Perspective: Colloidal Diffusion in Confined Geometries

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

PCCP

View Article Online


PERSPECTIVE View Journal | View Issue
Published on 19 May 2017. Downloaded by Nanyang Technological University on 9/15/2019 12:23:06 PM.

Colloidal diffusion in confined geometries


Cite this: Phys. Chem. Chem. Phys.,
Kim Nygård
2017, 19, 23632
Colloidal dispersions in confined geometries exhibit rich diffusive dynamics governed by an interplay
between particle–particle and particle–wall interactions. This perspective reviews recent selected
computational and experimental studies on diffusion of dense liquid-like dispersions in spatial
Received 18th April 2017, confinement, with an emphasis on general physical concepts rather than system-specific details. The
Accepted 19th May 2017
common thread is to analyse colloidal diffusion in confined geometries in terms of the local density
DOI: 10.1039/c7cp02497e experienced by the colloidal particles, viz. at the level of anisotropic pair densities, which have recently
become experimentally accessible.
rsc.li/pccp

1 Introduction to bulk;3 diffusivities, relaxation times, and viscosities may change


by orders of magnitude once a molecular liquid is confined on
It is well documented that nanoscopic confinement induces the nanometre scale. The mechanistic understanding of such
ordering of molecular liquids, which in turn strongly alters their scientifically intriguing confinement phenomena, and their use in
properties and even induces emergent phenomena.1 An early technological applications,4 has been an active area of research
textbook example is the phenomenon of oscillatory solvation for several decades.
forces between suspended surfaces;2 the degree of molecular Spatially confined colloidal dispersions exhibit many simi-
order between the suspended surfaces, and hence the pressure larities with nanoscopically confined molecular liquids, such
in the gap, depends on the surface separation, leading to an as confinement-induced microscopic ordering and ensuing
oscillatory solvation force between the surfaces as a function of oscillatory solvation forces.5 In terms of dynamic properties,
their separation. Another prominent feature of nanoscopically however, there are prominent differences between molecular
confined liquids is their markedly different dynamics compared liquids and colloidal dispersions. Notably, colloidal dynamics
is affected not only by direct particle interactions, but also by
Department of Chemistry and Molecular Biology, University of Gothenburg,
indirect, hydrodynamic interactions mediated by the solvent.
SE-41296 Gothenburg, Sweden. E-mail: kim.lj.nygard@gmail.com The resulting rich colloidal dynamics has been extensively
studied in bulk, as is evident from the comprehensive review
of ref. 6. Confining walls further complicate matters, and at the
Kim Nygård earned his PhD in moment colloidal diffusion in confinement is an active area of
experimental materials physics research.
in 2007 from the University of Colloidal dynamics in confined geometries is a broad topic.
Helsinki, Finland. After post- I have therefore decided to concentrate on a few selected
doctoral positions at the Paul scientific questions, focusing on general physical concepts
Scherrer Institut, Switzerland, rather than system-specific details. First, I only discuss diffusive
and the European Synchrotron dynamics of the colloidal particles in the absence of macroscopic
Radiation Facility, France, he density gradients, disregarding the much faster solvent dynamics.
took in 2011 up a position as Second, and most important, I concentrate on diffusion of dense
assistant professor at the colloidal dispersions confined between walls at a separation
University of Gothenburg, Sweden. of a few particle diameters, where the correlation between
In the summer of 2017 he will join confinement-induced microscopic ordering and colloidal diffusion
Kim Nygård the MAX IV Laboratory, University is prominent. I will not discuss closely related phenomena such as
of Lund, Sweden. During the last single-file diffusion7,8 the crossover from three- to two-dimensional
years his main research interest has been the microscopic structure, dynamics,9 or confinement-induced particle trapping.10 Finally, for
dynamics, and ensuing properties of soft matter at interfaces and in simplicity I only discuss rigid spherical particles dispersed in
spatial confinement. liquids, which are confined between planar solid walls, thereby

23632 | Phys. Chem. Chem. Phys., 2017, 19, 23632--23641 This journal is © the Owner Societies 2017
View Article Online

Perspective PCCP

neglecting e.g. diffusion of colloidal dispersions in mesoporous


matrices.11 Nevertheless, the concepts brought up may have a
broader range of application.
I am not the first one to review the dynamics of dense fluids
in confined geometries. For another recent example, I refer to
ref. 12. Here I therefore take a different angle. The temporally
averaged short-range neighbourhood of a colloidal particle is
Published on 19 May 2017. Downloaded by Nanyang Technological University on 9/15/2019 12:23:06 PM.

described in terms of the colloid’s pair density, or solvation


shell in the language of physical chemistry and chemical
physics. In stark contrast to bulk, dense confined fluids exhibit
anisotropic pair densities due to additional packing constraints Fig. 2 Schematic of a dense colloidal dispersion confined between planar
imposed by the confining walls.13–17 The common thread in solid walls at close separation of a few times the particle diameter. The
confinement-induced microscopic ordering of the particles is, in the first
this perspective will thus be to discuss various aspects of
approach, quantified by their density profile n(z) across the slit, as shown to
colloidal diffusion in confined geometries at this fundamental the right. The yellow sphere depicts a tagged or central particle, depending
level of anisotropic pair densities. on whether I discuss self- or wave-vector-dependent diffusion,
The rest of this perspective is organised as follows. In respectively.
Section 2 I briefly define the microscopic structure and dynamics
of confined colloidal dispersions, at the level of anisotropic pair
diameter, as schematically presented in Fig. 2. Here the colloidal
densities. In Sections 3 and 4 I discuss self- and wave-vector-
particles are depicted as spheres, while the solvent is only
dependent diffusion of colloidal dispersions in confined geometries,
implicitly assumed. The confining walls impose packing con-
respectively, emphasising their analysis at the level of pair densities.
straints on the particles, leading to their confinement-induced
Finally, in Section 5 I summarise the take-home messages of this
microscopic ordering;18 in the schematic figure the particles
perspective and provide a brief outlook.
order into three layers parallel to the confining walls. The
microscopic ordering is usually characterised in terms of the
2 Microscopic structure and dynamics particles’ ensemble-averaged number density profile n(z) across
the confining slit, as shown to the right. An obvious question
Let me begin by a brief comment about the confinement then emerges: how does the microscopic ordering modify the
geometry. For conceptual clarity I will only consider the simplest fluid’s diffusive properties?
conceivable case – two parallel planar and smooth solid walls at a It is important to realise, however, that while the density
separation H, as shown schematically in Fig. 1. In this geometry profile n(z) describes the distribution of particles across the
the colloidal dispersion exhibits planar symmetry, and to simplify confining slit, it does not describe the local distribution of
the notation I will keep the central particle in the in-plane origin particles around a given particle. The ensemble-averaged local
when discussing pair properties. I denote the directions parallel density around a central (or tagged) particle, say the yellow
and perpendicular to the confining walls by R and z, respectively, particle in Fig. 2, is instead given by the pair density n(r1)g(r,r1),
and the corresponding wave vector components by q|| and q>. where g is the pair distribution function, r the central particle’s
position, and r1 the position where one measures the local
2.1 Static pair density
density.† Since ng describes the local distribution of particles
Consider next a colloidal dispersion confined between solid experienced by the central particle, and hence provides means
planar walls at a short separation of a few times the particle to describe the central particle’s pair interactions, it is a
physically appealing starting point for analysing colloidal
diffusivity in confined geometries.
I emphasise that the pair density ng is a fundamental quantity
for understanding the properties of dense fluids, providing a
formal relationship between the fluid’s microscopic interactions
and its macroscopic properties. Moreover, it is experimentally
accessible in terms of the structure factor by scattering of
neutrons, X-rays, or light, allowing quantitative comparison
between experiment, theory, and simulations at an advanced
level. The pair density is thus the most important quantity for
characterising bulk fluids.19 However, due to the complexity of
Fig. 1 Schematic of the confinement geometry, with two parallel planar ng in confined geometries compared to bulk, as shown below,
and smooth solid walls separated by a distance H. The particles’ centres are
thus confined to the region H–s between the walls, where s denotes the
particles’ diameter. The z axis is oriented normal to the confining walls, and the † In the present geometry n(r1)g(r,r1) simplifies to n(z1)g(z,z1,R12), where z and z1
parallel direction is denoted by R. Wave vector components parallel and denote the perpendicular components of r and r1, respectively, and R12 the
perpendicular to the confining walls are given by q|| and q>, respectively. parallel component of |r  r1|.

This journal is © the Owner Societies 2017 Phys. Chem. Chem. Phys., 2017, 19, 23632--23641 | 23633
View Article Online

PCCP Perspective

confined fluids are only rarely analysed at the level of pair


densities.13,14,17,20,21 Nevertheless, computational studies on
confined fluids at this fundamental level have become more
common in recent years, including ion–ion correlation effects
in confined electrolytes,22 depletion interactions,23 tuneable
assembly of charged confined fluids,24 the glass transitions in
confined geometries,25 ordering mechanisms in confined
Published on 19 May 2017. Downloaded by Nanyang Technological University on 9/15/2019 12:23:06 PM.

fluids,16 and fluid inclusions in porous matrices.26 From an


experimental point of view, in turn, we have very recently
developed unique methodology for studying pair densities of
confined fluids in terms of structure factors, based on X-ray
scattering from colloid-filled micro- or nanofluidic channels.27–29
For a brief recent review on pair densities in confined fluids,
see ref. 30.
In Fig. 3 I exemplify the theoretical pair density ng for a
hard-sphere fluid confined between hard planar walls.15 Here I
show a relatively dense fluid in a narrow slit, in order to
highlight the effects of confinement, but similar although less
prominent features are also observed at lower particle concen-
trations and larger separations between the confining walls.
The data exhibit two particularly important features. First, the
anisotropic packing in confined fluids shows up as strongly
anisotropic pair densities ng. Second, the confined fluid’s pair
density depends strongly on the central particle’s position in
the slit. Both of these features are in stark contrast to the pair
densities of isotropic and translationally invariant bulk fluids.
In terms of the topic of the present perspective, the strongly
anisotropic and position-dependent pair density ng implies
likewise strongly anisotropic and position-dependent diffusivity,
as will be discussed below.

2.2 Intermediate scattering function


The pair density ng discussed above describes a temporally
averaged microscopic structure, and is thus a static quantity.
Its ‘‘dynamic’’ counterpart is given by the so-called Van Hove
function,31 which describes the ensemble-averaged pair corre-
lations in both space and time. For my purposes it is conve-
nient to instead define the spatiotemporal pair correlations in
reciprocal space in terms of the intermediate scattering func-
tion F(q,t), where q is the wave vector characterising density
fluctuations and t the time. Formally the intermediate scattering
function is given by6,32

F(q,t) = N1hdnq(0)dnq(t)i, (1)

where N denotes the number of particles in the system, dnq(t)


a Fourier component of number density fluctuations of wave
vector q, and h  i an ensemble average. In essence, F(q,t) is a
temporally resolved structure factor. Fig. 3 Anisotropic pair density n(z1)g(z,z1,R12) around a central particle,
A direct connection between the intermediate scattering calculated for a dense hard-sphere fluid confined between hard planar and
parallel walls at a short separation. The pair density is presented as a function of
function and the static pair densities of Section 2 is obtained
position parallel (R12) and normal (z1) to the confining walls, and the grey region
in the limit t - 0, where we obtain the structure factor, F(q,0) = depicts the excluded volume around the central particle. The confining wall
N1hdnq(0)dnq(0)i  S(q). For confined fluids the structure separation is H = 3.05s and the confined fluid is in contact with a bulk reservoir
factor is given in terms of static density distributions as of number density nb = 0.75s3 (bulk volume fraction fb E 0.39). The central
ðð particle is positioned either (a) in contact with one of the confining walls, (b) in a
1 density minimum, or (c) in the slit centre. Reprinted from K. Nygård et al.,
SðqÞ ¼ 1 þ nðrÞnðr1 Þhðr; r1 Þexp½iq  ðr  r1 Þdrdr1 ; (2) J. Chem. Phys., 2013, 139, 164701, with the permission of AIP Publishing.
N

23634 | Phys. Chem. Chem. Phys., 2017, 19, 23632--23641 This journal is © the Owner Societies 2017
View Article Online

Perspective PCCP

light scattering (SLS) using optical light6 or small-angle X-ray


scattering (SAXS) using X-rays.37,38 In fact, we have recently
demonstrated the feasibility of SAXS and XPCS from colloid-
filled micro- and nanofluidic containers, thus allowing us to
probe experimentally the microscopic structure27–29 and diffu-
sive dynamics39 of confined colloidal dispersions at the level of
anisotropic pair densities.
Published on 19 May 2017. Downloaded by Nanyang Technological University on 9/15/2019 12:23:06 PM.

The intermediate scattering function is a complicated quantity,


which contains a wealth of information on colloidal structure and
dynamics. Before I proceed with selected examples of colloidal
diffusion in confined geometries, I will therefore briefly mention a
few limiting cases of F(q,t):
 In the general case, F(q,t) is sensitive to collective
dynamics at length scales comparable to the reciprocal of the
wave vector. In such wave-vector-dependent diffusion, the local
density around a central particle, say the yellow sphere in Fig. 2,
relaxes due to the motion of all particles in the confined
system.
 For large wave vectors the contributions to F(q,t) from
distinct particles cancel out, and the data are dominated by
single-particle dynamics. In this limit one probes self-diffusion;
one focuses on a tagged particle, for example the yellow sphere
in Fig. 2, and follows its diffusion in the presence of the other
particles in the system and the confining walls. In practice this
limit is reached when q c qm, where qm is the position of the
primary peak in the structure factor S(q), such that S(q) has
essentially decayed to unity. In the case of Fig. 4(a), this limit
Fig. 4 (a) Anisotropic structure factor S(q) for the confined hard-sphere would be reached for q c 2ps1.
fluid of Fig. 3, shown as a function of wave vector components parallel (q8)  In the limit of short wave vectors, q { qm, F(q,t) probes
and normal (q>) to the confining walls. (b) Corresponding isotropic bulk collective diffusion at length scales much larger than the
structure factor as a function of wave vector magnitude q.
colloidal particles’ size (or any other characteristic length of
the system).
where h(r,r1) = g(r,r1)  1 denotes the total pair correlation In the following I will consider these different cases in more
function and the integrals are carried out over the space detail.
between the confining walls.‡ For convenience I exemplify in
Fig. 4 the structure factor of a dense fluid both in bulk and in
narrow confinement; whereas the isotropic bulk fluid exhibits 3 Self-diffusion
an isotropic structure factor, the strongly anisotropic pair
densities of the confined fluid, as seen in Fig. 3, shows up as Colloidal self-diffusion typically falls into three time regimes.6,32
a strongly anisotropic structure factor.15 Similar as for the In the short-time limit, t { s2/4D0 with D0 denoting the Stokes–
confined fluid’s anisotropic pair density ng discussed above, Einstein diffusion coefficient, the tagged particle diffuses only a
the anisotropic structure factor S(q) implies anisotropic small fraction of its size and the surrounding solvation shell has
diffusivity. essentially not changed. In this regime the central particle’s mean
An important feature of the intermediate scattering function square displacement (MSD) is linear in time. In the long-time
F(q,t) and the structure factor S(q) of eqn (1) and (2), respectively, limit, t c s2/4D0, the tagged particle has had time to collide many
is that they are directly accessible in scattering experiments. The times with other particles, thus exhibiting long-distance ‘‘random
former can be accessed, in terms of the normalised intermediate motion’’ and again a linear MSD. Finally, at intermediate times,
scattering function f (q,t)  F(q,t)/S(q), via temporally resolved t E s2/4D0, the tagged particle exhibits retarded diffusion due to
scattering experiments such as dynamic light scattering (DLS) in interactions with the distorted solvation shell, showing up as a
the optical regime33,34 or X-ray photon correlation spectroscopy sublinear MSD. In the next subsections I will focus on the long-
(XPCS) in the X-ray regime.35,36 The latter, in turn, can be probed time limit.
via temporally averaged scattering experiments, such as static The diffusive dynamics of colloidal dispersions at high
densities is governed by a complex interplay between direct
‡ For isotropic and translationally invariant bulk fluids eqn (2) simply yields the
particle interactions and indirect ones mediated by the solvent.
isotropic S(q) = 1 + nbh(q), where q = |q| and h(q) is the Fourier transform of the The latter, hydrodynamic interactions are both long-ranged and
isotropic total pair correlation function.19 nonadditive, and hence their effect on colloidal diffusion is in

This journal is © the Owner Societies 2017 Phys. Chem. Chem. Phys., 2017, 19, 23632--23641 | 23635
View Article Online

PCCP Perspective

general rather complicated.6,32 However, experiments imply


that hydrodynamic interactions may be negligible when con-
sidering self-diffusion in the long-time limit.40 Although the
following discussion on self-diffusion in confinement is largely
based on simulations containing direct particle interactions
only, the results should also be useful when predicting colloidal
diffusion in confined geometries.
Published on 19 May 2017. Downloaded by Nanyang Technological University on 9/15/2019 12:23:06 PM.

3.1 Slit-width-dependent self-diffusion


Let me first consider a colloidal dispersion confined between
planar solid walls, similar as depicted in Fig. 2, with a wall
separation H c s large compared to the characteristic length
scales of the colloid. In the vicinity of the solid walls the colloid
exhibits microscopic structure and diffusive dynamics different
from bulk, while in the slit centre it shows bulk diffusion. With
decreasing wall separation H the relative amounts of bulk and
Fig. 5 Dimensionless self-diffusion coefficient D8 parallel to the confining
interfacial colloid initially vary in a monotonic manner, and
walls of a dense hard-sphere fluid between planar hard walls, as obtained
accordingly the ensemble-averaged self-diffusion coefficient D from simulations. The blue circles depict D8 as a function of wall separation
also varies monotonically. Colloidal diffusion in this regime of H, when the average volume fraction of particles in the slit has been fixed at
large wall separation, sometimes coined weak confinement, is f = 0.40. The green dashed line shows the bulk self-diffusion coefficient at
thus governed by the single solid-fluid interfaces. For a brief the same volume fraction. The red squares represent the self-diffusion
coefficient in bulk, determined at thermodynamic states where the excess
discussion of diffusion in weak confinement, see ref. 12.
entropy sex equals the slit-width-dependent sex(H) of the confined fluid.
For sufficiently narrow confinement, also known as strong Data taken from ref. 41 and 43.
confinement, the wall-induced microscopic ordering of the
colloidal particles protrude throughout the confining slit.
In this regime the degree of ordering is governed by the scale with the pair excess entropy sex
(2), a hypothesis which has been
competition between the mutually incompatible preferred found to hold for both atomic and molecular liquids.53–57 The
packing of particles among themselves and the additional excess entropy scaling also seems to hold for colloidal dispersions
packing constraints imposed by both confining walls.16 During (at least over a limited range of particle concentrations), although
the last decade, extensive simulations on simple model fluids the exact interpretation in this case is still under debate.40,58
have shown that this competition also results in nonmonotonic During the last few years this kind of excess-entropy scaling has
confinement effects on colloidal diffusion.41–43 been extensively studied in terms of simple liquids’ so-called
In Fig. 5 I exemplify the nonmonotonic self-diffusivity (blue quasiuniversality;59,60 many simple liquids have been found to
circles) for a dense hard-sphere fluid as function of confining exhibit (nearly) invariant structure and dynamics along so-called
wall separation, obtained from simulations of ref. 41 and 43. configurational adiabats, i.e., lines in the thermodynamic phase
While these data were collected parallel to the confining walls, diagram where the excess entropy sex is constant.
qualitatively similar behaviour has been observed in the normal A series of simulation studies by Truskett et al. have shown
direction.42 Intriguingly, the self-diffusion coefficient D8(H) that the excess entropy sex indeed is a good predictor for
oscillates with increasing wall separation H – a first evidence dynamics in confined fluids.41,43,61–63 The close connection
of nonmonotonic confinement effects on colloidal diffusion. I between self-diffusivity and excess entropy is also evident in
emphasise that the volume fraction of particles in the slit was Fig. 5, where the red squares depict the bulk self-diffusion
kept constant in the simulations, i.e., the confined fluid was coefficient determined by fixing the excess entropy at each
not kept in equilibrium with a bulk reservoir, so the oscillatory point to the slit-width-dependent sex(H) of the confined fluid.
behaviour of D8(H) cannot be explained by simple density The semi-quantitative agreement with the oscillatory diffusive
arguments. I further note that the slit-width dependence of data of the confined fluid is striking. This observation implies
D8(H) closely follows that of crystallisation44–46 and vitrification,12,47 that local particle packing, as quantified by excess entropy sex
implying that local particle packing is governing the diffusive (or alternatively available volume43), also governs long-time
dynamics. self-diffusion of colloidal dispersions in confined geometries.
How can one rationalise this observation? Rosenfeld was the The slit-width-dependent long-time self-diffusivity has also
first to note that in simple bulk liquids the diffusivity scales been studied experimentally by confocal-microscopy-based particle
with the excess entropy sex = s  sid,48,49 i.e., the difference tracking of colloids between glass plates.64 Unfortunately, the wall
between the entropy of the system s and the ideal gas entropy separations H in this study were relatively large, hindering the
sid at the same thermodynamic conditions. As a further simplifi- observation of the predicted oscillatory diffusivity. Future experi-
cation, the excess entropy is reasonably well approximated by its mental verification of the intriguing oscillatory diffusivity and the
two-body or pair contribution,50–52 which is given in terms of the proposed entropy scaling would be an important advance in our
pair distributions. Thus one expects the bulk self-diffusivity D to understanding of colloidal diffusion in narrow confinement.

23636 | Phys. Chem. Chem. Phys., 2017, 19, 23632--23641 This journal is © the Owner Societies 2017
View Article Online

Perspective PCCP

3.2 Position-dependent self-diffusion


So far I have discussed oscillatory self-diffusion as a function
of confining wall separation. However, since the local density
around a central (or tagged) particle depends on the latter’s
position in the slit, as seen in Fig. 3, one also expects position-
dependent diffusivity. This is indeed seen in simulations,
exemplified in Fig. 6 (red squares) as the position-dependent
Published on 19 May 2017. Downloaded by Nanyang Technological University on 9/15/2019 12:23:06 PM.

self-diffusivity D>(z) for a confined hard-sphere fluid normal


to the planar confining hard walls.42 The data exhibit two
prominent features. First, D>(z) is reduced near the impene-
trable confining walls, where the normal diffusion is limited to
one direction. Second, and more interesting, D>(z) oscillates
across the confining slit, and these oscillations in D>(z) nearly
coincide with the oscillations in the confined fluid’s density
profile n(z) (blue circles). Clearly, the data evidence a strong
correlation between confinement-induced microscopic ordering
and position-dependent self-diffusivity.
In bulk hard-sphere dispersions self-diffusivity decreases
with increasing particle concentration. Naively one would
assume that the same holds in confined geometries. How can
one then understand the counterintuitive positive correlation
between D>(z) and n(z) of Fig. 6, where a large local density
corresponds to a large local diffusivity? Following the reasoning Fig. 7 Theoretical pair contribution to the excess entropy, determined for
of Mittal et al.,42 the local particle insertion probability,13 and the hard-sphere fluid of Fig. 3. (a) Position-dependent pair excess entropy
hence the local available volume,65 is proportional to the s(2)
ex (z) (red line) for a planar confining slit of width H = 5.0s, as determined

density profile n(z). Diffusing particles probe their neighbourhood using eqn (3). The fluid’s density profile n(z) across the confining slit (blue
for available space and hence diffusivity should be roughly line) is shown for comparison. (b) Pair excess entropy s(2) ex (red circles) as
a function of confining slit width H, obtained as the weighted average of
proportional to the local available space, or to the density profile s(2)
ex (z). The average number density nH in the slit (blue line) is shown for
as seen in Fig. 6. comparison.
A close inspection of Fig. 6 reveals a phase shift between
extrema in D>(z) and n(z), most notable around z = 1s. In line
with the discussion of Section 3.1, a local excess entropy may be density profile. One possibility is the position-dependent pair
a better predictor of position-dependent diffusivity than the excess entropy,66
ðð
sex
ð2Þ ðzÞ ¼ pkB Rnðz1 Þðg ln g  hÞdz1 dR; (3)

where kB is Boltzmann’s constant, the integration is again


carried out over all space between the confining walls, and I
have used the shorthand notation g = g(z,z1,R) (and similarly for
h = g  1). The pair excess entropy discussed above is retained
Ð .Ð
as a weighted average, sex
ð2Þ ¼ nðzÞs ex
ð2Þ ðzÞdz nðzÞdz. In Fig. 7(a)
I plot the theoretical position-dependent pair excess entropy
s(2)
ex (z) calculated for the confined hard-sphere fluid of Fig. 3,
together with the corresponding density profile n(z). Indeed,
ex (z) around z E 1s are shifted towards the
the extrema in s(2)
slit centre compared to the extrema in n(z), in line with the
phase shift of D>(z) in Fig. 6. In the spirit of this perspective, it
would be interesting to quantitatively compare D>(z) and s(2) ex (z),
thereby characterising how accurately the confined fluid’s local
Fig. 6 Microscopic structure and diffusion of a dense hard-sphere
diffusivity is predicted by particle packing at the level of pair
fluid between planar hard walls, as determined by simulations. The blue densities.§
circles depict the density profile n(z) across the confining slit, while the red
squares denote the dimensionless local self-diffusivity D>(z) normal to the § To the best of my knowledge, the connection between local pair excess entropy
confining walls. Data are shown for the wall separation H = 5.0s and the of eqn (3) and local diffusivity of confined fluids has prior only been briefly
average volume fraction f = 0.40 in the slit. Data taken from ref. 42. discussed in ref. 66.

This journal is © the Owner Societies 2017 Phys. Chem. Chem. Phys., 2017, 19, 23632--23641 | 23637
View Article Online

PCCP Perspective

Recent simulation studies have shown a rich behaviour of describing the effect of indirect interactions mediated by the
the position-dependent diffusivity.67 First, the self-diffusivity solvent. Note that while eqn (4) has been derived in the short-
D8(z) parallel to the confining walls does not show any dependence time limit, for bulk colloids the long-time diffusion has been
of the particle’s position in the slit, in stark contrast to the normal observed to exhibit almost the same wave-vector dependency
diffusivity D>(z). This observation has also been corroborated (apart from a scaling factor).69,70
experimentally by video microscopy on colloidal dispersions The functional form of eqn (4) is physically appealing.
confined between glass plates.64,68 Second, upon increasing the Density components corresponding to a maximum in S(q) are
Published on 19 May 2017. Downloaded by Nanyang Technological University on 9/15/2019 12:23:06 PM.

particles’ volume fraction to high densities, the positive correlation favourable in terms of the interaction free energy and therefore
between the normal diffusivity D>(z) and the number density decay slowly. This inverse relationship between D(q) and S(q)
profile n(z) evolves into a negative one, with local maxima (minima) is sometimes referred to as ‘de Gennes narrowing’,71 in spirit
in D>(z) approximately coinciding with local minima (maxima) of early work by de Gennes on the dynamics of atomic and
in n(z). To date no rationale for either the position-dependent molecular systems,72 although the underlying physics of
diffusivity or its concentration dependence has been proposed. colloidal diffusion and atomic/molecular dynamics differ.73
Let me close this subsection by returning to the slit-width- It should be noted, however, that the simple form of eqn (4) is
dependent excess entropy sex of the confined hard-sphere fluid. In conceiving, with the long-ranged and non-additive hydrodynamic
Fig. 7(b) I present the theoretical pair contribution s(2) ex versus slit interactions rendering H(q) a rather complicated function.
width H for the system of panel (a). For comparison, I also show In particular, while H(q) can be theoretically calculated with a
the average particle number density between the confining walls, reasonable accuracy in bulk,74–76 the presence of walls have so
Ð
nH ¼ H 1 nðzÞdz.15 In contrast to Fig. 5, where nH (and likewise far hindered its theoretical analysis in confinement.
f) was kept constant, these data were obtained by keeping the Note that the short-time wave-vector-dependent diffusion
confined fluid in equilibrium with a bulk reservoir, resulting in a coefficient D(q) and the Stokes–Einstein coefficient D0 can be
slit-width-dependent nH. The data exhibit two prominent effects. experimentally determined in photon correlation experiments
First, the general decrease in nH with decreasing slit width H is (DLS33,34 or XPCS35,36), whereas the structure factor S(q) can be
accompanied by an increase in s(2) ex . Second, and more interesting, determined in the temporally averaged limit (SLS6 or SAXS37,38).
we observe a phase shift between s(2) (2)
ex and nH versus H; sex exhibits By combining these results one thus obtains access to the
maxima (minima) for H slightly below integer (half-integer) multi- hydrodynamic function H(q) via eqn (4). For bulk colloids this
ples of the particle diameter, while for nH these occur at larger H is a standard approach,6 and it has also been applied to study
values. Clearly, in confinement the pair excess entropy is not the diffusion of dense colloidal dispersions near single solid
directly proportional to the number of particles in the slit, as walls by evanescent-wave DLS.77,78 For spatially confined dense
would be expected in analogy with bulk, but depends instead more colloids, in turn, previous work has mainly concentrated on the
subtly on particle packing at the level of pair densities via eqn (3). limiting cases of one- and two-dimensional fluids,7,8,79–81
rather than the strong confinement which is the focus here.
In the latter case a promising approach is to confine colloidal
4 Wave-vector-dependent diffusion dispersions in micro- or nanofluidic channel arrays, and use
In the preceding section I discussed self-diffusion in the long- X-rays for characterisation.82–85 We have recently extended this
time limit, with an emphasis on simulation studies. In the approach for simultaneous experimental determination of the
remainder of this perspective I will briefly consider the extension static structure factor27–30 and the wave-vector-dependent diffusion
to wave-vector-dependent diffusion, which has recently become coefficient39 of dense colloidal dispersions in confinement. In the
experimentally accessible.39 I foresee that this approach in the following I will exemplify these novel results.
coming years also will provide means to experimentally corro- In Fig. 8(a) I present the structure factor S(q), collected from
borate the simulation-based predictions discussed above. a confined colloid in three directions with respect to the
Wave-vector-dependent diffusion is in general a complicated confining walls.39 The structure factor exhibits strong aniso-
phenomenon, which I will not pursue here. Let me instead for a tropy similar to the data of Fig. 4(a); the primary peak in S(q) is
moment focus on short times, where the analysis is simplified. more prominent and shifted towards larger q values in the
In this limit the wave-vector-dependent diffusion coefficient intermediate direction (161 with respect to the confining walls)
D(q), which describes spatiotemporal density fluctuations of compared to the parallel (01) and diagonal directions (451).
wave vector q, retains the formally simple form6,32 In panel (b) I present for comparison the inverse of the wave-
vector-dependent diffusion coefficient, D1(q). Most importantly,
HðqÞ the qualitative agreement between the data of panels (a) and (b)
DðqÞ ¼ D0 : (4)
SðqÞ corroborate the inverse relationship between D(q) and S(q) of
Here D0 denotes the Stokes–Einstein diffusion coefficient of non- eqn (4), demonstrating the important contribution of direct
interacting particles, S(q) the structure factor describing the effect of particle interactions on wave-vector-dependent diffusion.8
direct particle interactions, and H(q) the hydrodynamic function¶
8 The data of Fig. 8(b) were not collected strictly in the short-time limit.39
¶ For clarity I use different fonts to make a distinction between the hydrodynamic However, very recent analysis in the short-time limit does yield a qualitatively
function H(q) and the slit width H. similar inverse relationship between D(q) and S(q).86

23638 | Phys. Chem. Chem. Phys., 2017, 19, 23632--23641 This journal is © the Owner Societies 2017
View Article Online

Perspective PCCP

5 Summary and outlook


In this perspective I have reviewed recent research on colloidal
diffusion in narrow confinement. The common thread has
been to analyse diffusion of confined colloids in terms of the
local density experienced by a tagged or central particle, namely
the anisotropic pair density. This approach highlights an
important take-home message of the present perspective –
Published on 19 May 2017. Downloaded by Nanyang Technological University on 9/15/2019 12:23:06 PM.

colloidal diffusion in confinement can be analysed using the


same concepts as in bulk, provided one makes the effort to
describe the anisotropic pair densities.
Colloidal diffusion in narrow confinement has been studied
extensively during the last decade by means of simulations.
This research has brought into evidence a wealth of interesting
diffusion phenomena, such as slit-width-dependent and position-
dependent long-time self-diffusion reviewed here. A particularly
important finding is that these diffusive properties of confined
fluids correlate with the (pair) excess entropy, implying that the
dynamics can be predicted based on particle packing. In order to
obtain mechanistic insights beyond such correlations between
self-diffusion coefficients and excess entropies (or their position-
dependent counterparts), however, a first-principles microscopic
theory of confined colloidal dynamics would be highly useful.
Fig. 8 Anisotropic wave-vector-dependent microscopic structure and A promising such candidate is the mode coupling theory,92 which
diffusive dynamics of a confined colloidal dispersion, as determined in has recently been extended to confined geometries.47,93,94 Since
X-ray scattering experiments on colloid-filled microfluidic containers. The this theory also yields the confined fluid’s intermediate scattering
data were collected for a hard-sphere-like colloid confined between
function,95 it should provide means for quantitative comparison
planar solid walls at a separation of 2.7s, and the confined colloid was
kept in equilibrium with a bulk reservoir of volume fraction fb = 0.168. with experiments.
(a) Structure factor S(q) and (b) inverse of the wave-vector-dependent An important point of this perspective is that we now have
diffusion coefficient D(q) of the confined colloid, measured parallel to the experimental access to confined colloids’ structure factors and
confining walls (01, red squares), diagonally across the confining slit (451, intermediate scattering functions, via X-ray scattering from
blue circles), and in an intermediate direction (161, green asterisks). Data
colloid-filled microfluidic containers.28,39 So far this approach
taken from ref. 39.
has been used to demonstrate an inverse relationship between
the confined colloid’s wave-vector-dependent short-time diffusion
The above discussion neglects the effect of indirect, hydro- coefficient and anisotropic structure factor, as discussed above.
dynamic interactions. In bulk colloids the hydrodynamic particle– At the moment, it will allow us to address many important
particle interactions result in a hydrodynamic function H(q) questions related to colloidal diffusion in confined geometries,
whose shape closely follows that of the structure factor.6 For ranging from the effect of hydrodynamic interactions to structural
dense confined colloids one expects two complications. First, arrest. With the advent of extremely brilliant sources,36 however,
hydrodynamic interactions between a central particle and its I also foresee that these studies are extended to large wave vectors,
anisotropic solvation shells (cf. Fig. 3) leads to anisotropic and hence self-diffusion. The possibility to simultaneously probe
ensemble-averaged hydrodynamic interactions, which in turn the confined colloid’s pair densities (via its anisotropic structure
should yield an anisotropic H(q) akin to the anisotropic S(q). factor) and its long-time self-diffusion coefficient will provide
Second, it has been known for a long time that a single solvated unique means to experimentally corroborate the aforementioned
particle is slowed down by the presence of solid walls,87,88 an correlation between slit-width-dependent self-diffusivity and
effect which has been observed experimentally both at a single (pair) excess entropy.
planar wall89,90 and in confinement between two planar walls.91 I have in this perspective focused on the microscopic
One thus expects the hydrodynamic function H(q) of a dense structure and diffusive dynamics of archetypical hard-sphere-
confined colloid to result from a convoluted combination like colloids confined between planar hard walls. The concepts
of anisotropic particle–particle and particle–wall interactions. brought up may have a broader range of application, however, as
As an initial step, we have very recently analysed the data of alluded to already in the introduction. Of particular interest is the
ref. 39 to obtain the first evidence of a strongly anisotropic common system of charge-stabilised colloids between charged
hydrodynamic function H(q), the shape of which is qualitatively walls. These systems exhibit qualitatively similar microscopic
similar to that of the structure factor S(q).86 More experimental ordering, at the level of anisotropic pair densities,27 as the hard-
and theoretical work is needed in order to understand this sphere fluid between hard walls. One may thus anticipate that
phenomenon. charged colloids in confinement also will exhibit qualitatively

This journal is © the Owner Societies 2017 Phys. Chem. Chem. Phys., 2017, 19, 23632--23641 | 23639
View Article Online

PCCP Perspective

similar diffusive properties as discussed above, although this 25 S. Mandal, S. Lang, M. Gross, M. Oettel, D. Raabe, T. Franosch
hypothesis needs to be corroborated in future studies. and F. Varnik, Nat. Commun., 2014, 5, 4435.
26 E. Lomba, C. Bores and G. Kahl, J. Chem. Phys., 2014, 141,
164704.
27 K. Nygård, D. K. Satapathy, J. Buitenhuis, E. Perret, O. Bunk,
Acknowledgements
C. David and J. F. van der Veen, Europhys. Lett., 2009,
I thank Jeetain Mittal, Gaurav Goel, and Thomas M. Truskett 86, 66001.
Published on 19 May 2017. Downloaded by Nanyang Technological University on 9/15/2019 12:23:06 PM.

for sharing their simulation data, and Roland Kjellander for 28 K. Nygård, R. Kjellander, S. Sarman, S. Chodankar, E. Perret,
comments on the manuscript. I am grateful to the Ruth and J. Buitenhuis and J. F. van der Veen, Phys. Rev. Lett., 2012,
Nils-Erik Stenäck foundation for financial support. 108, 037802.
29 K. Nygård, S. Sarman, K. Hyltegren, S. Chodankar, E. Perret,
J. Buitenhuis, J. F. van der Veen and R. Kjellander, Phys. Rev.
X, 2016, 6, 011014.
References
30 K. Nygård, Curr. Opin. Colloid Interface Sci., 2016, 22, 30.
1 M. Alcoutlabi and G. B. McKenna, J. Phys.: Condens. Matter, 31 L. Van Hove, Phys. Rev., 1954, 95, 249.
2005, 17, R461. 32 J. K. G. Dhont, An Introduction to Dynamics of Colloids,
2 J. N. Israelachvili, Intermolecular and Surface Forces, Academic Elsevier, Amsterdam, 1996.
Press, London, 3rd edn, 2011. 33 B. J. Berne and R. Pecora, Dynamic Light Scattering, Dover,
3 S. Granick, Science, 1991, 253, 1374. New York, 2000.
4 T. M. Squires and S. R. Quake, Rev. Mod. Phys., 2005, 77, 977. 34 P. N. Pusey, P. N. Segré, O. P. Behrend, S. P. Meeker and
5 S. H. L. Klapp, Y. Zeng, D. Qu and R. von Klitzing, Phys. Rev. W. C. K. Poon, Physica A, 1997, 235, 1.
Lett., 2008, 100, 118303. 35 G. Grübel and F. Zontone, J. Alloys Compd., 2004, 362, 3.
6 G. Nägele, Phys. Rep., 1996, 272, 215. 36 O. G. Shpyrko, J. Synchrotron Radiat., 2014, 21, 1057.
7 M. Kollmann, Phys. Rev. Lett., 2003, 90, 180602. 37 Small-Angle X-Ray Scattering, ed. O. Glatter and O. Kratky,
8 C. Lutz, M. Kollmann and C. Bechinger, Phys. Rev. Lett., Academic Press, London, 1982.
2004, 93, 026001. 38 T. Narayanan, Curr. Opin. Colloid Interface Sci., 2009, 14, 409.
9 S. Mandal and T. Franosch, Phys. Rev. Lett., 2017, 118, 39 K. Nygård, J. Buitenhuis, M. Kagias, K. Jefimovs, F. Zontone
065901. and Y. Chushkin, Phys. Rev. Lett., 2016, 116, 167801.
10 M. Krishnan, N. Mojarad, P. Kukura and V. Sandoghdar, 40 A. L. Thorneywork, R. E. Rozas, R. P. A. Dullens and
Nature, 2010, 467, 692. J. Horbach, Phys. Rev. Lett., 2015, 115, 268301.
11 T. O. E. Skinner, S. K. Schnyder, D. G. A. L. Aarts, J. Horbach 41 J. Mittal, J. R. Errington and T. M. Truskett, J. Chem. Phys.,
and R. P. A. Dullens, Phys. Rev. Lett., 2013, 111, 128301. 2007, 126, 244708.
12 F. Varnik and T. Franosch, J. Phys.: Condens. Matter, 2016, 42 J. Mittal, T. M. Truskett, J. R. Errington and G. Hummer,
28, 133001. Phys. Rev. Lett., 2008, 100, 145901.
13 R. Kjellander and S. Sarman, J. Chem. Soc., Faraday Trans., 43 G. Goel, W. P. Krekelberg, M. J. Pond, J. Mittal, V. K. Shen,
1991, 87, 1869. J. R. Errington and T. M. Truskett, J. Stat. Mech.: Theory Exp.,
14 B. Götzelmann and S. Dietrich, Phys. Rev. E: Stat. Phys., 2009, P04006.
Plasmas, Fluids, Relat. Interdiscip. Top., 1997, 55, 2993. 44 M. Schmidt and H. Löwen, Phys. Rev. E: Stat. Phys., Plasmas,
15 K. Nygård, S. Sarman and R. Kjellander, J. Chem. Phys., 2013, Fluids, Relat. Interdiscip. Top., 1997, 55, 7228.
139, 164701. 45 A. Fortini and M. Dijkstra, J. Phys.: Condens. Matter, 2006,
16 K. Nygård, S. Sarman and R. Kjellander, J. Chem. Phys., 2014, 18, L371.
141, 094501. 46 H. Löwen, J. Phys.: Condens. Matter, 2009, 21, 474203.
17 A. Härtel, M. Kohl and M. Schmiedeberg, Phys. Rev. E: Stat., 47 S. Lang, V. Botan, M. Oettel, D. Hajnal, T. Franosch and
Nonlinear, Soft Matter Phys., 2015, 92, 042310. R. Schilling, Phys. Rev. Lett., 2010, 105, 125701.
18 Fundamentals of Inhomogeneous Fluids, ed. D. Henderson, 48 Y. Rosenfeld, Phys. Rev. A: At., Mol., Opt. Phys., 1977, 15,
Marcel Dekker, New York, 1992. 2545.
19 J.-P. Hansen and I. R. McDonald, Theory of Simple Liquids, 49 Y. Rosenfeld, Chem. Phys. Lett., 1977, 48, 467.
Academic Press, Amsterdam, 3rd edn, 2006. 50 R. D. Mountain and H. J. Raveché, J. Chem. Phys., 1971, 55, 2250.
20 R. Kjellander and S. Marčelja, J. Chem. Phys., 1988, 88, 7138. 51 D. C. Wallace, J. Chem. Phys., 1987, 87, 2282.
21 D. Henderson, S. Sokolowski and D. Wasan, J. Stat. Phys., 52 A. Baranyai and D. J. Evans, Phys. Rev. A: At., Mol., Opt. Phys.,
1997, 89, 233. 1989, 40, 3817.
22 R. Kjellander, J. Phys.: Condens. Matter, 2009, 21, 424101. 53 M. Dzugutov, Nature, 1996, 381, 137.
23 V. Botan, F. Pesth, T. Schilling and M. Oettel, Phys. Rev. E: 54 J. R. Errington, T. M. Truskett and J. Mittal, J. Chem. Phys.,
Stat., Nonlinear, Soft Matter Phys., 2009, 79, 061402. 2006, 125, 244502.
24 J. W. Zwanikken and M. Olvera de la Cruz, Proc. Natl. Acad. 55 S. Ruchi, S. N. Chakraborty and C. Chakravarty, J. Chem.
Sci. U. S. A., 2013, 110, 5301. Phys., 2006, 125, 204501.

23640 | Phys. Chem. Chem. Phys., 2017, 19, 23632--23641 This journal is © the Owner Societies 2017
View Article Online

Perspective PCCP

56 Y. Yan, S. V. Buldyrev and H. E. Stanley, Phys. Rev. E: Stat., 78 V. N. Michailidou, J. W. Swan, J. F. Brady and G. Petekidis,
Nonlinear, Soft Matter Phys., 2008, 78, 051201. J. Chem. Phys., 2013, 139, 164905.
57 N. Jakse and A. Pasturel, Sci. Rep., 2016, 6, 20689. 79 B. Lin, S. A. Rice and D. A. Weitz, Phys. Rev. E: Stat. Phys.,
58 X. Ma, W. Chen, Z. Wang, Y. Peng, Y. Han and P. Tong, Plasmas, Fluids, Relat. Interdiscip. Top., 1995, 51, 423.
Phys. Rev. Lett., 2013, 110, 078302. 80 B. Lin, M. Meron, B. Cui and S. A. Rice, Phys. Rev. Lett., 2005,
59 T. S. Ingebrigtsen and T. B. S. J. C. Dyre, Phys. Rev. X, 2012, 94, 216001.
2, 011011. 81 B. Lin, B. Cui, X. Xu, R. Zangi, H. Diamant and S. A. Rice,
Published on 19 May 2017. Downloaded by Nanyang Technological University on 9/15/2019 12:23:06 PM.

60 J. C. Dyre, J. Phys.: Condens. Matter, 2016, 28, 323001. Phys. Rev. E: Stat., Nonlinear, Soft Matter Phys., 2014, 89,
61 J. Mittal, J. R. Errington and T. M. Truskett, Phys. Rev. Lett., 022303.
2006, 96, 177804. 82 O. Bunk, A. Diaz, F. Pfeiffer, C. David, C. Padeste,
62 J. Mittal, J. R. Errington and T. M. Truskett, J. Phys. Chem. B, H. Keymeulen, P. R. Willmott, B. D. Patterson, B. Schmitt,
2006, 111, 10054. D. K. Satapathy, J. F. van der Veen, H. Guo and G. H.
63 T. S. Ingebrigtsen, J. R. Errington, T. M. Truskett and Wegdam, Phys. Rev. E: Stat., Nonlinear, Soft Matter Phys.,
J. C. Dyre, Phys. Rev. Lett., 2013, 111, 235901. 2007, 75, 021501.
64 C. R. Nugent, K. V. Edmond, H. N. Patel and E. R. Weeks, 83 A. Diaz and J. F. van der Veen, Thin Solid Films, 2007,
Phys. Rev. Lett., 2007, 99, 025702. 515, 5645.
65 S. Sastry, T. M. Truskett, P. G. Debenedetti, S. Torquato and 84 D. K. Satapathy, O. Bunk, K. Jefimovs, K. Nygård, H. Guo,
F. H. Stillinger, Mol. Phys., 1998, 95, 289. A. Diaz, E. Perret, F. Pfeiffer, C. David, G. H. Wegdam and
66 J. L. Carmer, Local Structure and Dynamics of Complex Fluids, J. F. van der Veen, Phys. Rev. Lett., 2008, 101, 136103.
PhD thesis, University of Texas at Austin, 2013. 85 D. K. Satapathy, K. Nygård, O. Bunk, K. Jefimovs, E. Perret,
67 J. A. Bollinger, A. Jain, J. Carmer and T. M. Truskett, J. Chem. A. Diaz, F. Pfeiffer, C. David and J. F. van der Veen, Europhys.
Phys., 2015, 142, 161102. Lett., 2009, 87, 34001.
68 K. V. Edmond, C. R. Nugent and E. R. Weeks, Phys. Rev. E: 86 K. Nygård, J. Buitenhuis, M. Kagias, K. Jefimovs, F. Zontone
Stat., Nonlinear, Soft Matter Phys., 2012, 85, 041401. and Y. Chushkin, Phys. Rev. E, 2017, accepted.
69 P. N. Segré and P. N. Pusey, Phys. Rev. Lett., 1996, 77, 771. 87 H. Faxén, Ark. Mat. Astron. Phys., 1923, 17, 1.
70 P. Holmqvist and G. Nägele, Phys. Rev. Lett., 2010, 104, 058301. 88 H. Brenner, Chem. Eng. Sci., 1961, 16, 242.
71 G. Grübel, A. Madsen and A. Robert, in Soft Matter Charac- 89 P. Holmqvist, J. K. G. Dhont and P. R. Lang, Phys. Rev. E:
terization, ed. R. Borsali and R. Pecora, Springer, New York, Stat., Nonlinear, Soft Matter Phys., 2006, 74, 021402.
2008, p. 953. 90 P. Holmqvist, J. K. G. Dhont and P. R. Lang, J. Chem. Phys.,
72 P. G. de Gennes, Physica A, 1959, 25, 825. 2007, 126, 044707.
73 B. J. Ackerson, P. N. Pusey and R. J. A. Tough, J. Chem. Phys., 91 B. Lin, J. Yu and S. A. Rice, Phys. Rev. E: Stat. Phys., Plasmas,
1982, 76, 1279. Fluids, Relat. Interdiscip. Top., 2000, 62, 3909.
74 C. W. J. Beenakker and P. Mazur, Physica A, 1984, 126, 349. 92 W. Götze and L. Sjögren, Rep. Prog. Phys., 1992, 55, 241.
75 U. Genz and R. Klein, Physica A, 1991, 171, 26. 93 S. Lang, R. Schilling, V. Krakoviack and T. Franosch, Phys.
76 A. J. Banchio, J. Gapinski, A. Patkowski, A. Häußler, A. Flureasu, Rev. E: Stat., Nonlinear, Soft Matter Phys., 2012, 86, 021502.
S. Sacanna, P. Holmqvist, G. Meier, M. P. Lettinga and 94 S. Lang, R. Schilling and T. Franosch, Phys. Rev. E: Stat.,
G. Nägele, Phys. Rev. Lett., 2006, 96, 138303. Nonlinear, Soft Matter Phys., 2014, 90, 062126.
77 V. N. Michailidou, G. Petekidis, J. W. Swan and J. F. Brady, 95 S. Lang and T. Franosch, Phys. Rev. E: Stat., Nonlinear, Soft
Phys. Rev. Lett., 2009, 102, 068302. Matter Phys., 2014, 89, 062122.

This journal is © the Owner Societies 2017 Phys. Chem. Chem. Phys., 2017, 19, 23632--23641 | 23641

You might also like