1: \documentclass[10pt]{article}
2: %\usepackage{amsmath,amssymb}
3: \usepackage{multicol,citesort,epsfig,overcite}
4: \setlength{\oddsidemargin}{-0.4in}
5: \setlength{\evensidemargin}{-0.4in}
6: \setlength{\textwidth}{7.2in}
7: \setlength{\topmargin}{-0.2in}
8: \setlength{\headheight}{0in}
9: \setlength{\headsep}{0in}
10: \setlength{\textheight}{9.8in}
11: \setlength{\footskip}{.5in}
12: \setlength{\lineskip}{2pt}
13: \setlength{\lineskiplimit}{2pt}
14:
15:
16: \begin{document}
17:
18: \title{Nucleation of a non-critical phase in a fluid near a critical point}
19:
20:
21: \author{{\bf Richard P. Sear}\\
22: ~\\
23: Department of Physics, University of Surrey\\
24: Guildford, Surrey GU2 7XH, United Kingdom\\
25: email: r.sear@surrey.ac.uk}
26: %tel. +44 (0)1483 876793\\
27: %fax +44 (0)1483 876781}
28: %\date{}
29:
30: \maketitle
31:
32: \begin{abstract}
33: Phase diagrams of some globular proteins have a fluid-fluid
34: transition as well as a fluid-crystal transition.
35: Homogeneous nucleation of the crystal from the fluid phase
36: near the critical point
37: of the fluid-fluid transition is examined.
38: As the fluid-fluid critical point is approached, the number of
39: molecules in the critical nucleus, the nucleus at the top of the
40: free energy barrier to nucleation, is found to diverge as the isothermal
41: compressibility.
42: This divergence is due to a layer of the fluid phase of width
43: equal to the fluid's correlation length which surrounds
44: the core of the nucleus; the number of molecules in a crystalline
45: environment in the nucleus does not diverge.
46: The free energy barrier to nucleation remains finite but
47: its derivative with respect to the chemical potential is equal to minus
48: the number of molecules in the critical nucleus and so
49: diverges.
50: \end{abstract}
51:
52: %\newpage
53: \section{Introduction}
54:
55: The phase behaviour of a number of globular proteins has been studied
56: and in addition to the fluid-to-crystal transition there is a
57: metastable fluid--fluid transition
58: \cite{broide91,muschol97}, i.e., a transition between two
59: fluid phases differing in density which ends at a critical point.
60: This is analogous to the vapour--liquid transition of
61: simple substances such as water.
62: The transition is metastable: it exists within the fluid--crystal
63: coexistence region.
64: A fluid near the fluid-fluid critical point
65: is anomalous, essentially because the correlation length is very large.
66: Here we consider the free energy barrier $\Delta\Omega^*$
67: to crystallisation in a fluid which is close to the critical point.
68: Crystallisation starts by the nucleation
69: of a microscopic crystallite, this then grows to form a crystal.
70: The rate at which such microscopic crystallites form is proportional
71: to $\exp(-\Delta\Omega^*/kT)$ because their formation is
72: an activated process \cite{debenedetti,gunton}.
73: Here we study only the size and free energy
74: of the critical nucleus, we
75: leave consideration both of the dynamics of its formation and
76: of its growth to later work. We also limit ourselves
77: to temperatures above or equal to the critical temperature.
78: Our main finding is that the number
79: of molecules in the critical nucleus varies as $\chi_T$,
80: the isothermal compressibility, near the critical point.
81: As $\chi_T$ diverges at the critical point, so does the number of
82: molecules in the nucleus. Within our theory the number of
83: molecules in the critical nucleus is equal to minus the derivative
84: of the free energy barrier with respect to the chemical potential and
85: so near the critical point the barrier to nucleation is a rapidly
86: decreasing function of chemical potential.
87:
88: We focus on the universal aspects of critical nuclei near critical
89: points \cite{critnote}. It is well-known that near a critical point
90: fluids exhibit universal behaviour, behaviour which
91: is determined solely by the universality
92: class of the system \cite{chaikin}, here that of the three-dimensional
93: Ising model. Although in the centre of a critical nucleus the density
94: will be far from the critical density,
95: at the fringes of the nucleus far from the centre the density
96: will be close to the bulk density of the fluid, and there in the
97: fringes we find universal behaviour. We find that the fringe of the
98: nucleus dominates the number of molecules in the nucleus but not
99: the free energy. It does however dominate the derivative of the
100: free energy with respect to the chemical potential
101: as this is nothing but minus the number of molecules in the nucleus.
102: In order to focus on the universal aspects we use a simple
103: phenomenological theory and for simplicity we use a mean-field theory,
104: although such theories have well-known deficiencies near critical points
105: \cite{chaikin}. We use a simple theory first used for droplets
106: by Cahn and Hilliard \cite{cahn59}. It is variously called
107: Cahn-Hilliard theory, van der Waals-Cahn-Hilliard theory or
108: the square gradient approximation
109: \cite{cahn59,cahn58,widom85,evans89,debenedetti}.
110:
111: Numerical work on droplets near a critical point has been done
112: by Talanquer and Oxtoby \cite{talanquer98} using a theory similar to a
113: Cahn-Hilliard theory. This followed pioneering computer simulations
114: of a critical nucleus near a critical point by ten Wolde and
115: Frenkel \cite{tenwolde97}.
116: Talanquer and Oxtoby obtained nuclei
117: with very large numbers of molecules near a critical point but did
118: not perform the analytical analysis required to extract out the
119: scaling near a critical point. Although their theory
120: has an additional order parameter the scaling of the number of
121: molecules near the critical point in their model is almost
122: certainly the same as found here. See Refs.
123: \citen{haas00,dixit00,searxxx} for other recent theoretical work
124: on nucleation near a metastable transition.
125: Also, more than twenty years ago,
126: Widom \cite{widom85,widom77} used Cahn-Hilliard theory to look at the
127: (planar) interface between coexisting phases, where one of these
128: phases was at a critical point. He envisaged not a single
129: component system with fluid-fluid and fluid-crystal phase
130: transitions but a binary mixture with liquid-liquid demixing
131: and vapour-liquid phase transitions. However, the universal
132: aspects are the same for both systems.
133: Thus, our calculations for the radial
134: density profile of the nucleus are analogous to those for density
135: profile of the planar interface found by Widom. In addition,
136: our results are relevant to the nucleation of the vapour phase near the
137: critical point of a metastable liquid-liquid demixing transition.
138:
139:
140: In the next section we derive analytic expressions, within a mean-field
141: theory, for the density in the fringe of a nucleus in a near-critical fluid.
142: Then we go on to derive an expression for the number of molecules, and
143: hence the derivative of the free energy barrier near the critical point.
144: The third and final section is a conclusion.
145:
146:
147:
148:
149:
150: \section{Theory}
151:
152: The critical nucleus is at the top of the barrier, it is at a maximum
153: in the excess grand potential.
154: The excess grand potential is the grand potential with a nucleus minus that
155: without a nucleus. To start we require an expression for
156: the excess grand potential of a nucleus $\Delta\Omega$ as a functional
157: of the density function of the nucleus and then to find the density function
158: which extremises this functional. For a crystalline nucleus, a
159: rather complex functional is required, see for example the functional
160: used to calculate the fluid-crystal interfacial tension for hard
161: spheres \cite{ohnesorge94}.
162: However, the universal effects we are looking for here
163: occur at the fringe of the nucleus where the nucleus will have a density
164: close to that of the fluid. There a much simpler functional suffices,
165: such as those
166: used to calculate fluid-fluid interfacial tensions and the excess
167: grand potentials of fluid droplets in fluid phases.
168:
169: For a spherically symmetric fluid droplet in a fluid, the
170: density profile $\phi(r)$ is a function only of the distance $r$
171: from the centre of the droplet. The fluid
172: phase has a density $\rho$ and a chemical potential $\mu$;
173: $\phi$ is the density at a point
174: whereas $\rho$ is the bulk density of the fluid. Of
175: course $\phi(r\rightarrow\infty)=\rho$.
176: The standard Cahn-Hilliard expression for the
177: grand potential cost $\Delta\Omega$ of such a droplet, as a functional
178: of its density profile $\phi(r)$, is
179: \cite{cahn58,cahn59,widom85,evans89,debenedetti}
180: \begin{equation}
181: \Delta\Omega = \int\left[
182: \Delta\omega + \kappa\left(\nabla\phi\right)^2\right]
183: {\rm d}{\bf r},
184: \label{cahn}
185: \end{equation}
186: where
187: \begin{equation}
188: \Delta\omega(\phi)=f(\phi)-f(\rho)-\mu(\phi-\rho),
189: \label{omega}
190: \end{equation}
191: is the work required per unit volume to change the density from $\rho$
192: to $\phi$, at a chemical potential $\mu$.
193: $f(\phi)$ is the bulk Helmholtz free energy per unit volume of the
194: fluid at a density $\phi$.
195: The second term in Eq. (\ref{cahn})
196: within the brackets is the gradient term: the
197: grand potential cost due to variations in space of the density.
198: The gradient squared term is the lowest order term in a gradient
199: expansion and so is adequate when the density is slowly
200: varying. The coefficient, $\kappa$ of this term is
201: taken to be a constant. See
202: Refs. \citen{widom85,evans89,debenedetti} for its relation to the
203: intermolecular potential.
204:
205: The functional Eq. (\ref{cahn}) will be totally inadequate within the
206: crystalline core of a nucleus but we will not require either the density
207: function or the contribution to the grand potential of this core.
208: Thus, we will use Eq. (\ref{cahn}) as our functional and end up
209: with expressions which are the sum of two terms, one term from the core
210: for which our expression will be wrong but which we will not
211: evaluate, and another term from the fringe, for which Eq. (\ref{cahn})
212: is a reasonable (mean-field) approximation which we will evaluate.
213:
214: The critical nucleus is at the top of the free energy barrier
215: and so is at a maximum of $\Delta\Omega$. Thus, for the
216: critical nucleus we may set the functional derivative of
217: $\Delta\Omega$ with respect to the density profile $\phi(r)$
218: to zero,
219: \begin{equation}
220: \left(\frac{\partial \Delta\omega}{\partial\phi}\right)-2\kappa\nabla^2\phi
221: =0.
222: \label{temp}
223: \end{equation}
224: The curvature of the density profile at a point is proportional
225: to the derivative of the excess grand potential with respect to
226: the density at that point. Using Eq. (\ref{omega}) for $\Delta\omega$
227: in Eq. (\ref{temp}) we have
228: \begin{equation}
229: \mu(\phi)-\mu-2\kappa\nabla^2\phi =0,
230: \label{max}
231: \end{equation}
232: where the $\mu$ without an argument is the
233: chemical potential in the fluid, and $\mu(\phi)$ is the chemical
234: potential of a bulk fluid at a density $\phi$.
235: Once we have solved Eq. (\ref{max}) we can insert the solution
236: into Eq. (\ref{cahn}) to obtain the excess grand potential of the
237: critical nucleus, denoted by $\Delta\Omega^*$.
238:
239:
240: The excess number of molecules in the critical nucleus, $\Delta n^*$,
241: is the number of molecules with the nucleus present minus the number
242: without it. It may be obtained by integrating over the density profile,
243: \begin{equation}
244: \Delta n^*=
245: \int \Delta\phi(r) {\rm d}{\bf r},
246: \label{nstardef}
247: \end{equation}
248: where $\Delta\phi(r)=\phi(r)-\rho$ is the excess density at a point,
249: the density at a point minus the bulk density.
250: The derivative of $\Delta\Omega^*$ with respect
251: to the chemical potential of the fluid $\mu$ is, using
252: Eq. (\ref{cahn}),
253: \begin{equation}
254: \left(\frac{\partial\Delta\Omega^*}{\partial\mu}\right)_T=
255: -\int \Delta\phi(r) {\rm d}{\bf r}
256: =-\Delta n^*.
257: \label{nstar}
258: \end{equation}
259: Thus, the derivative of $\Delta\Omega^*$ with respect to
260: the chemical potential is simply $-\Delta n^*$.
261: This result, that the
262: derivative of the excess grand potential of the critical nucleus
263: is minus the excess number of molecules in the nucleus,
264: is often called the nucleation theorem
265: \cite{kashchiev82,viisanen93,bowles00}.
266:
267:
268: \subsection{The fringe of the nucleus}
269:
270: Equation (\ref{max}) can be solved numerically if the chemical potential
271: is known as a function of density. Here, we would like to
272: concentrate on the fringe of the critical nucleus. The
273: fringe is the outermost part of the nucleus, where
274: the density is near the density of the fluid, $\rho$.
275: We define the fringe as being that part of the nucleus which
276: is more than a distance $r_c$ from the centre.
277: The distance $r_c$ is such that $\Delta\phi(r\ge r_c)/\rho\ll 1$.
278: Fig. \ref{figprof} is
279: a schematic of the radial density profile of the nucleus.
280: %The fringe not only has a density near to $\phi$ but the density
281: %varies more slowly with $r$ than in the core of the nucleus.
282: Now, if the fractional density difference $\Delta\phi/\rho\ll 1$
283: we can use a Taylor expansion for $\mu(\phi)-\mu$,
284: \begin{equation}
285: \mu(\phi)-\mu=\Delta\phi
286: \left(\frac{\partial\mu(\phi)}{\partial\phi}\right)_{\phi=\rho}+\cdots
287: =\frac{\Delta\phi}{\rho^2\chi_T}+\cdots,
288: \label{taylor}
289: \end{equation}
290: where $\chi_T$ is the isothermal compressibility of the fluid (at
291: a density $\rho$). The isothermal compressibility is defined as
292: \cite{hansen86}
293: \begin{equation}
294: \chi_T^{-1}=\rho^2\left(\frac{\partial\mu}{\partial\rho}\right)_{V,T}.
295: \label{chi}
296: \end{equation}
297: Substituting Eq. (\ref{taylor}) into Eq. (\ref{max}) we have
298: \begin{equation}
299: \frac{1}{\rho^2\chi_T}\Delta\phi(r)-
300: 2\kappa\nabla^2\Delta\phi(r) =0 ~~~~~~~ \Delta\phi\ll\rho.
301: \label{helm}
302: \end{equation}
303: This is the Helmholtz equation, for $\Delta\phi$,
304: and the solution is a function
305: of the Ornstein-Zernike form,
306: \begin{equation}
307: \Delta\phi(r)=\rho\frac{\sigma}{r}\exp(-r/\xi),
308: \label{oz}
309: \end{equation}
310: with $\xi$ the correlation length of the fluid, given by
311: \begin{equation}
312: \xi^2=2\kappa\rho^2\chi_T.
313: \label{xi}
314: \end{equation}
315: To obtain Eq. (\ref{oz}) the boundary conditions
316: $\Delta\phi(r\rightarrow\infty)\rightarrow0$ and
317: $\Delta\phi(r_c)=\rho(\sigma/r_c)\exp(-r_c/\xi)$
318: were employed. $\sigma$ is a molecular length scale, a few nms for
319: proteins. $\Delta\phi/\rho$ will, as required
320: for Eq. (\ref{helm}), be small for $r\ge r_c$ provided
321: that $r_c$ is a few times $\sigma$ or more. We do not need to specify
322: $r_c$ beyond saying that it must be at least a few times $\sigma$.
323: This implies a core a few molecules across, which is reasonable.
324: From Eq. (\ref{oz}) we see that
325: the width of the fringe is, as we might have expected,
326: of the order of the correlation length $\xi$ in the surrounding fluid.
327: This width will thus diverge as the fluid approaches a critical point.
328:
329: \subsection{Near a critical point}
330:
331: We now consider a nucleus in a fluid which is near a critical point,
332: either at the critical density but just above the critical temperature,
333: or at the critical temperature but at a density near the
334: critical density. In either case as the critical point is approached the
335: isothermal compressibility, and
336: hence (see Eq. (\ref{xi})) $\xi$, diverges. Within mean-field theory,
337: along the critical isochore, i.e., with the density fixed at its value
338: at the critical point $\rho_{cp}$, $\chi_T$ scales with temperature
339: as \cite{chaikin,debenedetti}
340: \begin{equation}
341: \chi_T \sim (T-T_{cp})^{-1} ~~~~~~~~~\rho=\rho_{cp} ~~~T>T_{cp},
342: \label{isoch}
343: \end{equation}
344: where $T_{cp}$ is the temperature of the critical point.
345: The compressibility diverges as one over the
346: temperature difference to the critical point. An alternative
347: path to the critical point is along the critical isotherm.
348: We fix the temperature $T=T_{cp}$ and vary the density. Along the
349: critical isotherm $\chi_T$ varies as
350: \begin{equation}
351: \chi_T \sim (\rho-\rho_{cp})^{-2} ~~~~~~~~~ T=T_{cp}.
352: \label{isoth}
353: \end{equation}
354: The compressibility diverges as one over the square of the
355: density difference to the critical point.
356:
357: Putting our solution for $\Delta\phi(r>r_c)$, Eq. (\ref{oz}),
358: into Eq. (\ref{nstardef})
359: we find that when $\xi$ is very large and so
360: the fringe correspondingly large in volume, that the fringe
361: dominates $\Delta n^*$. Equation
362: (\ref{nstardef}) is easily evaluated by substituting Eq.
363: (\ref{oz}) for $\Delta\phi(r>r_c)$,
364: \begin{eqnarray}
365: \Delta n^*
366: &=&
367: \int_{r=0}^{r=r_c}
368: \Delta\phi(r) {\rm d}{\bf r}
369: +4\pi\rho\sigma\xi^2~~~~~\xi\gg r_c
370: \nonumber\\
371: &=&
372: \int_{r=0}^{r=r_c}
373: \Delta\phi(r) {\rm d}{\bf r}
374: +8\pi\sigma\kappa\rho^3\chi_T~~~~~\xi\gg r_c,
375: \label{ncrit}
376: \end{eqnarray}
377: where we used the relation between $\xi$ and $\chi_T$, Eq. (\ref{xi}).
378: The excess number of molecules $\Delta n^*$ is a sum of two terms.
379: The first term, the integral in Eq. (\ref{ncrit}), comes from
380: the core of the nucleus. It is of order
381: $(\rho_c-\rho)r_c^3$ where $\rho_c$ is the density of the
382: phase which is nucleating. It remains finite as the critical
383: point is approached. However, the second term
384: in Eq. (\ref{ncrit}), the contribution
385: of the fringe to the number
386: of molecules in the critical nucleus $\Delta n^*$, diverges.
387: As the critical point is approached the number of molecules
388: in the nucleus tends to infinity, as suggested by
389: Talanquer and Oxtoby \cite{talanquer98}. It does so as the compressibility
390: $\chi_T$. This is the central result of this work, and
391: is a general result independent of the nature of the phase which is
392: nucleating.
393: The fringe of the nucleus is at densities close to the fluid's
394: density and far from that in the core of the nucleus. So,
395: the divergence in $\Delta n^*$ is from a divergent number of molecules
396: at densities near the bulk fluid density: although they are part of
397: a nucleus of a crystalline phase they themselves are in a fluid
398: environment.
399: The number of molecules which are in a crystalline environment
400: does not diverge as the critical point is approached.
401: Thus, within mean-field theory, along the critical
402: isochore the number of molecules diverges as
403: $(T-T_{cp})^{-1}$ and along the critical isotherm as ($\rho-\rho_{cp})^{-2}$.
404:
405: Now we consider the contribution of the fringe to the grand potential
406: of the critical nucleus. We require the Taylor
407: expansion of $\Delta\omega$ about its value at the bulk density,
408: \begin{equation}
409: \Delta\omega(\phi)=\frac{1}{2}
410: \frac{(\Delta\phi)^2}{\rho^2\chi_T}+\cdots.
411: \label{quad}
412: \end{equation}
413: The first nonzero term in the Taylor expansion of $\Delta\omega$
414: is the quadratic term as the first density derivative of $\Delta\omega$
415: is zero at $\rho$. Now, splitting the integration in Eq. (\ref{cahn})
416: at $r_c$, and using the quadratic approximation, Eq. (\ref{quad}),
417: for $\Delta\omega$ for $r>r_c$, we have
418: \begin{equation}
419: \Delta\Omega =
420: \int_{r=0}^{r=r_c}\left[
421: \Delta\omega + \kappa\left(\nabla\phi\right)^2\right]{\rm d}{\bf r}+
422: \int_{r=r_c}^{r=\infty}\left[\frac{1}{2}
423: \frac{(\Delta\phi)^2}{\rho^2\chi_T}
424: + \kappa\left(\nabla\Delta\phi\right)^2\right]
425: {\rm d}{\bf r},
426: \label{cahnlin}
427: \end{equation}
428: where, for $r>r_c$,
429: we also substituted $\Delta\phi$ for $\phi$ in the gradient term.
430: After substituting Eq. (\ref{oz}) for $\Delta\phi$ in the
431: second integral,
432: we see that the contributions to the integral from both the
433: terms in the integrand {\em decreases} as a function of radial
434: distance $r$. Taking into account the factor of $r^2$
435: in ${\rm d}{\bf r}$ the integrand varies as $\exp(-2r/\xi)$ times a quadratic
436: polynomial in $1/r$: it is a monotonically decreasing function of $r$.
437: The farther we go out into the fringe of the nucleus the
438: smaller is the contribution to the excess grand potential,
439: the barrier to nucleation. This means that the barrier to nucleation
440: will be dominated by contributions from the core of the nucleus
441: where the behaviour is non-universal and our approximations are invalid.
442: Recall that our functional, Eq. (\ref{cahn}), is a very
443: poor approximation in the core and hence the first
444: integral in Eq. (\ref{cahnlin}) is not an accurate expression for
445: the contribution of the core to the excess grand potential.
446: We are thus unable to calculate the absolute excess
447: grand potential of a critical nucleus
448: near a critical point but we are of course able to calculate its
449: derivative with respect to the chemical potential as this is just
450: minus the excess number of molecules in the critical nucleus,
451: which is dominated by the fringe.
452:
453:
454: \section{Conclusion}
455:
456: Some globular proteins \cite{broide91,muschol97} have two transitions:
457: a fluid-crystal transition, call it transition $\alpha$, and
458: a fluid-fluid transition, call it transition $\beta$. Transition
459: $\alpha$ is always strongly first order but transition $\beta$
460: becomes continuous at the critical point. Here, we have studied
461: homogeneous nucleation associated with transition $\alpha$ near
462: a continuous transition $\beta$.
463: Systems near a continuous transition
464: show universal (within a universality class)
465: behaviour and are very susceptible to a perturbation ---
466: response functions such as the compressibility take large values.
467: The nucleus associated with homogeneous nucleation of a
468: transition $\alpha$ therefore perturbs the highly
469: susceptible fluid near transition $\beta$.
470: Because the fluid is near this continuous
471: transition, the response functions such as the isothermal
472: compressibility $\chi_T$, are large. The perturbation due to the
473: nucleus is therefore large, it extends out to the large distance $\xi$ and
474: involves a number of molecules which scales with $\chi_T$.
475: The free energy barrier to homogeneous nucleation of transition
476: $\alpha$ is then a rapidly decreasing function of chemical potential.
477: As suggested by ten Wolde and Frenkel \cite{tenwolde97}, the
478: continuous transition $\beta$ helps the dynamics of transition
479: $\alpha$.
480: This is true in general, not just for crystallisation near
481: a critical point. The nucleation of the vapour phase when
482: a mixture of liquids boils near a metastable critical point of liquid-liquid
483: demixing is another example. Note that there $\Delta n^*$ is negative
484: and will tend towards $-\infty$ not $+\infty$.
485:
486: The theory of section 2 is a simple mean-field theory. However,
487: our conclusion that $\Delta n^*\sim\chi_T$ near the critical
488: point may be valid beyond mean-field theory. If so then
489: the divergence of $\Delta n^*$ along a particular path in the
490: diagram, e.g., the critical isochore, will occur with whatever is the
491: correct critical exponent of $\chi_T$ along that path.
492: For example, for nucleation near a
493: critical point in the universality class of the Ising model
494: in three dimensions, $\chi_T$ scales as
495: $(T-T_{cp})^{-1.24}$ along the critical isochore not
496: $(T-T_{cp})^{-1}$ as mean-field theory predicts, see
497: Refs. \citen{debenedetti,chaikin}.
498:
499:
500:
501:
502: The starting point for most theoretical studies of nucleation is
503: classical nucleation theory \cite{debenedetti,gunton}.
504: Within classical
505: nucleation theory the critical nucleus is modeled as a small crystallite,
506: assumed to have bulk properties, separated from the surrounding
507: metastable fluid by an interface, assumed to have the same surface
508: tension as that between the coexisting bulk phases and to be thin,
509: no more than a couple of molecules thick. For a nucleus in a fluid
510: far from a fluid-fluid critical point, this assumption of a thin
511: interface with the bulk surface tension is not unreasonable.
512: The correlation lengths in the fluid and crystalline phases
513: will both be small and so will the width of the interface. Then if
514: the width of the interface is small with respect to the diameter
515: of the nucleus then approximating its surface tension by that
516: of the planar interface between coexisting phases is reasonable.
517: However, for a nucleus in a near-critical fluid, the correlation
518: length in the surrounding fluid $\xi$ may be very large.
519: So, for a crystal nucleus which is only a few lattice spacings
520: across the interface between it and the surrounding
521: fluid may be much thicker than the diameter of the crystalline
522: part of the nucleus. Then this interface will be very different from
523: that assumed within classical nucleation theory, and
524: so classical nucleation theory will yield only a poor estimate
525: of the barrier to nucleation.
526:
527:
528: It is a pleasure to thank R. Bowles for
529: useful discussions.
530: Work supported by EPSRC (GR/N36981).
531:
532:
533:
534:
535:
536: %\newpage
537: \begin{thebibliography}{99}
538:
539: \bibitem{broide91} M. L. Broide, C. R. Berland, J. Pande, O. O. Ogun
540: and G. B. Benedek, Proc. Nat. Acad. Sci. {\bf 88}, 5660 (1991).
541:
542: \bibitem{muschol97} M. Muschol and F. Rosenberger,
543: J. Chem. Phys. {\bf 107}, 1953 (1997).
544:
545: \bibitem{debenedetti} P. G. Debenedetti,
546: {\it Metastable Liquids}
547: (Princeton University Press, Princeton, 1996).
548:
549: \bibitem{gunton} J. D. Gunton, M. San Miguel and P. S. Sahni,
550: in {\it Phase Transitions} volume 8, edited by C. Domb and J. L. Lebowitz
551: (Academic Press, London, 1983).
552:
553: \bibitem{critnote}
554: Note that conventionally `critical'
555: is used to denote both a point in the phase diagram and the crystallite
556: at the top of the barrier although there is no connection between the
557: two uses of the word. Although this terminology is a little
558: unfortunate we will use it here and so we will be studying
559: a critical nucleus near a critical point.
560:
561: \bibitem{chaikin} P. M. Chaikin and T. C. Lubensky,
562: {\it Principles of Condensed Matter Physics}
563: (Cambridge University Press, Cambridge, 1995).
564:
565: \bibitem{cahn59} J. W. Cahn and J. E. Hilliard,
566: J. Chem. Phys. {\bf 31}, 688 (1959).
567:
568: \bibitem{cahn58} J. W. Cahn and J. E. Hilliard,
569: J. Chem. Phys. {\bf 28}, 258 (1958).
570:
571: \bibitem{widom85} B. Widom,
572: Chem. Soc. Rev. {\bf 14}, 121 (1985).
573:
574: \bibitem{evans89} R. Evans,
575: {\it Les Houches, Session XLVIII: Liquids at Interfaces},
576: edited by J. Charvolin, J. F. Joanny and J. Zinn-Justin
577: (Elsevier, Amsterdam, 1989)
578:
579: \bibitem{talanquer98} V. Talanquer and D. W. Oxtoby,
580: J. Chem. Phys. {\bf 109}, 223 (1998).
581:
582: \bibitem{tenwolde97} P. R. ten Wolde and D. Frenkel,
583: Science {\bf 277}, 1975 (1997).
584:
585: \bibitem{haas00} C. Haas and J. Drenth,
586: J. Phys. Chem. B {\bf 104}, 368 (2000).
587:
588: \bibitem{dixit00} N. M. Dixit and C. F. Zukoski,
589: J. Coll. Int. Sci. {\bf 228}, 359 (2000).
590:
591: \bibitem{searxxx} R. P. Sear,
592: cond-mat/9912199 (http://xxx.lanl.gov).
593:
594: \bibitem{widom77} B. Widom,
595: J. Chem. Phys. {\bf 67}, 872 (1977).
596:
597: \bibitem{ohnesorge94} R. Ohnesorge, H. L\"{o}wen and H. Wagner,
598: Phys. Rev. E {\bf 50}, 4801 (1994).
599:
600: \bibitem{kashchiev82} D. Kashchiev,
601: J. Chem. Phys. {\bf 76}, 5098 (1982).
602:
603: \bibitem{viisanen93} Y. Viisanen, R. Strey and H. Reiss,
604: J. Chem. Phys. {\bf 99}, 4680 (1993).
605:
606: \bibitem{bowles00} R. K. Bowles, R. McGraw, P. Schaaf, B. Senger,
607: J.-C. Voegel and H. Reiss,
608: J. Chem. Phys. {\bf 113}, 4524 (2000).
609:
610: \bibitem{hansen86} J.-P. Hansen, and I. R. McDonald,
611: {\it Theory of Simple Liquids} (Academic Press, London, 2nd Edition, 1986).
612:
613:
614: \end{thebibliography}
615:
616: %\newpage
617: \begin{figure}
618: \begin{center}
619: \caption{
620: \lineskip 2pt
621: \lineskiplimit 2pt
622: A schematic of the radial density profile of a droplet:
623: $\phi(r)$ is the density at a distance $r$ from the centre of the
624: droplet. The density of the surrounding fluid is $\rho$, and
625: $r_c$ is the (somewhat arbitrary) point where we divide
626: the density profile into a central part and an outer part.
627: }
628: \vspace*{0.1in}
629: \epsfig{file=prof.eps,width=3.0in}
630: \end{center}
631: \label{figprof}
632: \end{figure}
633:
634: \end{document}
635: