1: \documentclass[twocolumn,preprintnumbers,amsmath,amssymb]{revtex4}
2: \usepackage{graphicx,epsfig}
3: \usepackage{dcolumn}
4: \usepackage{bm}
5: %\parskip=0.5cm
6: \begin{document}
7:
8: \title{Anisotropic Local Stress and Particle Hopping in a Deeply
9: Supercooled Liquid}
10:
11: \author {Sarika Bhattacharyya and Biman Bagchi\footnote[1]
12: {For correspondence: bbagchi@sscu.iisc.ernet.in}}
13: \affiliation{ Solid State and Structural Chemistry Unit,
14: Indian Institute of Science,
15: Bangalore 560012, India}
16:
17: \begin{abstract}
18: The origin of the microscopic motions that lead to
19: stress relaxation in deeply supercooled liquid remains unclear.
20: We show that in such a liquid the stress relaxation is
21: locally anisotropic which can serve as the driving force
22: for the hopping of the system on its free energy surface.
23: However, not all hopping are equally effective in relaxing the
24: local stress, suggesting that diffusion can decouple from
25: viscosity even at local level. On the other hand, orientational relaxation
26: is found to be always coupled to stress relaxation.
27: \end{abstract}
28:
29: \maketitle
30: %\section{Introduction}
31: Dynamics of supercooled liquid show many fascinating properties,
32: namely the non-exponentiality in various relaxation functions of dynamical
33: variables,
34: like the stress, density, composition, polarization and
35: orientation\cite{sastry,
36: angell,arnab,parisi,lepo}.
37: One often observes a very slow power law decay in the intermediate
38: to long times\cite{fayerprl}.
39: Although these have drawn attention of experimentalists and theoreticians
40: and have been extensively studied, many aspects still remain ill-understood.
41: It is believed that the nature of the microscopic motion undergoes a
42: drastic change
43: at a temperature T$_{B}$ (a temperature substantially below
44: the melting temperature, T$_{m}$). At the temperature $T_{B}$, the
45: continuous hydrodynamic type motion, which is prevalent at higher temperature
46: changes to discontinuous motion. It is also believed that T$_{B}$ is close to
47: the temperature where effects of the underlying free energy landscape on the
48: dynamics are felt for the first time\cite{sastry}.
49: $T_{B}$ is also found to be close to the mode coupling
50: temperature, T$_{c}$.
51: This change in the nature of the microscopic motion is
52: believed to be the origin of the
53: experimentally observed $\alpha$-$\beta$ bifurcation \cite{jg}
54: and also the cross-over between the rotational and translational relaxation
55: times \cite{ediger}.
56:
57: Due to the complexity of the problem, computer simulations have played a key
58: role in augmenting our understanding in this area. In particular, simulations
59: allow one to directly look at the microscopic events.
60: The computer simulation studies of the stress auto correlation function
61: in the supercooled liquid could successfully reproduce the
62: power law behavior of the stress auto-correlation function
63: (SAF) \cite{arnab,parisi}.
64: However, in the deeply supercooled liquid, one finds that within the
65: simulation time, the relaxation,
66: after an initial decay (typically 10-20$\%$) becomes fully arrested.
67: The microscopic origin of the subsequent decay
68: is unclear.
69: The computer simulation studies further show that
70: the orientational and translational
71: hopping of particles are the only mode present and hence the stress
72: relaxation can happen only via hopping.
73: However, since the
74: relaxation time is much much longer (could be of the order of ms or sec),
75: the computer simulation results,
76: (which can explore mostly upto nano second regime and sometimes micro second
77: regime), cannot
78: include the effects of these hopping. Therefore, one cannot explore
79: the relationship between hopping and the {\em total} stress relaxation.
80:
81: The experiments on the other hand are successful in showing the
82: decay of the SAF. However, except a recent work using
83: single molecule spectroscopy \cite{van}, these experiments are macroscopic
84: and do not provide enough information of the microscopic
85: motions in the system.
86:
87: Majumdar \cite{majum} had earlier discussed the importance of
88: local relaxation modes (of wavelength less than the short range order)
89: in giving rise to non-exponentiality
90: in the stress relaxation function. This work discussed relaxation in
91: terms of relaxation within small regions, surfaces and also volumes,
92: with progressive lengthening of time scales.
93: However, in that analysis the
94: basic mechanism of relaxation was still assumed to be continuous.
95:
96: In this Letter we demonstrate for the first time that
97: in the deeply supercooled liquid (where hopping is the only
98: surviving large amplitude motion), there is
99: a close relationship between the {\em local} stress
100: and the
101: orientational and translational hopping.
102: The local SAF is anisotropic and is found to change drastically
103: during
104: the hopping, thus showing that the local stress and the hopping of
105: a particle are intimately connected. The anisotropy in the local stress
106: could be the driving force for hopping.
107: As the free energy of the
108: system can be expressed in terms of the position dependent stress in a
109: generalized Ginzburg-Landau formulation \cite{landau},
110: the change of the anisotropic
111: stress due to hopping should be regarded as the driving force for the
112: transitions of the system between different minima of the free energy surface.
113: However, not all hoppings are effective in relaxing the stress.
114:
115: Our solvent is represented by binary Lennard-Jones mixture, which has been
116: extensively studied \cite{sastry, sastry1, kob1,others} and is known
117: to be a good glass former,
118: and our solute probes are prolate ellipsoids.
119: Pressure is kept constant by Andersen's piston method \cite{andersen}
120: while in the case of temperature, a damped oscillator method has been
121: adopted which keeps temperature constant at
122: each and every time step \cite{brown}. The piston mass involved here is
123: $0.0027(m_{A}/\sigma_{A}^4)$ which is regarded as optimum \cite{brown,haile}.
124: The interaction between two
125: ellipsoids with arbitrary orientations is assumed to be given by the
126: Gay-Berne (GB) potential \cite{gb},
127: \begin{eqnarray}
128: U_{GB}& = &4\epsilon(\hat r,\hat u_1,\hat u_2)\Biggl[ \left( \frac{\sigma_{0}}
129: {r-\sigma(\hat r,\hat u_1,\hat u_2)+ \sigma_{0}} \right) ^{12}\\ \nonumber
130: &&-\left (\frac{\sigma_{0}}{r-\sigma(\hat r,\hat u_1,\hat u_2)+ \sigma_{0}}\right )^{6} \Biggr]
131: \end{eqnarray}
132: where $\hat u_1$ $\hat u_{2}$ are the axial vectors of the ellipsoids 1 and 2.
133: $\hat r$ is the
134: vector along
135: the intermolecular vector $r = r_2 - r_1$, where $r_1$ and $r_2$ denote the
136: centers of mass of ellipsoids 1 and 2 respectively. $\sigma(\hat r,\hat u_1,
137: \hat u_2)$ and
138: $\epsilon(\hat r,\hat u_1,\hat u_2)$ are the orientation-dependent range and
139: strength parameters
140: respectively. $\sigma$ and $\epsilon$ depend on the aspect ratio $\kappa$.
141: The minor axis of the ellipsoid is equal to the diameter of the larger
142: solvent and the major axis is 3 times that of the minor axis.
143: Finally, the interaction between a sphere and an ellipsoid is accounted for by
144: a modified GB-LJ potential given below
145: \begin{equation}
146: U_{Ei} = 4\epsilon_{Ei}\left [ \left(\frac{\sigma(\theta)_{Ei}}{r}
147: \right )^{12} - \left (\frac{\sigma(\theta)_{Ei}}{r}\right )^6\right ]
148: \end{equation}
149: \noindent where 'E' denotes the ellipsoids and 'i' can be 'A' or 'B'.
150: The expression for $\sigma(\theta)_{Ei}$ is available \cite{sbor}.
151:
152:
153: The ellipsoid in binary mixture system with the above mentioned potential is a
154: well behaved system and it can also exhibit the experimentally observed
155: anomalous viscosity dependence of the orientational correlation time
156: \cite{sbor}. Four ellipsoids were placed far from each other in a binary
157: mixture of 500 particles with number of 'A' particles, $N_{A}=400$ and number
158: of 'B' type particles $N_{B}=100$.
159: The reduced temperature is expressed as,
160: $T^{*}$(=$k_{B} T/\epsilon_{A}$), the reduced pressure as,
161: $P^{*}$(= $P\sigma^{3}_{A}/\epsilon_{AA}$).
162: and the reduced density as $\rho^{*}$(=$\rho \sigma_{A}^{3}$).
163: The time is scaled by $\tau =\sqrt{(m_{A}\sigma_{A}^{2}/\epsilon_{AA})}$.
164: The time step of the simulation is .002 $\tau$ and the system
165: is equilibrated for 1.5 $\times$ 10$^{5}$ steps and the data collection step
166: is 5 $\times$ 10$^{6}$. The studies have been performed at T$^{*}$=0.8 and the
167: P$^{*}$=6 and 10.
168:
169: At P$^{*}$ =6, both hopping and continuous motions exist in the system, thus
170: the stress relaxation could not be directly correlated with the hopping.
171: At P$^{*}$=10, only microscopic motion that survives is hopping.
172: In a recent study,
173: we have reported observation of correlated translational and
174: orientational hopping \cite{sab} at this pressure.
175: After extensive simulations we could find only two different kinds of motions.
176: The translational hopping was either associated with
177: correlated hopping of 5-6 nearest neighbors or it exhibited a
178: motion in a ring like tunnel. While it is possible that other types of
179: motion like isolated hopping can exist, we could not find them.
180: The hopping rate was found to be $2\times 10^{7}$ where both the type of
181: motions occurred with almost equal frequency.
182: In the following we
183: focus on the stress relaxation and its relation vis-a-vis hopping
184: at P$^{*}$=10 and T$^{*}$=0.8. The reduced density of the system is 1.41.
185:
186: In figure 1 we show two different kinds of spatial hopping observed in our
187: simulations. Both the hoppings are associated with orientational
188: hopping. In the inset we also plot the orientational time correlation
189: functions (OCF), before, during and after the hopping.
190: Figure 1(a) shows the trajectory of the first tagged
191: ellipsoid and inset shows its orientational time correlation function,
192: $<P_{2}(\hat u_{i}(0)\hat u_{i}(t))>/{<P_{2}(\hat u_{i}(0)\hat u_{i}(0))>}$,
193: were $P_{2}$ is the second order Legendre polynomial.
194: The hopping takes place in 20$\tau$ and the displacement is 0.5$\sigma$.
195: Here the ellipsoid hopping is accompanied by hopping of 5-6 of its nearest
196: neighbors. The OCF decays only during the period of hopping.
197: In figure 1 (b) the trajectory of the second tagged ellipsoid
198: and in the inset its orientational time correlation functions are shown.
199: Note that in this case the displacement of the particle is large
200: (1.1$\sigma$) and it also takes place over a very long period of time
201: (50 $\tau$). Here we find that
202: the tagged particle is moving in a ring like tunnel. Although the
203: orientational correlation function decays during the hopping,
204: it's decay is less when compared to that of the 1st tagged particle.
205:
206: \begin{figure}
207: %\epsfig[height=8cm,width=6cm,angle=-90]{fig1a.ps}
208: \epsfig{file=fig1a.eps,height=7cm,width=6cm,angle=-90}
209: %\vspace*{-0.8cm}
210: %\hspace*{-0.5cm}
211: \epsfig{file=fig1b.eps,height=7cm,width=6cm,angle=-90}
212: %\includegraphics[height=8cm,width=6cm,angle=-90]{fig1b.ps}
213: \caption{ (a) Shows the displacement of the 1st tagged ellipsoid over
214: 1000 $\tau$. There is a hopping of the ellipsoid around 300 $\tau$. In the
215: inset we plot the orientational correlation function obtained at different
216: intervals. (b) Same as figure 1 (a) but for the 2nd tagged ellipsoid.
217: The plots are at P$^{*}$=10 and T$^{*}$=0.8.}
218: \end{figure}
219:
220: The motions in a string like and ring like tunnel in a deeply supercooled
221: liquid have been
222: reported earlier by other authors, although they did not discuss this large
223: displacements over a long time for the ring like motion \cite{sharon}.
224:
225: In the following part we discuss in details the local stress
226: relaxation associated with these two different kinds of hoppings.
227: The local stress around the ellipsoid is obtained by summing over the
228: stress on the ellipsoid and its nearest neighbors.
229: The stress has six components and the stress auto-correlation function
230: is given by $<\sigma_{ij}(0)\sigma_{ij}(t)>$, where i,j=x,y,z.
231: In case of only Lennard-Jones fluid, $\sigma_{ij}=\sigma_{ji}$, but
232: for particles interacting via GB and GB-LJ potential this is not so.
233:
234:
235: Figures 2 (a) and (b) show the SAF around the
236: 1st tagged ellipsoid, before and after the period it is hopping,
237: respectively.
238: Before the hopping there is an anisotropy of the stress.
239: The xy and yx components
240: of the stress are much larger than the others
241: and also their correlations cease to decay.
242: {\it This anisotropy leads to a
243: hopping of the ellipsoid, mostly in the z direction}. During the hopping
244: there is a relaxation of the SAF in xy and yx components and after hopping
245: all the components relax. Note that the total stress in all the components
246: are also lower.
247:
248:
249: \begin{figure}
250: \epsfig{file=fig2a.eps,height=5cm,width=8cm}
251: %\includegraphics[height=5cm,width=8cm]{bfig2a.eps}
252: %\vspace*{-0.8cm}
253: \hspace*{-0.0cm}
254: \epsfig{file=fig2b.eps,height=5cm,width=8cm,angle=0}
255: %\includegraphics[height=5cm,width=8cm]{bfig2b.eps}
256: \caption{(a) The different components of the local stress
257: auto-correlation (obtained from the sum of the
258: stress of the 1st ellipsoid and its nearest neighbors)
259: function obtained before hopping, between 320-330 $\tau$.
260: (b)is the same as 2 (a) but the
261: components of the SAF are obtained after hopping between (340-350$\tau$).
262: The plots are at P$^{*}$=10 and T$^{*}$=0.8.}
263: \end{figure}
264:
265: Figures 3 (a) and (b) show the SAF around the
266: 2nd tagged ellipsoid, before and after the period it is hopping,
267: respectively.
268: Before the hopping there is an anisotropy of the stress. The yz and zy
269: components
270: of the stress are much larger than others, and
271: also their correlations cease to decay. {\it This leads to the hopping of
272: the ellipsoid, mostly
273: in the x direction}.
274: During and after the hopping there is an exchange of the stress. After the
275: hopping although the yz component of the stress relaxes and also the t=0
276: value of all the components reduces,
277: the SAF in the xz and zx components cease to decay.
278: Thus, this kind of motion in a ring like
279: tunnel does not lead to the relaxation of all the
280: components of the stress.
281: \begin{figure}
282: \epsfig{file=fig3a.eps,height=5cm,width=8cm,angle=0}
283: %\includegraphics[height=5cm,width=8cm]{bfig3a.eps}
284: \vspace*{-0.0cm}
285: \hspace*{-0.0cm}
286: \epsfig{file=fig3b.eps,height=5cm,width=8cm,angle=0}
287: %\includegraphics[height=5cm,width=8cm]{bfig3b.eps}
288: \caption{ (a)The different components of the local stress
289: auto-correlation (obtained from the sum of the
290: stress of the 2nd ellipsoid and its nearest neighbors)
291: function obtained before hopping, between 1590-1630 $\tau$.
292: (b) is the same as 3 (a) but the
293: components of the SAF are obtained after the hopping,
294: between 1670-1710$\tau$.
295: The plots are at P$^{*}$=10 and T$^{*}$=0.8.}
296: \end{figure}
297:
298: The orientational correlation function of the 1st tagged particle
299: relaxes more ( inset of figure 1 (a)) compared to that of the 2nd tagged
300: particle (inset of figure 1(b)) when computed in the respective
301: intervals where they are hopping.
302: From figure 2 and 3 we found that the local stress relaxation takes place
303: when the 1st ellipsoid is hopping where as when the second ellipsoid is
304: hopping although there is an exchange of stress between it's components, the local
305: SAF does not completely relax. There is a direct connection between the
306: local stress and the orientational relaxation functions, implying that
307: rotation and viscosity are coupled even in a localized region.
308:
309:
310: In order to understand what happens to the surrounding of the local
311: region when there is a relaxation of stress due to hopping we have
312: studied the stress
313: auto correlation function of a bigger region of 2 $\sigma$ around the
314: 1st ellipsoid.We found there are about 62-67 particles
315: in this region. Although there is an anisotropy of the
316: components of the stress in this larger region, this anisotropy cannot
317: be correlated with the
318: direction of hopping. Some of the components of SAF build up immediately
319: after hopping and in a later time it relaxes.
320: Similar analysis when done in a even bigger region shows that it takes
321: longer for the stress of that region to relax and also the effect of the
322: hopping is less.
323:
324:
325:
326: In conclusion we demonstrated that
327: the {\it direction} of the hopping of the tagged particle is
328: determined by the anisotropy in the stress.
329: Anisotropic stress relaxation is different when there is a many-particle
330: hopping and there is a motion in a ring like tunnel.
331: Although there is an exchange of stress between the components due to the
332: particle motion, the stress relaxation is less in a ring like motion.
333: Interestingly, the effect of hopping is found to spread over the adjoining region like
334: ripples with
335: the amplitude decreasing with increasing distance from the ellipsoid.
336: We note that in the case of the second tagged ellipsoid (Fig.1b)
337: although it
338: translates more, the stress relaxation during its hopping is less. Thus
339: suggesting that translational motion and viscosity are decoupled even in a
340: localized region. On the other hand, the orientational relaxation and also
341: the stress relaxation is more
342: for the first ellipsoid.
343: Thus suggesting that the orientational motion always remains
344: coupled to viscosity.
345: This is in agreement with the experimental results and in fact provides a
346: microscopic explanation of the results which are known for a long time.
347: There can be an apparent connection between the stress tensor and the
348: momentum circulation. Thus it is possible that the non-decaying SAF implies
349: that momentum circulation exists in a deeply supercooled liquid
350: %\cite {refer}.
351: Since the anisotropic stress contributes to the free energy of the system,
352: a change in the anisotropy drives the system from one free energy minimum to
353: the other. When the anisotropy in the stress disappears and all the SAF
354: relaxes then the system definitely moves to a lower free energy minimum.
355:
356: This work was supported by a grant from CSIR, India. We thank Srikanth Sastry
357: Arnab Mukherjee and Rajesh Murarka for helpful discussions.
358:
359:
360:
361: \begin{references}
362:
363: \bibitem{sastry} S. Sastry, P. G. Debenedetti and F. H. Stillinger, Nature
364: {\bf 393}, 554 (1998).
365:
366: \bibitem {angell} C. A. Angell, Chem. Rev. {\bf 90}, 523 (1990);
367: Luz-Maria Martinez and C. A. Angell, ( private communication)
368:
369: \bibitem {arnab} A. Mukherjee, S. Bhattacharyya, and B. Bagchi, {\bf 116},
370: 4577 (2001).
371:
372: \bibitem {parisi} L. Angelani, G. Parisi, G. Ruocco and G. Viliani,
373: Phys. Rev. Lett. {\bf 81}, 4648 (1998).
374:
375:
376: \bibitem {lepo} C. D. Michele and D. Leporini, Phys. Rev. E {\bf 63},
377: 036701 (2001); Phys. Rev. E {\bf 63}, 036702 (2001).
378:
379: \bibitem {fayerprl} G. Hinze, D. D. Brace, S. D. Gottke, and M. D. Fayer,
380: Phys. Rev. Lett. {\bf 84}, 2437 (2000); J. Chem. Phys. {\bf 113}, 3723 (2000).
381:
382: \bibitem{jg} G. P. Johari and M. Goldstein, J. Chem. Phys. {\bf 53}, 2372
383: (1970); ibid, {\bf 55}, 4245 (1971).
384:
385: \bibitem{ediger} M.D. Ediger, C. A. Angell and S. R. Nagel, J. Phys. Chem.
386: {\bf 100}; M. T. Cicerone and M. D. Ediger, J. Chem. Phys. {\bf 104}, 7210,
387: {\bf 103}, 5684 (1996); G. Heuberger and H. Sillescu, J. Chem.
388: Phys. {\bf 100}, 15255 (1996).
389:
390: \bibitem {van} L. A. Deschenes and D. A. Vanden Bout, Science {\bf 292}, (2001).
391:
392: \bibitem{majum} C. K. Majumdar, Solid State Com. {\bf 9} , 1087 (1971).
393:
394: \bibitem{landau} P. C. Hohenberg and B. I. Halperin,
395: Rev. Mod. Phys. {\bf 49 }, 435 (1977); L. D. Landau and
396: E. M. Lifshitz, {\it Theory of Elasticity} (Pergamon Press, U. K 1981).
397:
398: \bibitem {sastry1} S. Sastry, Phys. Rev. Lett. {\bf 85}, 590 (2000);
399: S. Sastry, Nature, {\bf 409}, 164 (2001).
400:
401:
402: \bibitem {kob1} W. Kob and H. C. Andersen, Phys. Rev. E, {\bf 51},
403: 4626 (1995) ; W. Kob and H. C. Andersen, Phys. Rev. Lett, {\bf 73},
404: 1376 (1994).
405:
406: \bibitem {others} K. Vollmayr, W. Kob and K. Binder, J. Chem. Phys.
407: {\bf 105}, 4714 (1996) ;
408: F. Sciortino, W. Kob and P. Tartagila, Phys. Rev. Lett. {\bf 83},
409: 3214 (1999) ;B. Coluzzi, G. Parisi and P. Verrocchio, J. Chem. Phys. {\bf 112},
410: 2933 (2000).
411:
412:
413: \bibitem{andersen} H.C. Andersen et al. Rapport d'activite
414: scientifique du CECAM, pp 82-115 [7.4, 7.4.3].
415:
416: \bibitem {brown} D. Brown and J. H. R. Clarke, Mol. Phys. {\bf 51}, 1243 (1984).
417:
418: \bibitem {haile} J. M. Haile and H. W. Graben, J. Chem. Phys. {\bf 73}, 2412 (1980)
419:
420: \bibitem{gb} J. G. Gay and B. J. Berne, J. Chem. Phys., {\bf 74}, 3316 (1981).
421:
422: \bibitem {sbor} S. Bhattacharyya and B. Bagchi, J. Chem. Phys.,
423: {\bf 115}, 9065 (2001).
424:
425: \bibitem {sab} S. Bhattacharyya, A. Mukherjee and B. Bagchi, J. Chem. Phys.,
426: {\bf 117},2741 (2002).
427:
428: \bibitem {sharon} C. Bennemann, C. Donati, J. Baschnagel, and S. C. Glotzer,
429: Nature (London) {\bf 399}, 246 (1999).
430:
431: %\bibitem {refer} We thank an anonymous referee for pointing this out.
432:
433: \end{references}
434: \end{document}
435:
436:
437:
438: