a reversible sol-gel transition [173].
Lipid bilayers and membranes constitute another class of lamellar polyions, which are flexible. Electrostatic in-
teractions renormalize the bending rigidity of these flexible membranes [51], a subject of ongoing work beyond the
17
scope of this review. In relation to the correlation-induced attraction between like-charged planes discussed earlier,
the fluidity of membranes provides an additional mechanism for attraction between membranes, resulting from the
lateral charge fluctuations within the planes of the latter [174].
IX. DISCRETE SOLVENT EFFECTS
So far all theoretical considerations of interacting electric double-layers were based on the PM, which ignores the
molecular nature of the solvent. The PM thus neglects excluded volume effects of the latter, as well as hydration
of ions and the expected reduction of the local dielectric constant near highly charged surfaces, due to polarization
of nearby water molecules. Neglect of these and other solvent effects is expected to be particularly inadequate in
situations where the spacing between two charged polyion surfaces is only of the order of a few molecular diameters.
To go beyond the PM level of description, simple models have been used to describe the solvent molecules (generally
water). The crudest model is to represent the latter by neutral hard spheres of appropriate diameter, while keeping a
macroscopic dielectric constant ǫ in the Coulombic interactions between ions; this model accounts only for excluded
volume effects. The next refinement is to consider hard spheres with embedded point dipoles, to account for the highly
polar nature of the solvent [8,175], while a reasonable local coordination of the solvent molecules can only be achieved
by adding higher order multipoles [176–178]. In MC or MD simulations, much more sophisticated pair potentials
may be used, involving three or more interaction sites on each water molecule, as in the widely used SPC/E potential
[179].
A purely HS solvent is implicit in the modified PB formulation of Kralj-Iglic and Iglic [180] and others [114,181].
As expected, the modified PB equation leads to a saturation of the counterion density at contact, for high surface
charge densities. The force between two charged surfaces in a mixture of charged and neutral hard spheres was calcu-
lated by Tang et al. [182] within a DFT generalizing the earlier theory of the same authors for a PM electrolyte [59];
the force is a strongly oscillatory function of the spacing between the plates, due to HS layering, and is rather insen-
sitive to surface charge and salt concentration. The latter observation, which holds when solvent, anions and cations
are of the same size, does not carry over to the more realistic case of unequal diameters, which has been investigated
within IHNC theory [183]. At low surface charge, the total force between platelets is reasonably approximated by a
superposition of the pure HS solvent ”hydration” force, and the electrostatic contribution of the ions, as calculated
within the PM, in the case of equal diameters [183,184].
The dipolar HS model for the solvent was first used to determine the structure of the electric double layer near
a single charged wall within the MSA [185,186]. More accurate and complete results on the planar electric double
layer in a dipolar solvent were obtained from numerical solutions of the coupled RHNC equations for the three density
profiles [187]. These calculations give Statistical Mechanics evidence for the reduction of the local dielectric constant
near the charged surface, and for significant electrostriction. A generic density functional, based on Rosenfeld’s suc-
cessful ”fundamental measure” theory for hard core fluids [188,189] has been put forward, which can be adapted to any
polyion geometry [190]; when applied to a single planar double layer, this theory shows a considerable enhancement
of the counterion density at contact, when a dipolar, rather than bare HS solvent is used.
The effective solvation force between two charged plates immersed in an ionic solution of charged and dipolar HS was
calculated within a quadratic free energy functional of the local density, charge density and polarization, generalizing
eqs. (24) and (7) (with direct correlation functions approximated by bulk MSA solutions) [191]. This calculation
provided the first convincing evidence of the strong influence of a granular (as opposed to continuous) solvent on the
solvation forces at short range.
The most complete investigation so far of the potential of mean force between two spherical polyions immersed
in an ionic solution with a ”realistic” solvent involving hard spheres with point dipoles and tetrahedral quadrupoles
[177], appears to be the work by Kinoshita et al [192], who solved the RHNC equations for this highly asymmetric
multicomponent ”mixture”, for ratios 10 ≤ a/as ≤ 30, where a and as are the polyion and solvent molecule radii.
While maintaining as fixed at 0.14nm, a value appropriate for water, these authors varied the counterion radius.
Their most striking finding is that larger counterions are more strongly adsorbed to the polyion in the presence of
a molecular solvent, leading to a greater reduction of the Coulomb repulsion between the polyions, and even to the
possibility of an effective attraction for monovalent counterions, when their radius exceeds as by more than 20%. The
trend towards a reduction of effective repulsion with increasing counterion size is the exact opposite of the prediction
of the PM, a clear illustration of the pitfalls of the latter in describing the short-range behaviour of solvation forces!
18
Obviously more work on discrete solvent models is needed.
X. OUTLOOK
Although the understanding of effective interactions between electric double layers has clearly advanced in the last
decade through a combination of new theoretical approaches, quantitative measurements and large-scale computer
simulations, there are still many open questions. Future research should focus on the influence of image charges [193]
due to dielectric discontinuities between the solvent and the container walls or the polyions themselves. Real samples
possess, moreover, an intrinsic polydispersity in size, charge and shape which becomes relevant for a quantitative
comparison between theory and experimental data. Another rapidly growing field concerns flexible polyelectrolyte
chains whose stiffness is governed by their persistence length [137]. A theoretical approach needs input from both
nonlinear screening and polymer theories. Colloids can also be stabilized by polyelectrolytes, and the resulting
effective interaction becomes qualitatively different from that between charged spheres, particularly near the overlap
concentration. The microscopic incorporation of the solvent is still in its early stage and full molecular theories
describing hydration forces and hydrogen bonding are highly desirable for aqueous suspensions. Lastly alternative
approaches, such as recent field theoretical formulations [194], might lead to additional insight.
ACKNOWLEDGEMENTS
We thank E.Allahyarov, D.Goulding, Y. Levin, P. Linse, V. Lobaskin, P. Pincus, and R. van Roij for helpful
comments.
[1] Israelachvili J. 1992. Intermolecular and Surface Forces. London : Academic Press. 2nd ed.
[2] Chen S.H. 1986. Ann. Rev. Phys. Chem. 37 : 351
[3] Evans R. 1991. In Fundamentals of Inhomogeneous Fluids, ed. D. Henderson, New York : Marcel Dekker
[4] Carnie S.L., Torrie G.M. 1987. In Adv. Chem. Phys., ed. I. Prigogine and S. R. Rice, 56 : 141 New York : Wiley
[5] Blum L., Henderson D. 1991. In Fundamentals of Inhomogeneous Fluids, ed. D. Henderson, New York : Marcel Dekker
[6] Attard P. 1996. In Adv. Chem. Phys., ed. I. Prigogine and S. R. Rice, 92 : 1, New York : Wiley
[7] Arora A.K., Tata B.V.R. 1996. Ordering and Phase Transitions in Charged Colloids. New York : VCH Publishers
[8] Hansen J.P., McDonald I.R. 1986. Theory of Simple Liquids. London : Academic Press. 2nd ed.
[9] Belloni L. 1985. Chem. Phys. 99 : 43
[10] Belloni L. 1986. J. Chem. Phys. 85 : 519
[11] Belloni L. 1988. J. Chem. Phys. 88 : 5143
[12] Khan S., Morton T., Ronis D. 1987. Phys. Rev. A 35 : 4295
[13] González-Mozuelos P., Carbajal-Tinoco M.D. 1998. J. Chem. Phys. 109 : 11074
[14] Lobaskin V., Linse P. 1999. J. Chem. Phys. 111 : 4300
[15] Patey G.N. 1980. J. Chem. Phys. 72 : 5763
[16] Wu J., Bratko D., Prausnitz J.M. 1998. Proc. Natl. Acad. Sci. USA 95 : 15169
[17] Gouy G. 1910. J. Phys. (France) 9 : 457
[18] Fuoss R.M., Katchalsky A., Lifson S. 1951. Proc. Natl. Acad. Sci. U.S.A. 37 : 579
[19] Marcus R.A. 1955. J. Chem. Phys 23 : 1057
[20] Rosenfeld Y., Schmidt M., Löwen H., Tarazona P. 1997. Phys. Rev. E55 : 4245
[21] Kjellander R., Marčelja S. 1988. J. Chem. Phys. 88 : 7129 (Erratum : 89 : 7649)
[22] Alexander S., Chaikin P.M., Grant P., Morales G. J., Pincus P., Hone D. 1984. J. Chem. Phys. 80 : 5776
[23] Groot R.D. 1991, J. Chem. Phys. 95 : 9191
[24] Stevens M.J., Falk M.L., Robbins M.O. 1996. J. Chem. Phys. 104 : 5209
[25] Gisler T., Schulz S.F., Borkovec M., Sticher H., Schurtenberger P., D’Aguanno B., Klein R. 1994. J. Chem. Phys. 101 :
992
[26] Ninham B.W., Parsegian V.A. 1971. J. Theor. Biol. 31 : 405
[27] Belloni L., Spalla O. 1997. J. Chem. Phys. 107 : 465
[28] von Grünberg H. H. 1999. J. of Colloid and Interface Science 219 : 339
[29] Härtl W., Versmold H., Wittig U. 1992. Langmuir 8 : 2885
19
[30] Lutterbach N., Versmold H., Reus V., Belloni L., Zemb T. 1999. Langmuir 15 : 337
[31] Lutterbach N., Versmold H., Reus V., Belloni L., Zemb T., Lindner P. 1999. Langmuir 15 : 345
[32] Belloni L. 1998. Colloids and Surfaces A 140 : 227
[33] Levin Y., Fisher M.E. 1996. Physica A225 : 164
[34] Tamashiro M.N., Levin Y., Barbosa M.C. 1998. Physica A 258 : 341
[35] Kjellander R., Mitchell D.J. 1994. J. Chem. Phys. 101 : 603
[36] Kjellander R., Mitchell D.J. 1997. Molec. Phys. 91 : 173
[37] Stillinger F.H., Lovett R. 1968. J. Chem. Phys 48 : 3858; 49 : 1991
[38] Attard P. 1993. Phys. Rev E48 : 3604
[39] Leote de Carvalho R.J.F., Evans R. 1994. Molec. Phys. 83 : 619
[40] Kjellander R., Ulander J. 2000. In Strongly Coupled Coulomb Systems, eds. C. Deutsch, M.M. Gombert. Les Ulis : EDP
Sciences
[41] Attard P. 1995. J. Phys. Chem 99 : 14174
[42] Robbins M., Kremer K., Grest G.S. 1988 J. Chem. Phys. 88 : 3286
[43] Hamaguchi S., Farouki R.T., Dubin D.H.E. 1996. J. Chem. Phys. 105 : 7641
[44] Palberg T., Mönch W., Bitzer F., Piazza R., Bellini T., 1995. Phys. Rev. Letters 74 : 4555
[45] van Roij R., Hansen J. P. 1997. Phys. Rev. Letters 79 : 3082
[46] van Roij R., Dijkstra M., Hansen J.P. 1999. Phys. Rev. E59 : 2010
[47] Yamanaka J., Yoshida H., Kog T., Ise N., Hashimoto T. 1998 Phys. Rev. Lett. 80 : 5806
[48] Yoshida H., Yamanaka J., Koga Ta., Koga Ta., Ise N., Hashimoto T. 1999. Langmuir 15 : 2684
[49] Kjellander R., Akesson T., Jönsson B., Marčelja S. 1992. J. Chem. Phys. 97 : 1424
[50] Guldbrand L., Jönsson B., Wennerström H., Linse P. 1984. J. Chem. Phys. 80 : 2221
[51] Andelman D. 1995. In Structure and Dynamics of Membranes, eds. R. Lipowsky and E. Sackmann, Amsterdam : Elsevier
[52] Attard P., Mitchell D.J., Ninham B.W. 1988. J. Chem. Phys. 89 : 4358
[53] Oosawa F. 1968. Biopolymers 6 : 1633
[54] Valleau J.P. Irkov R., Torrie G.M. 1991. J. Chem. Phys 95 : 520
[55] Pellenq R.J.-M., Caillol J.M., Delville A. 1997. J. Phys. Chem. B 101: 8584
[56] Greberg H., Kjellander R., Åkesson T. 1997. Molec. Phys. 92 : 35
[57] Greberg H., Kjellander R. 1998. J. Chem. Phys. 108 : 2940
[58] Stevens M.J. Robbins M.O. 1990. Europhys. Lett. 12 : 81
[59] Tang Z., Scriven L.E., Davis H.T. 1992. J. Chem. Phys. 97 : 9258
[60] Rouzina I., Bloomfield V.A. 1996. J. Phys. Chem. 100. 9977
[61] Sjöström L., Åkesson T., Jönsson B. 1996. Ber Bunsenges. Phys. Chem 100 : 889
[62] Borukhov I., Andelman D., Orland H. 1998. J. Phys. Chem. B. 103 : 5042
[63] Kékicheff P., Marčelja S., Senden T.J., Shubin V.E. 1993. J. Chem. Phys. 99 : 6098
[64] Hachisu S., Kobayashi Y., Kose A. 1973 . J. Colloid Interface Sci. 42 : 2
[65] Verwey E.J.W., Overbeek J.T.G. 1948. Theory of Stability of Lyophobic Colloids. Amsterdam : Elsevier.
[66] Yamanaka J., Yoshida H., Koga T., Ise N., Hashimoto T. 1999. Langmuir 15 : 4198
[67] Dosho S., Ise N., Ito K., Iwais S., Kitano H., Matsuoka H., Nakamura H., Okumura H., Ono T., Sogami I.S., Ueno Y.,
Yoshida H., Yoshiyama T. 1993. Langmuir 9 : 394
[68] Tata B.V.R., Yamahara E., Rajamani P.V., Ise N. 1997. Phys. Rev. Lett. 78 : 2660
[69] Ito K., Yoshida H., Ise N. 1994. Science 263 : 66
[70] Ise N., Konishi T., Tata B.V.R. 1999. Langmuir 15 : 4176
[71] Palberg T., Würth M. 1994. Phys. Rev. Lett 72 : 786
[72] Tata B.V.R., Rajalakshmi M., Arora A.K. 1992. Phys. Rev. Lett. 69 : 3778
[73] Vondermassen K., Bongers J., Mueller A., Versmold H. 1994. Langmuir 10 : 1351
[74] Crocker J.C., Grier D.G. 1994. Phys. Rev. Lett. 73 : 352
[75] Weiss J.W., Larsen A.E., Grier D.G. 1998. J. Chem. Phys. 109 : 8659
[76] Dufresne E.R. Grier D.G. 1999. J. Chem. Phys. 110 : 8845
[77] Löwen H., Madden P.A., Hansen J.P. 1992. Phys. Rev. Letters 68 : 1081
[78] Löwen H., Hansen J.P., Madden P.A. 1993. J. Chem. Phys. 98 : 3275
[79] Fushiki M. 1992. J. Chem. Phys. 97 : 6700
[80] Grimson M.J., Silbert M. 1991. Mol. Phys. 74 : 397
[81] Graf H., Löwen H. 1998. Phys. Rev. E57 : 5744
[82] van Roij R., Evans R. 1999. J. Phys. Cond. Matter : in press
[83] Baus M., Hansen J.P. 1980. Phys. Rep. 59 : 1
[84] Langmuir I. 1938. J. Chem. Phys. 6 : 873
[85] Warren P.B. 1999. To be published
[86] Sogami I., Ise N. 1984 J. Chem. Phys. 81 : 6320
[87] Levin Y. 1999. Physica A 265 : 432
20
[88] Grønbech-Jensen N., Beardmore K.M., Pincus P. 1998. Physica A 261 : 74
[89] Tehver R., Ancilotto A., Toigo F., Koplik J., Banavar J.R. 1999. Phys. Rev. E59 : R1335
[90] Löwen H., Kramposthuber G. 1993. Europhys. Lett 23 : 673
[91] Löwen H., D’Amico I. 1997. J. Phys.: Condensed Matter 9: 8879-8892
[92] D’Amico I., Löwen H. 1997. Physica A 237: 25-30
[93] Allahyarov E., Löwen H., Trigger S. 1998. Phys. Rev. Lett. 57 : 5818
[94] Allahyarov E., D’Amico I., Löwen H. 1998. Phys. Rev. Lett. 81 : 1334
[95] Löwen H., Allahyarov E., 1998. J. Phys. Cond. Matt. 10 : 4147
[96] Vlachy V. 1999. Annu. Rev. Phys. Chem. 50: 145
[97] Wu J.Z., Bratko D., Blanch H.W., Prausnitz J.M. 1999. J. Chem. Phys. 111 : 7084
[98] Linse P. 1999. J. Chem. Phys. 110: 3493-3501
[99] Linse P., Lobaskin V. 1999. Phys. Rev. Lett. 83 : 4208
[100] Grier D.G., Murray C. A. 1996. In Ordering and Phase Transitions in Charged Colloids, ed. A. K. Apcora and B.V.R.
Tata, New York : VCH Publishers
[101] Jancovici B. 1982 J. Stat. Phys. 29 : 263
[102] Kepler G.M., Fraden S. 1994. Phys. Rev. Lett. 73 : 356
[103] Crocker J.C., Grier D.G. 1996. Phys. Rev. Lett 77 : 1897
[104] Carbajal-Tinoco M.D., Castro-Román F., Arauz-Lara J.L. 1996. Phys. Rev E53 : 3745
[105] Rao K.S., Rajagopalan R. 1998. Phys. Rev. E57 : 3227
[106] Larsen A.E., Grier D.G. 1997. Nature 385 : 230
[107] Grier D.G. 2000. J. Phys. Cond. Matt. : in press
[108] Cruz de León G., Saucedo-Solario J. M., Arauz-Lara J.L. 1998 Phys. Rev. Lett. 81 : 1122
[109] Stillinger F. H. 1961. J. Chem. Phys. 35 : 1584
[110] Goulding D., Hansen J.P. 1998. Molec. Phys. 95 : 649
[111] Goulding D., Hansen J.P. 1999. Europhys. Lett. 46 : 407
[112] Denton A.M., Löwen H. 1998. Thin Solid Films 330 : 7
[113] Goulding D., Hansen J. P. 1999 : In New Approaches to Problems in Liquid State Theory, eds. C. Caccamo, J. P. Hansen
and G. Stell. Dordrecht : Kluwer.
[114] Hansen J. P., Goulding D., van Roij R. 2000. In Strongly Coupled Coulomb Systems, eds. C. Deutsch, M. M. Gombert,
Les Ulis : EDP Sciences.
[115] Bowen W.R., Sharif A.O. 1998. Nature 393 : 663
[116] Neu J.C. 1999. Phys. Rev. Lett. 82 : 1072
[117] Sader J.E., Chan D.Y.C. 1999. J. Colloid Interface Sci. 213 : 268
[118] Allahyarov E., I. D’Amico, H. Löwen 1999. Phys. Rev. E 60: 3199
[119] Maeda Y., Hachisu S. 1983. Colloid Surf. 6: 1
[120] Donkai N., Inagaka H., Kajiwara K., Urakawa H., Schmidt M. 1985. Macromol. Chem. 186: 2623
[121] Bugosh J. 1961. J. Phys. Chem. 65: 1791
[122] Buining P.A., Pathmamanoharan C., Jansen J.B.H., Lekkerkerker H.N.W. 1991. J. Am. Ceram. Soc. 74: 1303
[123] Wierenga A.M., Philipse A.P., Reitsma E.M. 1997. Langmuir 13: 6947-6950
[124] Folda T., Hoffmann H., Chanzy H., Smith P. 1988. Nature (London) 333: 55
[125] Nagy M., Keller A. 1989. Polym. Commun. 30: 133
[126] Hendrikx Y., Charvolin J., Raviso M., Liébert M.C., Holmes J. 1983. Phys. Chem. 87: 3991-3999
[127] Wu C.F., Chen S.H., Shih L.B., Lin J.S. 1988. Phys. Rev. Letters 61: 645
[128] Bawden F.C., Pirie N.W., Bernal J.D., Fankuchen I. 1936. Nature 138: 1051
[129] Fraden S., Hurd A.J., Meyer R.B., Cahoon M., Caspar D.L.D. 1985. Journal de Physique, Colloque C3: 85-113
[130] Graf C., Kramer H., Deggelmann M., Hagenbüchle M.,Johner C. et al. 1993. J. Chem. Phys. 98: 4920-4928
[131] Dogic Z., Fraden S. 1997. Phys. Rev. Letters 78: 2417
[132] Booy F.P., Fowler A. 1985. Int. J. Biol. Macromol. 7: 327
[133] Podgornik R., Strey H.H., Parsegian V.A. 1998. Current Opinion in Colloid Interface Science 3: 534-539
[134] Strey H.H., Podgornik R., Rau D.C., Parsegian V.A. 1998. Current Opinion Struc. Biol. 8: 309-313
[135] Tang J.X., Wong S., Tran P., Janmay P., 1996. Ber. Bunsenges. Phys. Chem. 100: 1
[136] Alfrey T., Berg P.W., Morawetz H. 1951. J. Polym. Sci. 7: 543
[137] Barrat J.L., Joanny J.-F. 1996. Advances in Chemical Physics, Volume XCIV: 1-66
[138] Manning G.S. 1996. Physica A 231: 236
[139] Deserno M., Holm C., May S. 1999. Macromolecules : in press
[140] Löwen H. 1994. J. Chem. Phys. 100: 6738-6749 .
[141] Ospeck M., Fraden S. 1998. J. Chem. Phys. 109: 9166
[142] Grønbech-Jensen N., Mashl R.J., Bruinsma R.F., Gelbart W.M. 1997. Phys. Rev. Letters 78: 2477
[143] Levin Y., Arenzon J.J.,Stilck J.F. 1999. Phys. Rev. Letters 83: 2680
[144] Ha B.Y., Liu S.J. 1999. Phys. Rev. Letters 83: 2681
21
[145] Ha B.Y., Liu S.J. 1997. Phys. Rev. Letters 79: 1289
[146] Arenzon J.J., Stilck J.F., Levin Y. 1999. Eur. Phys. J. B 12: 79
[147] Shklovskii B.I. 1999. Phys. Rev. Letters 82: 3268
[148] Ha B.Y., Liu A.J. 1999. Europhys. Letters 46: 624-630
[149] Ha B.Y., Liu A.J. 1999. Phys. Rev. E 60: 803
[150] Solis F.J., De la Cruz M.O. 1999. Phys. Rev. E 60: 4496
[151] Lyubartsev A.P., Tang J.X., Janmey P.A., Nordenskiöld L. 1998. Phys. Rev. Letters 81: 5465
[152] Ray J., Manning G.S. 1994. Langmuir 10: 2450
[153] Schneider J., Hess W., Klein R. 1985. J. Phys. A: Math. Gen. 18: 1221
[154] Graf H., Löwen H. 1999. Phys. Rev. E 59: 1932-1942
[155] Stevens M. 1999. Phys. Rev. Letters 82: 101-104
[156] Podgornik R., Parsegian V.A. 1998. Phys. Rev. Letters 80: 1560
[157] van Olphen H. 1977. Clay Colloid Chemistry. New York : Wiley. 2nd edition
[158] Dubois M., Zemb T., Belloni L., Delville A., Levitz P., Setton R. 1992. J. Chem. Phys. 96 : 2278
[159] Pellenq R.J.M., Caillol J.M., Delville A. 1997. J. Phys. Chem. 101 : 8584
[160] Delville A., Dajmeddine G., Pellenq R.J.M., Caillol J.M., Van Damme H. 1998. Langmuir 14 : 5077
[161] Boek E.S., Coveney P.V., Skipper N.T. 1995. J. Am. Chem. Soc. 117 : 12608
[162] Trizac E., Hansen J.P. 1997. Phys. Rev. E56 : 3137
[163] Leote de Carvalho R.J.F., Trizac E., Hansen J.P. 2000. Phys. Rev. E
[164] Delville A., 1999. J. Phys. Chem. B103 : 8296
[165] Hansen J.P., Trizac E. 1997. Physica A235 : 257
[166] Leote de Carvalho R.J.F., Trizac E., Hansen J.P. 1998. Europhys. Lett. 43 : 369
[167] Mourchid A., Delville A., Lambard J., Lécolier E., Levitz P. 1995. Langmuir 11 : 1942
[168] Pignon F., Magnin A., Piau J.M., Cabane B., Lindner P., Diat O. 1997. Phys. Rev. E56 : 3281
[169] Kroon M., Vos W.L., Wegdam G.H., 1998. Phys. Rev. E57 : 1962
[170] Bonn D., Kellay H., Tanaka H., Wegdam G.H., Meunier J. 1999. Langmuir 15 : 7534
[171] Kutter S., Hansen J.P., Sprik M., Boek E. 2000. J. Chem. Phys. (in press)
[172] Rowan D., Hansen J.P., Trizac E. 2000. to be published
[173] Dijkstra M., Hansen J.P., Madden P.A. 1997. Phys. Rev. E55 : 3044
[174] Pincus P.A., Safran S.A. 1999. Europhys. Lett. 42 : 103
[175] Levesque D., Weis J.J., Patey G.N. 1980. J. Chem. Phys. 72 : 1887
[176] Carnie S. L., Chan Y.C., Walker G.R. 1981. Molec. Phys. 43 : 1115
[177] Kusalik P.G., Patey G.N. 1988. J. Chem. Phys. 88 : 7715
[178] Blum L., Vericat F., Degreve L. 1999. Physica A256 : 396
[179] Berendsen H.J.C., Grigera J.R., Straatsma T.P. 1987. J. Phys. Chem. 91 : 6269
[180] Kralj-Iglič V., Iglič A. 1996. J. Phys. II (France) 6 : 477
[181] Borukhov J., Andelman D., Orland H. 1997. Phys. Rev. Lett. 79 : 435
[182] Tang Z., Scriven L.E., Davis H.T. 1994. J. Chem. Phys. 100 : 4527
[183] Otto F., Patey G.N. 1999. Phys. Rev. E60 : 4416
[184] Frink L.J.D., van Swol E. 1996. J. Chem. Phys. 105 : 2884
[185] Carnie S.L., Chan D.Y.C. 1980. J. Chem. Phys. 73 : 2949
[186] Blum L., Henderson D. 1981. J. Chem. Phys. 74 : 1902
[187] Diaz-Herrera E., Forstmann F. 1995. J. Chem. Phys. 102 : 9005
[188] Rosenfeld Y. 1993. J. Chem. Phys. 98 : 8126
[189] Rosenfeld Y., Schmidt M., Löwen H., Tarazona P. 1997. Phys. Rev. E55 : 4245
[190] Biben T., Hansen J.P., Rosenfeld Y. 1998. Phys. Rev E57 : R3727
[191] Augousti A.T., Rickayzen G. 1984. J. Chem. Soc. Faraday Trans. 2 80 : 141
[192] Kinoshita M., Iba S., Harada M. 1996 J. Chem. Phys. 105 : 2487
[193] Tandon S., Kesavamoorthy R., Asher S.A. 1998. J. Chem. Phys. 109: 6490
[194] Netz R.R., Orland H. 1999. Europhys. Letters 45: 726-732
22
σ
ε ε ε ε
-h/2 +h/2
σ σ
, ,
ε ε ε ε
-h/2 +h/2
FIG. 1. a,b
23
2.0
F
G F
1.5
ns[µM]
1.0
L
0.5
FCC
0.0
0.00 0.05 0.10 0.15
η
FIG. 2.
24
F
G G+L
F
+
L
10
FCC
ns[µM]
G+FCC
FCC
0
0.0 0.1 0.2
η
FIG. 3.
25
r
40
30
2a
20
F/F0
10
-10
-20
2.0 2.2 2.4 2.6 2.8 3.0 3.2
r/a
FIG. 4.
26
Figure captions
Figures 1a, 1b: Two different plate geometries.
Figure 2: Phase diagram for a charged suspensions in the η − ns plane with η = 4πna3 /3 and Z = 350, a = 380nm ,
ǫ = 13.9 (penthanol) and monovalent counterions at room temperature. Possible phases are fluid (F), gas (G), liquid
(L), and face-centered-cubic (FCC) crystals. Tie-lines denote coexistence conditions.
Figure 3: Same as Figure 2, but now for Z = 1000, a = 350nm, ǫ = 25.3 (ethanol). There is a triple point of
coexisting gas, liquid and crystal.
2 2
Figure 4: Effective force F in units of F0 = Za2e × 10−4 between two polyions versus reduced distance r/a. The
squares are for aqueous suspensions. The force is repulsive in good agreement with DLVO-theory (dashed line). The
circles are for a solvent with a strongly reduced dielectric constant ǫ and show attraction. The parameters are given
in detail in Ref. [94]. The inset shows a counterion configuration around two fixed polyions.
27

Effective Interactions Between Electric Double-Layers: Jean-Pierre Hansen

Uploaded by

Copyright:

Available Formats

Effective Interactions Between Electric Double-Layers: Jean-Pierre Hansen

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Effective Interactions Between Electric Double-Layers: Jean-Pierre Hansen

Uploaded by

Copyright:

Available Formats

Effective interactions between electric double-layers

D-40225 Düsseldorf, Germany; e-mail : [email protected]

II. MULTI-COMPONENT VERSUS EFFECTIVE ONE-COMPONENT DESCRIPTION

Important simplifications occur within the PM in a number of limiting situations.

A. DFT and effective interactions

δΩ ρ∗+ (r), ρ∗− (r)

Fex [ρ+ , ρ− ] = FCoul + Fcorr (7)

ρα (r) = ζα exp − Φdir

P = kB T [ρ+ (R) + ρ− (R)] (11)

IV. PLANAR GEOMETRY

P = Pkin + Pel + Pcoll (17)

where E(z) = −dψ(z)/dz is the

B. The effect of microion correlations

v(r) = −kB T log[g(r)] (23)

A. DLVO theory revisited

VI. POLYIONS IN CONFINED GEOMETRIES

VII. ROD-LIKE POLYIONS

VIII. LAMELLAR POLYIONS

IX. DISCRETE SOLVENT EFFECTS

You might also like