1: \documentclass[aps,prl,epsfig,twocolumn,showpacs]{revtex4}
2: %\usepackage{docs}
3: \usepackage{amsmath}
4: \usepackage{graphicx}
5:
6: \begin{document}
7: \title{ On the effective one-component description of highly asymmetric
8: hard-sphere binary fluid mixtures}
9: %\author { C. F. Tejero$^a$ and M. L\'{o}pez de
10: %Haro$^b$\\ {\em Centro de Investigaci\'{o}n en Energ\'{\i}a,
11: %U.N.A.M., Temixco, Morelos 62580 (M\'{e}xico)}}
12:
13: \author{C. F. Tejero}
14: \email[]{cftejero@fis.ucm.es}
15: \thanks{ permanent address: Facultad de Ciencias F\'{\i}sicas, Universidad Complutense de
16: Madrid, E-28040, Madrid, Spain}
17: \author{M. L\'opez de Haro}
18: \email[]{malopez@servidor.unam.mx}
19: \homepage[]{http://miquiztli.cie.unam.mx/xml/tc/ft/mlh/}
20: \affiliation{Centro de Investigaci\'{o}n en Energ\'{\i}a,
21: U.N.A.M., Temixco, Morelos 62580 (M{\'e}xico)}
22:
23:
24: \begin{abstract}
25: \indent {The phase diagram of a binary fluid mixture of highly
26: asymmetric additive hard spheres is investigated. Demixing is
27: analyzed from the exact low-density expansions of the
28: thermodynamic properties of the mixture and compared with the
29: fluid-fluid separation based on the effective one-component
30: description. Differences in the results obtained from both
31: approaches, which have been claimed to be equivalent, are pointed
32: out and their possible origin is discussed. It is argued that to
33: deal with these differences new theoretical approximations should
34: be devised.}
35: \end{abstract}
36: \pacs{05.70.Fh, 64.10.+h, 64.70.Ja, 64.75.+g}
37:
38:
39:
40: \maketitle
41:
42: It is well known that, due to the widely different time and length
43: scales involved, it is a very difficult task to describe a soft
44: matter complex fluid composed of mesoscopic (colloidal) particles
45: and microscopic (small) particles, i.e. a colloidal suspension.
46: Therefore people have struggled for years trying to devise
47: strategies that may render a tractable description. On the
48: theoretical side, a rather successful and often employed approach
49: in the colloids literature is that of coarse graining. In this
50: technique the idea is to integrate out the irrelevant degrees of
51: freedom (the small particles) to end up with a one-component
52: system of colloidal particles described by effective interactions
53: \cite{Likos}. One of the most celebrated effective interactions
54: results from the depletion mechanism first described by Asakura
55: and Oosawa \cite{Asakura}. Suppose that two colloidal hard spheres
56: are immersed in an ideal fluid of small hard spheres. When the
57: separation of the surfaces of two large spheres is less than the
58: diameter of the small ones, the depletion of the latter from the
59: gap between the colloids leads to an unbalanced osmotic pressure
60: which, in turn, results in an effective attraction between the two
61: large spheres. Although the Asakura-Oosawa depletion pair
62: potential is concerned only with the limit of extreme dilution,
63: the theory of the depletion mechanism has recently been reexamined
64: by also including the interactions between the small spheres
65: \cite{Gotzelmann,Roth}. In several of the recent studies the
66: depletion mechanism has been invoked as the possible driving force
67: for demixing, i.e. a phase transition whereby a binary fluid
68: mixture of hard spheres separates into two fluid phases of
69: different composition.
70:
71: The aim of this paper is to shed some more light on the usefulness
72: or limitations of the depletion pair potential picture, i.e. the
73: consideration of an effective one-component fluid of large spheres
74: with effective pair interactions for describing the phase behavior
75: of a binary fluid mixture of highly asymmetric additive hard
76: spheres \cite{DRE99}. Such an issue has recently been addressed by
77: simulations \cite{BK98,B99,LW99,MH01} but the analysis
78: unfortunately provides no definitive answer due to limitations in
79: the presently available simulation algorithms.
80:
81:
82: Recently \cite{LdHT04} we have analyzed the demixing transition
83: of a binary fluid mixture of $N= N_1 + N_2$ additive hard spheres
84: of diameters $\sigma_1$ and $\sigma_2$ ($\sigma_1>\sigma_2$). For
85: a fixed diameter ratio $\gamma= \sigma_2/\sigma_1$, the
86: thermodynamic properties of the mixture were described in terms of
87: the number density $\rho=N/V$, with $V$ the volume, the partial
88: number fraction of the big spheres $x=N_1/N$, and the total
89: packing fraction $\eta= \eta_1 + \eta_2$, where
90: $\eta_1=\pi\rho\sigma_1^3 x/6$ and
91: $\eta_2=\pi\rho\sigma_1^3(1-x)\gamma^3/6$ are the packing
92: fractions of the large and the small spheres. Demixing was
93: studied by starting from the exact low-density expansions of the
94: thermodynamic properties, using as input analytical expressions
95: \cite{Kihara1,Kihara2} and numerical computations
96: \cite{Saija,Enciso1,Enciso2,Wheatley,Yu,Vlasov} for the virial
97: coefficients so far reported in the literature (up to the sixth).
98: For all the size asymmetries considered we found that, by
99: successively incorporating in the density expansions a new exact
100: virial coefficient, the critical consolute point moves to higher
101: values of the pressure and of the total packing fraction as shown
102: in Fig. $1$. It is also seen in this figure that the rate of
103: convergence is slow and so definitive conclusions on the location
104: of the critical consolute point cannot yet be reached with the
105: available data on virial coefficients. We have also compared
106: these findings with the fluid-fluid separation resulting from
107: different empirical proposals for the equation of state
108: \cite{Coussaert,Hamad,Santos2}, that attempt to reproduce the
109: virial behavior and/or to comply with consistency conditions of
110: the contact values of the radial distribution functions. We have
111: realized that for historical reasons all these ``rescaled''
112: equations of state yield the exact second and third virial
113: coefficients but inherit the singularity of the Percus-Yevick
114: theory. We have shown that this apparently innocuous singularity
115: arbitrarily introduces, as compared to the exact low-density
116: expansion of the pressure, additional dependencies on $x$ through
117: the fourth, fifth, and so on virial coefficients. Such
118: dependencies make the location of the critical consolute point to
119: vary enormously from one equation of state to another, so that the
120: predictions stemming from each of them are indeed unreliable.
121:
122: \begin{figure}[tbp]
123: \includegraphics[width=.80 \columnwidth]{car1.eps}
124: \caption{Critical total packing fraction $\eta_c$ as a function of
125: the critical reduced pressure $p_{c}^{*}$, where $p^* \equiv \beta
126: p \sigma_{1}^3$, $\beta = 1/k_B T$, with $k_B$ Boltzmann's
127: constant, and $T$ the absolute temperature, as obtained from the
128: exact low-density expansions for three values of the diameter
129: ratio $\gamma$. Successive points (from left to right) indicate
130: the addition of a new exact virial coefficient, starting with the
131: second. Lines have been drawn to guide the eye. Dash-dotted line
132: and triangles: $\gamma=0.05$; dashed line and squares:
133: $\gamma=0.1$; solid line and circles: $\gamma=0.2$. \label{fig1}}
134: \end{figure}
135:
136:
137: The demixing transition of a binary mixture of additive hard
138: spheres may be analyzed using different thermodynamic planes. In
139: order to compare to well documented results in the effective
140: one-component fluid, we first consider the ($\eta_1-\eta_2$)
141: plane. In Fig. $2$ we present the binodals resulting from the
142: exact low-density expansions up to the sixth virial coefficient
143: for two size asymmetries, namely $\gamma=0.2$ and $\gamma=0.1$,
144: and up to the fifth virial coefficient for $\gamma= 0.05$. Note
145: that the density region spanned by each binodal cannot go beyond
146: the point where the total packing fraction occupied by the spheres
147: reaches for the dense phase the highest possible value. Since
148: these values are not known, we have drawn all the curves up to the
149: limiting (unphysical) value $\eta=1$. A comparison between these
150: binodals and parallel ones derived using the depletion mechanism
151: \cite{DRE99} indicates the following. For $\gamma=0.2$, the
152: low-density expansion predicts demixing, whereas there is no
153: fluid-fluid separation in the effective one-component fluid. The
154: qualitative behavior of the binodals for $\gamma= 0.1$ and
155: $\gamma= 0.05$ is, on the other hand, rather similar in both
156: approaches. However, there are two main differences: 1) the
157: values of $\eta_2$ for the mixture are much lower than the
158: corresponding ones in the effective one-component fluid; 2) the
159: critical consolute points already lie outside the range where the
160: binodals are predicted in the effective one-component fluid.
161:
162: \begin{figure}[tbp]
163: \includegraphics[width=.80 \columnwidth]{car2.eps}
164: \caption{ Binodals of asymmetric binary fluid mixtures of
165: additive hard spheres in the ($\eta_1$-$\eta_2$) plane as obtained
166: from the exact low-density expansions (continuous lines) for the
167: same values of the diameter ratio as considered in Fig. $1$.
168: Filled circles indicate the critical consolute points. The dashed
169: lines are the results of Ref. \protect\onlinecite{DRE99}. The
170: inset in the lower pannel is the curve for $\gamma= 0.05$ in a
171: magnified scale. \label{fig2}}
172: \end{figure}
173:
174:
175: Figure $3$ is yet another presentation of the binodals,
176: this time in the ($\eta_1- p$) plane, with $p$ the
177: pressure. These binodals follow from the low-density expansions
178: by keeping the same number of exact virial coefficients as in Fig.
179: 2. Since no values of the pressure for the coexisting phases have
180: been reported in the effective one-component fluid in
181: \cite{DRE99}, a direct comparison between both approaches is not
182: possible in this case. Nevertheless, the following comments are in
183: order. It is known that a one-component fluid with pairwise
184: interactions consisting of a repulsive part and a short-range
185: attraction can only produce in the ($\eta_1- p$) plane a van der
186: Waals-like fluid-fluid separation ending in an upper critical
187: point, the concavity of the binodals pointing downwards.
188: Moreover, it is well established that, by strongly reducing the
189: range of the attractions with respect to that of the repulsions,
190: the usual phase diagram with two stable transitions, namely
191: fluid-fluid and fluid-solid, changes and the fluid-fluid
192: transition becomes metastable with respect to the fluid-solid
193: transition \cite{Gast}. By a further reduction of the range of the
194: attractions, a stable isostructural solid-solid transition
195: develops at high densities in the phase diagram
196: \cite{Frenkel,Tejero}. The phase behavior found in Ref.
197: \cite{DRE99} follows indeed this trend. Note that although the
198: zero-body and one-body terms (the volume terms) in the effective
199: one-component description play no role in determining the phase
200: equilibria, they are however responsible for inverting the
201: concavity of the binodals from downwards to upwards, yielding the
202: same concavity as the one found in the true binary mixture. The
203: importance of the volume terms in the determination of the
204: complete phase diagram should therefore not be ignored
205: \cite{Likos}. On the other hand, the resulting phase diagrams
206: found in Ref. \cite{DRE99} also indicate that, for the depletion
207: pair potential used in this reference, the small repulsive barrier
208: that follows the attractive potential close to the surface of the
209: large sphere does not qualitatively modify the phase diagram that
210: would result by only considering a purely short-range attractive
211: potential. The predictions for the phase diagram in the effective
212: one-component fluid \cite{DRE99} for $\gamma=0.1$ (a metastable
213: fluid-fluid transition) and for $\gamma= 0.05$ (a metastable
214: fluid-fluid transition and a stable solid-solid transition) are
215: indeed at grips with the foregoing remarks.
216:
217: \begin{figure}[tbp]
218: \includegraphics[width=.80 \columnwidth]{car3.eps}
219: \caption{ Binodals of asymmetric binary fluid mixtures of
220: additive hard spheres in the ($\eta_1$-$p^*$) plane as obtained
221: from the exact low-density expansions for the same values of the
222: diameter ratio as considered in Figs. \protect\ref{fig1} and
223: \protect\ref{fig2}. Dash-dotted line: $\gamma=0.05$; dashed line:
224: $\gamma=0.1$; solid line: $\gamma=0.2$. Filled circles indicate
225: the critical consolute points. \label{fig3}}
226: \end{figure}
227:
228:
229: It should be clear that due to the fact that only a few exact
230: virial coefficients are available and that the radius of
231: convergence of the virial series in the case of mixtures is not
232: known, definite conclusions derived from the low-density
233: expansions cannot be drawn. Nevertheless, the observed trends lead
234: to the following considerations. As shown in Fig. 1, as one
235: increases the number of terms in the low-density expansions, both
236: the critical pressure and the critical total packing fraction
237: grow. It is not clear whether this feature will change by
238: including more exact virial coefficients (when available) in the
239: density expansions, but all the evidence so far makes it unlikely.
240: In any event, provided the trend obtained with the low-density
241: expansions is a true feature, two possible scenarios arise. One
242: may envisage the convergence of the critical constants to a
243: thermodynamic state with either a physical or an unphysical
244: critical total packing fraction. Indeed, the terms physical and
245: unphysical depend of the resulting solid phase at high densities.
246: For example, if the freezing transition corresponds to the partial
247: freezing of the large spheres, i.e. a face-centered cubic lattice
248: formed by the large spheres while the small spheres remain
249: disordered \cite{Coussaert}, the close packing fraction of the
250: large spheres $\eta_1^{\rm cp}\simeq 0.74$ defines the boundary
251: between physical and unphysical. In the first scenario the
252: demixing transition would be thermodynamically metastable with
253: respect to the freezing transition as found in previous studies,
254: while in the second no demixing would be present at all. At this
255: stage we have no way to decide which of the two scenarios is
256: correct since, by incorporating up to the fifth (sixth) virial
257: coefficient in the exact low-density expansions for $\gamma=0.05$
258: ($0.1$),
259: the critical packing fraction of the big
260: spheres $\eta_{1}^{c}$ is $\eta_{1}^{c}= 0.595 (0.557)$.
261: Concerning this point, a related analysis carried out with the
262: Boubl\'{\i}k-Mansoori-Carnahan- Starling-Leland (BMCSL) equation
263: of state \cite{Boublik} (which predicts no deximing) is
264: illustrative. This is shown in Fig. \ref{fig4}. The striking
265: feature to be noted is the excellent agreement of the results
266: obtained with the exact low-density expansion ({\em c.f.} Fig.
267: \ref{fig1}) and those obtained with the truncated BMCSL equation
268: of state. By including additional virial coefficients of the BCMSL
269: equation of state it is seen that $\eta_c\simeq \eta_{1}^{c}$
270: exceeds the close-packing fraction of the large spheres.
271:
272: \begin{figure}[tbp]
273: \includegraphics[width=.90 \columnwidth]{car4.eps}
274: \caption{Critical total packing fraction $\eta_c$ as a function of
275: the critical reduced pressure $p_{c}^{*}$ as obtained from the
276: truncation of the virial expansion of the BMCSL equation of state
277: for two of the values of the diameter ratio $\gamma$ considered in
278: Fig. \protect\ref {fig1}. Successive points (from left to right)
279: indicate that an additional (approximate) virial coefficient has
280: been retained in the truncation, starting with the second. Lines
281: have been drawn to guide the eye. Dash-dotted line and triangles:
282: $\gamma=0.05$; dashed line and squares: $\gamma=0.1$.
283: \label{fig4}}
284: \end{figure}
285:
286: In order to understand the differences between the phase diagrams
287: presented in Fig. 2 using the exact low-density expansions and the
288: effective one-component description, the following considerations
289: should be pointed out. For $\gamma= 0.2$ both approaches would be
290: equivalent whenever the critical packing fraction of the big
291: spheres resulting from the low-density expansions converges to an
292: unphysical value. As indicated above, for $\gamma= 0.05$ and
293: $\gamma= 0.1$ the critical consolute points obtained with a
294: reduced number of exact virial coefficients already lie outside
295: the range where the effective one-component description locates
296: the binodals. In these cases the effective one-component
297: description would only provide a qualitative approximation of the
298: true binary mixture if $\eta_{1}^{c}$ converges to a physical
299: value. Otherwise (convergence to an unphysical value) the
300: predictions of both approaches would be completely at odds with
301: each other in this basic issue. All these possibilities can
302: indeed only be tested by including more exact virial coefficients
303: in the density expansions. As a possible source for the
304: differences found between the exact density expansions and the
305: effective one-component description we suggest that it may be due
306: to the fact that, in order to determine in the latter the phase
307: diagram and to perform the direct simulations \cite{DRE99}, the
308: Carnahan-Starling equation of state is used for describing both
309: the big and the small hard spheres. Note that the
310: Percus-Yevick-type singularity of the Carnahan-Starling equation
311: of state may induce for each component different (through $\eta_1$
312: and $\eta_2$) spurious $x$-dependences similar to those discussed
313: in Ref. \onlinecite{LdHT04} in connection with equations of state
314: derived from ``rescaled'' virial expansions. These dependencies
315: would in turn lead to unreliable predictions on the location of
316: the demixing transition.
317:
318: Up to here and due to the involved approximations, none of the
319: arguments and results that we have discussed lead to a neat
320: conclusion concerning the claimed equivalence of the effective
321: one-component description vis a vis the true binary mixture
322: description of highly asymmetric hard spheres. Therefore, new
323: theoretical approaches to deal with this problem should be
324: sought. In this respect, systematic and controlled approximations
325: in the fluid regime, may profitably be undertaken. Of course these
326: are presently limited by the available virial coefficients. Also,
327: the role of reliable simulations capable of exploring the region
328: of interest can hardly be overemphasized. Irrespective of the
329: technical difficulties that no doubt are involved both in the
330: simulations and in the computation of additional virial
331: coefficients of highly asymmetric binary fluid mixtures of
332: additive hard spheres, we hope that the present paper may also
333: serve to stimulate efforts in this direction.
334:
335: C.F.T. acknowledges financial support from the Mi- nisterio de
336: Ciencia y Tecnolog\'{\i}a (Spain) Ref: BFM2001-1017-C03-03. This
337: work has also benefitted from the Collaboration Agreement between
338: Universidad Nacional Aut\'{o}noma de M\'{e}xico and Universidad
339: Complutense de Madrid.
340:
341:
342:
343: %\vspace{0.2cm}
344:
345: %$^a$ cftejero@fis.ucm.es; permanent address: Facultad de Ciencias
346: %F\'{\i}sicas, Universidad Complutense de Madrid, E-28040 Madrid
347: %(Spain)
348:
349: %$^b$ malopez@servidor.unam.mx;
350: %http://miquiztli.cie.unam.mx/xml/tc/ft/mlh/
351:
352:
353: \begin{references}
354:
355:
356: \bibitem{Likos} For a review see C. N. Likos, Physics Reports {\bf
357: 348}, 267 (2001).
358:
359: \bibitem{Asakura} S. Asakura and F. Oosawa, J. Chem. Phys. {\bf 22}, 1255
360: (1954).
361:
362: \bibitem{Gotzelmann} B. G\"{o}tzelmann, R. Evans, and S. Dietrich, Phys.
363: Rev. E {\bf 57}, 6785 (1998).
364:
365: \bibitem{Roth} R. Roth, R. Evans, and S. Dietrich, Phys. Rev. E {\bf 62}, 5360
366: (2000).
367:
368:
369: \bibitem{DRE99} M. Dijkstra, R. Roij and R. Evans,
370: Phys. Rev. Lett. {\bf 81}, 2268 (1998); {\em ibid} {\bf 82}, 117
371: (1999); Phys. Rev. E {\bf 59}, 5744 (1999).
372:
373: \bibitem{BK98} A. Buhot and W. Krauth, Phys. Rev. Lett. {\bf 80}, 3787
374: (1998).
375:
376: \bibitem{B99} A. Buhot, Phys. Rev. Lett. {\bf 82}, 960 (1999).
377: %A. A. Louis, Phys. Rev. Lett. {\bf 84}, 1840 (2000); A. Buhot,
378: %Phys. Rev. Lett. {\bf 84}, 1841 (2000).
379:
380: \bibitem{LW99} L. Lue and L. V. Woodcock, Mol. Phys. {\bf 96}, 1435
381: (1999).
382:
383: \bibitem{MH01} J. G. Malherbe and S. Amokrane, Mol. Phys. {\bf 99}, 355
384: (2001).
385:
386: \bibitem{LdHT04} M. L\'opez de Haro and C. F. Tejero, J. Chem.
387: Phys. {\bf 121}, 6918 (2004); arXiv: cond-mat/ 0408402 (2004).
388:
389:
390: \bibitem{Kihara1} T. Kihara, Rev. Mod. Phys. {\bf 27}, 412 (1955).
391:
392: \bibitem{Kihara2} T. Kihara and K. Miyoshi, J. Stat. Phys. {\bf 13}, 337
393: (1975).
394:
395: \bibitem{Saija} F. Saija, G. Fiumara and P. V. Giaquinta, Mol. Phys. {\bf 87},
396: 991 (1996).
397:
398: \bibitem{Enciso1} E. Enciso, N. G. Almarza, D. S. Calzas and M. A.
399: Gonz\'{a}lez, Mol. Phys. {\bf 92}, 173 (1997).
400:
401: \bibitem{Enciso2} E. Enciso, N. G. Almarza, M. A. Gonz\'{a}lez and F. J.
402: Bermejo, Phys. Rev. E {\bf 57}, 4486 (1998).
403:
404: \bibitem{Wheatley} R. J. Wheatley, F. Saija and P. V. Giaquinta, Mol. Phys.
405: {\bf 94}, 877 (1998).
406:
407: \bibitem{Yu} A. Yu. Vlasov, X. M. You and A. J. Masters, Mol. Phys. {\bf 100},
408: 3313 (2002).
409:
410: \bibitem{Vlasov} A. Yu. Vlasov and A. J. Masters, Fluid Phase Equilibria {\bf
411: 212}, 183 (2003).
412:
413: \bibitem{Coussaert} T. Coussaert and M. Baus, Phys. Rev. Lett. {\bf 79}, 1881
414: (1997); {\em ibid} {\bf 80}, 4832 (E) (1998); J. Chem. Phys. {\bf
415: 109}, 6012 (1998).
416:
417: \bibitem{Hamad} E. Z. Hamad, J. Chem. Phys. {\bf 103}, 3733 (1995).
418:
419: \bibitem{Santos2} A. Santos, S. B. Yuste and M. L\'{o}pez de Haro, Mol. Phys.
420: {\bf 96}, 1 (1999); J. Chem. Phys. {\bf 117}, 5785 (2002).
421:
422:
423: %\bibitem{Haro} Note that the Percus-Yevick-type singularity of the
424: %Carnahan-Starling equation of state induces for each component
425: %different spurious $x$-dependences similar to those discussed in
426: %Ref. \onlinecite{LdHT04} in connection with ``rescaled'' virial
427: %expansions.
428:
429: \bibitem{Gast} A. P. Gast, C. K. Hall, and W. B. Russel, J. Colloid. Interface
430: Sci. {\bf 96}, 251 (1983).
431:
432: \bibitem{Frenkel} P. Bolhuis and D. Frenkel, Phys. Rev. Lett. {\bf 72}, 2211
433: (1994).
434:
435: \bibitem{Tejero} C. F. Tejero, A. Daanoun, H. N. W. Lekkerkerker, and M. Baus,
436: Phys. Rev. Lett. {\bf 73}, 752 (1994).
437:
438: \bibitem{Boublik} T. Boubl\'{\i}k, J. Chem. Phys. {\bf 53}, 471 (1970); G. A.
439: Mansoori, N. F. Carnahan, K. E. Starling and T. W. Leland, J.
440: Chem. Phys. {\bf 54}, 1523 (1971).
441:
442:
443:
444: \end{references}
445:
446:
447: \end{document}
448: