1: \documentstyle[prl,aps,epsfig,amssymb]{revtex}
2: %\documentstyle[preprint,prl,aps,epsfig,amssymb]{revtex}
3: \begin{document}
4: \draft
5: %\tighten
6:
7: \title{Entropic torque}
8: %\wideabs{
9: \author{R. Roth$^{1,2}$, R. van Roij$^3$, D. Andrienko$^4$\thanks{present
10: address: Max-Planck-Institut f{\"u}r Polymerforschung, Achermannweg 10,
11: D-55128 Mainz, Germany}, K. R. Mecke$^{1,2}$, and S. Dietrich$^{1,2}$}
12: \address{$^1$Max-Planck-Institut f{\"u}r Metallforschung, Heisenbergstrasse 1,
13: D-70569 Stuttgart, Germany}
14: \address{$^2$Institut f{\"u}r Theoretische und Angewandte Physik,
15: Universit{\"a}t Stuttgart, Pfaffenwaldring 57, D-70569
16: Stuttgart, Germany}
17: \address{$^3$Institute for Theoretical Physics, Utrecht University,
18: Princetonpln 5, 3584 CC Utrecht, The Netherlands}
19: \address{$^4$H.H. Wills Physics Laboratory, University of Bristol, Tyndall
20: Avenue, Bristol BS8 1TL, United Kingdom}
21: \maketitle
22: \begin{abstract}
23: Quantitative predictions are presented of a depletion-induced torque and force
24: acting on a single colloidal hard rod immersed in a solvent of hard spheres
25: close to a planar hard wall. This torque and force, which are entirely of
26: entropic origin, may play an important role for the key-lock principle, where
27: a biological macromolecule (the key) is only functional in a particular
28: orientation with respect to a cavity (the lock).
29: \end{abstract}
30: \pacs{82.70.Dd,61.20.Gy,61.20.Ja}
31: %}
32:
33: The depletion effect, i.e., the effect that smaller particles in a colloidal
34: mixture are expelled from those regions where bigger particles are
35: sufficiently close together, leads to an effective force between the bigger
36: particles. In the case of hard-core interactions these effective forces are
37: purely entropic in origin \cite{Asakura54,Vrij76}. For the relatively
38: simple geometries of two large spheres or a sphere close to a planar or curved
39: wall in a solvent of small hard spheres this depletion force has been studied
40: in detail in theory \cite{Roth99,Goetzelmann99,Roth00}, simulation
41: \cite{Biben96,Dickman97}, and experiment \cite{Crocker99,Rudhardt98}. Depletion
42: forces are crucial for the phase behavior of colloidal system; they can drive
43: gas-liquid and liquid-solid phase separation, e.g., in colloid-polymer
44: \cite{Dijkstra99} and colloid-colloid mixtures \cite{Dijkstra98}. It has been
45: suggested \cite{r7} that depletion effects play an important role in the
46: biological ``key-lock'' mechanism, in which a nonspherical macromolecule (the
47: key) fits into a cavity (the lock) and forms a chemical bond, but only if
48: the key has a particular orientation. A robust performance of this mechanism
49: is possible if the biological environment is capable of passively transporting
50: the key, correctly oriented, to the lock. This requires a chemically unspecific
51: {\em force} to transport the center-of-mass of the macromolecule towards the
52: cavity and a {\em torque} to orient it. In order to be able to study the
53: relevance of the entropic contribution to this mechanism a quantitatively
54: reliable tool for predicting such an entropic torque is needed. Using density
55: functional theory and computer simulations, we show and predict quantitatively
56: that depletion effects can generate both a force and a torque on a nonspherical
57: particle in the vicinity of a planar wall thereby providing further evidence
58: for the relevance of depletion phenomena in key-lock mechanisms.
59:
60: The specific system we study is illustrated in Fig.\ref{fig:rod} and consists
61: of (i) an infinite planar hard wall located at $z=0$, (ii) a hard
62: spherocylinder with cylinder length $L$, diameter $\sigma$, at distance $z$
63: from the wall and forming an angle $\theta \in[0,\pi/2]$ with respect to the
64: wall normal, and (iii) a solvent of hard spheres with diameter $\sigma_s$
65: and number density $\rho_s$ in the bulk, i.e., far away from the wall. Since
66: we are considering only hard interactions the temperature $T$ of the system
67: plays only the role of an energy scale via $\beta=1/k_B T$. We are interested
68: in the effective rod-wall potential $W(z,\theta)$ induced by the spheres. The
69: force and the torque on the rod follow by differentiating $-W(z,\theta)$ with
70: respect to $z$ and $\theta$, respectively. The bare rod-wall interaction
71: restricts the center of mass of the rod to $z$ to $z\geq z_{min}(\theta)$ =
72: $(\sigma +L |\cos\theta|)/2$. The system, and hence $\beta W(z,\theta)$, is
73: completely characterized by the aspect ratio $L/\sigma$ of the rod, the
74: diameter ratio $\sigma/\sigma_s$, and the packing fraction
75: $\eta=\rho_s \pi \sigma_s^3/6$.
76:
77: Following Asakura and Oosawa \cite{Asakura54}, a first approximation
78: of the effective potential $W(z,\theta)$ can be obtained from the analysis of
79: the volume that is excluded to the centers of the solvent spheres by the wall
80: and the rod. The wall excludes the slab $0<z<\sigma_s/2$, and the rod the
81: volume of a spherocylinder of length $L$ and diameter $\sigma+\sigma_s$. If
82: these two contributions to the excluded volume overlap, i.e., for
83: $z<z_{min}+\sigma_s$ (see Fig.~\ref{fig:rod}) the volume accessible to the
84: centers of the spheres and hence the entropy of the solvent increases. This
85: gain of entropy translates directly into an effective, purely attractive force
86: acting on the center of the rod. An important ingredient of this so-called
87: Asakura-Oosawa approximation (AOa) is that the density profile of the spheres
88: is taken to be constant and equal to the bulk density $\rho_s$, i.e., all
89: correlations in the solvent are entirely neglected. The AOa therefore
90: underestimates the strength and the range of the force acting on the rod.
91:
92: Here we go beyond the AOa taking the sphere-sphere correlations into account.
93: We recall that the depletion potential equals the grand-potential change of
94: the (inhomogeneous) fluid of spheres in contact with the wall upon moving the
95: rod from $z\rightarrow\infty$ to a finite value of $z$, at a given $\theta$.
96: Within density functional theory (DFT), with
97: ${\cal F}_{ex}[\rho_s,\rho_r]$ the excess (over ideal) free energy functional
98: of a {\em mixture} of hard spheres and rods, one can write \cite{Roth00}
99: \begin{equation} \label{pot}
100: \beta W(z,\theta) = \lim_{\rho_r\to 0} [c_r^{(1)}(z\to\infty,\theta) -
101: c_r^{(1)}(z,\theta)],
102: \end{equation}
103: where $c^{(1)}_r(z,\theta)=-\delta \beta {\cal F}_{ex}[\rho_s,\rho_r]/
104: \delta \rho_r(z,\theta)$ is the direct one-body correlation function of the
105: rods \cite{Evans79} where $\rho_r(z,\theta)$ is the number density of rods for
106: a given orientation $\theta$. The functional we use is based on Rosenfeld's
107: fundamental measure theory (FMT) for mixtures of general convex hard bodies
108: \cite{Rosenfeld94}. FMT has proved to account accurately for both the structure
109: and the thermodynamics of inhomogeneous hard-sphere mixtures \cite{Rosenfeld89}
110: as compared with simulations. Recently, a FMT for a mixture of spheres and
111: needles of vanishing thickness was proposed \cite{Schmidt01}. Our extension to
112: include a nonspherical particle with finite volume (the rod) in the theory
113: requires the deconvolution of the Mayer-$f$ function of the rod-sphere
114: interaction into a set of orientation dependent weight functions
115: \cite{Rosenfeld94}. The details of this technically involved deconvolution
116: will be explained elsewhere \cite{Roth02}.
117:
118: We obtain the depletion potential in two steps. First we calculate the
119: unperturbed density profile $\rho_s(z)$ of the spheres close to the wall. This
120: is the equilibrium profile in the limit $\rho_r\to 0$, as required in
121: Eq.(\ref{pot}). The functional reduces, in this case, to the well-tested FMT
122: for spheres. In the second step we use $\rho_s(z)$ to evaluate
123: $c_r^{(1)}(z,\theta)$, using the deconvoluted Mayer-$f$ function. From
124: Eq.(\ref{pot}) we then obtain the depletion potential $W(z,\theta)$ for all
125: $z$ and $\theta$ (from the profile $\rho_s(z)$ alone!).
126:
127: We present results for $L/\sigma=10$, $\sigma/\sigma_s=1$, and a packing
128: fraction $\eta_s=(\pi/6)\sigma_s^3\rho_s=0.2239$ for the spheres.
129: Figure~\ref{fig:pot2D} displays the depletion potential $\beta W(z,\theta)$ as
130: function of $z$ and $\theta$. The first observation is that the potential is,
131: unlike the AOa, {\em not} monotonic; the hard-sphere correlations generate a
132: sequence of potential barriers and wells. The dashed and dotted curves in
133: Fig.~\ref{fig:pot2D} denote the positions of the minima and maxima of the
134: potential. We note that for small angles $\theta\ll 1$ the shape of the
135: rod-wall depletion potential $W(z,\theta)$ coincides almost perfectly with the
136: depletion potential $W_{ws}(z)$ between a sphere of diameter $\sigma$ and the
137: wall, i.e., $W(z,\theta\ll 1)\approx W_{ws}(z-L \cos|\theta|/2)$. For these
138: small angles the length of the rod $L$ is rather unimportant for details of
139: the depletion potential, as we have verified for various values of $L/\sigma$
140: and $\sigma/\sigma_s$. For large angles, however, the whole geometry of the
141: rod, i.e., $L/\sigma$ and $\sigma/\sigma_s$, is relevant, and the depth of the
142: depletion potential at contact becomes more negative as both $L/\sigma$ and
143: $\sigma/\sigma_s$ are increased. In addition the contact value of the depletion
144: potential can be further decreased by increasing $\eta_s$.
145:
146: Although certain general trends of the influence of the geometry on the shape
147: of the depletion potential can already be roughly understood within the simple
148: AOa \cite{Asakura54}, the correlations in the hard-sphere fluid are very
149: important and can lead to a quantitatively and even qualitatively different
150: behavior. In Fig.~\ref{fig:w0} we illustrate this by comparing the depletion
151: potential at contact as obtained within our DFT approach
152: ($W(z_{min}(\theta),\theta)$, full line) with the corresponding quantity within
153: the AOa ($W^{AOa}(z_{min}(\theta),\theta)$, dotted line). While for small
154: and large angles the contact value of the full depletion potential is more
155: negative than within the AOa, for intermediate values of $\theta$ packing
156: effects can shift the actual contact value above its corresponding AOa-value.
157: If the rod is almost parallel to the wall, i.e., for $\theta \lesssim \pi/2$
158: the deviation of the contact value calculated within the AOa from the
159: prediction of the full DFT is significant and much more pronounced than in the
160: well-studied wall-sphere or sphere-sphere geometry. The inset of
161: Fig.~\ref{fig:w0} shows that a rod in contact with the wall can exhibit a
162: metastable orientation for a certain angle $0<\theta_0<\pi/2$ while the AOa
163: predicts that this configuration is unstable. Although for the parameters
164: under consideration here the height of the potential barrier is rather small
165: ($0.007 k_B T$), we note that its value increases for smaller values of
166: $\sigma/\sigma_s$.
167:
168: As a consequence of the dependence of the rod-wall depletion potential on the
169: orientation $\theta$ for a given (fixed) distance $z$ of the center of the rod
170: from the wall an entropic torque acts on the rod which drives it into an
171: orientation with minimal depletion potential. This torque can be estimated
172: by replacing the spherocylinder by a dumbbell composed of two spheres with
173: diameter $\sigma$ connected by an infinitesimally thin but rigid wire of length
174: $L$. Since the sphere-wall depletion force $f_{sw}(z)$ acts on the two spheres
175: located at $z_\pm = z \pm (L/2) \cos \theta$ the corresponding torque with
176: respect to the center of mass is given by
177: \begin{equation} \label{simpletorque}
178: M_{db}(z,\theta) = \frac{L}{2} \sin\theta \left[ f_{sw}(z_-) -
179: f_{sw}(z_+) \right].
180: \end{equation}
181: It turns out that at low packing fractions of the hard-sphere solvent, or at
182: sufficiently large separations of the rod from the wall
183: Eq.~(\ref{simpletorque}) yields a semi-quantitative expression for the torque
184: acting on a spherocylinder.
185:
186: From the full rod-wall depletion potential $W(z,\theta)$ one can obtain the
187: torque by rotating the rod by an infinitesimal angle $d\theta$ around an axis
188: through the center of the rod in a direction characterized by the unit vector
189: ${\bf n}_\theta$ normal to the symmetry plane shown in Fig.~\ref{fig:rod}. The
190: corresponding change in the depletion potential is $d W(z,\theta)$ which can be
191: written as $d W=- \sum_i {\bf f}_i\cdot d{\bf r}_i$, i.e., as a sum of forces
192: ${\bf f}_i$ acting on the rod at positions ${\bf r}_i$ from the center with
193: $d {\bf r}_i=d \theta ({\bf r}_i\times {\bf n}_\theta$). It follows that the
194: torque ${\bf M}(z,\theta)=M(z,\theta) {\bf n}_\theta$ with respect to the
195: center of mass is given by
196: \begin{equation}
197: M(z,\theta) = - \frac{\partial W(z,\theta)}{\partial \theta}.
198: \end{equation}
199: The symmetry of the problem leads to $M(z,\theta)=0$ for $\theta=0$ and
200: $\theta=\pi/2$. A positive value of the torque acts on the rod as to
201: increase the angle $\theta$ (rotating it parallel to the wall) while a
202: negative value of $M$ leads to a decrease of $\theta$ (rotating it normal to
203: the wall). Some typical examples for the torque as a function of $\theta$ for
204: various values of $z$ are shown in Fig.~\ref{fig:torque}. Lines and symbols
205: denote results from DFT and molecular dynamics (MD) simulations, respectively.
206: The DFT predictions are in excellent agreement with the simulations. The
207: details of the simulations will be presented elsewhere \cite{Roth02}.
208:
209: For larger distances of the rod from the wall $W(z,\theta)$ exhibits local
210: minima and maxima in addition to the global minimum with corresponding zeros
211: of the torque (see Fig.~\ref{fig:torque} for $z=5.5 \sigma_s$ and
212: $z=4.5 \sigma_s$). A minimum (maximum) of $W(z,\theta)$ leads to a zero of
213: $M(z,\theta)$ with a negative (positive) gradient in $\theta$. Furthermore, the
214: torque can exhibit a cusp (see Fig.~\ref{fig:torque}) if the minimal distance
215: between the wall and the rod leaves space for precisely one solvent sphere.
216: This can be realized for distances
217: $\sigma_s+\sigma/2 \leq z \leq \sigma_s + (L+\sigma)/2$ and a corresponding
218: orientation $\theta_{cusp}=\arccos((2 (z-\sigma_s)-\sigma)/L)$.
219:
220: As the rod moves closer to the wall the modulus of the torque increases and
221: for small separations from the wall (see, e.g., $z=1.5 \sigma_s$ in
222: Fig.~\ref{fig:torque}) the torque vanishes only for $\theta=\pi/2$ which is,
223: however, only a metastable configuration. For such small and fixed values of
224: $z$ the entropic torque rotates the rod towards configurations with smaller
225: angles until the rod is in contact with the wall and has to stop its rotation.
226: Thus the rod reaches its most favorable configuration of lying parallel in
227: contact with the wall not by approaching the wall in a parallel configuration
228: but by touching the wall first at one end and then by decreasing the distance
229: of its center from the wall. The modulus of the maximum of the entropic torque
230: for the system considered here is of the order of about $20 k_B T$ rad$^{-1}$,
231: which corresponds to roughly $10^{-20}$ J rad$^{-1}$ at room temperature. The
232: strength of the torque increases for larger values of $L/\sigma$,
233: $\sigma/\sigma_s$, or $\eta_s$. For example, for $L/\sigma=20$,
234: $\sigma/\sigma_s=2$, and $\eta_s=0.2239$, the modulus of the maximum of the
235: entropic torque reaches a value of about $38.7 k_B T$ rad$^{-1}$. We note that
236: the maximum torque acts at small values of $z$ for which the AOa also would
237: predict the existence of a torque; however, its magnitude is largely enhanced
238: through the correlations in the solvent.
239:
240: With the results presented so far we can comment on some aspects of the path
241: in the $z$-$\theta$ plane a rod would take upon approaching the wall from the
242: bulk. Only if the rod comes sufficiently close to the wall to be subject to the
243: oscillations of the number density of the hard-sphere fluid as a function of
244: $z$, the entropic force and torque will act on it. The closer the rod gets to
245: the wall the higher are the potential barriers between the minima (see
246: Fig.~\ref{fig:pot2D}) the rod has to overcome by thermal motion in order to
247: move still closer. The potential barriers are, however, only moderate and easy
248: to overcome for small angles and increase for angles $\theta \gtrsim 60$
249: degrees so that it is not optimal to rotate the rod until it is parallel to the
250: wall since then the potential barrier for further approach is largest and for
251: the parameters considered here more than $2.3 k_B T$. On the other hand along
252: the line of the second minimum -- see the dashed lines in Fig.~\ref{fig:pot2D}
253: -- $\beta W$ has a similar shape as $\beta W(z_{min}(\theta),\theta)$ [see
254: Fig.~\ref{fig:w0}], so that the entropic torque will rotate the rod into the
255: local minimum at $\theta=\pi/2$ if $\theta$ is already sufficiently large. Once
256: the rod is in contact with the wall the torque will rotate it towards the
257: pronounced global minimum of $\beta W(z,\theta)$.
258:
259: We have presented quantitative predictions of the entropic torque acting on a
260: hard spherocylinder close to a hard planar wall in a solvent of small hard
261: spheres. Our DFT predictions are in excellent agreement with our MD
262: simulations. We find that the the depletion effect leads to a significant
263: entropic torque and it is tempting to speculate that this entropic torque can
264: play an important role in understanding the key-lock principle.
265:
266: It is a pleasure to thank R. Evans, M. Schmidt, C. Bechinger, and M. P. Allen
267: for stimulating discussions. This work is part of the Research program of the
268: ``Stichting voor Fundamenteel Onderzoek der Materie (FOM)'', which is
269: financially supported by the ``Nederlandse Organisatie voor Wetenschappelijk
270: Onderzoek (NWO)''. D.A. acknowledges the support of the Overseas Research
271: Students Grant, EPSRC grants GR/L89990, GR/M16023, INTAS grant 99-00312.
272: MD simulations used the GBMEGA program of the `Complex Fluids Consortium' with
273: computer time allocated at the CSAR facility.
274:
275: \begin{references}
276: \bibitem{Asakura54} S. Asakura and F. Oosawa, J. Chem. Phys. {\bf 22}, 1255
277: (1954); J. Polymer Sci. {\bf 33}, 183 (1958).
278: \bibitem{Vrij76} A. Vrij, Pure and Appl. Chem. {\bf 48}, 471 (1976).
279: \bibitem{Roth99} R. Roth, B. G\"otzelmann, and S. Dietrich, Phys. Rev. Lett.
280: {\bf 83}, 448 (1999).
281: \bibitem{Goetzelmann99} B. G\"otzelmann, R. Roth, S. Dietrich, M. Dijkstra,
282: and R. Evans, Europhys. Lett. {\bf 47}, 398 (1999); R. Roth, doctoral thesis,
283: Bergische Universit\"at Wuppertal (1999).
284: \bibitem{Roth00} R. Roth, R. Evans, and S. Dietrich, Phys. Rev. E {\bf 62},
285: 5360 (2000).
286: \bibitem{Biben96} T. Biben, P. Bladon, and D. Frenkel, J. Phys.: Condens.
287: Matter {\bf 8}, 10799 (1996).
288: \bibitem{Dickman97}R. Dickman, P. Attard, and V. Simonian, J. Chem. Phys.
289: {\bf 107}, 205 (1997).
290: \bibitem{Crocker99} J.C. Crocker, J.A. Matteo, A.D. Dinsmore, and
291: A.G. Yodh, Phys. Rev. Lett. {\bf 82}, 4352 (1999).
292: \bibitem{Rudhardt98} D. Rudhardt, C. Bechinger, and P. Leiderer, Phys. Rev.
293: Lett. {\bf 81}, 1330 (1998).
294: \bibitem{Dijkstra99} M. Dijkstra, J.M. Brader, and R. Evans, J. Phys.:
295: Condens. Matter {\bf 11}, 10079 (1999).
296: \bibitem{Dijkstra98} M. Dijkstra, R. van Roij, and R. Evans, Phys. Rev. Lett.
297: {\bf 81}, 2268 (1998); {\em ibid} {\bf 82}, 117 (1999); Phys. Rev. E {\bf 59},
298: 5744 (1999).
299: \bibitem{r7} see, e.g., M. Kinoshita and T. Oguni, Chem. Phys. Lett. {\bf 351},
300: 79 (2002) and references therein.
301: \bibitem{Evans79} R. Evans, Adv. Phys. {\bf 28}, 143 (1979).
302: \bibitem{Rosenfeld94} Y. Rosenfeld, Phys. Rev. E {\bf 50}, R3318 (1994);
303: Mol. Phys. {\bf 86}, 637 (1995).
304: \bibitem{Rosenfeld89} Y. Rosenfeld, Phys. Rev. Lett. {\bf 63}, 980 (1989);
305: J. Chem. Phys. {\bf 98}, 8126 (1993).
306: \bibitem{Schmidt01} M. Schmidt, Phys. Rev. E {\bf 63}, 050201-1 (2001).
307: \bibitem{Roth02} R. Roth, R. van Roij, D. Andrienko, K. R. Mecke, and S.
308: Dietrich, unpublished.
309: \end{references}
310:
311: \newpage
312:
313: \begin{figure}
314: \centering\epsfig{file=rod.eps,width=0.75\linewidth}
315: \vspace{0.5cm}
316: \caption{\label{fig:rod} A spherocylinder of length $L$ and diameter $\sigma$
317: at angle $\theta$ and distance $z$ relative to a
318: a planar hard wall at $z=0$. The minimal value of $z$
319: is $z_{min}(\theta)=(\sigma+L |\cos\theta|)/2$. The rod is immersed in
320: a solvent of hard spheres of diameter $\sigma_s$ and number density $\rho_s$
321: far from the wall.}
322: \end{figure}
323:
324: \begin{figure}
325: \centering\epsfig{file=pot2D.ps,width=0.75\linewidth}
326: \caption{\label{fig:pot2D} Depletion potential $\beta W(z,\theta)$ for the
327: spherocylinder shown in Fig.~\ref{fig:rod} for $L/\sigma=10$,
328: $\sigma/\sigma_s=1$, and a packing fraction
329: $\eta_s=\rho_s \pi \sigma_s^3/6=0.2239$ of the hard-sphere solvent. $\theta$
330: is measured in degrees. The dashed and dotted lines represent the positions of
331: local minima and maxima, respectively.}
332: \end{figure}
333:
334: \begin{figure}
335: \centering\epsfig{file=w0.ps,width=0.75\linewidth}
336: \vspace{0.5cm}
337: \caption{\label{fig:w0} The contact value $\beta W(z_{min}(\theta),\theta)$ as
338: a function of $\theta$ (in degrees) as calculated within DFT (full line) and
339: within the AOa (dotted line) for the system described in Fig.~\ref{fig:pot2D}.
340: The inset shows that a rod in contact with the wall can exhibit a metastable
341: orientation for $\theta=\theta_0\approx 8$ degrees with a barrier height
342: $\beta \Delta W=0.007$.}
343: \end{figure}
344:
345: \begin{figure}
346: \centering\epsfig{file=torque.ps,width=0.75\linewidth}
347: \caption{\label{fig:torque} The torque $\beta M(z,\theta)$ as function of
348: $\theta$ for various values of $z$. For reasons of clarity the curves for
349: $z=4.5$, $3.4$, $\dots$ are shifted downward by $-5$, $-10$, $\dots$,
350: respectively. The horizontal lines $M=0$ are indicated partially. The
351: parameters of the system are the same as in Fig.~\ref{fig:pot2D}. Because of
352: the hard wall at $z=0$ the torque is defined only for
353: $\theta \geq \arccos(2 z/L-\sigma/L)$ if $z<(\sigma+L)/2$. $M>0$ $(<0)$
354: corresponds to a force which tends to align the rod parallel (normal) to
355: the wall. The lines denote our DFT results which are in excellent agreement
356: with our simulations (symbols). The error bars of the simulations are of the
357: order of the symbol size.}
358: \end{figure}
359:
360: \end{document}