1: %\documentclass[aps,preprint,groupedaddress,showpacs]{revtex4}

2: %\documentclass[aps,preprint,superscriptaddress]{revtex4}

3: \documentclass[aps,twocolumn,groupedaddress,showpacs]{revtex4}

4: %\documentclass[aps,twocolumn,superscriptaddress,showpacs]{revtex4}

5: %\documentclass[twocolumn,prb,groupedaddress,showpacs,preprintnumbers,amsmath,amssymb,floatfix]{revtex4}

6: \usepackage{amsmath}

7: \usepackage{amssymb}

8: \usepackage{amsfonts}

9: \usepackage[dvips]{graphicx}

10: \usepackage[]{caption}

11: \usepackage[]{epsfig}

12: \bibliographystyle{apsrev}

13: \begin{document}

14: 

15: \title{Reduction of the hydrophobic attraction between charged solutes

16: in water}

17: 

18: \author{J. Dzubiella}

19: \email[e-mail address: ] {jd319@cam.ac.uk}

20: \affiliation{University Chemical Laboratory,

21: Lensfield Road,

22: Cambridge CB2 1EW,

23: United Kingdom}

24: \author{J.-P. Hansen}

25: \affiliation{University Chemical Laboratory,

26: Lensfield Road,

27: Cambridge CB2 1EW,

28: United Kingdom}

29: \date{\today}

30: 

31: \begin{abstract}

32: We examine the effective force between two nanometer scale solutes in

33: water by Molecular Dynamics simulations. Macroscopic considerations

34: predict a strong reduction of the hydrophobic attraction between

35: solutes when the latter are charged. This is confirmed by the

36: simulations which point to a surprising constancy of the effective

37: force between oppositely charged solutes at contact, while like

38: charged solutes lead to significantly different behavior between

39: positive and negative pairs. The latter exhibit the phenomenon

40: of ``like-charge attraction" previously observed in some colloidal

41: dispersions. 

42: \end{abstract}

43: 

44: \pacs{82.70.Uv,87.16.Ac,61.20Ja,68.08.Bc}

45: 

46: \maketitle

47: When apolar solutes are dispersed in water, they exert effective,

48: solvent induced forces on each other, generally referred to as

49: hydrophobic attraction. For large solutes, say globular proteins, the

50: mechanism for hydrophobic interactions can be traced back to solvent

51: depletion (or ``drying") from the volume bounded by opposite surfaces

52: of the solutes \cite{wallquist:jpc,lum:jpc}. This mechanism, which is

53: reminiscent of polymer-induced depletion attraction between colloidal

54: particles on larger (mesoscopic) scales \cite{likos:physrep}, is most

55: pronounced when the solvent is near liquid-vapour coexistence

56: \cite{lum:jpc}. However, many biomolecular solutes are charged, and in

57: this work we present evidence, based on extensive Molecular Dynamics

58: (MD) simulations, that the electric field due to the charges carried

59: by the solutes leads to a considerable reduction of the hydrophobic

60: attraction. 

61: 

62: Such a reduction may be qualitatively understood from macroscopic

63: considerations. It is a well-known fact that a polar liquid, like

64: water, will rise inside a vertical condenser to minimize the overall

65: electrostatic energy (the rise being limited by gravity on a

66: macroscopic scale) \cite{landau}. Transposing this situation to the

67: nanometric scale, consider two parallel plate-like solutes of area

68: $A$, separated by a distance $D$, inside a polar solvent of relative

69: dielectric permittivity $\epsilon$, carrying opposite surface charges

70: $\pm\sigma$. Neglecting edge effects, the electric field between the

71: plates is $E_{0}/\epsilon$ with $E_{0}=\sigma/2\epsilon_{0}$. We require the

72: difference in the grand potential between the situations where the liquid

73: solvent ($l$) or its vapour ($g$) fill the volume $AD$ between the

74: two plates:

75: \begin{eqnarray}

76: \Omega_{\alpha}=-P_{\alpha} AD + 2\gamma_{w\alpha}A +

77: \frac{1}{2}\epsilon_{0}\frac{E^{2}_{0}}{\epsilon_{\alpha}}AD;\;\;\;\;\; \alpha=l,g

78: \end{eqnarray}

79: where $P_{\alpha}$ is the pressure of phase $\alpha$ and

80: $\gamma_{w\alpha}$ the surface tension between phase $\alpha$ and the

81: plate (``wall"). Consider a state close to phase coexistence at

82: temperature $T$, and let $\delta\mu=\mu-\mu_{\rm sat}$ be the positive

83: deviation of the chemical potential from its saturation value. Expanding

84: the $P_{\alpha}$ to linear order in $\delta\mu$ around their common

85: value at saturation, one easily arrives at the following expression for

86: the difference in grand potentials per unit area:  

87: \begin{eqnarray}

88: \frac{\Omega_{l}-\Omega_{g}}{A}&=&(\rho_{ g}-\rho_{

89:   l})\delta\mu D+2(\gamma_{ w l}-\gamma_{w

90:   g})\\&+&\frac{\epsilon_{0}}{2}E^{2}_{0}(\frac{1}{\epsilon_{

91:   l}}-\frac{1}{\epsilon_{g}})D \nonumber

92: \end{eqnarray}

93: At the ``drying" transition between the plates,

94: $\Omega_{l}-\Omega_{g}=0$ and

95: $\gamma_{wl}-\gamma_{wg}=\gamma_{lg}\equiv\gamma$. Since

96: $\rho_{g}\ll\rho_{l}$, and

97: $\epsilon_{l}\equiv\epsilon\gg\epsilon_{g}\simeq 1$, eq. (2) yields

98: the following expression for the critical distance $D_{\rm c}$ between the

99: plates at which drying occurs:

100: \begin{eqnarray} 

101: D_{\rm c}\simeq\frac{2\gamma}{\rho_{l}\delta\mu+\frac{\epsilon_{0}}{2}E^{2}_{0}}

102: \end{eqnarray}

103: For strong electric fields, ($E_{0}\lesssim 10^{10}$V/m corresponding

104: to surface charges $\sigma\lesssim e/{\rm nm}^{2}$), the electrostatic

105: term in the denominator is typically 10 times larger than the

106: $\delta\mu$ term in the vicinity of gas-liquid coexistence, and

107: leads to values of $D_{\rm c}$ of the order of a few \AA. This strong

108: reduction of $D_{\rm c}$ hints at a considerable weakening of the hydrophobic

109: interaction between two solutes when the latter are charged. Note that

110: within our macroscopic model this reduction is due to the overall

111: electric field, not to any Hydrogen-bonding of the solvent molecules

112: to hydrophilic ``patches" on the solute surface.

113: 

114: In order to confirm the qualitative prediction of the schematic model,

115: we have carried out extensive MD simulations of two spherical solutes

116: immersed in a bath of SPC/E water molecules \cite{berendsen:jpc}. The

117: solutes are spheres of radius $R$ which repel the solvent

118: molecules by a repulsive $\epsilon (r-R)^{-12}$, where $r$ is the distance from

119: the solute center to the oxygen atom of a water molecule;

120:  the energy scale $\epsilon$ is chosen such that the O atom

121:  experiences an energy $k_{\rm B}T$ at a distance of $r-R=$1\AA~ from 

122: the solute surface. The simulation cell is a cube of length up to

123: $L=40$\AA, containing up to 2000 water molecules, depending on the

124: solute size; solutes are placed at fixed positions on the body diagonal of the

125: simulation cell. The box dimensions are chosen such that the surface

126: to surface distance to the nearest image solute is at least 20\AA. 

127: 

128: The MD simulations are carried out with the DLPOLY2 package

129: \cite{dlpoly}, using the Verlet algorithm \cite{frenkelsmit}, with a

130: timestep of 2fs. The Berendsen barostat and thermostat

131: \cite{berendsen:jcp} were used to maintain the SPC/E water at a

132: pressure of 1 bar and a temperature $T=300$K. All electrostatic

133: interactions were calculated using particle-mesh Ewald summations

134: \cite{essmann:jcp}. 

135: \begin{figure}

136:   \begin{center}

137: \includegraphics[width=8.6cm]{fig1.ps}

138:     \caption{Density profiles of the water molecules around two

139:       neutral spherical solutes of radius $R=10$\AA~ (a) and two oppositely

140:       charged spherical solutes of radius $R=10$\AA~ carrying a

141:       charge $\pm qe$ with (b) $q=2$, (c) $q=5$, and (d)

142:       $q=10$. The surface-to-surface distance in all cases is

143:       $s=4$\AA. In the contour plots  dark regions show low density

144:       regions while high densities are plotted bright. The panels

145:       below the contour plots show the water density $\rho$ scaled

146:       with water bulk density $\rho_{0}$ in a cylinder of radius

147:       $R_{c}=5$\AA, coaxial with the center-to-center line of the

148:       solutes.}    

149: \label{fig1}

150: \end{center}

151: \end{figure}

152: We first consider the case of uncharged solutes. The water density

153: profile is illustrated in frame (a) of Fig. 1 for the case of solutes

154: of radius $R=10$\AA~ and a surface-to-surface distance along the $z$-axis

155: joining the centers $s=4$\AA. The upper part of the frame shows

156: a density contour plot coded by variable shades of grey. The lower part

157: shows density profiles along the center-to-center to axis $z$, averaged over a

158: coaxial-cylindrical volume of radius $5$\AA. The density profiles

159: show a considerable depletion of the solvent within a radial distance

160: of 5\AA~ from the center-to-center axis, reminiscent of the

161: observations of Wallquist 

162: and Berne for flatter solutes \cite{wallquist:jpc}.  As the

163: surface-to-surface distance $s$ is increased for fixed radius $R$, the

164: water molecules penetrate into the region between opposite solute

165: surfaces, as signalled by a rapid increase of the central peak (around

166: $z=0$) in the density profiles. When $s\approx6.5$\AA, the solvent

167: layers around an isolated solute are hardly disturbed by the presence

168: of the other solute.

169: 

170: \begin{figure}

171:   \begin{center}

172: \includegraphics[width=8cm,angle=0.,clip]{fig2.eps}

173:     \caption{Simulation results (symbols) of the mean force,

174:       $F^{*}=\beta F$\AA, between

175:       two neutral ($q=0$) spheres in SPC/E water. Error bars 

176:       are omitted for clarity; the error is estimated to be $\Delta

177:       F^{*}=0.4$, comparable to the symbol size. The

178:       lines are guide to the eye. Results are plotted for solute

179:       radii  $R=3$\AA~ (circles), $R=5$\AA~ (squares), $R=8$\AA~

180:       (diamonds), $R=10$\AA~ (triangles pointing up), and $R=12$\AA~ (triangles

181:       pointing left). The inset shows the integrated force (potential of mean

182:       force) in obvious order.}  

183: \label{fig2} 

184: \end{center}

185: \end{figure}

186: The mean effective force acting on each of the solutes in the presence

187: of the second at a surface-to-surface distance $s$ is calculated by

188: averaging the total force due to all solvent molecules over the

189: configurations generated by MD runs extending over typically 1ns. This

190: average force obviously goes to zero at large distances $s$ and for

191: symmetry reasons, it is directed along the center-to-center

192: axis. Examples for several radii 3\AA~$\leq R \leq 12$\AA~ are shown in

193: Fig. 2. The largest radii are of the order of the size of small

194: globular proteins or of oil-in-water micelles. As expected from a

195: depletion mechanism, the force is attractive and its contact values

196: and range increase with $R$. The potentials of mean force $w(s)$ may

197: be calculated for each $R$ by integrating the force. The resulting

198: potentials are shown in the inset to Fig. 2. They closely resemble

199: results obtained for polymer-induced depletion potentials between

200: spherical colloids, albeit on different length and energy scales

201: \cite{wallquist:jpc,louis:jcp}. Note that the force at contact,

202: $F(0)$, scales roughly with $R$. This may be rationalized by a simple

203: consideration of the potential of mean force for plates ($R=\infty$)

204: at contact, $w(0)=-2\gamma$ (where $\gamma$ is now the plate-solvent

205: surface tension which differs little from the liquid-gas surface

206: tension \cite{huang:jpc}) and an application of the Derjaguin

207: approximation, valid for weakly curved substrates (i.e. large $R$)

208: \cite{louis:jcp}; this leads to the estimate $F(s=0)=-2\pi R\gamma$,

209: which indeed predicts linear scaling; the value $\gamma\simeq 0.05\,{\rm

210: J/m^{2}}$, extracted from the MD value of $F(0)$ for the largest $R$,

211: is reasonably close to the liquid-vapour surface tension of  

212: water under normal conditions ($\gamma=0.073 \,{\rm J/m^{2}}$). The

213: agreement must be considered quite satisfactory in view of the

214: roughness of the estimate. 

215: 

216: \begin{figure}

217:   \begin{center}

218: \includegraphics[width=8cm,angle=0.,clip]{fig3.eps}

219:     \caption{Density profiles of water oxygen and hydrogen atoms

220:       (inset) around one isolated solute with radius $R=10$\AA~ and central charge $q=0$

221:       (circles), $q=-10$ (squares), and $q=10$ (diamonds).}

222: \label{fig3}

223: \end{center}

224: \end{figure}

225: 

226: \begin{figure}

227:   \begin{center}

228: \includegraphics[width=8cm,angle=0.,clip]{fig4.eps}

229:     \caption{Same as Fig. 2 but now the mean force, $F^{*}=\beta F$\AA,

230:       is for oppositely charged spheres of

231:       $R=10$\AA~ and different central charges $\pm qe$ are shown:

232:       $q=0$ (circles), $q=2$ (squares), $q=5$ (diamonds), $q=10$

233:       (triangles up). The dashed line represents the electrostatic

234:       force between 2 periodically repeated solutes with opposite

235:       charges $q=\pm$ 10 in a continuous solvent with permittivity

236:       $\epsilon=80$.  The inset shows the resulting potentials of mean

237:       force; the contact values $w(s=0)$ increase with $q$.}

238: \label{fig4}

239: \end{center}

240: \end{figure}

241: We now turn to the charged solutes. Frames (b)-(d) in Fig. 1 show

242: water density profiles in the vicinity of two spheres carrying

243: opposite electric charges $\pm qe$ at their center (opposite charges

244: ensure overall charge neutrality without any need for counterions). As

245: $q$ increases from zero (frame (a)), water is seen to penetrate between

246: the two solutes, the central peak around $z=0$ in the density profiles

247: increases rapidly and its amplitude reaches roughly the bulk density

248: of water when $q=10$. Note that this central peak is asymmetrically

249: split, indicating the presence of two hydration layers which differ

250: somewhat depending on their association with the anionic or cationic

251: solute. This difference is also evident in the contact values of the

252: outside surfaces of the solutes, and is a consequence of

253: the different arrangements of the water dipoles around the solutes

254: induced by the local electric fields. The asymmetry of the profiles

255: can be rationalized by inspecting the density profiles of O and H

256: atoms around isolated solutes, plotted in Fig. 3. The hydration shell 

257: is more sharply defined  around the cationic than around the anionic

258: solute. The water dipoles tend to point radially away from the cation,

259: while the opposite configuration is more favourable around

260: anions. Note also that driving water into a narrowly confined region

261: under the action of a strong electric field (here between oppositely

262: charged solutes) is an effect reminiscent of that observed 

263: in recent MD simulations of ion permeation of hydrophobic nanopores

264: \cite{joe:channel}.  

265: 

266: The resulting mean forces between solutes are plotted for $q=0,2,5$

267: and 10, as functions of the surface-to-surface distance $s$ in

268: Fig. 4 together with corresponding potentials of mean force. The mean

269: force includes the direct Coulomb interaction between the two solutes

270: (with proper account for the periodic images), which is in fact an

271: order of magnitude larger that the total mean force. At large

272: distances hydrophobic interactions become negligible and the force

273: should tend to $-q^{2}e^{2}/(4\pi\epsilon_{0}\epsilon r^{2})$, where

274: $r=2R+s$ and $\epsilon$ is the dielectric permittivity of bulk SPC/E

275: water; the corresponding curve is also shown in Fig. 4.

276: 

277: \begin{figure}

278:   \begin{center}

279: \includegraphics[width=8cm,angle=0.,clip]{fig5.eps}

280:     \caption{Same as Fig. 2 but now the mean force, $F^{*}=\beta

281:       F$\AA, is for different signs of

282:       the  charged spheres of radius $R=10$\AA~ and fixed central charges

283:       $qe=\pm 5e$ are shown: oppositely charged (+ --,circles), equally

284:       charged with $q=-5$ (-- --,squares), equally charged with $q=5$

285:       (++,diamonds).}  

286: \label{fig5}

287: \end{center}

288: \end{figure}

289: The most striking result illustrated in Fig. 4 is the near

290: independence of the force at contact, $s=0$, with respect to

291: solute charge. From the density profiles in Fig. 1 the hydrophobic

292: attraction is expected to be reduced but this reduction is almost

293: exactly compensated by the Coulomb attraction between solutes in the

294: presence of the solvent. As $q$ increases, the initial slope of the

295: effective force increases. The potential of mean force (shown in the

296: inset of Fig. 4) exhibits a contact value which increases with $q$,

297: indicating that the reduction of hydrophobic attractive free energy

298: and the reduction of dielectric screening due to restructuring of

299: water by the strong electric field of the solutes clearly outweigh the

300: increase in bare Coulomb attraction between the latter. Simulations

301: calculating the forces at and  near contact for $q=7$ and $q=15$, not 

302: shown in Fig. 4, confirm this trend. Note that the potential of

303: mean force curve for $q=10$ shows more long range attraction as

304: compared to the smaller $q$ data due to the increased electrostatic 

305: attractive interaction.The eye-catching kink in the force for $q=$10,

306: at a distance $s\approx 1$\AA~ is reproducible, and is probably

307: related  to the pronounced shell structure of water molecules around

308: highly charged solutes, illustrated in Fig. 3. While for neutral (and

309: weakly charged) solutes, the O and H density profiles show little

310: structure, they are sharply peaked at a distance  $s\approx 1$\AA~ of

311: the O atoms from the solute surface. This would lead to a complete

312: shared hydration layer, and consequently to a kink in the force versus

313: distance curve, between 2 flat solutes separated by $s=2$\AA~. This

314: critical separation is shifted to shorter distances due to 

315: the curvature of spherical solutes. 

316:  

317: In view of this delicate balance between various interactions, we have

318: also examined the case of equally charged solutes. In this case

319: monovalent counterions (Na$^{+}$ or Cl$^{-}$) were included to ensure overall

320: charge neutrality. The situation is summarized in Fig. 5 for solutes

321: of radius $R=10$\AA~ and charge $q=\pm5$. The water density profiles

322: (not shown here) are symmetric with respect to $z=0$ for equally

323: charged solutes, but differ substantially when going from a pair of

324: anions to a pair of cations. In the latter case water is much more

325: structured into well defined hydration shells, as shown in

326: Fig. 3. This difference is 

327: reflected in the effective forces and potentials shown in

328: Fig. 5. While the force and potential of mean force are practically

329: zero at all distances in the case of cationic solutes, anionic solutes

330: attract each other significantly, showing that residual hydrophobic attraction

331: overcomes the repulsion between like charges, which may be more

332: efficiently reduced by the local permittivity of water in the

333: immediate vicinity of the anionic solutes. 

334: 

335: To summarize, we have shown that electric charges carried by

336: nanometer-scale solutes have a profound influence on hydrophobic

337: interactions. Our MD simulations confirm that for neutral solutes

338: hydrophobic attraction is intimately linked to solvent depletion, and

339: the contact value of the hydrophobic force is directly related to the

340: surface tension of the pure solvent. Solvent depletion is suppressed by

341: the electric field due to any charge carried by the solute, but for

342: oppositely charged solutes the resulting loss of hydrophobic

343: attraction is compensated by their mutual Coulomb attraction, leading

344: to a nearly constant contact force. The effective force between

345: equally charged solutes depends on the sign of their charge. Anionic

346: solutes exhibit a striking ``like-charge attraction", a subject of

347: considerable debate in the field of mesoscopic charge-stabilized

348: colloids \cite{hansen:review}. Coulombic and hydrophobic effects are intimately

349: entangled, and their inter-relation is expected to play an important

350: role in the analysis of protein aggregation and related biomolecular

351: mechanisms. 

352: 

353: The authors are grateful to R. Evans and H. L{\"o}wen for useful clarifications. JD

354: acknowledges the financial support of EPSRC within the Portfolio Grant

355: RG37352. 

356: 

357: 

358: \begin{thebibliography}{10}

359: \expandafter\ifx\csname bibnamefont\endcsname\relax

360:   \def\bibnamefont#1{#1}\fi

361: \expandafter\ifx\csname bibfnamefont\endcsname\relax

362:   \def\bibfnamefont#1{#1}\fi

363: \expandafter\ifx\csname url\endcsname\relax

364:   \def\url#1{\texttt{#1}}\fi

365: \expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi

366: \expandafter\ifx\csname bibinfo\endcsname\relax \def\bibinfo#1#2{#2}\fi

367: \expandafter\ifx\csname eprint\endcsname\relax \def\eprint#1{#1}\fi

368: 

369: \bibitem{wallquist:jpc}

370: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Wallquist}} \bibnamefont{and}

371:   \bibinfo{author}{\bibfnamefont{B.~J.} \bibnamefont{Berne}},

372:   \bibinfo{journal}{J. Phys. Chem.} \textbf{\bibinfo{volume}{99}},

373:   \bibinfo{pages}{2893} (\bibinfo{year}{1995}).

374: 

375: \bibitem{lum:jpc}

376: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Lum}},

377:   \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Chandler}}, \bibnamefont{and}

378:   \bibinfo{author}{\bibfnamefont{J.~D.} \bibnamefont{Weeks}},

379:   \bibinfo{journal}{J. Phys. Chem. B} \textbf{\bibinfo{volume}{103}},

380:   \bibinfo{pages}{4570} (\bibinfo{year}{1999}).

381: 

382: \bibitem{likos:physrep}

383: \bibinfo{author}{\bibfnamefont{C.~N.} \bibnamefont{Likos}},

384:   \bibinfo{journal}{Phys. Rep.} \textbf{\bibinfo{volume}{348}},

385:   \bibinfo{pages}{267} (\bibinfo{year}{2001}).

386: 

387: \bibitem{landau}

388: \bibinfo{author}{\bibfnamefont{L.~D.} \bibnamefont{Landau}},

389:   \bibinfo{author}{\bibfnamefont{E.~M.} \bibnamefont{Lifshitz}},

390:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{L.~P.}

391:   \bibnamefont{Pitaevski}}, \emph{\bibinfo{title}{Electrodynamics of Continious

392:   Media}} (\bibinfo{publisher}{Butterwort-Heinemann}, \bibinfo{year}{1993}),

393:   \bibinfo{note}{second edition}.

394: 

395: \bibitem{berendsen:jpc}

396: \bibinfo{author}{\bibfnamefont{H.~J.~C.} \bibnamefont{Berendsen}},

397:   \bibinfo{author}{\bibfnamefont{J.~R.} \bibnamefont{Grigera}},

398:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{T.~P.}

399:   \bibnamefont{Straatsma}}, \bibinfo{journal}{J. Phys. Chem.}

400:   \textbf{\bibinfo{volume}{91}}, \bibinfo{pages}{6269} (\bibinfo{year}{1987}).

401: 

402: \bibitem{dlpoly}

403: \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{Smith}} \bibnamefont{and}

404:   \bibinfo{author}{\bibfnamefont{T.~R.} \bibnamefont{Forester}}

405:   (\bibinfo{year}{1999}), \bibinfo{note}{the DLPOLY\_2 User Manual}.

406: 

407: \bibitem{frenkelsmit}

408: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Frenkel}} \bibnamefont{and}

409:   \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Smit}},

410:   \emph{\bibinfo{title}{Understanding Molecular Simulation: From Algorithms to

411:   Applications}} (\bibinfo{publisher}{Academic Press}, \bibinfo{year}{1996}).

412: 

413: \bibitem{berendsen:jcp}

414: \bibinfo{author}{\bibfnamefont{H.~J.~C.} \bibnamefont{Berendsen}},

415:   \bibinfo{author}{\bibfnamefont{J.~P.~M.} \bibnamefont{Postma}},

416:   \bibinfo{author}{\bibfnamefont{W.~F.} \bibnamefont{van Gunsteren}},

417:   \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{DiNola}}, \bibnamefont{and}

418:   \bibinfo{author}{\bibfnamefont{J.~R.} \bibnamefont{Haak}},

419:   \bibinfo{journal}{J. Chem. Phys.} \textbf{\bibinfo{volume}{81}},

420:   \bibinfo{pages}{3684} (\bibinfo{year}{1984}).

421: 

422: \bibitem{essmann:jcp}

423: \bibinfo{author}{\bibfnamefont{U.}~\bibnamefont{Essmann}},

424:   \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{Perera}},

425:   \bibinfo{author}{\bibfnamefont{M.~L.} \bibnamefont{Berkowitz}},

426:   \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{Darden}}, \bibnamefont{and}

427:   \bibinfo{author}{\bibfnamefont{H.~L. L.~G.} \bibnamefont{Pedersen}},

428:   \bibinfo{journal}{J. Chem. Phys.} \textbf{\bibinfo{volume}{103}},

429:   \bibinfo{pages}{8577} (\bibinfo{year}{1995}).

430: 

431: \bibitem{louis:jcp}

432: \bibinfo{author}{\bibfnamefont{A.~A.} \bibnamefont{Louis}},

433:   \bibinfo{author}{\bibfnamefont{P.~G.} \bibnamefont{Bolhuis}},

434:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{J.~P.}

435:   \bibnamefont{Hansen}}, \bibinfo{journal}{J. Chem. Phys}

436:   \textbf{\bibinfo{volume}{117}}, \bibinfo{pages}{1893} (\bibinfo{year}{2002}).

437: 

438: \bibitem{huang:jpc}

439: \bibinfo{author}{\bibfnamefont{D.~M.} \bibnamefont{Huang}},

440:   \bibinfo{author}{\bibfnamefont{P.~L.} \bibnamefont{Geissler}},

441:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Chandler}},

442:   \bibinfo{journal}{J. Phys. Chem. B} \textbf{\bibinfo{volume}{105}},

443:   \bibinfo{pages}{6704} (\bibinfo{year}{2001}).

444: 

445: \bibitem{joe:channel}

446: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{Dzubiella}},

447:   \bibinfo{author}{\bibfnamefont{R.~J.} \bibnamefont{Allen}}, \bibnamefont{and}

448:   \bibinfo{author}{\bibfnamefont{J.~P.} \bibnamefont{Hansen}}

449:   (\bibinfo{year}{2003}), \bibinfo{note}{to be published}.

450: 

451: \bibitem{hansen:review}

452: \bibinfo{author}{\bibfnamefont{J.~P.}~\bibnamefont{Hansen}} \bibnamefont{and}

453:   \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{L{\"o}wen}},

454:   \bibinfo{journal}{Ann. Rev. Phys. Chem.} \textbf{\bibinfo{volume}{51}},

455:   \bibinfo{pages}{209} (\bibinfo{year}{2000}).

456: 

457: \end{thebibliography}

458: 

459: \end{document}

460: 

461: 

462: 

463: 

464: 

465: 

466: