1: %\documentclass[prl,preprint]{revtex4}
2: %\documentclass[pre]{revtex4}
3: \documentclass[pre,twocolumn,showpacs]{revtex4}
4: \usepackage{amsmath,amssymb}
5: \usepackage{epsfig}
6:
7:
8: \begin{document}
9:
10: \newlength{\mylen}
11: \setlength{\mylen}{\textwidth}
12: \addtolength{\mylen}{-1cm}
13: \newcommand{\bea}{\begin{eqnarray}}
14: \newcommand{\eea}{\end{eqnarray}}
15: %\newcommand{\spaceint}[2]{\int_{#1} d^3 #2 \;}
16: \newcommand{\vect}[1]{\mathbf{#1}}
17: %\newcommand{\vat}{V^{\rm att}}
18: %\newcommand{\di}{\displaystyle}
19: %\newcommand{\rp}{r_{||}}
20: \newcommand{\eps}{\epsilon}
21: %\newcommand{\ep}{\varepsilon_\Pi}
22: %\newcommand{\ef}{\varepsilon_F}
23: %\newcommand{\emix}{\varepsilon_{\rm m}}
24: %\newcommand{\ev}{\varepsilon_{\rm rep}}
25: %\newcommand{\Ezp}{E_{z,0^+}}
26: %\newcommand{\Ezm}{E_{z,0^-}}
27: %\newcommand{\Ep}{E_{||,0}}
28: %\newcommand{\ro}{r_{0,{\rm ref}}}
29: %\newcommand{\agt}{\gtrsim}
30: %\newcommand{\alt}{\lesssim}
31:
32:
33: \title{Multipole expansion of the electrostatic interaction between charged colloids at interfaces}
34:
35: \author{A. Dom{\'i}nguez$^1$, D. Frydel$^2$\ and M. Oettel$^3$}
36: \thanks{Supported by the German Science Foundation (DFG) via the Collaborative
37: Research Center SFB--TR6 ``Colloids in External Fields'', project section D6.}
38: %\email{oettel@mf.mpg.de}
39: \affiliation{$^1$F\'\i sica Te\'orica, Universidad de Sevilla, Apdo.1065, E-41080
40: Sevilla, Spain}
41: \affiliation{$^2$Max--Planck--Institut f\"ur Metallforschung, Heisenbergstr.~3, D--70569 Stuttgart, Germany, and }
42: \affiliation{Institut f\"ur Theoretische und Angewandte
43: Physik, Universit\"at Stuttgart, Pfaffenwaldring 57, D--70569 Stuttgart, Germany}
44: \affiliation{$^3$Johannes Gutenberg Universit\"at Mainz, Institut f{\"ur} Physik,
45: WA 331, D--55099 Mainz, Germany}
46:
47:
48:
49:
50: \begin{abstract}
51: The general form of the electrostatic potential around an arbitrarily charged
52: colloid at a flat interface
53: between a dielectric and a screening phase (such as air and water, respectively) is
54: analyzed in terms of a multipole expansion. The leading term is isotropic in the
55: interfacial plane and varies with $d^{-3}$ where $d$ is the in--plane distance
56: from the colloid. The effective interaction potential between two arbitrarily
57: charged colloids is likewise isotropic
58: and $\propto d^{-3}$, thus generalizing the dipole--dipole repulsion first
59: found for point charges at water interfaces. Anisotropic, attractive interaction terms can arise
60: only for higher powers $d^{-n}$ with $n \ge 4$.
61: %
62: The relevance of these findings for recent experiments is discussed.
63:
64: \end{abstract}
65:
66: \pacs{82.70.Dd}
67:
68: \maketitle
69:
70: \section{Introduction}
71:
72: %{\em Introduction.--}
73: The self--assembly of stably trapped,
74: sub-$\mu$m colloidal particles at water--air or water--oil interfaces
75: has gained much interest in recent years.
76: %These particles are trapped irreversibly
77: %at the interface if water wets the colloids only partially.
78: For the
79: specific case of charge--stabilized colloids at interfaces, the {\em repulsive} part of their mutual
80: interaction resembles a dipole--dipole interaction at large
81: separations.
82: This may be understood theoretically by approximating the colloid
83: as a point charge located either right at the interface \cite{Stil61,Hur85}
84: (i.e. assuming charges on the colloid--water interface)
85: and/or above the interface \cite{Ave00a} (i.e. assuming charges on the colloid--air/oil
86: interface).
87: Additionally, the formation of metastable mesostructures with such colloids point
88: to the possible existence of intercolloidal {\em attractions} far
89: beyond the range of van--der--Waals forces \cite{Gom05,Che05,Che06}, however,
90: care must be taken to avoid contaminations of the interface which lead
91: to colloid mesostructures with similar appearance \cite{Fer04}.
92: %According to
93: %Refs.~\cite{Ghe97,Ghe01,Gar98a,Gar98b,Sta00,Que01,Gom05}, polystyrene spheres
94: %(radii $R=0.25\dots 2.5$ $\mu$m)
95: % on flat water--air interfaces using deionized water exhibit
96: %spontaneous formation
97: %of complicated metastable mesostructures. They are consistent with the presence
98: %of an attractive, secondary
99: %minimum in the effective intercolloidal potential at separations $d/R\approx 3\dots 20$ with
100: %a depth of a few $k_B T$.
101: Previous work \cite{For04,Oet05,Oet05a,Wue05,Dom07} aimed at relating this attractive
102: minimum to capillary interactions
103: due to interfacial deformations caused by a {\em homogeneous} surface charge on the
104: colloids but with no conclusive answer.
105: In recent work \cite{Che05,Che06},
106: it was experimentally shown that the charge--carrying surface groups used
107: for charge--stabilizing polystyrene colloids are actually distributed quite
108: inhomogeneously and patchily over the colloid surface. Thus it was
109: speculated in Refs.~\cite{Che05,Che06} that through this inhomogeneous
110: charge distribution like--charged colloids could acquire effective dipole
111: moments in the interface plane and attractive electrostatic interactions of dipole--dipole
112: type could arise which might overcome the repulsion at shorter distances.
113:
114: Motivated by the finding of inhomogeneous surface charge on colloids, we
115: extend the asymptotic results for the electrostatic potential and interaction of
116: point charges at water interfaces \cite{Hur85} to the general case of an
117: arbitrary, localized colloidal charge distribution using a multipole expansion.
118: The presence of the interface leads to restrictions in the
119: multipole coefficients of the potential around a single colloid and of the interaction energy
120: between two colloids. In particular, we
121: find that the leading term in the effective interaction energy between two colloids at
122: lateral distance $d$ is isotropic in the
123: interfacial plane, repulsive and $\propto
124: d^{-3}$ regardless of the inhomogeneities of the charge distribution in the colloids.
125: % if the colloids are spherical and inhomogeneously charged.
126: Angular dependencies enter the effective interaction potential only in higher orders. %, starting from $d^{-4}$.
127:
128: \section{Electrostatics at water interfaces}
129:
130: %\label{sec:water}
131:
132: \subsection{A toy model: water as a perfect conductor}
133:
134: %{\em A toy model: water as a perfect conductor.--}
135: For a quick insight on the
136: effect of an interface on the multipole expansion of the electrostatic potential,
137: we consider the water phase being a perfect conductor.
138: %, i.e.the Debye screening length $\kappa^{-1} \to 0$.
139: The flat interface is located at
140: $z=0$ and the colloid is modelled by
141: a fixed charge distribution $\rho_{\rm C}(\vect r)$ above the water phase. %which is zero inside the water phase ($z<0$).
142: The boundary condition at $z=0$ simply implies that there
143: is no tangential (or in--plane) electric field and the potential for $z>0$ can be obtained with the method of image charges.
144: % results from a charge configuration made of the colloid charge and a mirror image charge
145: % of opposite sign in the water phase.
146: Therefore, the effective (real + image) charge distribution is
147: spatially localized and can be enclosed in a ball of finite radius $R$
148: (see Fig.~\ref{fig:geom} with $\kappa^{-1} \to 0$). In standard
149: spherical coordinates $(s,\theta,\varphi)$ measured from the center of
150: this ball, the potential in the upper phase for $s>R$ can be written
151: as a multipole expansion (in the remainder of the paper, the $+$($-$) index
152: will refer to evaluation in the upper(lower) phase):
153: % We use standard spherical coordinates [$(x,y,z)=s\,(\sin\theta\cos\varphi,\,
154: % \sin\theta\sin\varphi,\, \cos\theta)$] and unnormalized spherical harmonics
155: % $Y_{lm}(\theta,\varphi)=P_l^m(\cos\theta)\,e^{im\varphi}$ with $P_l^m(x)$ denoting the associated
156: % Legendre polynomials.
157: % The multipole expansion for the potential in phase I ($z>0$) exists since
158: % the charge and image charge distribution is localized and it reads:
159: \bea
160: \label{eq:mp}
161: \Phi_+ (s,\theta,\varphi) %\sim \sum_{lm} \Phi_{{\rm I},lm}
162: = \sum_{\ell m} a_{\ell m} s^{-\ell-1} Y_{\ell m}(\theta,\varphi)\;,
163: \eea
164: in terms of normalized spherical harmonics $Y_{\ell m}$.
165: %
166: The boundary condition of vanishing in--plane electric field at the interface ($\theta=\pi/2$)
167: %($\propto s^{-l-2}\, Y_{lm}(\pi/2,\varphi)$)
168: implies
169: $a_{\ell m}=0$ for $\ell+m$ even. Thus, the monopole vanishes ($a_{00}=0$)
170: as well as the in--plane dipole ($a_{1\,\pm 1}=0$), and the leading decay is described generically by a dipole perpendicular to the interface ($a_{10}\neq 0$).
171: % The off--diagonal quadrupole
172: % corresponding to $a_{2\,\pm 1}$ is in general non-zero but could be reduced to zero
173: % by a specific choice of the origin and the in-plane orientation of the coordinate
174: %system. The electrostatic interaction energy of the colloid with a second one
175: % located at an in--plane distance $\vect d=(d_x,d_y)=d(\cos\varphi,\sin\varphi)$ and described by a charge density
176: % $\rho_{{\rm C},2}(\vect r)$ is given by
177: Consider a second, identical colloid located at an in--plane position
178: $\vect d=(d_x,d_y)$. %=d(\cos\varphi,\sin\varphi)$.
179: The total potential $\Phi_+$ is now the linear superposition of the
180: single--particle potentials $\Phi_+^{0}$ by each colloid, and the
181: electrostatic energy of the two--particle configuration is
182: %
183: \bea
184: % \label{eq:Ugen}
185: U = U^{0} + \int d^3 r\, \rho_{\rm C}(\vect r) \,
186: \Phi_+^{0} ({\vect r}+{\vect d})\;,
187: \eea
188: %
189: where $U^{0}$ is the energy in the limit $d\to\infty$.
190: %(``self--energy'' of the colloids).
191: Taylor expanding $\Phi_+^{0}$ about ${\vect r}={\vect 0}$ one obtains to
192: leading order in $1/d$
193: %
194: \bea
195: \label{eq:Utoy}
196: U - U^{0} \sim \frac{p_{z}^2}{2 d^{3}} \qquad (d\to\infty) ,
197: \eea
198: %
199: where $p_z =2 \int d^3 r \, z\, \rho_{\rm C}(\vect r) = a_{10}
200: \sqrt{3/(4\pi)}$ is the dipolar moment of $\rho_{\rm C}$ and its image
201: charge in the direction normal to the interface.
202: This dipole--dipole interaction energy differs by a factor of one-half from the textbook result because
203:
204: \begin{figure}
205: \begin{center}
206: \epsfig{file=fig_geom.eps, width=\columnwidth}
207: \end{center}
208: \caption{Geometrical configuration of the electrostatic problem: the
209: potential %$\Phi(\vect r)$
210: is calculated in the domain outside the sphere of radius $R$, which
211: encloses the colloidal particle. The flat interface at $z=0$
212: separates an upper dielectric phase (dielectric constant $\eps_+$)
213: from a lower electrolytic phase (dielectric constant $\eps_-$,
214: Debye screening length $\kappa^{-1}$).}
215: \label{fig:geom}
216: \end{figure}
217:
218: \subsection{Water as a conductor with linear screening}
219:
220: %{\em Water as a conductor with linear screening.--}
221: The
222: image charge construction in the case of perfectly conducting water
223: provides an intuitive explanation of
224: the origin of the normal dipole
225: and the absence of an in--plane dipole.
226: In the following we demonstrate that this finding still holds in the
227: more realistic case of water being an electrolyte and the
228: colloidal particle having an arbitrary shape, possibly protruding into
229: the region $z<0$, with given charge distribution and dielectric
230: properties.
231: %
232: Assuming linear screening, the electrostatic potential satisfies $(\Delta
233: -\kappa_\pm^2)\Phi_{\pm}(\vect r)=0$ with $\kappa_+=0$ and
234: $\kappa_-=\kappa$ being the inverse screening length in water. Using
235: standard cylindrical coordinates $(r,z,\varphi)$, we search for a solution outside
236: a ball of radius $R$ whose center is the coordinate origin and which encloses the colloid
237: (see Fig.~\ref{fig:geom})
238: with the
239: boundary conditions that the potential (i) vanishes at infinity, (ii)
240: reduces to a given potential $\Phi_R(\theta,\varphi)$ at the surface of the ball
241: $s=R$, (iii) is continuous at the interface $z=0$, and (iv) that the associated
242: electric displacement perpendicular to the interface is continuous,
243: i.e.,
244: %
245: \bea
246: \label{eq:electricD}
247: \eps_+ \left.\frac{\partial\Phi_+}{\partial z}\right|_{z=0} & = &
248: \eps_- \left.\frac{\partial\Phi_-}{\partial z}\right|_{z=0}
249: \qquad (r>R) .
250: \eea
251: %
252: The function $\Phi_R(\theta,\varphi)$ is determined by the
253: solution of the electrostatic problem inside the ball and
254: contains the relevant information on the precise geometrical and
255: electric properties of the particle.
256: %
257: By decomposing the problem in the full domain into the solution of
258: problems in simpler domains (the exterior of the sphere $s=R$ and each
259: of the halfspaces defined by $z=0$ (details can be found in
260: App.~\ref{sec:app}), one can finally write the solution as the
261: superposition $\Phi_\pm (\vect r) = \Phi^{\rm cyl}_\pm (r,z,\varphi) +
262: \Phi^{\rm sph}_\pm (s,\theta,\varphi)$, where the contribution
263: $\Phi^{\rm cyl}_\pm (r,z,\varphi)$ (using cylindrical coordinates) is given by
264: %
265: \bea
266: \label{eq:cyl}
267: \Phi^{\rm cyl}_\pm (r,z,\varphi) &=& \sum_{m=-\infty}^{+\infty} e^{i m\varphi}
268: \int_0^\infty \!\!\! dq\, A_m (q) J_{|m|}(q r) \, {\rm e}^{- K_{\pm}z} \nonumber \\
269: % & & \left( K_\pm=\pm \sqrt{q^2+\kappa_\pm^2} \right) ,
270: %(K_+=-q, \quad K_{-}=\sqrt{q^2+\kappa^2})
271: \eea
272: with $K_\pm=\pm \sqrt{q^2+\kappa_\pm^2}$,
273: and the contribution $\Phi^{\rm sph}_\pm$ (using spherical coordinates) reads
274: \bea
275: \label{eq:sph}
276: \Phi^{\rm sph}_\pm (s,\theta,\varphi) &=& \sum_{\ell=0}^\infty \sum_{m=-\ell}^{+\ell}
277: C_{\ell m}^\pm {\cal R}_\ell^\pm (s) Y_{\ell m}(\theta,\varphi) \nonumber \\
278: & & \left( {\cal R}_\ell^\pm (s) = s^{\ell} \frac{d^\ell}{(s\, ds)^\ell} \left[\frac{{\rm e}^{-\kappa_\pm s}}{s} \right] \right).
279: \eea
280: The coefficients $C_{\ell m}^\pm$ are given by
281: \bea
282: \label{eq:Ccoeff}
283: % C_{\ell m}^\pm & = & \pm [1 - (-1)^{\ell-m}]
284: % \int_0^{2\pi} d\varphi \int_{(1\mp 1) \pi/4}^{(3\mp 1)\pi/4} d\theta\, \sin\theta
285: % \, Y_{\ell m}^* (\theta,\varphi) \times\mbox{} \nonumber \\
286: % & & \mbox{}\times \left[ \Phi_R (\theta,\varphi) - \Phi_{\pm}^{\rm cyl} (r=R\sin\theta, z=R\cos \theta,\varphi) \right] ,
287: C_{\ell m}^\pm & = & [1 - (-1)^{\ell-m}]
288: \int_0^{2\pi} d\varphi \int_{0}^{\pm 1} d(\cos\theta)
289: \, Y_{\ell m}^* (\theta,\varphi) \times\mbox{} \nonumber \\
290: & & \mbox{} \left[ \Phi_R (\theta,\varphi) - \Phi_{\pm}^{\rm cyl} (r=R\sin\theta, z=R\cos \theta,\varphi) \right] , \qquad
291: \eea
292: %
293: such that $\Phi^{\rm sph}_\pm =0$ at $z=0$, $\Phi_\pm (s=R) = \Phi_R
294: (\theta,\varphi)$, and the boundary conditions (i)--(iii) are satisfied
295: automatically.
296: %
297: The coefficients $A_m(q)$ in the expression for
298: $\Phi_\pm^{\rm cyl}$ (Eq.~(\ref{eq:cyl})) must be chosen to enforce the
299: boundary condition~(\ref{eq:electricD}). This condition can be
300: extended to the region $0<r<R$ by continuing the fields
301: $\Phi_\pm(\vect r)$ with any virtual solution into the interior of the ball, $s<R$.
302: The precise form of the continuation is irrelevant, since the solution
303: outside the ball depends only on the potential at the surface of the ball,
304: $\Phi_R(\theta,\varphi)$ (Faraday's cage effect).
305: % This also means that the solution for $A_m(q)$ is not unique, but
306: % depends on the arbitrary continuation.
307: Thus, by using orthonormality and closure of the set of
308: Bessel functions $\{J_{|m|}(q r)\}_{q\in [0,\infty)}$ in the domain
309: $0<r<\infty$, Eq.~(\ref{eq:electricD}) can be solved for the
310: coefficients $A_m(q)$:
311: %
312: \bea
313: \label{eq:Acoeff}
314: A_m(q) & = & \frac{q \,\sum_{\ell=0}^\infty \left[
315: \eps_+ \hat{C}^+_{\ell m} \gamma^+_{\ell m}(q) -
316: \eps_- \hat{C}^-_{\ell m} \gamma^-_{\ell m}(q)
317: \right]}{\eps_+ q + \eps_-\sqrt{q^2+\kappa^2}} , \qquad
318: \eea
319: with %the coefficients
320: $
321: \hat{C}^\pm_{\ell m} := - {\rm e}^{-i m\varphi}
322: \partial_\theta Y_{\ell m}(\theta=\pi/2,\varphi) \, C^\pm_{\ell m}
323: $
324: and %the functions
325: $
326: \gamma^\pm_{\ell m} (q) := \int_R^{+\infty} dr\; {\cal R}^\pm_\ell(r)\, J_{|m|}(qr)
327: $,
328: which are the Hankel transforms of the radial dependence of the
329: spherical part $\Phi_\pm^{\rm sph}$ (see Eq.~(\ref{eq:sph})) continued into the region
330: $s<R$ by zero. Eq.~(\ref{eq:Acoeff}) is not the explicit expression
331: for the coefficients $A_m(q)$ because they appear implicitly also in
332: the coefficients $C^\pm_{\ell m}$, see Eq.~(\ref{eq:Ccoeff}), but it
333: does provide their dependence on $q$. In particular, for $\ell-m$ odd
334: (i.e., when $C^\pm_{\ell m}\neq 0$), the functions
335: $\gamma^\pm_{\ell m}(q)$ possess a Taylor expansion around $q=0$ with the
336: lowest term being of order $q^{|m|}$, so that %as $q\to 0$ one has
337: \begin{equation}
338: \label{eq:Aexp}
339: A_m(q) = \sum_{j=0}^\infty a_{jm} q^{j} , \qquad
340: \textrm{with $a_{jm}=0$ if } j\leq |m| .
341: \end{equation}
342: The existence of a Taylor expansion in $q$ of the coefficients $A_m(q)$
343: allows to extract the large--$r$ behavior of the potential and the $z$--component of the
344: electric field at the
345: interface. Introducing the factors
346: \bea
347: {\cal J}_{jm} & := & \lim_{z\to 0} \int_0^\infty dp \, p^j J_{|m|}(p) {\rm e}^{-z p}
348: = \frac{2^j\, \Gamma\left(\frac{|m|+j+1}{2}\right)}{\Gamma\left(\frac{|m|-j+1}{2}\right)} \nonumber
349: % = \left\{
350: % \begin{array}[c]{cl}
351: % 0 & \textrm{if $j-|m|$ odd and $j-|m|\geq 1$}, \\
352: % & \\
353: % \neq 0 & \textrm{if $j-|m|$ even or $|m|>j$} .
354: % \end{array}\right.
355: \eea
356: and inserting the expansion~(\ref{eq:Aexp}) into the
357: corresponding definitions of the fields, one
358: obtains \footnote{These are asymptotic
359: expansions. There are also exponentially decaying terms which cannot
360: be recovered from an expansion like Eq.~(\ref{eq:Aexp}).}
361: \bea
362: \Phi_\pm(r,\varphi,z=0) & \sim & \sum_{j=0}^\infty\;\sum_{m=-j+1}^{j-1}
363: \frac{{\rm e}^{i m \varphi}}{r^{j+1}} \, a_{jm} {\cal J}_{jm} , \\
364: & \sim & \mbox{} - \frac{a_{20}}{r^3} -
365: \frac{3}{r^4} \sum_\pm a_{3\pm 1} {\rm e}^{\pm i \varphi} \cdots ,
366: %\left[ a_{31}{\rm e}^{i \varphi} + a_{3-1}{\rm e}^{-i \varphi}\right] \cdots , %O(1/r^5) ,
367: \nonumber
368: \eea
369: \bea
370: \left.\frac{\partial\Phi_+}{\partial z}\right|_{z=0}
371: & \sim & \sum_{j,m}
372: % \sum_{j=0}^\infty \sum_{m=-j+1}^{j-1}
373: \frac{{\rm e}^{i m \varphi}}{r^{j+2}} \, \left[ r^{j+1} {\cal R}^+_j(r) \hat{C}^+_{jm}
374: - a_{jm} {\cal J}_{j+1,m}\right] , \nonumber \\
375: & \sim &
376: \frac{a_{10} + \hat{C}^+_{10}}{r^3} +
377: \frac{3}{r^4} \sum_\pm (a_{2\pm 1}+\hat{C}^+_{2\pm 1}) {\rm e}^{\pm i \varphi} \cdots , \nonumber
378: %\left[ (a_{21} + \hat{C}^+_{21}){\rm e}^{i \varphi} \right. \nonumber \\
379: %& & \left. \mbox{} + (a_{2-1} + \hat{C}^+_{2-1}) {\rm e}^{-i \varphi} \right] + \cdots , \nonumber
380: \eea
381: where we have used that $C^+_{jm}=0$ if $|m|=j$ and $\hat{C}^+_{jm}=0$
382: whenever ${\cal J}_{j+1,m}=0$.
383: %
384: Therefore, both the potential and the normal component of the electric field at the interface
385: are asymptotically dominated by an {\em angular--independent}
386: decay $\propto 1/r^3$; anisotropic behavior arises only in subleading terms.
387: %
388: By continuity, this conclusion also holds asymptotically for the fields at a fixed
389: height $h$ above or below the interface ($r\gg|h|$).
390:
391:
392: This result is not exclusive of the single--particle configuration: if
393: there are several particles at the interface, one can surround each of
394: them by a ball of radius $R$ and the solution $\Phi(\vect r)$ of the
395: electrostatic problem will be written as a superposition of
396: single--particle potentials determined by the total potential at the
397: surface of each ball (in general different from the potential
398: $\Phi_R(\theta,\varphi)$ in the single--particle configuration). For
399: each of these single--particle potentials the expansion~(\ref{eq:Aexp})
400: still holds, since it does not depend on the precise value of
401: the potential at the balls. \\
402:
403: \subsection{An illustrative 2d example}
404:
405: %{\em An illustrative 2d example:}
406: We calculated the electrostatic
407: potential for an inhomogeneously charged cylinder at an air--water interface
408: (see the inset of Fig.~\ref{fig1} for some definitions). Because of its two--dimensional nature,
409: this problem is amenable to a numerical treatment. Here, the multipole expansion at the interface
410: gives $\Phi_\pm (z=0) \sim a_0 \ln |x| + p_x/x + q_{xx}/x^2 + \dots $ and $\left.\partial \Phi_+/\partial z\right|_{z=0} \sim p_z/x^2 + \dots$.
411: The numerical solution for $\Phi_\pm$ shows
412: that the in--plane dipole term $\propto x^{-1}$ is absent and the asymptotic expansion
413: starts with the quadrupolar term (see Fig.~\ref{fig1}).
414: The asymptotics for $\left.\partial \Phi_+/\partial z\right|_{z=0}$ (not shown) also
415: contains the term $\propto x^{-2}$, which is interpreted as the effect of a counter--ion generated
416: dipole component $p_z$ perpendicular to the interface. These findings, most notably
417: the absence of $p_x$, match the previous ones in three dimensions. \\
418:
419:
420: \begin{figure}
421: \begin{center}
422: \epsfig{file=phi_2d.eps, width=7cm}
423: \end{center}
424: \caption{Potential along the interface for an inhomogeneously charged cylinder half--immersed in water.
425: Maxwell's equations hold in the upper phase and the cylinder, the Poisson--Boltzmann equation
426: holds in the lower phase.
427: The parameters are given in the inset, the numerical calculations have been done using
428: FEMLAB.}
429: \label{fig1}
430: \end{figure}
431:
432:
433: \subsection{The effective interaction energy}
434:
435: The free energy functional of a multiparticle configuration in the
436: linear screening approximation reads \cite{ShHo90}
437: \bea
438: F[\Phi] = \int d^3 r\, \left[ \rho_{\rm C}({\vect r}) \Phi %({\vect r})
439: - \frac{ \eps({\vect r})}{8\pi} \left[
440: \kappa^2({\vect r}) \Phi^2 + |\nabla\Phi|^2 \right]
441: \right] ,
442: \eea
443: %
444: where the charge density $\rho_{\rm C}({\vect r})$ is localized on the
445: colloidal particles. This includes the electrostatic energy as well as the entropy associated to the ion distribution.
446: The extremum of $F[\Phi]$ provides the field equation in thermal
447: equilibrium, $\nabla\cdot[\eps\nabla\Phi] = \eps\kappa^2\Phi -
448: 4\pi\rho_C$. With the help of this equation, the free energy in
449: equilibrium simplifies to
450: \bea
451: \label{eq:totalF}
452: F_{\rm eq}(\{ {\vect x}_\alpha \}) =
453: \frac{1}{2} \int d^3 r\, \rho_{\rm C}({\vect r}) \Phi ({\vect r}) ,
454: \eea
455: %
456: which is known as the ''potential of mean force'' for the degrees of
457: freedom ${\vect x}_\alpha$ (position of the center of a ball of
458: radius $R$ enclosing the $\alpha$-th particle).
459: One may decompose
460: $F_{\rm eq} = F^{0} + \delta F$, where $F^{0}$ is the equilibrium free
461: energy in the limit $|{\vect x}_\alpha - {\vect x}_\beta|\to\infty$
462: (isolated particles).
463: %
464: The total potential can be similarly written as $\Phi = \sum_\alpha
465: \Phi^{0}_\alpha + \delta\Phi$, where $\Phi^{0}_\alpha({\vect r})$
466: denotes the potential field generated by the $\alpha$-th particle in
467: isolation and $\delta \Phi({\vect r}; \{{\vect x}_\alpha\})$ is the
468: total perturbation induced by the presence of other particles. Due to
469: the linear nature of the problem, the perturbation $\delta \Phi$,
470: %
471: \bea
472: \delta \Phi ({\vect r}) = \sum_{\alpha\neq\beta} \int_{|{\vect r}'-{\vect x}_\alpha|<R} d^3 r' \,
473: G_{\alpha\beta} ({\vect r}, {\vect r'})
474: \Phi^{0}_\beta ({\vect r}') ,
475: \eea
476: %
477: can be written
478: in terms of a generalized %, operator--valued
479: susceptibility $G_{\alpha\beta}({\vect r}, {\vect r}')$ depending on
480: the precise shape and charge distribution of the particles.
481:
482: Since $\Phi^{0}({\vect r})$ near the interface exhibits asymptotically
483: an isotropic decay $\propto 1/r^3$,
484: $\delta \Phi({\vect r}; \{{\vect x}_\alpha\})$
485: depends {\em only} on $d_{\alpha\beta}= |{\vect x}_\alpha - {\vect
486: x}_\beta|$ (and not on the orientation of
487: ${\vect x}_\alpha - {\vect x}_\beta$) in the asymptotic limit
488: $d_{\alpha\beta}\to\infty$.
489: Furthermore, $\delta \Phi$ is rescaled by a factor
490: $\lambda^{-3}$ if all distances $d_{\alpha\beta}$ are rescaled
491: simultaneously by a factor $\lambda$.
492: From Eq.~(\ref{eq:totalF})
493: the same property holds for $\delta F (\{{\vect x}_\alpha\})$.
494: In particular, for a
495: two--particle configuration this yields an asymptotic
496: potential of mean force of the form
497: %
498: \bea
499: \label{eq:U2part}
500: F_{\rm eq} (d) - F^0 \sim \frac{B}{d^3} \qquad (d\to\infty) ,
501: \eea
502: %
503: and the constant $B$ is positive for like particles.
504: In analogy with Eq.~(\ref{eq:Utoy}), it is natural to
505: interpret this expression as the interaction energy
506: between two effective dipoles perpendicular to the interface.
507:
508:
509: \section{Discussion and Conclusion}
510:
511: %{\em Discussion and Conclusion.--}
512: We have shown that the form of the multipole expansion of the potential around a charged colloid
513: and of the effective interaction energy between two colloids trapped at a water interface
514: is qualitatively different
515: %leads to certain restrictions in powers and orientation dependence of the multipole terms
516: from the situation in bulk.
517: The dominating interaction terms can be qualitatively understood by assuming
518: water to be a perfect conductor.
519: %If we assume the colloidal shape to be symmetric with respect to
520: %rotations around the interface normal,
521: The leading interaction term between the colloids a distance $d$ apart is of dipole--dipole type
522: ($\propto d^{-3}$) and isotropic in the
523: interfacial plane.
524: In other words, even if the charges on the colloid surface are distributed
525: arbitrarily
526: the counterions arrange themselves such that asymptotically
527: the configuration corresponds to an effective dipole perpendicular to the interface.
528: Orientation--dependent interactions and thus possible attractions for like--charged
529: colloids only arise
530: in subleading order.
531:
532: This is in marked contrast to the analysis of the experiment reported in
533: Refs.~\cite{Che05, Che06}.
534: Motivated by the experimentally found inhomogenous surface charge, it was
535: pictorially suggested (see Fig.~1 in Ref.~\cite{Che05})
536: that spontaneous fluctuations in the colloid's orientation would generate
537: (via an instantaneously equilibrating counterion cloud)
538: effective in--plane dipoles $\vect p_i$ with corresponding interactions
539: $\propto [d^2 ({\vect p}_1\cdot{\vect p}_2) -3 ({\vect d}\cdot {\vect p}_1)({\vect d}\cdot {\vect p}_2)]/d^5$.
540: %\bea
541: % \label{eq:ufluc}
542: % U_{\rm fluc}(d) \propto \frac{ -
543: % 3 ({\vect d}\cdot {\vect p}_1)({\vect d}\cdot {\vect p}_2)}{d^5}
544: %\eea
545: After averaging over the orientation fluctuations, such an interaction would lead to
546: an effective, isotropic attraction competing with the isotropic dipole--dipole
547: repulsion. According to the model worked out by the authors, the total interaction potential would exhibit an attractive minimum due to
548: the effect of the fluctuating in--plane dipoles at rather small distances ($d \simeq 2.2$ colloid radii $R_{\rm C}$, so small that already the use of a pure dipole--dipole interaction casts serious doubts on the reliability of the model).
549: %
550: The analysis in Ref.~\cite{Zho07} purported to support this picture is
551: actually incomplete and just states that no monopolar term arises,
552: without entering into a systematic analysis of constraints on higher
553: order multipoles. In another note \cite{Zho07a} the existence of
554: the Taylor expansion of the coefficients $A_m(q)$ around $q=0$ (see
555: Eq.~(\ref{eq:Aexp})) was doubted on which the asymptotic analysis of the
556: electrostatic potential and field is based. The present explicit
557: proof of the analyticity of the coefficients $A_m(q)$ should disperse such doubts.
558:
559:
560:
561: The results of our work imply that
562: asymptotically an in--plane dipolar interaction cannot arise if the
563: counterions are equilibrated (see Eq.~(\ref{eq:U2part})). Consequently one cannot
564: expect asymptotically relevant attractions from the orientational fluctuations of the
565: colloids. However, for smaller $d$ the asymptotic $1/d$ expansion is likely to break
566: down. For small colloid radius, $R_{\rm C} \ll \kappa^{-1}$, this becomes relevant when
567: $d \sim \kappa^{-1}$: in this case the screening clouds of the colloids overlap
568: and the interaction falls off exponentially with $d$ before crossing over to the algebraic decay \cite{Hur85,Dom07}. For large
569: colloid radius, $R_{\rm C} \gg \kappa^{-1}$, the precise shape and charge distribution of the colloids will determine the interaction whenever $d\sim R_{\rm C}$.
570: % effect of the ``holes" dug out by the
571: % colloids become relevant if the shortes distance between the colloid surfaces $d-2R \sim R$
572: % (see the discussion at the end of Sec.~\ref{sec:water}).
573: Certainly, for both regimes a more elaborate numerical analysis of the anisotropy
574: in the electrostatic interactions is required to assess whether fluctuations in the
575: orientation of the colloids may lead to attractions.
576: However, even in that case their existence is doubtful looking at the general
577: results on the absence of like--charge attraction in confined geometries \cite{Tri99}.
578: % In any case, the thermal
579: % average will again lead to isotropy in the attractive terms, as opposed to the
580: % claims of orientation--sensitivity made in Refs.~\cite{Che05,Che06}.
581: In any case, the results from the model studied in
582: Refs.~\cite{Che05,Che06} are not reliable since the model presupposes
583: an interaction energy which does not satisfy the correct
584: asymptotic decay given by Eq.~(\ref{eq:U2part}).
585:
586: %\vfill
587:
588:
589:
590: \begin{appendix}
591:
592: \section{Solution of the electrostatic problem}
593:
594: \label{sec:app}
595:
596: We consider the potential $\Phi_\pm({\vect r})$ in the domain shown in
597: Fig.~\ref{fig:geom} given as the solution to the following problem :
598: \begin{eqnarray*}
599: \nabla^2 \Phi_+ & = & 0 , \qquad {\vect r} \in \{ s>R, z>0 \} , \\
600: \nabla^2 \Phi_- & = & \kappa^2 \Phi_- , \qquad {\vect r} \in \{ s>R, z<0 \} , \\
601: \Phi_+ (s=R, \theta, \varphi) & = & \Phi_R (\theta,\varphi) ,
602: \qquad 0<\theta<\frac{\pi}{2} , \\
603: \Phi_- (s=R, \theta, \varphi) & = & \Phi_R (\theta,\varphi) ,
604: \qquad \frac{\pi}{2}<\theta<\pi , \\
605: \Phi_+ (r,\varphi,z=0) & = & \Phi_- (r,\varphi,z=0), \qquad r>R , \\
606: \epsilon_+ \left.\frac{\partial\Phi_+}{\partial z}\right|_{z=0} & = &
607: \epsilon_- \left.\frac{\partial\Phi_-}{\partial z}\right|_{z=0}, \qquad
608: r>R , \\
609: |\Phi({\vect r})| & < & \infty, \qquad |{\vect r}|\to\infty .
610: \end{eqnarray*}
611: Here, $\Phi_R(\theta,\varphi)$ is the potential at the surface of the
612: ball $s=R$ and is assumed to be given. In order to solve this problem,
613: we split it in two auxiliary problems,
614: one for each halfspace: \\
615:
616: \begin{figure}[h]
617: \begin{center}
618: \epsfig{file=fig_geomUP.eps, width=7cm}
619: \end{center}
620: \caption{Domain of definition of the problem {\bf UP}.}
621: \label{fig:geomUP}
622: \end{figure}
623: \noindent %
624: {\bf Problem UP} in the domain $s>R$ and $z>0$, see
625: Fig.~\ref{fig:geomUP}:
626: \begin{eqnarray*}
627: \nabla^2 \Phi_+ & = & 0 , \qquad {\vect r} \in \{ s>R, z>0 \} , \\
628: \Phi_+ (s=R, \theta, \varphi) & = & \Phi_R (\theta,\varphi) ,
629: \qquad 0<\theta<\frac{\pi}{2} , \\
630: \Phi_+ (r,\varphi,z=0) & = & F_0 (r,\varphi),
631: \qquad r>R , \\
632: |\Phi_+ ({\vect r})| & < & \infty, \qquad |{\vect r}|\to\infty .
633: \end{eqnarray*}
634: \newline
635:
636: \begin{figure}[h]
637: \begin{center}
638: \epsfig{file=fig_geomLOW.eps, width=7cm}
639: \end{center}
640: \caption{Domain of definition of the problem {\bf LOW}.}
641: \label{fig:geomLOW}
642: \end{figure}
643: \noindent %
644: {\bf Problem LOW} in the domain $s>R$ and $z<0$, see
645: Fig.~\ref{fig:geomLOW}:
646: \begin{eqnarray*}
647: \nabla^2 \Phi_- & = & \kappa^2 \Phi_- , \qquad {\vect r} \in \{ s>R, z<0 \} , \\
648: \Phi_- (s=R, \theta, \varphi) & = & \Phi_R (\theta,\varphi) ,
649: \qquad \frac{\pi}{2}<\theta<\pi , \\
650: \Phi_- (r,\varphi,z=0) & = & F_0 (r,\varphi),
651: \qquad r>R , \\
652: |\Phi_- ({\vect r})| & < & \infty, \qquad |{\vect r}|\to\infty .
653: \end{eqnarray*}
654: Here $F_0(r,\varphi)$ (=potential at the interface $z=0$) is an
655: auxiliary function which will be eventually determined by the boundary
656: condition~(\ref{eq:electricD}). Each of these problems can in turn be
657: decomposed in simpler problems, one with boundary conditions imposed
658: only at the plane $z=0$ and one with boundary conditions imposed only
659: at the ball $s=R$: \\
660:
661: \begin{figure}[h]
662: \begin{center}
663: \epsfig{file=fig_geomUPcyl.eps, width=7cm}
664: \end{center}
665: \caption{Domain of definition of the problem {\bf UP-cyl}.}
666: \label{fig:geomUPcyl}
667: \end{figure}
668: \noindent %
669: {\bf Problem UP-cyl} in the domain $z>0$, see Fig.~\ref{fig:geomUPcyl}:
670: \begin{eqnarray*}
671: \nabla^2 \Phi_+^{\rm cyl} & = & 0 , \qquad {\vect r} \in \{ z>0 \} , \\
672: \Phi_+^{\rm cyl} (r,\varphi,z=0) & = & F_0 (r,\varphi)
673: + {\cal F}_0 (r,\varphi) ,
674: \qquad 0< r , \\
675: |\Phi_+^{\rm cyl} ({\vect r})| & < & \infty, \qquad |{\vect r}|\to\infty .
676: \end{eqnarray*}
677: \newline
678: \begin{figure}[h]
679: \begin{center}
680: \epsfig{file=fig_geomUPsph.eps, width=7cm}
681: \end{center}
682: \caption{Domain of definition of the problem {\bf UP-sph}.}
683: \label{fig:geomUPsph}
684: \end{figure}
685: \noindent %
686: {\bf Problem UP-sph} in the domain $s>R$, see Fig.~\ref{fig:geomUPsph}:
687: \begin{eqnarray*}
688: \nabla^2 \Phi_+^{\rm sph} & = & 0 , \qquad {\vect r} \in \{ s>R \} , \\
689: \Phi_+^{\rm sph} (R,\theta,\varphi) & = & \Phi_R (\theta,\varphi) -
690: \Phi_+^{\rm cyl} (R,\theta,\varphi) ,
691: \quad 0 < \theta < \frac{\pi}{2} , \\
692: \Phi_+^{\rm sph} (R,\theta,\varphi) & = & \mbox{} -
693: \Phi_+^{\rm sph} (R,\pi-\theta,\varphi) ,
694: \qquad \frac{\pi}{2} < \theta < \pi , \\
695: |\Phi_+^{\rm sph} ({\vect r})| & < & \infty,
696: \qquad |{\vect r}|\to\infty .
697: \end{eqnarray*}
698: Here the function ${\cal F}_0 (r,\varphi)$ verifies ${\cal F}_0
699: (r>R,\varphi)=0$ and is an otherwise arbitrary smooth function which
700: continues the potential $F_0 (r,\varphi)$ into the region $r<R$ of the
701: plane $z=0$. As discussed in the main text, ${\cal F}_0 (r,\varphi)$
702: is just an intermediary auxiliary function whose precise choice is
703: ultimately irrelevant for the determination of the total potential
704: outside the ball $s=R$.
705: %
706: With the choice of boundary condition for $\Phi_+^{\rm sph}$ at $s=R$
707: it is clear that $\Phi_+^{\rm sph}=0$ at $z=0$ and therefore
708: \begin{displaymath}
709: \Phi_+ ({\vect r}) = \Phi_+^{\rm cyl} + \Phi_+^{\rm sph} , \qquad
710: {\rm if}\quad {\vect r} \in \{ s>R, z>0 \}.
711: \end{displaymath}
712: Analogously, the problem {\bf LOW} can be decomposed into a problem
713: {\bf LOW-cyl} and a problem {\bf LOW-sph} and
714: \begin{displaymath}
715: \Phi_- ({\vect r}) = \Phi_-^{\rm cyl} + \Phi_-^{\rm sph} , \qquad
716: {\rm if}\quad {\vect r} \in \{ s>R, z<0 \}.
717: \end{displaymath}
718: Each of these simpler problems is now amenable to an analytical
719: solution, provided by Eqs.~(\ref{eq:cyl})--(\ref{eq:Ccoeff}). The
720: auxiliary functions $F_0$ and ${\cal F}_0$ are absorbed in the unknown
721: coefficients $A_m(q)$ in Eq.~(\ref{eq:cyl}), which are then determined
722: by Eq.~(\ref{eq:electricD}), the only boundary condition of the
723: original problem not taken into account by the stepwise process of
724: decomposing the problem into simpler ones.
725:
726: \end{appendix}
727:
728:
729:
730:
731:
732:
733:
734: \vspace*{1cm}
735:
736: \begin{thebibliography}{99}
737:
738: \bibitem{Stil61} F.~H.~Stillinger Jr.,
739: % The electrostatic interaction bteween interfacial colloidal particles,
740: J.~Chem.~Phys.~{\bf 35}, 1584 (1961).
741:
742: \bibitem{Hur85} A.~Hurd,
743: % The electrostatic interaction bteween interfacial colloidal particles,
744: J.~Phys.~A {\bf 18}, L1055 (1985).
745:
746: \bibitem{Ave00a} R.~Aveyard {\em et al.},
747: % R.~Aveyard, J.~H.~Clint, D.~Nees, and V.~N.~Paunov,
748: % Compression and structure of monolayers of charged latex particles at air/water and octane/water interfaces,
749: Langmuir {\bf 16}, 1969 (2000).
750:
751: %\bibitem{Ghe97} F.~Ghezzi and J.~C.~Earnshaw,
752: % Formation of meso--structures in colloidal monolayers,
753: % J.~Phys.: Condens.~Matt. {\bf 9}, L517 (1997).
754:
755: %\bibitem{Ghe01} F.~Ghezzi {\em et al.},
756: % F.~Ghezzi, J.~C.~Earnshaw, M.~Finnis, and M.~McCluney,
757: % Pattern Formation in colloidal monolayers at the Air-Water Interface,
758: % J.~Colloid Interface Sci. {\bf 238}, 433 (2001).
759:
760: %\bibitem{Gar98a} J.~Ruiz-Garc\'\i a and B.~I.~Ivlev,
761: % Formation of colloidal clusters and chains at the air/water interface,
762: % Mol.~Phys. {\bf 95}, 371 (1998).
763:
764: %\bibitem{Gar98b} J.~Ruiz-Garc\'\i a, R. G\'amez-Corrales, and B.~I.~Ivlev,
765: % Formation of two-dimensional colloidal voids, soap froths, and clusters,
766: % Phys.~Rev.~E {\bf 58}, 660 (1998).
767:
768: %\bibitem{Que01} M.~Quesada--P\'erez {\em et al.},
769: % M.~Quesada--P\'erez, A.~Moncho--Jord\'a, F.~Mart\'\i nez--L\'opez, and R.~Hidalgo--Alvarez,
770: % Probing interaction forces in colloid monolayers: Inversion of structural data,
771: % J.~Chem.~Phys. {\bf 115}, 10897 (2001).
772:
773: \bibitem{Gom05} O.~G{\'o}mez-Guzm{\'a}n and J.~Ruiz-Garc{\'i}a,
774: % Attractive interactions between like--charged colloidal particles at the air/water interface,
775: J.~Colloid Interface Sci. {\bf 291}, 1 (2005).
776:
777: \bibitem{Che05} W.~Chen, S.~Tan, T.-K.~Ng, W.~T.~Ford, and P.~Tong,
778: % Long--ranged attractions between charged polystyrene spheres at aqueous interfaces,
779: Phys.~Rev.~Lett. {\bf 95}, 218301 (2005).
780:
781: \bibitem{Che06} W.~Chen, S.~Tan, T.-K.~Ng, W.~T.~Ford, and P.~Tong,
782: %
783: Phys.~Rev.~E {\bf 74}, 021406 (2006).
784:
785: \bibitem{Fer04} J.~C.~Fern{\'a}ndez--Toledano, A.~Moncho--Jord{\'a}, F.~Mart{\'i}nez--L{\'o}pez, and
786: R.~Hidalgo--Alvarez,
787: % Spontaneous formtaion of of mesostructures in colloidal monolayers trapped at the air-water interface: A simple explanation,
788: Langmuir {\bf 20}, 6977 (2004).
789:
790:
791: %\bibitem{Sta00} D.~Stamou, C.~Duschl, and D.~Johannsmann,
792: % Long--range attraction between colloidal spheres at the air--water interface: the consequence of an irregular meniscus,
793: % Phys.~Rev.~E {\bf 62}, 5263 (2000).
794:
795: \bibitem{For04} L.~Foret and A.~W{\"u}rger,
796: % Electric--field induced capillary interaction of charged particles at a polar interface,
797: Phys.~Rev.~Lett.~{\bf 92}, 058302 (2004). %cond-mat/0310657.
798:
799: \bibitem{Oet05} M.~Oettel, A.~Dom{\'i}nguez, and S.~Dietrich,
800: % Effective capillary interaction of spherical particles at fluid interfaces,
801: Phys.~Rev.~E {\bf 71}, 051401 (2005). %cond-mat/0411329.
802:
803: \bibitem{Oet05a} M.~Oettel, A.~Dom{\'i}nguez, and S.~Dietrich,
804: % Attractions between charged colloids at water interfaces,
805: J.~Phys.: Condens.~Matter {\bf 17}, L337 (2005).
806:
807: \bibitem{Wue05} A.~W{\"u}rger and L.~Foret,
808: % Capillary atraction of colloidal particles at an aqueous interface,
809: J.~Phys.~Chem.~B {\bf 109} 16435 (2005).
810:
811: \bibitem{Dom07} A.~Dom{\'i}nguez, M.~Oettel, and S.~Dietrich
812: % Theory of ...
813: J.~Chem.~Phys., in press (2007).
814: % preprint {\tt arXiv:0706.0013}.% [cond-mat.soft]}.
815:
816:
817: % \bibitem{Gra81} I.~Gradstein and I.~P.~Ryshik, {\em Tables}, Deutsch Frankfurt 1981, 6.621.
818:
819: \bibitem{ShHo90} K.~A.~Sharp and B.~Honig,
820: %Calculating total electrostatic energies with the nonlinear Poisson-Boltzmann equation,
821: J.~Phys.~Chem {\bf 94}, 7684 (1990).
822:
823: \bibitem{Zho07} Y.~Zhou and T.-K.~Ng,
824: % Origin of long--ranged attraction between like-charged particles at water-air interface,
825: {\tt arXiv:cond-mat/0703667}.
826:
827: \bibitem{Zho07a} T.-K.~Ng and Y.~Zhou,
828: % Comment to "Multipolar expansion of the electrostatic interaction between charged colloids at interfaces",
829: {\tt arXiv:0708.2518}.
830:
831: \bibitem{Tri99} E.~Trizac and J.-L.~Raimbault,
832: % Long-range electrostatic interactions between like-charged colloids: Steric and confinement effects,
833: Phys.~Rev.~E {\bf 60}, 6530 (1999).
834:
835: \end{thebibliography}
836:
837:
838: \end{document}
839: