1: \documentclass[aps,preprint,prb]{revtex4}
2: %\documentclass[aps,twocolumn,prl,showpacs]{revtex4}
3: \usepackage{graphicx}
4: \usepackage{epsfig}
5: \usepackage{amsmath}
6: \usepackage{graphics}
7: \begin{document}
8:
9: \author{M. F. Gelin}
10: \author{D. S. Kosov}
11:
12: \affiliation{Department of Chemistry and Biochemistry,
13: University of Maryland,
14: College Park,
15: 20742,
16: USA}
17:
18:
19: \title{Molecular reorientation in hydrogen-bonding liquids:
20: through algebraic $\sim t^{-3/2}$ relaxation toward exponential decay}
21:
22: \begin{abstract}
23: We present a model for the description of orientational relaxation
24: in hydrogen-bonding liquids. The model contains two relaxation parameters
25: which regulate the intensity and efficiency of dissipation, as well
26: as the memory function which is responsible for the short-time relaxation
27: effects. It is shown that the librational portion of the orientational
28: relaxation is described by an algebraic $\sim t^{-3/2}$ contribution,
29: on top of which more rapid and non-monotonous decays caused
30: by the memory effects are superimposed. The long-time behavior of the orientational
31: relaxation is exponential, although non-diffusional. It is governed
32: by the rotational energy relaxation. We apply the model to interpret
33: recent molecular dynamic simulations and polarization pump-probe experiments
34: on $HOD$ in liquid $D_{2}O$ [C. J. Fecko et al, J. Chem.
35: Phys. 122, 054506 (2005)].
36: \end{abstract}
37: \maketitle
38:
39: \section{Introduction}
40:
41: The recent advances in nonlinear ultrafast polarization-sensitive
42: spectroscopy \cite{zew01} make it possible to monitor molecular
43: rotation in hydrogen-bonding liquids in real
44: time \cite{tok04,tok05,fay05,bak05,bak00,fay04,lau98,kei99,wie04,bra02}.
45: Due to the enormous complexity of the problem, which is exacerbated by many-body effects and multitudes of the time scales involved, experimental data alone are insufficient
46: for understanding the underlying dynamics.
47: Nowadays, molecular dynamic has become a standard tool for studying molecular
48: reorientation in liquids \cite{AlTi}. Furthermore, one can even use \textit{ab
49: initio} molecular dynamics (in which the density functional
50: theory is invoked to describe molecular electronic structure and inter- and intramolecular
51: forces are calculated on-the-fly) \cite{marx03,par96} or centroid
52: molecular dynamics (which accounts for quantum effects/corrections)
53: \cite{kus04}. On the other hand, there exists a plenty of "old" phenomenological models of molecular reorientation in gases and
54: liquids. We mention the small-angle rotational diffusion model \cite{fav60,hub70,hun70},
55: the jump diffusion model \cite{val73,cuk72,cuk74,gel98}, the friction model
56: \cite{ste63,McCo,LB84}, the Gaussian cage model \cite{ste84,ste85},
57: the itinerant oscillator model \cite{gri84}, along with the more sophisticated
58: memory function approach (\cite{BurTe,gri84,ste84a,kiv88,det80,key72}
59: and references therein), the extended diffusion models \cite{gor66,sack,rid69,ste72,bra78,McCl77,bul84,gel98a,con76},
60: the rotational Fokker-Planck equation \cite{sack,rid69,hub72,McCo,mor82,McCl87,gel96a},
61: the confined rotator model \cite{kal81}, the Steele model \cite{ste81},
62: the Keilson-Storer model (KSM) \cite{sack,BurTe,gel97,gel97a,gel00,gel96,gel01},
63: the fluctuating/stochastic cage model \cite{fre95,BurTe,bur83,pol04,pol05},
64: and the generalized Langevin equations/normal mode approach \cite{str00,BurTe,gri84,ste84a}. Furthermore, the model has been elaborated
65: \cite{bra02}, which accounts for the effects of rotation-vibration
66: coupling in ensembles of hydrogen-bonding molecules on the time-resolved
67: pump-probe signals. Very recently, the generalized jump model of water reorientation has been suggested \cite{hyn06}.
68: The models, of course, rely upon a simplified picture of molecular rotation. However, in contrast with molecular dynamics simulations, they
69: get a deep insight into physics of molecular reorientation, give a clear perception of rotational relaxation and provide us with explicit formulas for the pertinent correlation functions (CFs).
70:
71: The aim of the present paper is to develop a simple and physically sound model of molecular reorientation in hydrogen-bonding liquids. The model is intended to supply experimentalists with a simple theory to interpret and to fit their data and to clarify
72: the interconnection of orientational relaxation and hydrogen bond making/breaking processes. In hydrogen-bonding liquids, the angular momentum
73: CFs exhibit pronounced oscillations and orientational CFs (OCFs) display a rapid short-time decay followed by a slower (sometimes oscillatory)
74: pattern which transforms gradually into a monotonous exponential relaxation. By incorporating the proper description of the memory effects into the
75: KSM framework, we developed a non-Markovian generalization of thereof, NKSM. Within the NKSM, we derived analytical expressions
76: for the angular momentum and energy CFs, as well as simple recursive expressions for OCFs.
77:
78:
79: The paper is structured as follows. The NKSM is formulated in Sec.II.
80: The explicit expressions for the angular momentum CF, rotational energy CF and OCFs are presented and discussed in Sec.III.
81: Sec.IV contains illustrative calculations of various NKSM CFs and comparisons
82: with the results of molecular dynamic simulations and polarization pump-probe experiments
83: on $HOD$ in $D_{2}O$ at a room temperature \cite{tok05}. A brief summary of the main findings can
84: be found in Sec. V. Appendix A contains the explicit formulas for
85: the calculation of spherical and linear rotor OCFs within the NKSM.
86: Analytical expressions for OCFs in a particular case of ``perfect''
87: librations are obtained and discussed in Appendix B.
88:
89: A few words about the notation and conventions. (i) The reduced variables
90: are used throughout the article: time, angular momentum and energy
91: are measured in units of $\sqrt{I/(k_{B}T})$, $\sqrt{Ik_{B}T}$ and
92: $k_{B}T$, respectively. Here $k_{B}$ is the Boltzmann constant,
93: $T$ is the temperature, and $I$ is a characteristic moment of inertia
94: of the molecule so that $\tau_{r}=\sqrt{I/(k_{B}T})$ is the averaged
95: period of free rotation. (ii) All Laplace-transformed operators
96: are denoted by tilde, viz. $\tilde{f}(s)=\int_{0}^{\infty}dt\exp\{-st\} f(t)$
97: for $\forall$ $f(t)$. (iii) Repeated dummy Greek indexes imply summation
98: over $x,\, y$ and $z$.
99:
100:
101: \section{The model}
102:
103: We start with a formally exact Zwanzig-type master equation, which
104: can be derived from the general $N$-particle rotation-translational
105: Liouville equation by applying the projection operator technique \cite{fre75,eva78,gel98}
106:
107: \begin{equation}
108: \partial_{t}\rho(\mathbf{J},\mathbf{\Omega},t)=-i\hat{\Lambda}(\mathbf{J},\mathbf{\Omega})\rho(\mathbf{J},\mathbf{\Omega},t)-\int_{0}^{t}dt'\hat{C}(\mathbf{J},\mathbf{\Omega},t-t')\rho(\mathbf{J},\mathbf{\Omega},t').\label{kin1}\end{equation}
109: Here $\rho(\mathbf{J},\mathbf{\Omega},t)$ is the single particle
110: probability density function, $\mathbf{J}$ is the angular momentum
111: in the molecular frame, $\mathbf{\Omega}$ are the Euler angles which
112: specify orientation of the molecular frame with respect to the laboratory
113: one. The free-rotor Liouville operator consists of the two contributions,
114: \begin{equation}
115: \hat{\Lambda}(\mathbf{J},\mathbf{\Omega})=\hat{\Lambda}_{\mathbf{\Omega}}+\hat{\Lambda}_{\mathbf{J}},\label{str}\end{equation}
116: which describe, respectively, the angular momentum driven reorientation
117: and the angular momentum change during free rotation:\begin{equation}
118: \hat{\Lambda}_{\mathbf{\Omega}}=I_{\alpha}^{-1}J_{\alpha}\hat{L}_{\alpha},\,\,\,\hat{\Lambda}_{\mathbf{J}}=-i\varepsilon_{\alpha\beta\gamma}I_{\beta}^{-1}J_{\alpha}J_{\beta}\partial_{J_{\gamma}}.\label{str1}\end{equation}
119: $I_{\alpha}$ are the main moments of inertia, $\hat{L}_{\alpha}$
120: are the angular momentum operators in the molecular frame. For linear
121: and spherical rotors, $\hat{\Lambda}_{\mathbf{J}}\equiv0$.
122:
123: The relaxation operator $\hat{C}$ assumes the form\begin{equation}
124: \hat{C}(\mathbf{J},\mathbf{\Omega},t)=\hat{C}(\mathbf{J})g(t),\label{LiHo}\end{equation}
125: $g(t)$ being the memory function which is normalized to unity, $\int_{0}^{\infty}dtg(t)=\tilde{g}(0)=1$.
126: All the formulas derived in the present paper are valid for any functional
127: form of $g(t)$. Since a simple exponential memory function is known
128: to exaggerate oscillatory effects in the angular momentum CF and OCFs
129: (see, e.g., \cite{det80,ste84a,gel97a,LB84}), the two-exponential
130: memory function will be adopted for making all specific calculations,
131: viz.,
132:
133: \begin{equation}
134: g(t)=\sigma\lambda_{1}\exp\{-\lambda_{1}t\}+(1-\sigma)\lambda_{2}\exp\{-\lambda_{2}t\},\label{g(t)}\end{equation}
135: \begin{equation}
136: \tilde{g}(s)=\frac{\sigma\lambda_{1}}{s+\lambda_{1}}+\frac{(1-\sigma)\lambda_{2}}{s+\lambda_{2}}.\label{g(s)}\end{equation}
137: The parameters $\lambda_{i}$ regulate the memory effects of the two
138: contributions, and $\sigma$ controls their relative significance.
139: If we let both $\lambda_{1}$ and $\lambda_{2}$ tend to infinity, then $g(t)\rightarrow\delta(t)$
140: and the Markovian limit is recovered.
141:
142: Since molecules are massive inertial particles, the relaxation operator
143: $\hat{C}(\mathbf{J})$ is assumed to be $\mathbf{\Omega}$-independent
144: \cite{foot3}. This is tantamount to the statement that molecular reorientation
145: is driven by the (time-dependent) angular momenta, whose relaxation,
146: in turn, is governed by operator (\ref{LiHo}). This assumption is
147: consistent with classical molecular dynamics simulations,
148: in which one integrates equations of motion of the kind \[
149: \partial_{t}D^{j}(\Omega)=-i\hat{\Lambda}_{\mathbf{\Omega}}D^{j}(\Omega),\,\,\,\partial_{t}\mathbf{J}=-i\hat{\Lambda}_{\mathbf{J}}\mathbf{J}+\mathbf{N},\]
150: $D^{j}(\Omega)$ being the Wigner D-functions \cite{var89} and $\mathbf{N}$
151: being the torque acting on a chosen molecule from its neighbours.
152: As is demonstrated below, $\mathbf{N}$ is essentially non-Gaussian
153: and non-Markovian.
154:
155: The operator $\hat{C}$ can further be represented in the general
156: form \cite{BurTe}\begin{equation}
157: \hat{C}(\mathbf{J})\rho(\mathbf{J},\mathbf{\Omega},t)=-\nu\{\rho(\mathbf{J},\mathbf{\Omega},t)-\int d\mathbf{J}'T(\mathbf{J}|\mathbf{J}')\rho(\mathbf{J}',\mathbf{\Omega},t)\}.\label{Ci}\end{equation}
158: The rate $\nu$ determines the dissipation strength, and the relaxation
159: kernel $T$ obeys the normalization
160:
161: \begin{equation}
162: \int d\mathbf{J}T(\mathbf{J}|\mathbf{J}')=1\label{norm}\end{equation}
163: and the detailed balance \begin{equation}
164: T(\mathbf{J}|\mathbf{J}')\rho_{B}(\mathbf{J}')=T(\mathbf{J}'|\mathbf{J})\rho_{B}(\mathbf{J}),\label{DetBal}\end{equation}
165:
166:
167: \begin{equation}
168: \rho_{B}(\mathbf{J})=(2\pi)^{-3/2}\exp\{-J_{\alpha}^{2}/(2I_{\alpha})\}\label{Boltz}\end{equation}
169: being the equilibrium rotational Boltzmann distribution.
170:
171: To proceed further, we adopt the KSM parametrization of the relaxation
172: kernel \cite{BurTe,KS}:
173:
174: \begin{equation}
175: T(\mathbf{J}|\mathbf{J}')=\prod_{a=x,y,z}T_{a}(J_{a}|J_{a}'),\label{KS}\end{equation}
176: \[
177: T_{a}(J_{a}|J_{a}')=[2\pi I_{a}(1-\gamma_{a}^{2})]^{-1/2}\exp\{-(J_{a}-\gamma_{a}J_{a}')^{2}/[2I_{a}(1-\gamma_{a}^{2})]\}.\]
178: Here the parameters $-1\leq\gamma_{a}\leq1$ determine the relaxation
179: mechanisms. When $\gamma_{a}=1$, then $T_{a}(J_{a}|J_{a}')$$=\delta(J_{a}-J_{a}')$
180: and $\hat{C}(\mathbf{J})=0$. The Fokker-Planck relaxation operator
181: is recovered in the limit $\gamma_{a}\rightarrow1$, $\nu\rightarrow\infty$,
182: $\nu(1-\gamma_{a})\rightarrow\nu_{a}=\mathrm{const}$. If $\gamma_{a}=0,$
183: intermolecular interactions are so strong that they {}``immediately''
184: restore an equilibrium Boltzmann distribution in the molecular ensemble
185: ($T(\mathbf{J}|\mathbf{J}')\rightarrow\rho_{B}(\mathbf{J})$). Therefore, the
186: KSM contains the J-diffusion model \cite{gor66,sack,rid69,ste72,bra78,McCl77,bul84,gel98a,con76}
187: and the rotational Fokker-Planck equation \cite{sack,rid69,hub72,McCo,mor82,McCl87}
188: as special cases.
189:
190: By letting $\gamma_{a}=-1$, one gets $T_{a}(J_{a}|J_{a}')$ $=\delta(J_{a}+J_{a}')$.
191: Thus the magnitude of the angular momentum is preserved but its direction
192: is reversed. This regime ($\gamma_{a}\simeq-1$, a molecule rotates
193: back and forth within the cage formed by its nearest neighbors) is
194: expected to be of particularly relevance for hydrogen-bonding liquids.
195: Such physical picture of molecular reorientation in liquids is inherent
196: in a number of theoretical approaches, in which the influence of the
197: nearest neighbors on the selected molecule is modeled by external
198: potentials with several minima
199: \cite{ste84,ste85,kal81,kus77,kam81,deb88,eva95,bur94,lit72,fre55},
200: or by fluctuating torques and structures \cite{gri84,AlTi,fre95}. Within
201: the present approach, the fluctuating cage potential is not introduced
202: explicitly, but its influence is taken into account dynamically via kernel (\ref{KS}).
203:
204: Before embarking at particular calculations, it is useful to estimate
205: values of the parameters $\nu,\,\gamma_{a}$ (Eq. (\ref{KS})) and
206: $\lambda_{i}$ (Eq. (\ref{g(t)}) for hydrogen-bonding liquids. Since
207: the molecules are assumed to undergo hindered librations, one expects
208: $\nu\gg1$ and $\gamma_{a}\sim-1$. The memory effects are supposed
209: to be quite significant ($\lambda_{i}\sim1$).
210:
211:
212: \section{Correlation functions}
213:
214: After the explicit form of the relaxation operator (\ref{KS}) has
215: been determined, the master equation (\ref{kin1}) can be invoked
216: to calculate any rotational and/or orientational CF of interest.
217: To study the evolution of any quantity, which
218: depends solely on the angular momentum, we integrate Eq. (\ref{kin1})
219: over $\mathbf{\Omega}$ and obtain the reduced master equation
220:
221: \begin{equation}
222: \partial_{t}\rho(\mathbf{J},t)=-i\hat{\Lambda}_{\mathbf{J}}\rho(\mathbf{J},t)-\int_{0}^{t}dt'g(t-t')\hat{C}(\mathbf{J})\rho(\mathbf{J},t').\label{kinJ}\end{equation}
223: OCF of the rank $j$ is defined through the Wigner D-functions as
224: follows:
225:
226: \begin{equation}
227: G^{j}(t)\equiv<D^{j}(\Omega(t))D^{j}(\Omega(0))^{-1}>\equiv\int d\mathbf{J}G^{j}(\mathbf{J},t).\label{OCF}\end{equation}
228: After the insertion of the above definition into Eq. (\ref{kin1})
229: one obtains the following equation: \begin{equation}
230: \partial_{t}G^{j}(\mathbf{J},t)=-i(\Lambda_{\mathbf{\Omega}}+\hat{\Lambda}_{\mathbf{J}})G^{j}(\mathbf{J},t)-\int_{0}^{t}dt'g(t-t')\hat{C}(\mathbf{J})G^{j}(\mathbf{J},t').\label{OCF1}\end{equation}
231: Here operator $\Lambda(\mathbf{J})$ is determined by Eqs. (\ref{str})
232: and (\ref{str1}), in which the angular momentum operators $\hat{L}_{\alpha}$
233: are replaced by their matrix elements $L_{\alpha}^{j}$ over the D-functions:
234:
235: \begin{equation}
236: (L_{x}^{j})_{kl}\pm i(L_{y}^{j})_{kl}=\delta_{k,l\mp1}\{(j\pm l)(j\mp l+1)\}^{1/2},\,\,(L_{z}^{j})_{kl}=l\delta_{kl};\,\,-j\leq k,l\leq j.\label{Jxyz}\end{equation}
237: Eqs. (\ref{kinJ}) and (\ref{OCF1}) can be solved numerically in
238: case if a general asymmetric top molecule. To make the presentation
239: simpler, we restrict ourselves to the consideration of spherically
240: symmetric molecules. The corresponding formulas are much more elucidating
241: and convenient to analyse, since the relaxation operator $\hat{C}(\mathbf{J})$
242: is described by only two dynamic parameters (the intensity, $\nu$,
243: and the efficiency, $\gamma$, of dissipation) and a spherical molecule
244: possesses a single moment of inertia, $I$. This theory can be applied
245: to asymmetric tops as well. Indeed, in the hindered rotation limit
246: ($\tau_{J}\ll1$), a molecule librates back and forth in its cage,
247: and every single libration reorients the molecule to a small angle.
248: Using the explicit form of the operator $\hat{\Lambda}_{\mathbf{\Omega}}$
249: (\ref{str1}), it is easy to demonstrate that the averaged inertial
250: reorientation angle of an asymmetric top around its $z$-axis equals
251: $(t/\tau_{r})^{2}\ll1$, where \begin{equation}
252: \tau_{r}=\sqrt{I/(k_{B}T}),\,\,\, I^{-1}=(I_{x}^{-1}+I_{y}^{-1})/2.\label{Ixy}\end{equation}
253: This corresponds to the rotation of the spherical molecule with the
254: effective moment of inertia $I$. Thus, all the formulas obtained
255: below can be used for asymmetric top molecules by making use of the
256: substitution (\ref{Ixy}), provided one is interested in the orientational
257: relaxation of the tensor with nonzero components along the molecular
258: $z$-axis. If the quantity under study possesses nonzero components
259: along several axes of the main moments of inertia (like $OH$ stretch
260: in $HOD$), the above analysis remains true if the effective moment
261: of inertia $I$ is modified accordingly. On the contrary, free inertial
262: rotation of spherical, symmetric and asymmetric tops is very different
263: \cite{ste69,blo86,lei86}.
264:
265: Starting from Eq. (\ref{kinJ}), it is straightforward to derive the explicit formulas for the Laplace
266: transformations of the angular momentum and rotational energy CFs:
267:
268: \begin{equation}
269: C_{J}(t)=\frac{\left\langle \textbf{JJ}(t)\right\rangle }{\left\langle \textbf{J}^{2}\right\rangle },\,\,\,\tilde{C}_{J}(s)=\frac{1}{s+\nu_{J}\tilde{g}(s)},\label{CJ}\end{equation}
270: \begin{equation}
271: C_{E}(t)=\frac{\left\langle \textbf{J}^{2}\textbf{J}^{2}(t)\right\rangle -\left\langle \textbf{J}^{2}\right\rangle ^{2}}{\left\langle \textbf{J}^{4}\right\rangle -\left\langle \textbf{J}^{2}\right\rangle ^{2}},\,\,\,\tilde{C}_{E}(s)=\frac{1}{s+\nu_{E}\tilde{g}(s)}.\label{CE}\end{equation}
272: Here the rates \begin{equation}
273: \nu_{J}=\nu(1-\gamma),\,\,\nu_{E}=\nu(1-\gamma^{2})\label{NuJE}\end{equation}
274: determine the angular momentum and rotational energy integral relaxation
275: times:\begin{equation}
276: \tau_{J}=\int_{0}^{\infty}dtC_{J}(t)=\tilde{C}_{J}(0)=\nu_{J}^{-1},\label{TauJ}\end{equation}
277: \begin{equation}
278: \tau_{E}=\int_{0}^{\infty}dtC_{E}(t)=\tilde{C}_{E}(0)=\nu_{E}^{-1}.\label{TauE}\end{equation}
279:
280: As to the OCFs, the solution of Eq. (\ref{OCF1})
281: can be given in terms of recursive relationships (or, equivalently,
282: in terms of continued fractions) for any value of the relaxation parameters $\nu$ and $\gamma$, as
283: well as for any memory function $g(t)$, see Appendix A.
284:
285: \section{ Illustrative calculations}
286: In this Section, we present and discuss the results of representative calculations of the angular momentum CFs (\ref{CJ}), energy CFs (\ref{CE}), and OCFs (\ref{Gb0}) - (\ref{sisi}) within the NKSM.
287: The free rotation period (\ref{Ixy}) has been taken as $\tau_{r}=\sqrt{I/(k_{B}T})=93$fs.
288: This corresponds to $HOD$ at $T=300$ K ($I_{x}=2.63$, $I_{y}=1.85$,
289: $I_{z}=0.72$ $a.m.u.\times \AA^{2}$).
290:
291: We start from the angular momentum (\ref{CJ}) and energy (\ref{CE})
292: CFs. If the two-exponential memory function (\ref{g(t)}) is used, the
293: Laplace images $\tilde{C}_{J}(s)$ and $\tilde{C}_{E}(s)$ can be inverted into the time domain by solving the pertinent cubic equation. The CFs in the time domain are thus determined by
294: linear combinations of one real and two complex conjugated exponentials.
295: The results of representative calculations are depicted in Fig.
296: 1. A decrease of the second memory parameter, $\lambda_{2}$, causes
297: a typical transformation of the angular momentum CF, which reflects
298: a passage from simple to hydrogen-bonding liquids \cite{LB84}. Since
299: CFs $C_{J}(t)$ attain negative values, their actual decay occur at
300: a timescale of several hundreds of femtosecond, which is much longer
301: than the integral relaxation time $\tau_{J}=\nu_{J}^{-1}=4.65$ fs.
302: The rotational energy CFs are monotonous and decay at a much longer
303: time scale of $10\div15$ ps, which is a direct manifestation of the
304: librational motion ($\gamma\sim-1$, $\tau_{J}\ll\tau_{E}$).
305:
306: Let us turn to the study of orientational relaxation.
307: To get a qualitative feeling
308: of the influence of the relaxation efficiency $\gamma$ on molecular
309: reorientation, let us concentrate on the long-time behavior of OCFs
310: and consider the Markovian limit, $g(t)\rightarrow\delta(t)$. As
311: has been established in \cite{gel96,gel01}, the KSM predicts that
312: the smaller is $\gamma$, the slower is the OCF decay. Thus, for a
313: fixed angular momentum relaxation rate $\nu_{J}$, the rotational
314: Fokker-Planck equation ($\gamma=1$) predicts the most rapid orientational
315: relaxation, while the limit of perfect librations ($\gamma=-1$) corresponds
316: to the slowest orientational relaxation. This is clearly seen in Fig.
317: 2.
318:
319: As is demonstrated in Appendix B, Eq. (\ref{OCF1}) can be solved
320: analytically for OCFs in case of perfect forward-backward librations ($\gamma=-1$
321: ) in the hindered rotation limit ($\nu\gg1$):
322:
323: \begin{equation}
324: G^{j}(t)=\frac{1}{2j+1}\left(1+2\sum_{k=1}^{j}\left[1+\frac{k^{2}}{\nu}t\right]^{-3/2}\right).\label{Gt2}\end{equation}
325: On the scale of Fig. 2, the exact solution of Eq. (\ref{OCF1}) in
326: the limit of $\gamma=-1$ and the approximate one, which is delivered
327: by Eq. (\ref{Gt2}), are indistinguishable. Thus the in-cage librations
328: manifest themselves through a slow algebraic $t^{-3/2}$ decay of
329: the OCF. This behavior is caused by the angular momentum reversion
330: ($\gamma=-1$) and has nothing in common with long-time hydrodynamic
331: tails of the angular velocity CFs of Brownian particles (see \cite{fel97,mas97,AlTi}
332: and references therein) \cite{foot2}. It is interesting to point
333: out that the OCFs calculated within the M- \cite{bha85,McCl77} and
334: E- \cite{gel98a} diffusion models exhibit similar ($\sim t^{-3/2}$
335: ) long time tails, which are commonly regarded as unphysical. Nonetheless,
336: these models were successfully invoked to reproduce {}``experimental''
337: OCFs, which were obtained through the inversion of IR and Raman spectra
338: into the time domain \cite{bha85,dre79,gel98a,jon81,rot75}. The present
339: consideration reveals that this success is not accidental, since the
340: conservation of the magnitude of the angular momentum (M- diffusion)
341: or rotational energy (E- diffusion) mimics the description of in-cage
342: librations via the KSM kernel (\ref{KS}), which in the limit $\gamma\sim-1$
343: conserves both of these quantities.
344:
345: Fig. 3 illustrates the influence of the memory effects, which modify
346: the short-time behavior of the OCFs. The mechanism of this influence
347: is uncovered by the general expression \cite{McCo}\begin{equation}
348: G^{j}(t)\approx1-j(j+1)\int_{0}^{t}dt'(t-t')C_{J}(t'),\label{GtShort}\end{equation}
349: which relates the short-time OCF behavior with the time evolution
350: of the angular momentum CF. That is why the parameters which have
351: been used for the calculation of the OCFs in Fig. 3 and the angular
352: momentum CFs in Fig. 1a are the same. Generally, the memory effects
353: speed up the short-time OCF decay. If the angular momentum CF is highly
354: oscillatory, these oscillations show up in OCFs also (compare dotted
355: lines in Figs. 1a and 3).
356:
357: OCFs depicted in Fig. 3 look qualitatively similar to those obtained
358: via computer simulations \cite{AlTi,ski03,tok04,tok05,kus04,kus94}.
359: A more quantitative comparison of the simulated and NKSM OCFs is presented
360: in Fig. 4. It depicts the results of molecular dynamic simulations
361: of the first (upper dotted line) and second (lower dotted line) rank OCFs
362: for $HOD$ in $D_{2}O$ at a room temperature using the SPC/E model for water\cite{tok05} along with the fits obtained within the NKSM (full lines). Detail of the molecular dynamics simulation protocol can be found in ref.\cite{tok04,eav04}
363:
364: The present theory is seen to reproduce the simulated OCFs quite well.
365: We were unable, however, to fit the first and second rank OCFs by
366: the same set of the NKSM parameters. This is not unexpected: OCFs
367: of different ranks are affected by the cage potentials in a different
368: way, which is determined by the potential symmetry \cite{fre95}.
369: The dielectric friction, which governs orientational relaxation in
370: polar systems is also known to be rank-dependent \cite{bag94}. The
371: values of $\nu$, $\gamma$, $\lambda_{i}$ and $\sigma$ which are
372: extracted from the angular momentum CF deliver, in fact, certain averaged
373: values. It is nonetheless quite remarkable that the parameters which
374: have been used for the calculation of the second rank OCF are identical
375: to those which have been used for the computation of the angular momentum
376: CF (Fig. 1a, full line). This latter CF looks very similar to the
377: simulated one \cite{kus04}.
378:
379: Fig. 5 shows the normalized anisotropy extracted from the pump-probe
380: signal \cite{tok05} (dotted line). It deviates quite significantly
381: from the simulated second-rank OCF (our best fit to this OCF is plotted
382: here for the sake of comparison). As has been pointed out in \cite{tok05,tok04,ski03},
383: simulations normally predict faster, in comparison with experiment,
384: anisotropy decays. This is caused, perhaps, by ignoring either the
385: water polarizability \cite{ber02,ich99} or quantum effects \cite{kus04}.
386: We are not attempting to resolve this controversy here. Note merely
387: that the experimental anisotropy can be fitted quite well within the
388: NKSM (full line), with the set of parameters which predict more {}``librational''
389: reorientation ($\gamma$ is closer to $-1$) and more pronounced memory
390: effects ($\lambda_{i}$ is smaller).
391:
392: Let us return back to Fig. 4. The OCFs calculated within the standard
393: diffusion equation, \begin{equation}
394: G^{j}(t)=\exp\{-j(j+1)\tau_{J}t\},\label{dif}\end{equation}
395: which reproduces the long-time limit of the first cumulant formula,
396: \begin{equation}
397: G^{j}(t)=\exp\{-j(j+1)\int_{0}^{t}dt'(t-t')C_{J}(t')\},\label{GtShort1}\end{equation}
398: are seen to deviate significantly from the simulated/NKSM OCFs. Despite
399: the simulated/NKSM OCFs do exhibit the long-time exponential behavior,
400: and despite the rotational motion is definitely hindered ($\tau_{J}\ll1$),
401: the diffusion equation cannot reproduce these OCFs. The failure of
402: the diffusion equation has quite an evident explanation. If we assume
403: that the molecules undergo {}``perfect'' librations ($\gamma=-1$
404: within the present theory), then the long-time behavior of the OCF
405: in the hindered rotation limit is described by Eq. (\ref{Gt2}) rather
406: than by the small-angle diffusion equation (\ref{dif}). Since the
407: actual librations are not {}``perfect'' ($\gamma$ is close to but
408: less than $-1$) one expects deviations from Eq. (\ref{Gt2}). As is
409: seen from Fig. 6, this is indeed the case. This Figure reproduces
410: the Markovian limits of the KSM OCFs from Figs. 4 and 5, along with
411: their counterparts calculated via Eq. (\ref{Gt2}). The short-to-intermediate-time
412: resemblance of the OCFs is quite remarkable.
413:
414: On the other hand, the long-time behavior of the simulated/KSM OCFs
415: is seen to be exponential and is not reproduced by Eq. (\ref{Gt2}).
416: This hints at a possibility that rotational energy relaxation might
417: be responsible for the long time exponential decay of the simulated/NKSM
418: OCFs. This hypothesis is corroborated by the observation that the
419: fit of the simulated and NKSM OCFs of the first and second rank (Fig.
420: 4) delivers different values of the relaxation parameters $\nu$ and
421: $\gamma$. However, the quantity $\nu_{E}=\nu(1-\gamma^{2})=0.027$
422: turns out to be the same for both $j=1$ and $2$. This gives the
423: estimated value of $3.4$ ps for the rotational energy relaxation
424: time $\tau_{E}$. This value correlates with the experimentally measured
425: long time anisotropy decay times of $2\div3$ ps \cite{tok04,tok05,bak00,bak05,fay05,wie04}.
426:
427: Note, finally, that a close interrelation between the hydrogen bond
428: dynamics and rotation dynamics has repeatedly been emphasized in the
429: literature. The NKSM values of the {}``persistence time'' of the
430: oscillatory angular momentum CF (Fig. 1a) and the rotational energy
431: relaxation time, $\sim150$fs and $3.4$ps, correlate quite well with
432: the estimations for the continuous and intermitted hydrogen bond lifetimes
433: \cite{ski03,luz00}. This observation is consistent with the physical
434: picture of molecular rotation underlying the NKSM approach. The in-cage
435: librations, which can cause bond breakings, manifest themselves on
436: the time scale of the {}``persistence time'' of the angular momentum
437: CF. On the other hand, the bond rupturings are accompanied by subsequent
438: bond reformings, since the molecule does not leave its local cage.
439: A ``true'' cleavage of the bond can occur on the $\tau_{E}$ timescale
440: since, within the NKSM description, the rotational energy relaxation
441: occurs due to a hopping to a new position of the local equilibrium
442: or restructuring the local potential well.
443:
444:
445: \section{Conclusion}
446: We have developed a NKSM description of the orientational relaxation
447: in hydrogen-bonding liquids. Within the NKSM, molecular rotation is
448: governed by two relaxation parameters $\nu$ and $\gamma$ (which
449: describe the intensity and mechanism of dissipation), as well as by
450: the memory function $g(t)$ (\ref{g(t)}), which is responsible for
451: the short-time dynamics. Alternatively, the relaxation parameters
452: $\nu$ and $\gamma$ are uniquely determined through the angular momentum
453: and energy relaxation times $\tau_{J}$ (\ref{TauJ}) and $\tau_{E}$
454: (\ref{TauE}). Once a set of the parameters is selected, the NKSM
455: allows to calculate any rotational CF or OCF of interest. Keeping
456: in mind a considerable success of the KSM in reproducing molecular
457: reorientation in gases \cite{BurTe,gel00}, the results of the present
458: work demonstrate that the NKSM can be used for the description
459: and interpretation of the orientational relaxation in a condensed
460: phase, from rarefied gases with binary collisions, through dense fluids
461: to hydrogen-bonding liquids.
462:
463: The NKSM suggests the short-time relaxation of the OCFs in the hydrogen-bonding
464: liquids is described by an algebraic $\sim t^{-3/2}$ contribution. This algebraic behavior is
465: modified by more rapid and non-monotonous dynamics, which is induced by the memory effects.
466: The long-time decay
467: of the OCFs is exponential, although non-diffusional. It is governed
468: by the rotational energy relaxation time, $\tau_{E}$.
469: Our results are contrary to standard belief that
470: the angular momentum CF determines molecular reorientation
471: in the hindered rotation limit, and the first cumulant expression,
472: Eq. (\ref{GtShort1}), delivers the leading contribution into the
473: OCF. Our results indicate, that knowing $C_{J}(t)$ is not enough
474: to predict OCFs for hydrogen-bonding liquids, since the long-time
475: behavior of OCFs is governed by the rotational energy CF, $C_{E}(t)$.
476:
477: It is a conventional practice to fit various experimental or simulated CFs
478: via a linear combination of several real (if the CF decays monotonously)
479: or complex (if the CF exhibits oscillatory behavior) exponents. According
480: to the present analyses, the OCF in hydrogen-bonding liquids contains an algebraic $\sim t^{-3/2}$ contribution.
481: This finding suggests that the following fitting formulas for the angular momentum CF, \begin{equation}
482: C_{J}(t)=a_{1}\exp\{-\nu_{1}t\}+\exp\{-\nu_{2}t\}(a_{2}\cos(\Omega t)+a_{3}\sin(\Omega t)),\label{FitCJ}\end{equation}
483: and for the OCF,\[
484: G^{j}(t)=b_{1}\exp\{-\nu_{1}t\}+\exp\{-\nu_{2}t\}(b_{2}\cos(\Omega t)+b_{3}\sin(\Omega t))\]
485: \begin{equation}
486: +b_{4}\exp\{-\nu_{3}t\}\left[1+b_{5}t\right]^{-3/2}+b_{6}\exp\{-\nu_{4}t\},\label{Fit}\end{equation}
487: can be more physically motivated ($a_{i},\, b_{i}$, $\nu_{i}$ and
488: $\Omega$ being certain real-valued parameters). Indeed, Eq. (\ref{FitCJ})
489: is nothing else than a formal solution of Eq. (\ref{CJ}) in the case
490: of two-exponential memory function (\ref{g(t)}). As to the OCF (\ref{Fit}),
491: the first three terms with coefficients $b_{1},\, b_{2},\, b_{3}$
492: describe the angular momentum induced short-time rapid decay and oscillations.
493: The term which is proportional to $b_{4}$ is responsible for the algebraic
494: contribution, and the last term governs the long-time exponential
495: decay.
496:
497: \begin{acknowledgments}
498: The authors are grateful to Joseph Loparo for sending them the numerical
499: data on the simulated OCFs and measured anisotropies, which have
500: been published in \cite{tok05}. M. F. G. thanks Alexander Blokhin
501: for numerous stimulating and useful discussions.
502: \end{acknowledgments}
503:
504:
505: \section{Appendix A. Recursive relations for linear and spherical top OCFs}
506:
507: After being Laplace transformed, Eq. (\ref{OCF1}) reads:\begin{equation}
508: -\rho_{B}(\mathbf{J})+s\tilde{G}^{j}(\mathbf{J},s)=-i(\Lambda_{\mathbf{\Omega}}+\hat{\Lambda}_{\mathbf{J}})\tilde{G}^{j}(\mathbf{J},t)-\nu \tilde{g}(s)\{\tilde{G}^{j}(\mathbf{J},s)-\int d\mathbf{J}'T(\mathbf{J}|\mathbf{J}')\tilde{G}^{j}(\mathbf{J}',s)\}.\label{OCFL}\end{equation}
509: Since Eq. (\ref{OCFL}) depends on $s$ parametrically, the method
510: of its solution in the Markovian limit ($\tilde{g}(s)=1$), which has been
511: been developed in \cite{sack,gel96,gel96a,gel00}, is directly applicable
512: to the present case also. One has merely consider the complex quantity
513: $\nu g(s)$ as the generalized relaxation rate. We therefore present
514: the final expressions for the calculation of the first and second
515: rank OCFs.
516:
517: Their Laplace images of the spherical top OCFs can be calculated via
518: the formula\begin{equation}
519: \tilde{G}^{j}(s)=(1+2b_{0})/s.\label{Gb0}\end{equation}
520: For $j=1$, the coefficient $b_{0}$ can be retrieved from the simple
521: three-term recursive formula \cite{sack,gel96}\begin{equation}
522: \frac{1}{s}\delta_{m0}=\frac{4m+10}{\sigma_{m+1}}b_{m+1}-\left\{ \frac{2m+3}{\sigma_{m+1}}+\frac{2m+2}{\sigma_{m}}+\zeta_{m}\right\} b_{m}+\frac{m}{\sigma_{m}}b_{m-1},\label{Lj1}\end{equation}
523: $\delta_{m0}$ being the Kronecker delta. The value of $b_{0}$ for
524: the second rank OCF can be extracted from the system of coupled recursive
525: relations for the coefficients $b_{m}$ and $d_{m}$ \cite{gel96}:\begin{equation}
526: \frac{1}{s}\delta_{m0}=-\left\{ \frac{6}{\sigma_{m}}+\zeta_{m}\right\} b_{m}+\frac{12}{\sigma_{m+1}}b_{m+1}\label{Lj2a}\end{equation}
527: \[
528: -\frac{4m}{\sigma_{m}}d_{m-1}+\left\{ \frac{8m-14}{\sigma_{m}}+\frac{8m-31}{\sigma_{m+1}}\right\} d_{m}+\frac{-16m^{2}+74m+12}{(m+1)\sigma_{m+1}}d_{m+1};\]
529:
530:
531: \textcompwordmark{}
532:
533: \textcompwordmark{}\begin{equation}
534: 0=\frac{m}{\sigma_{m}}b_{m-1}-\left\{ \frac{2m-1}{\sigma_{m}}+\frac{2m-1}{\sigma_{m+1}}\right\} b_{m}+\frac{4m}{\sigma_{m+1}}b_{m+1}\label{Lj2b}\end{equation}
535: \[
536: +\frac{5m}{\sigma_{m}}d_{m-1}-\left\{ \frac{14}{\sigma_{m}}+\frac{4+10m}{\sigma_{m+1}}+\zeta_{m}\right\} d_{m}+\frac{4m(9+5m)}{(m+1)\sigma_{m+1}}d_{m+1}.\]
537: Here \begin{equation}
538: \sigma_{m}\equiv s+\nu \tilde{g}(s)(1-\gamma^{2m}),\,\,\,\zeta_{m}\equiv s+\nu \tilde{g}(s)(1-\gamma^{2m+1}).\label{sisi}\end{equation}
539: The solution of the recursive relations for the first rank OCF (\ref{Lj1})
540: can be expressed in the continued fraction form \cite{sack}, while
541: the solution of Eqs. (\ref{Lj2a}) and (\ref{Lj2b}) for the second-rank
542: OCF can be given in terms of the matrix $2\times2$ continued fractions
543: \cite{ris}.
544:
545: The first and second rank OCFs for linear rotors can be evaluated
546: very similarly, through the simple three-term recursive formulas.
547: Namely, the Laplace images of the OCFs can also be computed via Eq.
548: (\ref{Gb0}). For $j=1$, the coefficient $b_{0}$ must be determined
549: by the formula \cite{sack,gel00} \[
550: \frac{1}{s}\delta_{m0}=\frac{4m+8}{\sigma_{m+1}}b_{m+1}-\left\{ \frac{2m+2}{\sigma_{m+1}}+\frac{2m+2}{\sigma_{m}}+\zeta_{m}\right\} b_{m}+\frac{m}{\sigma_{m}}b_{m-1},\]
551: while the second-rank OCF ($j=2$) can be computed through \cite{gel00}\[
552: \frac{3}{s}\delta_{m0}=\frac{16m+32}{\sigma_{m+1}}b_{m+1}-\left\{ \frac{8m+10}{\sigma_{m+1}}+\frac{8m+6}{\sigma_{m}}+\zeta_{m}\right\} b_{m}+\frac{4m}{\sigma_{m}}b_{m-1}.\]
553:
554:
555:
556: \section{Appendix B. Orientational relaxation in case of perfect angular momentum
557: reorientation}
558:
559: If we neglect the memory effects ($g(t)\rightarrow\delta(t)$) and
560: put $\gamma=-1$, then Eq. (\ref{OCF1}) can be solved analytically:
561:
562: \begin{equation}
563: G^{j}(t)=\frac{1}{2j+1}\int_{0}^{\infty}dJ\rho_{B}(J)\label{Gt1}\end{equation}
564: \[
565: \times\left(1+\exp\{-\nu t\}\sum_{k=1}^{j}\left[(1+\nu/\omega_{k})\exp\{\omega_{k}t\}+(1-\nu/\omega_{k})\exp\{-\omega t\}\right]\right).\]
566: The frequencies are explicitly defined as follows\begin{equation}
567: \omega_{k}=\sqrt{\nu^{2}-k^{2}J^{2}}.\label{Wk}\end{equation}
568: Several important properties of Eq. (\ref{Gt1}) are to be discussed.
569: OCF (\ref{Gt1}) possesses a stationary asymptote: $G^{j}(t\rightarrow\infty)=(2j+1)^{-1}$,
570: which is identical to the free OCF asymptote. This can be easily understood:
571: reversion of the angular momentum is equivalent to the reversion of
572: the sense of molecular rotation. Therefore a sequence of forward-backward
573: rotations is equivalent to a single free rotation. This asymptote
574: is solely caused by dynamic effects (in-cage librations, compare with
575: \cite{eva95}) rather than by external potentials (see, e.g., \cite{sza84,ham05,fay05}).
576: The librational motion itself is caused, of course, by in-cage potentials
577: but they do not enter explicitly into our analysis. If $\nu\rightarrow0$,
578: Eq. (\ref{Gt1}) reproduces the free spherical top OCF. If we take
579: the opposite limit $\nu\rightarrow\infty$, then $G^{j}(t)\rightarrow1$
580: since a large number of small-angle forward-backward rotations causes
581: no net reorientation. In the hindered rotation limit ($\tau_{J}\ll1$)
582: Eq. (\ref{Gt1}) reduces to (\ref{Gt2}).
583:
584: Using the explicit form of the free linear rotor OCF \cite{ste69}
585: one can easily derive the linear rotor counterpart of Eq. (\ref{Gt2}):\begin{equation}
586: G^{j}(t)=\left(d_{00}^{j}(\frac{\pi}{2})\right)^{2}+2\sum_{k=1}^{j}\left(d_{0k}^{j}(\frac{\pi}{2})\right)^{2}\left[1+\frac{k^{2}}{\nu_{l}}t\right]^{-1},\label{Gtlin}\end{equation}
587: $d_{km}^{j}(\beta)$ being the reduced Wigner function \cite{var89}.
588: It is well known that the exponent $d$ of the long-time hydrodynamic
589: tails $t^{-d/2}$ of the angular velocity CFs of Brownian particles
590: is determined by the dimensionality of the rotation space: $d=3$
591: for any spherical, symmetric or asymmetric top while $d=2$ for a
592: linear rotor (\cite{fel97,mas97} and references therein). This general
593: statement holds true in the present case also, and OCF (\ref{Gtlin})
594: possesses a $\sim t^{-1}$ tail. This means that the algebraic contribution
595: to OCF (\ref{Gtlin}) decays slower than its $t^{-3/2}$ counterpart
596: in the spherical top OCF (\ref{Gt2}). Furthermore, $d_{00}^{j}(\pi/2)=0$
597: for odd $j$. Thus, the odd-ranked OCFs do not possess the stationary
598: contribution and decay faster then the even-ranked OCFs.
599:
600: \clearpage
601:
602: \begin{thebibliography}{20}
603: \bibitem{zew01}J. S. Baskin and A. H. Zewail, J. Phys. Chem. A 105, 3680 (2001).
604: \bibitem{lau98}R. Laenen, C. Rauscher and A. Laubereau, Phys. Rev. Lett. 80, 2622
605: (1998).
606: \bibitem{kei99}C. Ronne, P.-O. Astrand and S. O. Keiding, Phys. Rev. Lett. 82, 2888
607: (1999).
608: \bibitem{tok04}J. J. Loparo, C. J. Fecko, J. D. Eaves, S. T. Roberts, and A. Tokmakoff,
609: Phys. Rev. B 70, 180201(R) (2004).
610: \bibitem{tok05}C. J. Fecko, J. J. Loparo, S. T. Roberts, and A. Tokmakoff, J. Chem.
611: Phys. 122, 054506 (2005).
612: \bibitem{bak00}H.-K. Neinhuys, R. A. van Santen and H. J. Bakker, J. Chem. Phys.
613: 112, 8487 (2000).
614: \bibitem{bak05}Y. L. A. Rezus and H. J. Bakker, J. Chem. Phys. 123, 114502 (2005).
615: \bibitem{fay04}T. Steinel, J. B. Asbury, J. Zheng and M. D. Fayer, J. Phys. Chem.
616: A 108, 10957 (2004).
617: \bibitem{fay05}H.-S. Tan, I. R. Piletic, and M. D. Fayer, J. Chem. Phys. 122, 174501
618: (2005).
619: \bibitem{wie04}D. Cringus, S. Yeremenko, M. S. Pshenichnikov and D. A. Wiersma, J.
620: Phys. Chem. B 108, 10387 (2004).
621: \bibitem{bra02}G. Gallot, S. Bratos, S. Pommeret, N. Lascoux, J.-Cl. Leicknam, M.
622: Kozinski, W. Amir and G. M. Gale, J. Chem. Phys. 117, 11301 (2002).
623: \bibitem{AlTi}M. P. Allen and D. J. Tildesley, Computer Simulation of Liquids (Clarendon
624: Press, Oxford, 1991).
625: \bibitem{par96}M. Sprik, J. Hutter and M. Parrinello, J. Chem. Phys. 105, 1142 (1996).
626: \bibitem{marx03}A. D. Boese, A. Chandra, J. M. L. Martin, and D. Marx, J. Chem. Phys.
627: 119, 5965 (2003).
628: \bibitem{kus04}L. Hernndez de la Pea and P. G. Kusalik, J. Chem. Phys. 121, 5992
629: (2004).
630: %\bibitem{LB80}R. M. Lynden-Bell, Chem. Phys. Lett. 70, 477 (1980).
631: \bibitem{fav60}L. D. Favro, Phys. Rev. 119, 53 (1960).
632: \bibitem{hub70}P. S. Hubbard, J. Chem. Phys. 52, 563 (1970).
633: \bibitem{hun70}W. T. Huntres, Adv. Magn. Res. 40, 1 (1970).
634: \bibitem{val73}K. A. Valijev and E. N. Ivanov, Uspekhi Fiz. Nauk 109, 31 (1973).
635: \bibitem{cuk72}R. I. Cukier and K. Lakatos-Lindenberg, J. Chem. Phys. 57, 3427 (1972).
636: \bibitem{cuk74}R. I. Cukier, J. Chem. Phys. 60, 734 (1974).
637: \bibitem{gel98}A. P. Blokhin and M. F. Gelin, Physica A 251, 469 (1998).
638: \bibitem{ste63}W. A. Steele, J. Chem. Phys. 38, 2404 (1963); ibid 38, 2411 (1963).
639: \bibitem{McCo}G. W. Ford, J. T. Lewis and J. McConnell, Phys. Rev. A. 19, 907 (1979).
640: \bibitem{LB84}R. M. Lynden-Bell, in Molecular liquids, edited by A. J. Barnes, W.
641: J. Orville-Thomas and J. Yarwood (NATO ASI Series C, V. 135, 1984),
642: P. 501.
643: \bibitem{ste84}R. M. Lynden-Bell and W. A. Steele, J. Phys. Chem. 88, 6514 (1984).
644: \bibitem{ste85}V. N. Kabadi and W. A. Steele, J. Phys. Chem. 89, 1467 (1985).
645: \bibitem{gri84}W. Coffey, M. Evans and P. Grigolini. Molecular diffusion and spectra
646: (John Wiley \& Sons, 1984).
647: \bibitem{ste84a}W. A. Steele, in Molecular liquids, edited by A. J. Barnes, W. J.
648: Orville-Thomas and J. Yarwood (NATO ASI Series C, V. 135, 1984), P.
649: 111.
650: \bibitem{key72}D. Kivelson and T. Keyes, J. Chem. Phys. 57, 4599 (1972).
651: \bibitem{kiv88}D. Kivelson and R. Miles, J. Chem. Phys. 88, 1925 (1988).
652: \bibitem{det80}E. Detyna, K. Singer, J. V. L. Singer and A. J. Taylor, Mol. Phys.
653: 41, 31 (1980).
654: \bibitem{BurTe}A. I. Burshtein and S. I. Temkin. Spectroscopy of Molecular Rotations
655: in Gases and Liquids (Cambridge University Press, Cambridge, 1994).
656: \bibitem{gor66}R. G. Gordon, J. Chem. Phys. 44, 1830 (1966).
657: \bibitem{sack}R. A. Sack, Proc. Phys. Soc. B. 70, 402 (1957); ibid 70, 414 (1957).
658: \bibitem{rid69}M. Fixman and K. Rider, J. Chem. Phys. 51, 2425 (1969).
659: \bibitem{ste72}A. G. St. Pierre and W. A. Steele, J. Chem. Phys. 57, 4638 (1972).
660: \bibitem{bra78}J.-Cl. Leicknam, Y. Guissani and S. Bratos, J. Chem. Phys. 68, 3380
661: (1978).
662: \bibitem{McCl77}R. E. D. McClung, Adv. Mol. Rel. Int. Proc. 10, 83 (1977).
663: \bibitem{bul84}T. E. Bull and W. Egan, J. Chem. Phys. 81, 3181 (1984).
664: \bibitem{gel98a}A. P. Blokhin and M. F. Gelin, Khim. Fiz. 17, No 12, 108 (1998).
665: \bibitem{con76}M. Constan, R. Fauquembergue and P. Descheerde, J. Chem. Phys 64,
666: 667 (1976).
667: \bibitem{hub72}P. S. Hubbard, Phys. Rev. A. 6, 2421 (1972).
668: \bibitem{mor82}A. Morita, J. Chem. Phys. 76, 3198 (1982).
669: \bibitem{McCl87}D. H. Lee and R. E. D. McClung, Chem. Phys. 112, 23 (1987).
670: \bibitem{gel96a}A. P. Blokhin and M. F. Gelin, Physica A 229, 501 (1996).
671: \bibitem{kal81}V. I. Gaiduk and Y. P. Kalmykov, J. Chem. Soc., Faraday Trans. 2 77,
672: 929 (1981).
673: \bibitem{ste81}W. A. Steele, Mol. Phys. 43, 141 (1981).
674: \bibitem{gel97}A. P. Blokhin and M. F. Gelin, Khim. Fiz. 16, No 1, 39 (1997); ibid
675: 16, No 1, 50 (1997).
676: \bibitem{gel97a}A. P. Blokhin and M. F. Gelin, J. Phys. Chem. B 101, 236 (1997).
677: \bibitem{gel00}M. F. Gelin, J. Phys. Chem. A 104, 6150 (2000).
678: \bibitem{gel96}A. P. Blokhin and M. F. Gelin, Molec. Phys. 87, 455 (1996).
679: \bibitem{gel01}A. P. Blokhin and M. F. Gelin, J. Molec. Liq. 93, 47 (2001).
680: \bibitem{bur83}Y. A. Serebrennikov, S. I. Temkin, A. I. Burshtein, Chem. Phys. 81,
681: 31 (1983).
682: \bibitem{fre95}A. Polimeno, G. J. Moro and J. H. Freed, J. Chem. Phys. 102, 8094
683: (1995); ibid 104, 1090 (1996).
684: \bibitem{pol04}G. J. Moro and A. Polimeno, J. Phys. Chem. B 108, 9359 (2004).
685: \bibitem{pol05}A. Magro, D. Frezzato, A. Polimeno, G. J. Moro, R. Chelli and R. Righini,
686: J. Chem. Phys. 123, 124511 (2005).
687: \bibitem{str00}J. Jang and R. M. Stratt, J. Chem. Phys. 112, 7524 (2000); ibid 112,
688: 7538 (2000).
689: \bibitem{hyn06} D. Laage and J. T. Hynes, Science 311, 832 (2006).
690: \bibitem{eva78}G. T. Evans, Mol. Phys. 36, 65 (1978).
691: \bibitem{fre75}L.-P. Hwang and J. H. Freed, J. Chem. Phys. 63, 118 (1975).
692: \bibitem{foot3}For a more general (but complicated) description see \cite{eva78,BurTe,gel98}.
693: \bibitem{var89}D. A. Varshalovich, A. N. Moskalev and V. K. Hersonski. Quantum Theory
694: of Angular Momentum (World Scientific, Singapore, 1989).
695: \bibitem{KS}J. Keilson and J. E. Storer, Quart. Appl. Math., 10, 243 ( 1952).
696: \bibitem{eva95}S. Tang and G. T. Evans, J. Chem. Phys. 103, 1553 (1995).
697: \bibitem{fre55}J. Frenkel, Kinetic theory of fluids (Dover, New York, 1955).
698: \bibitem{lit72}F. J. Bartoli and T. A. Litovitz, J. Chem. Phys. 56, 413 (1972).
699: \bibitem{kus77}J. N. Kushick, J. Chem. Phys. 67, 2068 (1977).
700: \bibitem{kam81}E. Praestgaard and N.G. van Kampen, Mol. Phys. 43, 33 (1981).
701: \bibitem{deb88}S. K. Deb, Chem. Phys. 120, 225 (1988).
702: \bibitem{bur94}Yu. Georgievskii and A. I. Burshtein, J. Chem. Phys. 101, 10858 (1994).
703: \bibitem{ste69}A. G. St. Pierre and W. A. Steele, Phys. Rev. 184, 172 (1969).
704: \bibitem{blo86}A. P. Blokhin, Vesti AN BSSR, Ser. fiz.-mat. 2, 70 (1986); ibid 4,
705: 77 (1986).
706: \bibitem{lei86}M. Aguado-Gomez and J.-Cl. Leicknam, Phys. Rev. A 34, 4195 (1986);
707: ibid A 35, 286 (1987).
708: \bibitem{fel97}B. Cichocki and B. U. Felderhof, J. Chem. Phys. 107, 291 (1997).
709: \bibitem{mas97}A. J. Masters, J. Chem. Phys. 107, 292 (1997).
710: \bibitem{foot2}Strictly speaking, these hydrodynamic long-time tails are associated
711: with local structures and confining potentials in liquids, but these
712: effects are evidently not taken into account within the NKSM description.
713: \bibitem{bha85}S. K. Deb and K. V. Bhagwat, Chem. Phys. 98, 251 (1985).
714: \bibitem{rot75}W. G. Rothschild, G. J. Rosasco, R. C. Livingston, J. Chem. Phys.
715: 62, 1253 (1975).
716: \bibitem{jon81}S. Perry, T. W. Zerda, J. Jonas, J. Chem. Phys. 75, 4214 (1981).
717: \bibitem{dre79}C. Dreyfus, C. Breuillard, T. Nguyen-Tan, R. Ouillon, Chem. Phys.
718: Lett. 62, 246 (1979).
719: \bibitem{kus94}I. M. Svischev and P. G. Kusalik, J. Phys. Chem. 98, 728 (1994).
720: \bibitem{ski03}C. P. Lawrence and J. L. Skinner, J. Chem. Phys. 118, 264 (2003).
721: \bibitem{eav04} J. D. Eaves, PhD thesis, Massachusetts Institute of Technology, 2004.
722: \bibitem{bag94}S. Ravichandran and B. Bagchi, J. Phys. Chem. 98, 2729 (1994).
723: \bibitem{ber02}H. Hu, H. A. Stern and B. J Berne, J. Phys. Chem. B 106, 2054 (2002).
724: \bibitem{ich99}A. Chandra and T. Ichiye, J. Chem. Phys. 111, 2701 (1999).
725: \bibitem{luz00}A. Luzar, J. Chem. Phys. 113, 10633 (2000).
726: \bibitem{ris}H. Risken. The Fokker-Planck Equation (Springer, Berlin, 1984).
727: \bibitem{ham05}J. Helbing, K. Nienhaus, G. U. Nienhaus, P. Hamm, J. Chem. Phys. 122,
728: 124505 (2005).
729: \bibitem{sza84}A. Szabo, J. Chem. Phys. 81, 150 (1984).
730: \end{thebibliography}
731:
732: \clearpage
733:
734: \begin{figure}
735: \includegraphics[keepaspectratio,totalheight=10cm,angle=270]{Figure1}
736: \caption{Angular momentum (a) and energy (b) CFs for $\nu=16$, $\gamma=-0.99915$
737: (that is $\nu_{J}=32$ and $\nu_{E}=0.027$), $\sigma=0.2$ and $\lambda_{1}=0.7$.
738: The dashed, full and dotted lines correspond to $\lambda_{2}=1000$,
739: $10$ and $3$, respectively. On the scale of the figure, the rotational
740: energy CFs for $\lambda_{2}=$ $10$ and $3$ are indistinguishable.
741: }
742: \end{figure}
743:
744: \clearpage
745:
746: \begin{figure}
747: \includegraphics[keepaspectratio,totalheight=12cm,angle=270]{Figure2}
748: \caption{ The second rank OCFs in the Markovian limit for the relaxation
749: rate rate $\nu_{J}=32$. From bottom to top, the curves correspond
750: to $\gamma=1$, $-0.99$, $-0.999$, $-0.9999$ and $1$.
751: }
752: \end{figure}
753:
754: \clearpage
755:
756: \begin{figure}
757: \includegraphics[keepaspectratio,totalheight=12cm,angle=270]{Figure3}
758: \caption{
759: The influence of the memory effects on the second rank OCFs.
760: $\nu_{J}=32$, $\gamma=$ $-0.999$, $\sigma=0.2$ and $\lambda_{1}=0.7$;
761: $\lambda_{2}=1000$ (dashed lines), $10$ (full lines) $3$ (dotted
762: lines). The dash-dotted curve depicts the OCF in the Markovian limit.}
763: \end{figure}
764:
765: \clearpage
766: \begin{figure}
767: \includegraphics[keepaspectratio,totalheight=12cm,angle=270]{Figure4}
768: \caption{
769: Comparison of the simulated OCFs with those calculated within
770: the NKSM. The dotted lines correspond to the first (upper
771: curve) and second (lower curve) rank OCFs which were simulated for
772: $HOD$ in liquid $D_{2}O$ at a room temperature \cite{tok05}. The black
773: curves are computed for $\nu_{J}=46$, $\gamma=$ $-0.999415$, $\sigma=0.2$,
774: $\lambda_{1}=0.4$, $\lambda_{2}=10$ ($j=1$); $\nu_{J}=32$, $\gamma=$
775: $-0.99915$, $\sigma=0.2$, $\lambda_{1}=0.7$, $\lambda_{2}=10$
776: ($j=2$). The OCFs calculated via the diffusion equation (\ref{dif})
777: are depicted by dashed lines. }
778: \end{figure}
779:
780: \clearpage
781: \begin{figure}
782: \includegraphics[keepaspectratio,totalheight=12cm,angle=270]{Figure5}
783: \caption{ Comparison of the experimental OCFs with those calculated
784: within the NKSM. The dotted line reproduces the best fit
785: to the experimental anisotropy decay \cite{tok05}. The solid black
786: curve shows the second-rank OCF computed within the NKSM for $\nu_{J}=63$,
787: $\gamma=$ $-0.99995$, $\sigma=0.2$, $\lambda_{1}=0.3$, $\lambda_{2}=10$.
788: The dashed line reproduces the second rank NKSM OCFs from Fig. 4. }
789: \end{figure}
790:
791: \clearpage
792: \begin{figure}
793: \includegraphics[keepaspectratio,totalheight=10cm,angle=270]{Figure6}
794: \caption{
795: Elucidation of the algebraic contributions into OCFs. For
796: each couple of the curves, the upper one is calculated via Eq. (\ref{Gt2})
797: and the lower one is computed within the Markovian limit of the NKSM.
798: The dotted ($j=2$, $\nu_{J}=32$, $\gamma=$ $-0.99915$) and dashed
799: ($j=1$, $\nu_{J}=46$, $\gamma=$ $-0.999415$) curves correspond
800: to the best-fit NKSM OCFs from Fig. 4, and the solid curves ($j=2$,
801: $\nu_{J}=63$, $\gamma=$ $-0.99995$) correspond to the best-fit
802: NKSM OCFs from Fig. 5.}
803: \end{figure}
804:
805:
806:
807: \end{document}
808: