1: \documentclass[preprint,superscriptaddress,showpacs,
2: preprintnumbers,amssymb,aps,prl]{revtex4}
3: \usepackage{graphicx,dcolumn,bm,amsmath}
4:
5: \begin{document}
6:
7: \title{Influence of hydrodynamic interactions
8: on lane formation in oppositely charged driven colloids}
9:
10: \author{M. Rex}
11: \affiliation{Institut f\"ur Theoretische Physik II: Weiche Materie,
12: Heinrich-Heine-Universit\"at D\"{u}sseldorf,
13: Universit{\"a}tsstra{\ss}e 1, D-40225 D\"{u}sseldorf,
14: Germany}
15:
16: \author{H. L{\"o}wen}
17: \affiliation{Institut f\"ur Theoretische Physik II: Weiche Materie,
18: Heinrich-Heine-Universit\"at D\"{u}sseldorf,
19: Universit{\"a}tsstra{\ss}e 1, D-40225 D\"{u}sseldorf,
20: Germany}
21:
22: \date{\today}
23:
24: % The correct dates will be entered by Springer
25: %
26: \begin{abstract}
27: The influence of hydrodynamic interactions on lane formation of
28: oppositely charged driven colloidal suspensions is investigated using
29: Brownian dynamics computer simulations performed on the Rotne-Prager
30: level of the mobility tensor. Two cases are considered, namely
31: sedimentation and electrophoresis. In the latter case the Oseen
32: contribution to the mobility tensor is screened due to the opposite
33: motion of counterions. The simulation results are compared to that
34: resulting from simple Brownian dynamics where hydrodynamic
35: interactions are neglected. For sedimentation, we find that
36: hydrodynamic interactions strongly disfavor laning. In the
37: steady-state of lanes, a macroscopic phase separation of lanes is
38: observed.
39: This is in marked contrast to the simple Brownian case where a finite
40: size of lanes was obtained in the steady-state. For strong Coulomb
41: interactions between the colloidal particles a lateral square lattice
42: of oppositely driven lanes is stable similar to the simple Brownian
43: dynamics. In an electric field, on the other hand, the behavior is
44: found in qualitative and quantitative accordance with the case of
45: neglected hydrodynamics.
46: %
47: \pacs{82.70.Dd, 61.20.Ja, 64.70.Dv, 05.70.Ln}
48: \end{abstract} %end of abstract
49: %
50: \maketitle
51: %
52: \section{Introduction}
53: \label{intro}
54:
55: The dynamics of colloidal particles dispersed in a fluid solvent
56: is quite different from the ballistic motion of molecular systems
57: which is described by Newton's law
58: \cite{pusey91,loewen_pra_1991,naegele_pr_1996,vermant_jpcm_2005}.
59: The viscous solvent both damps
60: the motion of a colloidal particle and leads to kicks of the solvent
61: molecules with the colloidal particle leading to Brownian motion
62: if the time-scales of the molecular solvent is much faster than that
63: of the diffusive motion of the colloidal particles. In concentrated suspensions,
64: a dragged colloidal particle influences the motion of another particles
65: via the solvent flow field.
66: These so-called hydrodynamic interaction is typically long-ranged.
67: While it can be neglected in colloidal suspensions of very small volume fractions,
68: it induces significant corrections in the equilibrium and nonequilibrium dynamics
69: of colloidal suspensions \cite{naegele_pr_1996,pesche_epl_2000}.
70: The equilibrium structures and phase boundaries,
71: on the other hand, are unaffected by hydrodynamic interactions.
72:
73:
74: In the past years a simple nonequilibrium phase transition
75: \cite{katz_prb_1983,katz_jsp_1984,schmittmann_1995,loewen_jpcm_2001}
76: has been discussed in a binary mixture of colloidal particles which
77: are driven by a constant external field
78: \cite{dzubiella_pre_2002,netz_epl_2003b,dzubiella_epl_2003,dzubiella_fd_2003}.
79: The drive is different on the two particle species and could arise
80: from gravity and from an external electric field in the case of
81: charged colloidal suspensions.
82: Brownian dynamics computer simulations with neglected hydrodynamic
83: interactions strongly support the scenario that - as a function of the
84: driving strength - the mixtures undergoes a transition from a mixed
85: steady-state with anisotropic correlations towards a steady-state
86: where macroscopic lanes are formed.
87: The transitions has been found for oppositely driven repulsive mixtures
88: \cite{dzubiella_pre_2002,netz_epl_2003b,dzubiella_epl_2003,dzubiella_fd_2003,delhommelle_pre_2005,pandley_ijmpc_2003,koppl_prl_2006}
89: in two and three spatial dimensions and it seems to be a first-order
90: nonequilibrium transition with a significant hysteresis in an order
91: parameter which detects laning \cite{dzubiella_pre_2002}.
92: The general scenario occurs also in pedestrian dynamics
93: \cite{helbing_pa_2006,helbing_njp_2003} and in granular matter
94: \cite{ehrhardt_pre_2005,coniglio_prl_2005}.
95:
96:
97: Recently, lane formation was observed in real-space
98: experiments by Leunissen et al \cite{leunissen_nature_2005}.
99: Equimolar mixtures of oppositely charged colloidal particles
100: were prepared which form binary ionic crystals
101: \cite{hynninen_prl_2006_b,hynninen_prl_2006_a}.
102: These crystals were exposed to a strong
103: external electric field and the dynamics of lane formation was watched
104: by confocal microscopy. Subsequently extensive Brownian dynamics simulations
105: were carried out to map the nonequilibrium phase diagram
106: \cite{rex_pre_2007}. These simulations assumed a Yukawa pair
107: interaction between the particles
108: and neglected hydrodynamic interactions completely.
109: A wealth of different steady-state structures was detected.
110: In particular different orderings were found
111: in the plane perpendicular to the drive including
112: a square, triangular or rhombic crystalline lattice of lanes,
113: a network structure with a finite structural length and intermediate chain
114: formation of lanes \cite{rex_pre_2007}.
115: A rough estimate of the experimental parameters used in Ref.\
116: \cite{leunissen_nature_2005} reveals that the laned state observed
117: experimentally indeed falls into the region where laning is expected
118: to occur by Brownian dynamics computer simulations.
119: %\cite{loewen_sendai_2008}.
120:
121:
122: In this paper we address the question how hydrodynamic interactions
123: influence the scenario and the steady-state diagram of lane formation.
124: Our motivation to do so is twofold: First, the experiments, of course,
125: contain hydrodynamic interactions in their full glory, and therefore
126: an inclusion of hydrodynamic interactions is needed for a quantitative
127: comparison.
128: Second, there is a principal need to understand in which direction
129: hydrodynamics influence lane formation.
130: In particular it is known \cite{long_epje_2001} that the leading
131: long-ranged term in the mobility pair tensor is screened if an
132: electric field is applied since a charged colloidal particles is
133: surrounded by counterions of opposite charge.
134: That makes the hydrodynamic interaction significantly different from
135: e.g. sedimentation induced by different buoyant masses of oppositely
136: charged colloids where the action of gravity on the microions can
137: safely be neglected.
138: In the latter case, the leading part in the mobility tensor at large
139: interparticle separation is the {\it unscreened} Oseen tensor.
140: It would be interesting to explore how far the steady-state is
141: affected by hydrodynamic interactions in both cases of sedimentation
142: and electrophoresis.
143:
144: We use Brownian dynamic computer simulations and include hydrodynamic
145: interaction by using mobility tensors on the Rotne-Prager level.
146: Both cases of sedimentation and electrophoresis are studied separately
147: with an unscreened resp.\ screened version of the mobility tensor.
148: The steady-state phase diagrams and the drift velocity are simulated.
149: The simulation data are compared to that obtained by simple
150: Brownian dynamics where hydrodynamic interactions are neglected.
151: For sedimentation, we find that hydrodynamic interactions strongly
152: disfavor laning.
153: In the steady-state of lanes, a macroscopic phase separation of lanes
154: is observed, i.e.\ the sickness of the lanes are of the system size.
155: This is in marked contrast to the simple Brownian case where a finite
156: size of lanes was obtained in the steady-state.
157: For strong Coulomb interactions between the colloidal particles a
158: lateral square lattice of oppositely driven lanes is stable similar to
159: the simple Brownian dynamics.
160: In an electric field, on the other hand, the behavior is found in
161: qualitative and quantitative accordance with the case of neglected
162: hydrodynamics.
163: All lateral structures are reproduced and the topology of the
164: steady-state phase diagram is unchanged.
165:
166: The paper is organized as follows: in section II, we describe our
167: model and simulation scheme for sedimentation and electrophoresis.
168: Brownian dynamics simulation results are presented in section III. We
169: conclude in section IV.
170:
171: \section{The Model}
172: \label{sec:model}
173: We perform Brownian dynamic simulations to study an equimolar binary
174: mixture of $2N=1024$ oppositely charged colloidal particles of
175: diameter $\sigma$ dissolved in a solvent fluid of shear viscosity
176: $\eta$ at temperature $T$ and volume fraction $\phi =
177: 2N\pi\sigma^{3}/6l^{3}$ exposed to an external driving field, where
178: $l$ is the dimension of a cubic simulation box having periodic
179: boundary conditions.
180: Henceforth, $\sigma$ serves as the unit of length and
181: $k_{\mathrm{B}}T$, the thermal energy, as the energy unit of the
182: system.
183: To mimic the experiments by Leunissen et al
184: \cite{leunissen_nature_2005,royall_jcp_2006} the particles interact
185: with an effective screened Coulomb potential (or Yukawa potential)
186: plus a steric repulsion $V_{\mathrm{h}}$:
187: \begin{equation}
188: \label{eq:yukawa_potential}
189: V(r_{ij}) = V_{0}\frac{Z_{i}Z_{j}}{(1+\kappa\sigma/2)^2}
190: \frac{e^{(-\kappa\sigma(r_{ij}/\sigma-1))}}{r_{ij}/\sigma}+
191: V_{\mathrm{h}}(r_{ij})
192: \end{equation}
193: with $V_{0}=50k_{\mathrm{B}}T$ the strength of the interaction potential and
194: $Z_{i}=\pm 1$ the sign of the charge of particle $i$.
195: $r_{ij}=|\mathbf{r}_{i}-\mathbf{r}_{j}|$ denotes the distance between
196: particle $i$ and $j$, where $\mathbf{r}_{i}$ is the coordinate vector.
197: The inverse screening length $\kappa$ governs the range of the
198: interaction and is determined by the salt concentration of the
199: solution.
200: The steric repulsion between the particles, that prevents the
201: system from collapsing, is approximated by a repulsive (shifted and
202: truncated) Lennard-Jones potential
203: \begin{equation}
204: \label{eq:hardcore_potential}
205: V_{\mathrm{h}}(r_{ij}) = \left\{
206: \begin{array}{ll}
207: \epsilon \left [ \left (
208: \frac{\sigma}{r_{ij}}\right)^{12} - \left (
209: \frac{\sigma}{r_{ij}}\right)^{6} +\frac{1}{4}\right] &
210: \textrm{if $r_{ij} \leq 2^{1/6}\sigma$}\\
211: 0 & \textrm{else},
212: \end{array}
213: \right.
214: \end{equation}
215: with $\epsilon = 4V_{0}/(1+\kappa\sigma/2)^2$.
216: The constant external driving field that acts in opposite directions
217: on the two different particle species reads as
218: \begin{equation}
219: \label{eq:f_ext}
220: \mathbf{F}_{i}^{\mathrm{ext}} = Z_{i}f\mathbf{e}_{z},
221: \end{equation}
222: where $\mathbf{e}_{z}$ is the unit vector along the $z$ direction and
223: $f= 150k_\mathrm{B} T/\sigma$ is the strength of the external force.
224: The external force is supposed to stem from either an electric field
225: or a gravitational field accompanied with different buoyant masses of
226: the oppositely charged particles.
227: Though the external force Eq.\ (\ref{eq:f_ext}) may in both cases be
228: identical, if the charges and/or buoyant masses are chosen accordingly,
229: the hydrodynamic interactions are not.
230:
231: The algorithm used to simulate the diffusive Brownian motion of the
232: colloidal particles was proposed by Ermak and McCammon
233: \cite{ermak_jcp_1978}.
234: Here, the translational displacements of the particles are
235: deemed to occur in time steps of fixed length $\Delta t$ and the
236: update algorithm is given by \cite{allen_tildesley_book}:
237: \begin{equation}
238: \label{eq:integration_algorithm}
239: \mathbf{r}_{i}(t+\Delta t)=\mathbf{r}_{i}(t)+
240: \Delta t \sum_{j=1}^{N}\left\{\frac{\mathbf{D}_{ij}(t)}{k_{\mathrm{B}}T}
241: \cdot \mathbf{F}_{j}(t) +
242: \mathbf{\nabla}_{\mathbf{r}_{j}}\cdot \mathbf{D}_{ji}(t)\right\} +
243: \Delta{\mathbf r}^{{\mathrm G}}_{i},
244: \end{equation}
245: where $\mathbf{D}_{ij}$ denotes the diffusion tensor field depending on the
246: positions of the particles at time $t$.
247: The random displacements $\Delta{\mathbf r}^{{\mathrm G}}_{i}$ are
248: chosen from a joint Gaussian distribution with mean and covariant
249: matrix \cite{allen_tildesley_book}
250: \begin{equation}
251: \label{eq:mean}
252: \langle \Delta{\mathbf r}^{{\mathrm G}}_{i} \rangle_{\mathrm{G}} =
253: 0;\;\;\;\;
254: \langle \Delta{\mathbf r}^{{\mathrm G}}_{i}
255: \Delta{\mathbf r}^{{\mathrm G}}_{j} \rangle_{\mathrm{G}} =
256: 2{\mathbf D}_{ij}\Delta t,
257: \end{equation}
258: where $\langle \ldots \rangle_{\mathrm{G}} $ denotes the average over
259: the Gaussian noise distribution.
260: $\mathbf{F}_{i}(t)$, $i=1,...,N$, comprises the nonhydrodynamic
261: forces due to interparticle interactions, determined by the gradient
262: of the interaction potentials in Eq.\ (\ref{eq:yukawa_potential}) and
263: Eq.\ (\ref{eq:hardcore_potential}), and the external force
264: $\mathbf{F}_{i}^{\mathrm{ext}}$, Eq.\ (\ref{eq:f_ext}), acting onto
265: particle $i$.
266:
267: Hydrodynamic interactions are included in the simulation through the
268: mobility tensor $\bm{\mu}_{ij}=\mathbf{D}_{ij}/k_{\mathrm{B}}T$.
269: In a first approach we neglect hydrodynamic interactions
270: completely to asses its effect on the system.
271: In that case the diffusion tensor is given by Stoke's law
272: in diagonal form
273: \begin{equation}
274: \label{eq:without}
275: \gamma \bm{\mu}_{ij}=
276: \delta_{ij}\mathbf{1},
277: \end{equation}
278: with friction $\gamma=3\pi \eta \sigma_{\mathrm{H}}$, where
279: $\sigma_{\mathrm{H}}$ is the hydrodynamic diameter.
280: In the sedimentation and electrophoresis situation we approximate the
281: mobility tensor by two-body interactions.
282: In this approximation the divergence in Eq.\
283: (\ref{eq:integration_algorithm}) vanishes always
284: \cite{wajnryb_pa_2004}.
285: When studying the sedimentation, the buoyant masses of the oppositely
286: charged particles are supposed to be such that the same force acts on
287: the two species but in opposite directions.
288: The action of gravity on the microions can safely be neglected.
289: Therefore, in sum no net force is acting on the solvent and overall it
290: remains quiescent.
291: Then, for a pair of spheres of hydrodynamic diameter
292: $\sigma_{\mathrm{H}}$ the mobility tensor is approximated by the
293: well-known Rotne-Prager expression \cite{rotne_jcp_1969}
294: \begin{equation}
295: \label{eq:rotne_prager_tensor}
296: \gamma \bm{\mu}^{\mathrm{RP}}_{ij}=
297: \delta_{ij}\mathbf{1}+(1-\delta_{ij})\left[
298: \frac{3\sigma_{\mathrm{H}}}{8}\mathbb{O}({\mathbf r}_{ij}) +
299: \frac{\sigma_{\mathrm{H}}^{3}}{16}\mathbb{Q}({\mathbf r}_{ij})
300: \right ],
301: \end{equation}
302: where
303: \begin{equation}
304: \label{eq:oseen_tensor}
305: \mathbb{O}({\mathbf r}) = \frac{1}{|\mathbf{r}|}
306: ({\mathbf 1}+\hat{\mathbf{r}}\otimes\hat{\mathbf{r}}); \;\;\;\;
307: \mathbb{Q}({\mathbf r}) = \frac{1}{|\mathbf{r}|^{3}}
308: ({\mathbf 1}-3\hat{\mathbf{r}}\otimes\hat{\mathbf{r}}),
309: \end{equation}
310: with the unit vector $\hat{\mathbf{r}}=\mathbf{r}/|\mathbf{r}|$,
311: $\otimes$ a dyadic product, and $\delta_{ij}$ Kronecker's symbol.
312: On this level of approximation we incorporate all interactions up to
313: $O((\sigma_{\mathrm{H}}/r)^{3})$.
314: Higher order contributions such as many-body, coupling between
315: rotational and translational motions, and lubrication forces are
316: neglected.
317: The leading term in Eq.\ (\ref{eq:rotne_prager_tensor}) is given by
318: $\mathbb{O}({\mathbf r})$ which is of the order of $1/|\mathbf{r}|$
319: for large distances.
320:
321: However, when regarding the electrophoresis the mobility tensor
322: has to be altered since forces induced by the surrounding counterions
323: into the solvent are in sum equal to the force induced by a colloidal
324: particle.
325: Thus, the solvent flow stemming from the drag on the counterions
326: cannot be neglected as is done in the sedimentation case.
327: It results in an effective screening of the leading far-distance term
328: of the hydrodynamic interactions between the colloidal particles as
329: Long and Ajdai have shown \cite{long_epje_2001}.
330: Their mobility tensor $\bm{\mu}^{\mathrm{LA}}_{ij}$ reads as
331: \begin{eqnarray}
332: \label{eq:long_tensor}
333: \nonumber
334: \gamma \bm{\mu}^{\mathrm{LA}}_{ij}&=&
335: \delta_{ij}\mathbf{1}+\frac{3\sigma_{\mathrm{H}}}{4} (1-\delta_{ij})
336: \left[
337: \frac{e^{-\kappa r_{ij}}}{r_{ij}} \left(\left(1+\frac{1}{\kappa
338: r}+\frac{1}{\kappa^{2}r^{2}}\right) \mathbf{1} -\right.\right.\\
339: &&\left.\left.\left(\frac{1}{3}+\frac{1}{\kappa
340: r}+\frac{1}{\kappa^{2}r^{2}}\right)
341: 3\hat{\mathbf{r}}_{ij}\otimes\hat{\mathbf{r}}_{ij}\right)
342: + \frac{1}{\kappa^{2}}\mathbb{Q}({\mathbf r}_{ij})
343: \right ].
344: \end{eqnarray}
345: Here, the leading order term is $\mathbb{Q}({\mathbf r}_{ij})$ which
346: decays as $1/|\mathbf{r}^{3}|$.
347: To account for the polymer coating on the colloidal particles that
348: gives rise to the steric repulsion we choose
349: $\sigma_{\mathrm{H}}=0.9\sigma$ throughout this paper.
350:
351: More sophisticated simulation techniques for spherical particles in an
352: unbounded space including lubrication approximation for particles in
353: close proximity and multipolar expansion methods are available
354: \cite{durlofsky_jfm_1987,ladd_jcp_1988,cichocki_jcp_1994,sierou_jfm_2001,banchio_jcp_2003}.
355: However, in the electrophoresis where the hydrodynamic
356: interactions of the counterions become important explicit simulations
357: of all colloidal particles and their counterions -- $110$ per
358: colloidal particle in the experiments by Leunissen et al -- are still
359: beyond computational means.
360: Therefore, we adopted the calculations of Long and Ajdari to our
361: simulations and compare it to the sedimentation problem on the same
362: level of accuracy, i.e.\ the Rotne-Prager level.
363: To our best knowledge this is the first Brownian dynamic simulations
364: with the Long and Ajdari mobility term \ref{eq:long_tensor}.
365:
366: Both mobility tensors \ref{eq:rotne_prager_tensor} and
367: \ref{eq:long_tensor} are long-ranged and thus require an
368: Ewald-like summation in simulations analogous to Coulomb and
369: dipole-dipole interactions.
370: Details on the summation and discussions about appropriate boundary
371: conditions to the system can be found elsewhere
372: \cite{smith_pa_1987,smith_fdsc_1987,bradey_jfm_1988b,beenakker_jcp_1986}.
373: We applied the scheme suggested by Beenakker \cite{beenakker_jcp_1986}
374: and adapted it for $\bm{\mu}^{\mathrm{LA}}_{ij}$ accordingly.
375: The square root of the diffusion tensor, needed when calculating the
376: random displacements in Eq.\ (\ref{eq:mean}), are obtained
377: from a Cholesky decomposition:
378: \begin{equation}
379: \label{eq:decomposition}
380: \mathbf{D} = \mathbf{L}\cdot\mathbf{L}^{\mathrm{T}},
381: \end{equation}
382: where $\mathbf{L}$ is a lower triangular matrix and
383: $\mathbf{L}^{\mathrm{T}}$ is its transpose.
384: A suitable time scale for our system is $\tau_{\mathrm{B}} =
385: \gamma \sigma^{2}/k_{\mathrm{B}}T$.
386: The equations of motion including the external field are
387: numerically solved using a finite time step $\Delta
388: t=2\cdot10^{-5}\tau_{\mathrm{B}}$ in all simulations.
389: Statistics were gathered after an initial relaxation period of
390: $20\tau_{\mathrm{B}}$.
391: The starting configuration of all simulations was a homogeneous
392: mixture.
393:
394:
395: \section{Results}
396: \label{sec:results}
397: \subsection{Order parameter and steady-state phase diagrams}
398: To assess the effect of hydrodynamic interactions on the
399: lane behavior of oppositely charged colloidal particles we study a
400: set of volume fractions $\phi$ and inverse screening lengths
401: $\kappa^{*}=\kappa\sigma$ and map out nonequilibrium steady-state
402: phase diagrams for all three situations: hydrodynamic interactions
403: neglected (A), electrophoresis (B), and sedimentation (C).
404:
405: A state of lane is thereby identified by a laning order parameter that
406: is defined through
407: \begin{equation}
408: \label{eq:orderparameter}
409: \Phi=\frac{1}{2N}\left
410: \langle\sum_{i=1}^{2N}\Phi_{i}\right \rangle_{t},
411: \end{equation}
412: where the angular brackets $\langle...\rangle_{t}$ denote a time average.
413: The local order parameter $\Phi_{i} =
414: (n_{\mathrm{l}}-n_{\mathrm{o}})^2/(n_{\mathrm{l}}+n_{\mathrm{o}})^2$
415: is assigned to every particle $i$, where the numbers $n_{\mathrm{l}}$
416: and $n_{\mathrm{o}}$ are the number of like charged particles and
417: oppositely charged particles, respectively, whose projections of
418: distance onto the plane perpendicular to the field are smaller than a
419: suitable cut-off length scale $z_{c}$.
420: $\Phi_{i}$ is equal to $1$ if all particles within this distance
421: criterion are of the same kind and zero if
422: $n_{\mathrm{l}}=n_{\mathrm{o}}$, i.e.\ a homogeneous mixture.
423: We chose for convenience $z_{c}=\frac{3}{4}\sigma$ to detect all lanes
424: starting from a single queue of particles.
425: In what follows we will use a threshold: for
426: $\Phi\geq1/2$ we call the configuration a state of lanes while in the
427: opposite case ($\Phi<1/2$) we call it a state without lanes.
428:
429: We observe that lanes form different structures in the plane
430: perpendicular to the driving direction for different values of
431: $\kappa^{*}$ and $\phi$.
432: We find lanes placed on a square or triangular lattice, a network-like
433: structure (reminiscent of a bicontinous microemulsion or
434: microphase-separated system), coexistence regimes of the same, and
435: macroscopically separated lanes.
436: The resulting nonequilibrium steady-state phase diagrams are shown in
437: Figures \ref{fig:phasediagram1}, \ref{fig:phasediagram2}, and
438: \ref{fig:phasediagram3}.
439: They are accompanied with typical simulation snapshots of the
440: projection of all particle coordinates onto the $xy$-plane of the
441: respective situation.
442: \begin{figure}[t]
443: \begin{center}
444: \includegraphics[width=8cm, clip=true, draft=false]{./fig1}
445: \caption{(Color online) Nonequilibrium steady-state phase diagram for
446: a constant driving force of strength $f= 150k_\mathrm{B} T/\sigma$
447: with hydrodynamic interactions neglected accompanied by a typical
448: simulation snapshot of the projection of the particle coordinates
449: onto the plane perpendicular to the driving field for each different
450: state.
451: The lines between the phases are a guide for the eye.}
452: \label{fig:phasediagram1}
453: \end{center}
454: \end{figure}
455: %\end{figure*}
456: %\begin{figure*}
457: \begin{figure}[t]
458: \begin{center}
459: \includegraphics[width=8cm, clip=true, draft=false]{./fig2}
460: \caption{(Color online) Same as Fig. \ref{fig:phasediagram1}
461: but for electrophoresis with hydrodynamic interactions taken into
462: account through $\bm{\mu}_{ij}^{\mathrm{LA}}$ in Eq.\
463: (\ref{eq:long_tensor}).
464: The phase diagram reveals only minor differences as compared to the
465: the case of neglected hydrodynamic interactions in Fig.\
466: \ref{fig:phasediagram1}.}
467: \label{fig:phasediagram2}
468: \end{center}
469: \end{figure}
470: \begin{figure}[t]
471: \begin{center}
472: %\end{figure*}
473: %\begin{figure*}
474: \includegraphics[width=8cm, clip=true, draft=false]{./fig3}
475: \caption{(Color online) Same as Fig. \ref{fig:phasediagram1}
476: but for sedimentation with hydrodynamic interactions taken into
477: account through $\bm{\mu}_{ij}^{\mathrm{RP}}$ in Eq.\
478: (\ref{eq:rotne_prager_tensor}).
479: The phase diagram shows significant changes as compared to Fig.\
480: \ref{fig:phasediagram1} and Fig.\ \ref{fig:phasediagram2}.}
481: \label{fig:phasediagram3}
482: \end{center}
483: \end{figure}
484: %\end{figure*}
485:
486: What can be seen at first sight is that the qualitative behavior of
487: situation (A) and (B) in Figures \ref{fig:phasediagram1} and
488: \ref{fig:phasediagram2} is almost identical with only subtle
489: differences, while on the other hand the phase behavior changes
490: drastically for situation (C), Figure \ref{fig:phasediagram3}.
491: In the latter the whole phase diagram is altered and the diversity of
492: phases found is reduced compared to the first two cases.
493:
494: \subsection{Comparison of simulation results for neglected
495: hydrodynamic interactions and electrophoresis}
496: \label{sec:el}
497: In this subsection we briefly describe the two phase diagrams in
498: Figures \ref{fig:phasediagram1} and \ref{fig:phasediagram2} and their
499: differences, beginning at low volume fractions and ending at high
500: ones, and then dwell on the third diagram, Fig.\
501: \ref{fig:phasediagram3}, thereafter in subsection \ref{sec:sed}.
502: A more ample discussion on how different phases are identified and
503: what structural correlations they exhibit can be found in a previous
504: work of the authors on the same system with hydrodynamic interactions
505: neglected but for a slightly different driving strength and larger
506: systems \cite{rex_pre_2007}.
507: Here, we find for situation (A) virtually the same results as in the
508: previous work with only one difference, namely that we do not
509: encounter a rhombic phase for $\phi=0.4$ and $\kappa^{*}=1,2,3$.
510:
511: For very low volume fraction, $\phi\lesssim 0.01$, in both systems the
512: correlations between the particles are not sufficient to form lanes at
513: all.
514: Thus, the systems are in a phase of no-lanes.
515: Only for very low salt concentration, i.e.\ small $\kappa^{*}$, where
516: the electrostatic coupling between the colloidal particles is strong,
517: we find a coexistence region between lanes and no-lanes.
518: Here, the region with no-lanes consists of voids, where hardly any
519: particle is found.
520: The structure of the lane region, on the other hand, is different in the
521: two situations.
522: For situation (A) the corresponding snapshot in Fig.\
523: \ref{fig:phasediagram1} reveals fixed lattice points while the
524: snapshot in Fig.\ \ref{fig:phasediagram2} situation (B) shows a
525: network-like structure.
526: For situation (A), an initial configuration with lanes placed on a square
527: lattice separated from a completely depleted region is stable in
528: simulations, as well.
529: Thus, we assume that in situation (A) the lanes/no-lanes phase is a
530: transient state toward a complete square lattice and no-lane phase
531: separation.
532: Hydrodynamic interactions destroy the coexistence phase for $\phi=0.01$.
533: It only occurs in a denser system with $\phi=0.1$ whereas in situation
534: (A) this state shows already up at $\phi=0.01$.
535: Additionally, the voids are more pronounced in the latter case
536: compared to situation (B).
537: Upon increasing $\kappa^{*}\geq2$ for $\phi=0.1$ in both situations we
538: find a network-like structure whose characteristic spacing is
539: increasing with increasing $\kappa^{*}$.
540: For situation (A) there is also a small coexistence region between
541: network and square lattice at $\kappa^{*}=2$.
542:
543: To obtain a quantitative measure of the characteristic spacing in the
544: network structure we determine a structure factor perpendicular to the
545: driving field of like-charged particles.
546: The steady-state partial structure factor has been calculated by
547: evaluating the expression
548: \begin{equation}
549: \label{eq:Sofk}
550: S_{\mathrm{AA}}(k) = 1+\rho\hat{h}_{\mathrm{AA}}(k),
551: \end{equation}
552: %\begin{eqnarray}
553: % \label{eq:Sofk}
554: % \nonumber
555: % S_{\perp}(k) &=& \frac{1}{2}\left(
556: % N^{-1}\langle\rho^{+}(k)\rho^{+}(-k)\rangle_{t} +
557: % N^{-1}\langle\rho^{-}(k)\rho^{-}(-k)\rangle_{t} \right )\\
558: % \nonumber
559: % &=&\frac{1}{2N} \left(\left\langle\left[\sum_{i=1}^{N}
560: % \cos\mathbf{k}\cdot\mathbf{r}_{i\;\perp}\right]^{2}+
561: % \left[\sum_{i=1}^{N}\sin\mathbf{k}\cdot\mathbf{r}_{i\;\perp}\right]^{2}
562: % \right\rangle_{t}+\right.\\
563: % &&\left.\left\langle\left[\sum_{i=N+1}^{2N}
564: % \cos\mathbf{k}\cdot\mathbf{r}_{i\;\perp}\right]^{2}+
565: % \left[\sum_{i=N+1}^{2N}\sin\mathbf{k}\cdot\mathbf{r}_{i\;\perp}\right]^{2}
566: % \right\rangle_{t}\right ),
567: %\end{eqnarray}
568: %$\rho^{\pm}(k)$ is the Fourier transform of the number density of the
569: %positively and negatively charged particle species, respectively, in
570: %the plain perpendicular to the driving field.
571: with $\rho=6\phi/\pi\sigma^{3}$ the number density and the wave vector
572: $k=|\mathbf{k}|$, where $\mathbf{k}=(2\pi/l)(k_{x},k_{y})$ and
573: $k_{x}$, $k_{y}$ are integers.
574: $\hat{h}_{\mathrm{AA}}(k)$ is the Fourier transform of the total correlation
575: function $h_{\mathrm{AA}}(\mathbf{r}_{\perp})=g_{\mathrm{AA}}(r_{\perp})-1$ with
576: $\mathbf{r}=(\mathbf{r}_{\perp},z)$ and
577: \begin{equation}
578: \label{eq:gofrperp}
579: g_{\mathrm{AA}}(r_{\perp}) = \frac{1}{\rho N}
580: \left \langle
581: \sum_{\substack{i,j\\ (Z_{i}=Z_{j},\; i\ne j)}}^{2N}
582: \delta(\mathbf{r}_{\perp}-|\mathbf{r}_{i\;\perp}-\mathbf{r}_{j\;\perp}|)
583: \delta(z_{i}-z_{j})
584: \right\rangle_{t},
585: \end{equation}
586: where $\delta(\mathbf{x})$ denotes Dirac's delta-distribution.
587: \begin{figure}[t]
588: \begin{center}
589: \includegraphics[width=8cm, clip=true, draft=false]{./fig4}
590: \caption{Partial structure factor $S_{\perp}(k)$ of like-charged
591: particles perpendicular to the driving field for $\kappa^{*}=10$ and
592: $\phi=0.1$ for hydrodynamics neglected and electrophoresis.
593: The pre-peak at $k^{*}_{0}=k_{0}\sigma$ indicates an additional length
594: scale of the structure in the network-like phase.
595: The inset shows the position $k^{*}_{0}$ of the pre-peak as a function
596: of the inverse screening length $\kappa^{*}$ for a fixed volume
597: fraction of $\phi=0.1$.}
598: \label{fig:sk}
599: \end{center}
600: \end{figure}
601: An example of the steady-state partial structure factors for
602: $\kappa^{*}=10$ and $\phi=0.1$ for both situation, (A) and (B), is
603: shown in Fig. \ref{fig:sk}.
604: One clearly observes a pronounced pre-peak at the wave number
605: $k_{0}$ in both cases.
606: A pre-peak in the structure factor is an indication of an additional
607: mesoscopic length scale as typical for bicontinous networks, such as
608: e.g.\ microemulsions \cite{teubner_jcp_1987,gompper_prl_1989}.
609: In the inset we additionally present the position $k_{0}$ of the
610: pre-peak as a function of the inverse screening length.
611: It is evident from the picture that the characteristic spacing is
612: indeed growing with increasing $\kappa^{*}$.
613: We find hardly any difference between situation (A) and (B).
614:
615: For $\phi=0.2$ an additional phase for small inverse screening length
616: shows up in both phase diagrams.
617: Oppositely driven lanes are placed on a square lattice with an
618: alternating charge pattern.
619: The formation of this lattice structure can be qualitatively
620: understood from an effective interaction between oppositely charged
621: driven lanes which has a short-ranged repulsive and a long-ranged
622: attractive interaction.
623: The former stems from the friction between oppositely driven
624: particles while the later results form the Coulomb interaction.
625: The square lattice then reduces the electrostatic energy of the system
626: because each particle has only oppositely charged neighbors.
627: For increasing salt concentrations we encounter a coexistence region
628: between the square lattice and the network-like phase and finally end up
629: in a pure network-like phase.
630: The phase diagram is in both situations very similar, only the borders
631: of the transitions are slightly shifted.
632: In the electrophoresis case the network-like structure is preferred to
633: the square lattice.
634:
635: For a higher volume fractions of $\phi=0.3$ a coexistence regime
636: between a triangular lattice and a network-like structure is found.
637: The lattice-points in the triangular phase are rather randomly
638: decorated with different charges.
639: Here, the short range repulsion plays the dominant role compared to
640: the electrostatic interaction.
641: It enforces a triangular lattice due to packing effects although
642: electrostatically it is strongly disfavored because like-charged
643: particles necessarily occupy lattice points next to each other.
644: Again, hydrodynamic interactions slightly shift the phase boundaries
645: towards the network-like structure.
646:
647: For the highest volume fraction studied, $\phi=0.4$, both phase
648: diagrams show exactly the same behavior.
649: Here, the short range repulsions dictates the phase behavior for
650: nearly all salt concentration but for $\kappa^{*}=1$ and enforces
651: lanes to be placed on a triangular lattice.
652: Only for $\kappa^{*}=1$, where electrostatic interactions are
653: prominent, a square lattice is preferred.
654: In principle a square lattice is possible up to the packing of a
655: simple cubic lattice of $\phi=0.52$.
656:
657: In summary we observe very similar behavior in both situations.
658: The observed differences can be qualitatively explained by the fact
659: that hydrodynamic interactions disfavor lanes driven oppositely past
660: each other.
661:
662: \subsection{Sedimentation}
663: \label{sec:sed}
664: Regarding sedimentation, Fig.\ \ref{fig:phasediagram3}, the whole
665: phase diagram exhibits {\it only} three different phases.
666: For volume fractions $\phi \leq 0.1$ we do not find lane formation for
667: all inverse screening length studied.
668: For increasing volume fractions and strong electrostatic interactions,
669: $\kappa^{*}\leq2$, first the square lattice at $\phi\approx0.3$, that
670: is also present in the previous two situations, is recovered and then
671: the system reenters a region with no-lanes for $\phi=0.4$.
672: This behavior nicely illustrates the competition between hydrodynamic
673: interactions disfavoring lanes driven oppositely past each other and
674: the electrostatic interactions favoring a square lattice.
675: Only for the small regime around $\phi\approx0.3$ the electrostatics
676: succeeds the hydrodynamic interactions and enforces a square lattice.
677: For all other volume fractions laning is destroyed.
678: However, for stronger salt concentrations, where the Coulombic
679: coupling is reduced, we discover a situation that
680: is not present in the previous situations (A) and (B), namely a region
681: in which only two big completely separated lanes.
682: We call this state {\it phase separated}.
683: %, because its structure is reminiscent of that of two phase separated fluids.
684: In that case the long ranged hydrodynamic interactions prescribe the
685: structure and the short ranged Yukawa interaction plays its role only
686: at the rough interface of the two phases.
687: From our simulations we conclude the lanes are separated by half of the
688: box length.
689:
690: \subsection{Drift velocity}
691: \label{sec:vd}
692: Now, we study the influence of hydrodynamic interaction on the drift
693: velocity along the field direction that is defined as follows
694: \begin{equation}
695: \label{eq:vel}
696: v^{2}:=\lim_{t\rightarrow\infty}\frac{\left\langle\left[(\mathbf{r}_{i}(t)
697: - \mathbf{r}_{i}(0))\cdot\mathbf{e}_{z}\right]^{2}\right\rangle}
698: {t^{2}}.
699: \end{equation}
700: This entity measures the mean-square displacement of each particle
701: in the nonequilibrium steady-state.
702: A study on the effect of hydrodynamics on the drift velocity of like
703: charged colloidal particles was carried out by Watzlawek and N\"agele
704: \cite{watzlawek_jcis_1999}.
705: We study two cases, first we fix the volume fraction at $\phi=0.3$ and
706: vary the inverse screening length and afterward vice versa for
707: $\kappa^{*}=1$.
708:
709: In Fig.\ \ref{fig:v} we display $v^{*}=v\tau_{\mathrm{B}}/\sigma$ for
710: a fixed volume fraction $\phi=0.3$ as a function of the inverse
711: screening length $\kappa^{*}$ for all three situations.
712: \begin{figure}[t]
713: \begin{center}
714: \includegraphics[width=8cm, clip=true, draft=false]{./fig5}
715: \caption{Average dimensionless drift velocity
716: $v^{*}=v\tau_{\mathrm{B}}/\sigma$ in drive direction as a function
717: of the inverse screening length $\kappa^{*}$ at $\phi=0.3$ for
718: Brownian dynamic simulations with hydrodynamic interactions
719: neglected, taken into account through $\bm{\mu}_{ij}^{\mathrm{LA}}$,
720: and $\bm{\mu}_{ij}^{\mathrm{RP}}$.}
721: \label{fig:v}
722: \end{center}
723: \end{figure}
724: For all cases the drift velocity increases with
725: decreasing Coulomb coupling because oppositely charged colloids
726: attract each other while driven in opposite direction and lanes
727: mutually retard each other.
728: For very strongly screened particles where this friction is less
729: important all three curves reveal approximately the same value of
730: $v\approx 130\tau_{\mathrm{B}}/\sigma$.
731: Accordingly, this value is close to the drift velocity of $v =
732: 150\tau_{\mathrm{B}}/\sigma$ for a system of infinite dilution
733: subjected to the same driving force.
734: While in (A) and (B) $v$ grows gradually, in the sedimentation curve
735: we encounter a jump in the drift velocity between $\kappa^{*}=3$ and
736: $\kappa^{*}=4$.
737: This coincides with the transition from the no-lane regime to the phase
738: separated regime, see the phase diagram Fig.\ \ref{fig:phasediagram3}.
739: On the other hand for $\kappa^{*}\leq2$, where we find a square lattice,
740: the drift velocity is similar to $\kappa^{*}=3$.
741: From that we conclude that the phase separated state of lanes supports
742: particle transport while lanes placed on a square lattice enforced by
743: strong Coulombic interactions slows down particle transportation.
744: A further interesting feature is that curves for (A) and (B)
745: intersect between $\kappa^{*}=1$ and $\kappa^{*}=2$ and that the screened
746: hydrodynamic interaction enhance the drift velocity for larger inverse
747: screening length.
748: The same is true for the unscreened hydrodynamic interactions in the
749: sedimentation for $\kappa^{*}\geq6$.
750: \begin{figure}[t]
751: \begin{center}
752: \includegraphics[width=8cm, clip=true, draft=false]{./fig6}
753: \caption{Average dimensionless drift velocity
754: $v^{*}=v\tau_{\mathrm{B}}/\sigma$ in drive direction as a function
755: of the volume fractions $\phi$ at $\kappa^{*}=1$ for
756: Brownian dynamic simulations with hydrodynamic interactions
757: neglected, taken into account through $\bm{\mu}_{ij}^{\mathrm{LA}}$,
758: and $\bm{\mu}_{ij}^{\mathrm{RP}}$.}
759: \label{fig:v2}
760: \end{center}
761: \end{figure}
762: When studying the drift velocity for a fixed inverse screening length
763: but for varying volume fractions in Fig.\ \ref{fig:v2} we find again
764: an intersection point of the curves for situations (A) and (B).
765: Here, the drift velocity in the electrophoresis reaches an
766: approximately constant value of $v^{*}\approx112$ for $\phi=0.1 - 0.4$
767: whereas it gradually decreases when hydrodynamic interactions
768: are neglected.
769: For the sedimentation the drift velocity decreases monotonically.
770: In contrast to the case of varying salt concentration we do not
771: encounter a jump in the sedimentation drift velocity when entering the
772: square lattice at $\phi=0.3$ and reentering the no-lane regime at
773: $\phi=0.4$.
774:
775: \section{Conclusions}
776: In conclusion the influence of hydrodynamic interactions on lane
777: formation of opposite charged colloids driven by an electric field or
778: by gravity was investigated by Brownian dynamics computer
779: simulations.
780: Hydrodynamic interactions were included on the Rotne-Prager level.
781: For an electric field, the leading Oseen term is screened due to the
782: presence of counterions.
783: The latter fact has lead to very similar steady state phase diagrams
784: for an electric field as a driving source than that in the simple case
785: of neglected hydrodynamic interactions.
786: Various steady state were obtained as a function of the colloidal
787: density and the range of the interaction.
788: They can qualitatively be understood in terms of a competition of the
789: mutual Coulomb attraction and friction of sliding lanes.
790: At high densities the lateral structure is crystalline, the crystal is
791: either triangular as dictated by packing at high densities and high
792: screening or square-like at low-screening which minimizes the Coulomb
793: attractive energy.
794: On the other hand, in sedimentation where the two colloidal species
795: have the same buoyant mass up to a relative sign, friction of sliding
796: lanes is strongly enhanced leading to macroscopic separation of
797: lanes.
798:
799: The steady-state phase diagram can in principle be verified in
800: real-space experiments of charged suspensions which are driven in an
801: electric field or sedimenting \cite{leunissen_nature_2005}.
802: It would be interesting to construct a microscopic theory for the lane
803: transitions which includes the lateral crystalline structure.
804: The instability analysis within a dynamical density functional theory
805: as applied to the case of equal charges in two spatial dimensions
806: \cite{dzubiella_epl_2003,dzubiella_pre_2004} should in principle be
807: generalizable to the case of oppositely charged particles.
808:
809: Finally more sophisticated simulations schemes are needed in order to
810: go beyond the Rotne-Prager level of approximation used in this paper.
811: Among the various promising approaches are the stochastic rotation
812: dynamics code \cite{padding_pre_2006,padding_prl_2004}, a lattice
813: Boltzmann theory including hydrodynamics
814: \cite{lobaskin_njp_2004,chatterji_jcp_2005,capuani_jcp_2006,chatterji_jcp_2007,lobaskin_prl_2007}
815: and counterion flow or the recently developed fluid particle dynamics
816: methods \cite{kodama_jpcm_2004,kim_prl_2006,kim_mts_2005}.
817:
818:
819: \section*{ACKNOWLEDGMENTS}
820:
821: We thank M. Leunissen, A. van Blaaderen, R. Yamamoto, A. Louis, and J.
822: Padding for helpful remarks and the DFG (SFB TR6, project section D1)
823: and the Graduiertenf{\"o}rderung of the Heinrich-Heine-Universit\"at
824: D\"usseldorf for financial support.
825:
826:
827: \begin{thebibliography}{53}
828:
829: \bibitem{pusey91}
830: P.N. Pusey, in \emph{Liquids, Freezing and Glass Transition}, edited by J.P.
831: Hansen, D.~Levesque, J.~Zinn-Justin (North-Holland, Amsterdam, 1991)
832:
833: \bibitem{loewen_pra_1991}
834: H.~L{\"o}wen, J.P. Hansen, J.N. Roux, Phys.\ Rev.\ A \textbf{44}, 1169 (1991)
835:
836: \bibitem{naegele_pr_1996}
837: G.~N{\"a}gele, Phys.\ Rep. \textbf{272}, 215 (1996)
838:
839: \bibitem{vermant_jpcm_2005}
840: J.~Vermant, M.J. Solomon, J.\ Phys.: Condens.\ Matter \textbf{17}, R187 (2005)
841:
842: \bibitem{pesche_epl_2000}
843: R.~Pesch\'e, G.~N{\"a}gele, Europhys.\ Lett. \textbf{51}, 584 (2000)
844:
845: \bibitem{katz_prb_1983}
846: S.~Katz, J.L. Lebowitz, H.~Spohn, Phys.\ Rev.\ B \textbf{28}, 1655 (1983)
847:
848: \bibitem{katz_jsp_1984}
849: S.~Katz, J.L. Lebowitz, H.~Spohn, J.\ Stat.\ Phys. \textbf{34}, 497 (1984)
850:
851: \bibitem{schmittmann_1995}
852: B.~Schmittmann, R.K.P. Zia, \emph{Statistical mechanics of driven diffusive
853: systems} (Acadamic Press, London, 1995)
854:
855: \bibitem{loewen_jpcm_2001}
856: H.~L{\"o}wen, J.\ Phys.: Condens.\ Matter \textbf{13}, R415 (2001)
857:
858: \bibitem{dzubiella_pre_2002}
859: J.~Dzubiella, G.P. Hoffmann, H.~L{\"o}wen, Phys.\ Rev.\ E \textbf{65}, 021402
860: (2002)
861:
862: \bibitem{netz_epl_2003b}
863: R.R. Netz, Europhys.\ Lett. \textbf{65}, 616 (2003)
864:
865: \bibitem{dzubiella_epl_2003}
866: J.~Chakrabarti, J.~Dzubiella, H.~L{\"o}wen, Europhys.\ Lett. \textbf{61}, 415
867: (2003)
868:
869: \bibitem{dzubiella_fd_2003}
870: H.~L{\"o}wen, J.~Dzubiella, Faraday Discuss. \textbf{123}, 99 (2003)
871:
872: \bibitem{delhommelle_pre_2005}
873: J.~Delhommelle, Phys.\ Rev.\ E \textbf{71 (1)}, 016705 (2005)
874:
875: \bibitem{pandley_ijmpc_2003}
876: R.B. Pandey, J.F. Gettrust, R.~Seyfarth, L.A. Cueva-Parra, Int.\ J.\ Mod.\
877: Phys.\ C \textbf{14 (7)}, 955 (2003)
878:
879: \bibitem{koppl_prl_2006}
880: M.~K{\"o}ppl, P.~Henseler, A.~Erbe, P.~Nielaba, P.~Leiderer, Phys.\ Rev.\ Lett.
881: \textbf{97}, 208302 (2006)
882:
883: \bibitem{helbing_pa_2006}
884: R.~Jiang, D.~Helbing, P.K. Shuklai, Q.S. Wu, Physica A \textbf{368 (2)}, 567
885: (2006)
886:
887: \bibitem{helbing_njp_2003}
888: R.~Kolbl, D.~Helbing, N.\ J.\ Phys. \textbf{5}, 48 (2003)
889:
890: \bibitem{ehrhardt_pre_2005}
891: G.C.M.A. Ehrhardt, A.~Stephenson, P.M. Reis, Phys.\ Rev.\ E \textbf{71}, 041301
892: (2005)
893:
894: \bibitem{coniglio_prl_2005}
895: M.P. Ciamarra, A.~Coniglio, M.~Nicodemi, Phys.\ Rev.\ Lett. \textbf{94}, 188001
896: (2005)
897:
898: \bibitem{leunissen_nature_2005}
899: M.E. Leunissen, C.G. Christova, A.P. Hynninen, C.P. Royall, A.I. Campbell,
900: A.~Imhof, M.~Dijkstra, R.~van Roij, A.~van Blaaderen, Nature \textbf{437},
901: 235 (2005)
902:
903: \bibitem{hynninen_prl_2006_b}
904: A.P. Hynninen, M.E. Leunissen, A.~van Blaaderen, M.~Dijkstra, Phys.\ Rev.\
905: Lett. \textbf{96 (1)}, 018303 (2006)
906:
907: \bibitem{hynninen_prl_2006_a}
908: A.P. Hynninen, C.G. Christova, R.~van Roij, A.~van Blaaderen, M.~Dijkstra,
909: Phys.\ Rev.\ Lett. \textbf{96 (13)}, 138308 (2006)
910:
911: \bibitem{rex_pre_2007}
912: M.~Rex, H.~L{\"o}wen, Phys.\ Rev.\ E \textbf{75}, 051402 (2007)
913:
914: \bibitem{long_epje_2001}
915: D.~Long, A.~Ajdari, Eur.\ Phys.\ J.\ E \textbf{4}, 29 (2001)
916:
917: \bibitem{royall_jcp_2006}
918: C.P. Royall, M.E. Leunissen, A.P. Hynninen, M.~Dijkstra, A.~van Blaaderen, J.\
919: Chem.\ Phys. \textbf{124}, 2447061 (2006)
920:
921: \bibitem{ermak_jcp_1978}
922: D.L. Ermak, J.\ Chem.\ Phys. \textbf{69}, 1352 (1978)
923:
924: \bibitem{allen_tildesley_book}
925: M.P. Allen, D.J. Tildesley, eds., \emph{Computer Simulation of Liquids}
926: (Clarendon Press Oxford, Oxford, 1989)
927:
928: \bibitem{wajnryb_pa_2004}
929: E.~Wajnryb, P.~Szymczak, B.~Cichocki, Physica A \textbf{335}, 339 (2004)
930:
931: \bibitem{rotne_jcp_1969}
932: J.~Rotne, S.~Prager, J.\ Chem.\ Phys. \textbf{50}, 4831 (1969)
933:
934: \bibitem{durlofsky_jfm_1987}
935: L.~Durlofsky, J.F. Brady, G.~Bossis, J.\ Fluid Mech. \textbf{180}, 21 (1987)
936:
937: \bibitem{ladd_jcp_1988}
938: A.J.C. Ladd, J.\ Chem.\ Phys. \textbf{88}, 5051 (1988)
939:
940: \bibitem{cichocki_jcp_1994}
941: B.~Cichocki, B.U. Felderhof, K.~Hinsen, E.~Wajnryb, J.~B{\l}iawzdziewicz, J.\
942: Chem.\ Phys. \textbf{100}, 3780 (1994)
943:
944: \bibitem{sierou_jfm_2001}
945: A.~Sierou, J.F. Brady, J.\ Fluid Mech. \textbf{448}, 115 (2001)
946:
947: \bibitem{banchio_jcp_2003}
948: A.J. Banchio, J.F. Brady, J.\ Chem.\ Phys. \textbf{118}, 10323 (2003)
949:
950: \bibitem{smith_pa_1987}
951: E.R. Smith, I.K. Snook, W.~van Megen, Physica A \textbf{143A}, 441 (1987)
952:
953: \bibitem{smith_fdsc_1987}
954: E.R. Smith, Faraday Discuss.\ Chem.\ Soc. \textbf{83}, 193 (1987)
955:
956: \bibitem{bradey_jfm_1988b}
957: J.F. Brady, R.J. Phillips, J.C. Lester, G.~Bossis, J.\ Fluid Mech.
958: \textbf{195}, 257 (1988)
959:
960: \bibitem{beenakker_jcp_1986}
961: C.W.J. Beenakker, J.\ Chem.\ Phys. \textbf{85}, 1581 (1986)
962:
963: \bibitem{teubner_jcp_1987}
964: M.~Teubner, R.~Strey, J.\ Chem.\ Phys. \textbf{87}, 3195 (1987)
965:
966: \bibitem{gompper_prl_1989}
967: G.~Gompper, M.~Schick, Phys.\ Rev.\ Lett. \textbf{62}, 1647 (1989)
968:
969: \bibitem{watzlawek_jcis_1999}
970: M.~Watzlawek, G.~N{\"a}gele, Jour.\ Colloid Interface Sci. \textbf{214}, 170
971: (1999)
972:
973: \bibitem{dzubiella_pre_2004}
974: J.~Chakrabarti, J.~Dzubiella, H.~L{\"o}wen, Phys.\ Rev.\ E \textbf{70}, 012401
975: (1 (2004)
976:
977: \bibitem{padding_pre_2006}
978: J.T. Padding, A.A. Louis, Phys.\ Rev.\ E \textbf{74}, 031402 (2006)
979:
980: \bibitem{padding_prl_2004}
981: J.T. Padding, A.A. Louis, Phys.\ Rev.\ Lett. \textbf{93}, 220601 (2004)
982:
983: \bibitem{lobaskin_njp_2004}
984: V.~Lobaskin, B.~D{\"u}nweg, N.\ J.\ Phys. \textbf{6}, 54 (2004)
985:
986: \bibitem{chatterji_jcp_2005}
987: A.~Chatterji, J.~Horbach, J.\ Chem.\ Phys. \textbf{122}, 184903 (2005)
988:
989: \bibitem{capuani_jcp_2006}
990: F.~Capuani, I.~Pagonabarraga, D.~Frenkel, J.\ Chem.\ Phys. \textbf{124}, 124903
991: (2006)
992:
993: \bibitem{chatterji_jcp_2007}
994: A.~Chatterji, J.~Horbach, J.\ Chem.\ Phys. \textbf{126}, 064907 (2007)
995:
996: \bibitem{lobaskin_prl_2007}
997: V.~Lobaskin, B.~D{\"u}nweg, M.~Medebach, T.~Palberg, C.~Holm, Phys.\ Rev.\
998: Lett. \textbf{98}, 176105 (2007)
999:
1000: \bibitem{kodama_jpcm_2004}
1001: H.~Kodama, K.~Takeshita, T.~Araki, H.~Tanaka, J.\ Phys.: Condens.\ Matter
1002: \textbf{16}, L115 (2004)
1003:
1004: \bibitem{kim_prl_2006}
1005: K.~Kim, Y.~Nakayama, R.~Yamamoto, Phys.\ Rev.\ Lett. \textbf{96}, 208302 (2006)
1006:
1007: \bibitem{kim_mts_2005}
1008: K.~Kim, R.~Yamamoto, Macromol.\ Theory Simul. \textbf{14}, 278 (2005)
1009:
1010: \end{thebibliography}
1011:
1012: % BibTeX users please use
1013: %\bibliographystyle{epj}
1014: %\bibliography{/home/rexm/Documents/bibliography_pre}
1015:
1016: \end{document}
1017:
1018: % end of file template.tex
1019:
1020: