1: \documentclass[prl,twocolumn,showpacs]{revtex4}

2: \usepackage{graphicx,amsmath}

3: 

4: %% 21 june 2006 : revised version after referee reports

5: 

6: %%%%%%%%%%%%%%%%%%%%

7: %% PRIVATE MACROS %%

8: %%%%%%%%%%%%%%%%%%%%

9: 

10: \newcommand{\figwidth}{0.7\columnwidth}

11: \newcommand{\eq}[1]{Eq.(\ref{#1})}

12: \newcommand{\fig}[1]{Fig.~\ref{#1}}

13: \newcommand{\avg}[1]{ {\langle #1 \rangle} }

14: \newcommand{\olcite}[1]{Ref.~\onlinecite{#1}}

15: 

16: \newcommand{\rc}{\rho_{\rm cr}}

17: \newcommand{\rl}{\rho_{\rm L}}

18: \newcommand{\rg}{\rho_{\rm G}}

19: \newcommand{\zc}{z_{\rm c}}

20: \newcommand{\zp}{z_{\rm p}}

21: \newcommand{\zpcr}{z_{\rm p,cr}}

22: \newcommand{\lc}{l_{\rm c}}

23: \newcommand{\sigmap}{\sigma_{\rm p}}

24: \newcommand{\nc}{N_{\rm c}}

25: \newcommand{\np}{N_{\rm p}}

26: \newcommand{\rhoc}{\rho_{\rm c}}

27: \newcommand{\pc}{P_L(\rhoc|\zp,\zc)}

28: \newcommand{\ucr}{U_4^\star}

29: 

30: \newcommand{\ZPCR}{3.877}

31: 

32: %%%%%%%%%%%%%%%%%%%%%

33: %% DOC STARTS HERE %%

34: %%%%%%%%%%%%%%%%%%%%%

35: 

36: \begin{document}

37: 

38: \title{coexistence diameter in two-dimensional colloid-polymer 

39: mixtures}

40: 

41: \author{R. L. C. Vink and H. H. Wensink}

42: 

43: \affiliation{Institut f\"ur Theoretische Physik II, Heinrich Heine

44: Universit\"at D\"usseldorf, Universit\"atsstra{\ss}e 1, 40225

45: D\"usseldorf, Germany}

46: 

47: \date{\today}

48: 

49: \begin{abstract}

50: 

51: We demonstrate that the {\it law of the rectilinear coexistence diameter} 

52: in two-dimensional (2D) mixtures of {\it non-spherical} colloids and 

53: non-adsorbing polymers is violated. Upon approach of the critical point, 

54: the diameter shows logarithmic singular behavior governed by a term $t \ln 

55: t$, with $t$ the relative distance from the critical point. No sign of a 

56: term $t^{2\beta}$ could be detected, with $\beta$ the critical exponent of 

57: the order parameter, indicating a very weak or absent Yang-Yang anomaly. 

58: Our analysis thus reveals that non-spherical particle shape alone is {\it 

59: not} sufficient for the formation of a pronounced Yang-Yang anomaly in the 

60: critical behavior of fluids.

61: 

62: \end{abstract}

63: 

64: %% 05.70.Jk 	Critical point phenomena

65: %% 64.60.Fr 	Equilibrium properties near critical points, critical exponents

66: %% 64.70.Fx 	Liquid-vapor transitions

67: 

68: \pacs{05.70.Jk, 64.60.Fr, 64.70.Fx}

69: 

70: \maketitle

71: 

72: Critical phenomena have been studied extensively for decades. 

73: Nevertheless, a few important issues remain controversial. One 

74: longstanding problem concerns the critical behavior of the so-called 

75: coexistence diameter of fluids undergoing a phase transition from liquid 

76: to gas (the coexistence diameter is defined as the average density of the 

77: two coexisting phases). In 1964, Yang and Yang showed that the divergence 

78: of the constant-volume specific heat at the critical point implies that 

79: {\it either} $d^2 p /d T^2$ or $d^2 \mu / d T^2$ or {\it both} diverge 

80: \cite{yang.yang:1964}. Here, $T$ is the temperature, $p$ the pressure, and 

81: $\mu$ the chemical potential. A remarkable consequence, realized only 

82: recently by Fisher and co-workers \cite{fisher.orkoulas:2000}, is that the 

83: divergence of $d^2 \mu / d T^2$ implies that the coexistence diameter 

84: $\delta$ gains an additional term $t^{2\beta}$ which, assuming Ising 

85: universality, dominates the previously recognized term $t^{1-\alpha}$ 

86: \cite{widom.rowlinson:1970, mermin:1971}. As usual, $t$ is the relative 

87: distance from the critical point; $\alpha$ and $\beta$ are the critical 

88: exponents of the specific heat and the order parameter, respectively. The 

89: coexistence diameter near the critical point thus reads as

90: %%

91: \begin{equation}\label{eq:coex}

92:   \delta_{\alpha \neq 0} = \rc \left( 1 + A_{2\beta} t^{2\beta}

93: 	+ A_{1-\alpha} t^{1-\alpha} + A_1 t \right),

94: \end{equation}

95: %%

96: with $\rc$ the critical concentration, and non-universal amplitudes $A_i$. 

97: The divergence of $d^2 \mu / d T^2$ is called a Yang-Yang (YY) anomaly, 

98: and $A_{2\beta} \neq 0$ if one is present. For {\it symmetric} fluids, 

99: such as the Widom-Rowlinson mixture \cite{widom.rowlinson:1970}, $d^2 \mu 

100: / d T^2$ remains finite, in which case $A_{2\beta}=0$ \cite{vink:2006}. 

101: However, realistic fluids are typically {\it asymmetric}, in which case a 

102: YY-anomaly cannot be ruled out.

103: 

104: Indeed, evidence for a YY-anomaly was found in simulations of asymmetric 

105: three-dimensional (3D) fluids \cite{orkoulas.fisher.ea:2001, 

106: kim.fisher.ea:2003}. Experiments on propane and carbon dioxide also show 

107: evidence for a weak YY-anomaly \cite{orkoulas.fisher.ea:2000}, while for 

108: He-3 the situation is less clear \cite{anisimov.zhong.ea:2004, 

109: hahn.weilert.ea:2004}. However, for 3D fluids, the analysis is extremely 

110: difficult. Assuming 3D Ising universality with $\alpha \approx 0.109$ and 

111: $\beta \approx 0.326$ \cite{fisher.zinn:1998}, there are two singular 

112: terms in \eq{eq:coex} which need to be distinguished not only from each 

113: other, but also from the leading analytic background term $A_1 t$. A 

114: direct observation of the term $t^{2\beta}$ is consequently very 

115: difficult. Only very recently, after it was recognized that the amplitudes 

116: $A_{1-\alpha}$ and $A_1$ are coupled, could both singular terms in 

117: \eq{eq:coex} be resolved from experimental data \cite{anisimov.wang:2006}. 

118: 

119: Nevertheless, due to the competition between terms in \eq{eq:coex}, 

120: investigations of the coexistence diameter in 3D fluids remain 

121: challenging. An attractive alternative, where the competition between 

122: singular terms is less severe, is to consider a fluid where the specific 

123: heat diverges logarithmically at the critical point, implying $\alpha=0$. 

124: The critical behavior of the diameter is then given by

125: %%

126: \begin{equation}\label{eq:coex2d}

127:   \delta_{\alpha=0} = \rc \left( 1 + A_{2\beta} t^{2\beta}

128: 	+ A_0 \, t \ln t + A_1 t \right).

129: \end{equation}

130: %%

131: In other words, the term $t^{1-\alpha}$ is replaced by a (much weaker) 

132: {\it logarithmic} singularity \cite{anisimov:2006}. In order to detect the 

133: YY-anomaly, one thus needs to distinguish power law behavior from 

134: logarithmic behavior, which is expected to yield a more pronounced 

135: numerical signature. To this end, one could consider a 2D fluid, where 

136: $\alpha=0$ and $\beta=1/8$ (assuming 2D Ising universality). Note that 

137: mean-field systems do not qualify, despite having $\alpha=0$. For 

138: mean-field systems, $\alpha=0$ corresponds to a finite discontinuity in 

139: the specific heat, yielding a purely rectilinear diameter 

140: \cite{anisimov:2006}. Still, even though the numerical analysis of a 2D 

141: fluid with Ising critical behavior may become more simple, this is no 

142: guarantee that a YY-anomaly will be found. The absence or presence of a 

143: YY-anomaly is a non-universal feature: $A_{2\beta}$ may well be zero! It 

144: is not completely clear which features in a fluid determine the strength 

145: of the YY-anomaly. Obviously, the fluid must be asymmetric. In addition, 

146: there are experimental indications that molecular shape and symmetry, in 

147: particular departures from spherical form, are important contributing 

148: factors \cite{orkoulas.fisher.ea:2000}. From these considerations, it 

149: appears that the ``minimal'' fluid in which a YY-anomaly may most easily 

150: be found should (1) be two-dimensional and exhibit a 2D Ising critical 

151: point (2) be asymmetric, and (3) contain non-spherical particles.

152: 

153: In this work, we will investigate a fluid with precisely these properties, 

154: and focus on the critical behavior of its diameter. We consider a 2D 

155: version of the colloid-polymer model of Asakura and Oosawa (AO) 

156: \cite{asakura.oosawa:1954}, but generalized to non-spherical colloids. In 

157: the original AO model, colloids and polymers are treated as spheres in 3D, 

158: assuming hard-core interactions between colloid-colloid and 

159: colloid-polymer pairs, while polymer-polymer pairs can interpenetrate 

160: freely (the AO model is thus clearly asymmetric). Since the polymers may 

161: overlap freely, their translational entropy is increased significantly 

162: when the colloids group together. Hence, there is an effective (depletion) 

163: attraction between the colloids. Provided polymer concentration and size 

164: are sufficiently large, the attraction is strong enough to drive phase 

165: separation in the AO model, whereby the system splits-up into a 

166: colloid-rich (polymer-poor) phase, and a colloid-poor (polymer-rich) 

167: phase. As expected for systems with short-ranged interactions, the 

168: corresponding unmixing critical point belongs to the 3D Ising universality 

169: class \cite{vink.horbach:2004*1}. Hence, it is anticipated, although one 

170: should plan to check, that the AO model in 2D will exhibit a 2D Ising 

171: critical point. The effect of non-spherical particle shape is incorporated 

172: by modeling the colloids not as spheres, but as line segments. Our minimal 

173: model is thus a 2D mixture of colloidal line segments of length $\lc$, and 

174: effective polymer disks of diameter $\sigmap$, interacting via AO 

175: potentials. In other words, overlaps between line segments, as well as 

176: overlaps between line segments and polymer disks are forbidden, while the 

177: polymer disks may overlap freely.

178: 

179: The aim of this work is to check if a YY-anomaly in this model can be 

180: found. We will do so using computer simulation and finite size scaling 

181: (FSS) in the grand canonical ensemble. In this ensemble, the total area of 

182: the system $A$, the temperature $T$, and the colloid (polymer) fugacity 

183: $\zc$ ($\zp$) are fixed, while the number of colloids ($\nc$) and polymers 

184: ($\np$) fluctuates. The thermal wavelength is set to unity, such that 

185: $\zp$ reflects the average concentration $\np/A$ a pure phase of polymers 

186: would have (recall that such a phase is simply an ideal gas). The 

187: remaining lengths are expressed in units of $\lc$. The colloid-to-polymer 

188: size ratio is set to $\sigmap / \lc = 0.95$. At the coexistence colloid 

189: fugacity, it is expected that the mixture phase separates into a colloid 

190: poor phase (the gas) and a colloid rich phase (the liquid), provided the 

191: polymer fugacity exceeds a critical value $\zpcr$. The phase separation is 

192: thus driven by $\zp$, which therefore plays a role analogous to inverse 

193: temperature in gas-liquid transitions of simple fluids. The relative 

194: distance from the critical point is written as $t \equiv \zp/\zpcr-1$, and 

195: $\rg$ ($\rl$) denotes the concentration $\rhoc \equiv \nc/A$ of colloids 

196: in the gas (liquid) phase. A natural order parameter is the density gap 

197: $\Delta \equiv (\rl-\rg)/2$, while the coexistence diameter reads as 

198: $\delta \equiv (\rl+\rg)/2$. In the limit $t \to 0$, we expect critical 

199: power law behavior $\Delta \propto t^\beta$ for the order parameter, and 

200: \eq{eq:coex2d} for the coexistence diameter.

201: 

202: \begin{figure}

203: \begin{center}

204: \includegraphics[clip=,width=\figwidth]{phase}

205: 

206: \caption{\label{phase} {\it upper frame:} Binodals of the 2D 

207: colloid-polymer mixture of this work, obtained using mean-field theory 

208: (MF) and simulation (SIM); circles mark critical points. {\it lower 

209: frame:} Order parameter $\Delta$ as function of $t$, obtained using 

210: simulation and FSS; a value $\zpcr = \ZPCR$ in $t$ was used.}

211: 

212: \end{center}

213: \end{figure}

214: 

215: For our minimal model, the critical point must be located first, in 

216: particular the value of $\zpcr$. Next, 2D Ising universality of the 

217: critical point must be established, after which the diameter can be 

218: investigated. To ``guide'' the simulations, the phase diagram is obtained 

219: approximately first, using a simple mean-field (MF) theory based on a 

220: free-volume approach for 2D rod-polymer mixtures. The free-volume fraction 

221: and the free energy for a pure system of 2D lines are derived from scaled 

222: particle theory \cite{lekkerkerker.stroobants:1994}. The free energy is 

223: exactly the same as the one obtained within Onsager's second virial theory 

224: \cite{lekkerkerker.stroobants:1994}. The resulting binodal is shown in 

225: \fig{phase}. The theory predicts the critical point at $\rc^{\rm MF} 

226: \approx 0.550$ and $\zpcr^{\rm MF} \approx 2.054$. No nematic ordering of 

227: the colloids is predicted in the direct vicinity of the critical point: 

228: the coexisting phases are isotropic. Since the theory ignores critical 

229: fluctuations, it is expected that $\zpcr^{\rm MF}$ underestimates the true 

230: value $\zpcr$ significantly. Nevertheless, the theoretical result is 

231: important because it provides an indication in which regime the (time 

232: consuming) simulations need to be carried out. The simulations are 

233: performed in the grand canonical ensemble, on a 2D square of size $A = L 

234: \times L$ using periodic boundary conditions. We measure the distribution 

235: $\pc$, defined as the probability of observing a system with colloid 

236: concentration $\rhoc$, at fugacities $\zp$ and $\zc$, with $L$ the system 

237: size. The insertion and removal of particles is performed using a cluster 

238: move \cite{vink.horbach:2004*1}, combined with a biased sampling scheme 

239: \cite{virnau.muller:2004} to overcome the free energy barrier separating 

240: the phases, and histogram reweighting \cite{ferrenberg.swendsen:1989}. To 

241: obtain a single distribution, around 100 CPU hours for a small system 

242: ($L=22$), and 350~h for a large system ($L=30$), are required.

243: 

244: \begin{figure}

245: \begin{center}

246: \includegraphics[clip=,width=\figwidth]{cumulant}

247: 

248: \caption{\label{cumulant} Cumulant analysis near the critical point. The 

249: top frame shows $U_4$ as function of $\zp$ for various system sizes $L$ as 

250: indicated. The lower frame shows the cumulant slope $Y_1$ at $\zpcr$ as 

251: function of $L$.}

252: 

253: \end{center}

254: \end{figure}

255: 

256: \begin{figure}

257: \begin{center}

258: \includegraphics[clip=,width=\figwidth]{scaling}

259: 

260: \caption{\label{scaling} {\it upper frame:} Order parameter scaling curve 

261: $y=f(x)$ (solid line). Following convention \cite{kim.fisher.ea:2003}, the 

262: scaling curve is raised to a negative exponent with $\phi = 1/\beta$, 

263: where the 2D Ising value $\beta=1/8$ was used. Also shown is the small $x$ 

264: limiting form $y=1+x/2$ (dashed line). {\it lower frame:} Coexistence 

265: diameter scaling curve $y'=g(x')$ (solid line). The dashed line shows the 

266: small $x'$ limiting form $y'=x'/2$.}

267: 

268: \end{center}

269: \end{figure}

270: 

271: A standard route to obtain $\zpcr$ in simulations, is to measure the 

272: $L$-dependence of the cumulant $U_4 = \avg{m^2}^2 / \avg{m^4}$ along some 

273: path in the $(\rhoc,\zp)$-plane. Here, $m = \rhoc - \avg{\rhoc}$, and 

274: $\avg{\cdot}$ denote grand canonical averages. The cumulant becomes 

275: system-size independent at the critical point \cite{binder:1981}. Plots of 

276: $U_4$ as function of $\zp$ for different system sizes $L$ are expected to 

277: show a common intersection point, leading to an estimate of $\zpcr$. 

278: Moreover, the value of the cumulant $\ucr$ at the intersection point is 

279: universal, and this gives an indication of the universality class. The 

280: result is shown in the upper frame of \fig{cumulant}. The data were 

281: obtained using the colloid fugacity at which $\avg{m^2}$ is maximized 

282: \cite{orkoulas.fisher.ea:2001}. From the intersections, we obtain $\zpcr = 

283: 3.881 \pm 0.005$, where the error reflects the scatter in the various 

284: intersection points. At the intersection point $\ucr \approx 0.85$, which 

285: is very close to the accepted 2D Ising value $\ucr \approx 0.856$ 

286: \cite{kamieniarz.blote:1993} (horizontal line in \fig{cumulant}). While 

287: this already suggests 2D Ising universality, additional confirmation is 

288: obtained from the critical exponents. We consider $\beta$ and $\nu$, with 

289: $\nu$ the critical exponent of the correlation length. Here, $\nu$ is 

290: extracted from the cumulant slope $Y_1 \equiv d U_4 / d \zp$ {\it at} the 

291: critical value of $\zp$. One expects that $Y_1 \propto L^{1/\nu}$, with 

292: $L$ the system size. The lower frame of \fig{cumulant} shows $Y_1$ as 

293: function of $L$, where the above estimate of $\zpcr$ was used. The line is 

294: a linear fit through the origin, which describes the data very well, and 

295: thus confirms the 2D Ising value $\nu=1$. To obtain $\beta$, we apply the 

296: FSS algorithm of \olcite{kim.fisher.ea:2003}, using system sizes 

297: $L=20-30$. Starting with $\zp$ significantly {\it above} its 

298: critical value, the algorithm proceeds by plotting $U_4$ as function of 

299: the average colloid concentration $\avg{\rhoc}$. The resulting plot 

300: reveals two minima, located at $\rho^-$ and $\rho^+$, with respective 

301: values $Q^-$ and $Q^+$ at the minima. Defining the quantities $Q_{\rm min} 

302: = (Q^+ + Q^-)/2$, $x = Q_{\rm min} \ln (4/e Q_{\rm min} )$, and $y = 

303: (\rho^+ - \rho^-) / (2 \Delta)$, the points $(x,y)$ obtained for different 

304: system sizes $L$ should, in the limit far away from the critical point, 

305: collapse onto the line $y=1+x/2$. Recall that $\Delta$ is the order 

306: parameter in the thermodynamic limit at the considered fugacity $\zp$, 

307: precisely the quantity of interest, which may thus be obtained by fitting 

308: until the best collapse onto $1+x/2$ occurs. In the next step, $\zp$ is 

309: chosen closer to the critical point, the points $(x,y)$ are calculated as 

310: before, but this time $\Delta$ is chosen such that the new data set 

311: joins smoothly with the previous one, yielding an estimate of the order 

312: parameter at the new fugacity. This procedure is repeated as closely as 

313: possible to the critical point, where $\Delta$ vanishes. The output of the 

314: algorithm, $\Delta$ as function of $\zp$, is then fitted to $\Delta 

315: \propto t^\beta$, in order to estimate $\zpcr$ and $\beta$. We obtain 

316: $\zpcr = \ZPCR \pm 0.001$, which is consistent with the (less precise) 

317: analysis of \fig{cumulant}. Shown in the lower frame of \fig{phase} is the 

318: order parameter as function of $t$, on double logarithmic scales. The line 

319: has slope $\beta=1/8$, and confirms the 2D Ising exponent in the 

320: simulation data. The FSS algorithm also yields the order parameter scaling 

321: curve $y=f(x)$ \cite{kim.fisher.ea:2003}, with $x$ and 

322: $y$ defined as above, shown for completeness in the upper frame of 

323: \fig{scaling}. Away from the critical point ($x \to 0$), the scaling curve 

324: has the limiting form $y=1+x/2$; at the critical point, the scaling curve 

325: diverges. The significance of the scaling curve is its universal 

326: character: all systems with a 2D Ising critical point should yield a 

327: scaling curve for the order parameter similar to the one shown here 

328: \cite{kim.fisher.ea:2003}.

329: 

330: \begin{figure}

331: \begin{center}

332: \includegraphics[clip=,width=\figwidth]{diam}

333: 

334: \caption{\label{diam} {\it upper frame:} Coexistence diameter $\delta$ as 

335: function of $t$ (circles). The dashed line is a fit to \eq{eq:coex2d}. 

336: {\it lower frame:} $\kappa$ as function of $\ln t$ (circles). The straight 

337: line (dashed) confirms the logarithmic nature of the divergence. In both 

338: plots, $\zpcr=\ZPCR$ in $t$ was used.}

339: 

340: \end{center}

341: \end{figure}

342: 

343: At this point, sufficient evidence for 2D Ising universality has been 

344: provided. Since the order parameter is a scalar, and since the 

345: interactions are short-ranged, there are in any case no theoretical 

346: grounds to contemplate a different universality class. Hence, we will now 

347: consider the coexistence diameter. To obtain the diameter, the FSS 

348: algorithm of \olcite{kim.fisher.ea:2003} is used. Similarly, also for the 

349: coexistence diameter, a scaling curve $y'=g(x')$ is generated, based on 

350: {\it different} quantities $x'$ and $y'$ defined in 

351: \olcite{kim.fisher.ea:2003}. In contrast to the order parameter, the 

352: scaling curve of the diameter is {\it not} universal. For our model, the 

353: corresponding curve is shown in the lower frame of \fig{scaling}. For 

354: small $x'$, it correctly approaches the exact limiting form $y'=x'/2$ 

355: \cite{kim.fisher.ea:2003}. The curvature at $x' \gg 0$ already suggests 

356: singular behavior. According to \eq{eq:coex2d}, this may reflect the 

357: YY-anomaly, or logarithmic behavior, or both. To quantify this, the 

358: diameter itself is shown in the upper frame of \fig{diam}. The quality of 

359: the data is such that $\delta$ can be resolved down to $t \approx 0.0015$. 

360: A fit to \eq{eq:coex2d} yields $\rc = 0.9270 \pm 0.0006$, $A_{2\beta} 

361: \approx 0$, $A_0=-0.29 \pm 0.01$ and $A_1 = 0.76 \pm 0.03$, where the 

362: error reflects the scatter resulting from the range over which the fit is 

363: performed. The fit describes the data perfectly well, without the need for 

364: a term $t^{2\beta}$, indicating {\it logarithmic} singular behavior. In 

365: other words, despite the non-spherical particle shape in our model, the 

366: present analysis does {\it not} reveal a YY-anomaly. For completeness, in 

367: the lower frame of \fig{diam}, the derivative $\kappa \equiv d\delta/dt$ 

368: is plotted as function of $\ln t$. In case of singular behavior, $\kappa$ 

369: is expected to diverge as $t \to 0$. The (logarithmic) divergence is 

370: clearly visible. Finally, by combining the coexistence diameter and order 

371: parameter data, the binodal in the thermodynamic limit was constructed, 

372: see the top frame of \fig{phase}.

373: 

374: In conclusion, we have shown that 2D colloid-polymer mixtures, with 

375: non-spherical colloids, do not display a pronounced YY-anomaly, at least 

376: not for the colloid-to-polymer size ratio $q=0.95$ considered by us. 

377: Although the diameter becomes singular upon approach of the critical 

378: point, the singularity is logarithmic, and well described by the 

379: theoretically expected term $t \ln t$. While it has been suggested 

380: \cite{orkoulas.fisher.ea:2000} that non-spherical particle shape may be an 

381: important contributing factor to the formation of the YY-anomaly, this 

382: seems not to be the case for our 2D model. If a YY-anomaly is present in 

383: our 2D model nevertheless, it is very weak, and negligible down to $t 

384: \approx 0.0015$ accessible in our simulations. In contrast, our results 

385: may be compatible with the very recent \olcite{anisimov.wang:2006}, where 

386: it is argued that a YY-anomaly is expected when $\rho^\star = \rc a_{\rm 

387: i}$ is small, where $a_{\rm i}$ represents the typical interaction volume. 

388: For our model, $a_{\rm i} \approx q^2$, and so $\rho^\star \approx 0.8$. 

389: This value even exceeds $\rho^\star \approx 0.75$ of the Widom-Rowlinson 

390: mixture \cite{widom.rowlinson:1970}, for which no YY-anomaly was detected 

391: either \cite{vink:2006}. In order to detect a YY-anomaly in our 2D model, 

392: it seems that smaller size ratios $q$ are required; this could be a topic 

393: for further simulations. It is tempting to speculate if the 2D model 

394: considered in this work can also be realized experimentally. Colloidal 

395: particles, due to their mesoscopic size, pose many advantages over atomic 

396: fluids. This has already enabled the investigation of critical phenomena 

397: in 3D, whereby the particles are visualized directly using confocal 

398: microscopy. Other applications may be found in order-disorder phase 

399: transitions in adsorbed monolayers of atoms or small molecules at 

400: surfaces.

401: 

402: \acknowledgments

403: 

404: This work was supported by the {\it Deutsche Forschungsgemeinschaft} under 

405: the SFB-TR6/D3. HHW also acknowledges the {\it Alexander von Humboldt} 

406: foundation. We thank M. A. Anisimov and K. Binder for valuable comments.

407: 

408: \bibstyle{revtex}

409: \bibliography{mainz}

410: 

411: \end{document}

412: