cond-mat0001030/rh.tex
1: \tolerance=10000
2: \documentstyle[preprint,prl,aps,epsf]{revtex}
3: \baselineskip 0.9truecm
4: \large
5: \begin{document}
6: \draft
7: \title
8:       {
9:        Relaxation processes in harmonic glasses?
10:       }
11: \author{
12:         G.~Ruocco$^{1}$,
13: 	F.~Sette$^{2}$,
14:         R.~Di Leonardo$^{1}$,
15: 	G.~Monaco$^{2}$,\\
16:         M.~Sampoli$^{3}$,
17:         T.~Scopigno$^{4}$ and
18:         G.~Viliani$^{4}$
19:        }
20: \address{
21:          $^1$
22:          INFM and Dipartimento di Fisica, Universit\'a di L'Aquila,
23: 	 I-67100, L'Aquila, Italy. \\
24:          $^2$
25:          European Synchrotron Radiation Facility, 
26: 	 B.P. 220, F-38043 Grenoble Cedex, France. \\
27:          $^3$
28:          INFM and Dipartimento di Energetica, Universit\'a di Firenze,
29: 	 I-50139, Firenze, Italy. \\
30:          $^3$
31:          INFM and Dipartimento di Fisica, Universit\'a di Trento,
32: 	 I-3805, Povo, Trento, Italy. 
33:         }
34: 
35: \date{\today}
36: \maketitle
37: \begin{abstract}
38: A relaxation process, with the associated phenomenology 
39: of sound attenuation and sound velocity dispersion, is found 
40: in a simulated {\it harmonic} Lennard-Jones glass. We propose to 
41: identify this process with the so called {\it microscopic} 
42: (or {\it instantaneous}) relaxation process observed in real 
43: glasses and supercooled liquids. A model based on the memory 
44: function approach accounts for the observation, and allows to 
45: relate to each others: 1) the characteristic time and strength 
46: of this process, 2) the low frequency limit of the dynamic structure 
47: factor of the glass, and 3) the high frequency sound attenuation 
48: coefficient, with its observed quadratic dependence on the momentum 
49: transfer.
50: \end{abstract}
51: \vskip 1cm
52: \pacs{PACS numbers:63.50.+x, 61.43.-j, 61.43.Fs}
53: 
54: 
55: In a recent paper W.~G\"otze and M.R.~Mayr \cite{Goetze} adapted 
56: the Mode Coupling Theory \cite{MCT} to glassy phase. This theory, as 
57: shown numerically for a hard-spheres model, accounts for many of the 
58: features found in the dynamic structure factor, $S(Q,\omega)$, 
59: of glasses. Among these: {\it i)} The existence of propagating 
60: excitations (sound waves), with an almost linear momentum transfer 
61: ($Q$) dependence of their excitation energy $\Omega(Q)$, up to $Q$ 
62: values that are a significant fraction of $Q_o$, the maximum of the 
63: static structure factor $S(Q)$. {\it ii)} The quadratic dependence 
64: on $Q$ of the excitations broadening  $\Gamma(Q)$, and therefore of 
65: sound attenuation. {\it iii)} The temperature insensitivity of 
66: $\Gamma(Q)$. {\it iv)} The development in the $S(Q,\omega)$ at 
67: large $Q$ ($Q/Q_o$$\approx$0.3) of a secondary excitation band, 
68: at frequencies below the Brillouin peak. The theory also predicts 
69: two features not yet experimentally detected: {\it a)} A positive 
70: dispersion of the sound velocity and {\it b)} an intensity "gap" 
71: in the low frequency region of the $S(Q,\omega)$. In spite of 
72: the success of this approach, it is still of great interest to 
73: investigate the physical origin of these phenomena, and, in 
74: particular, whether they are related to the topological disorder 
75: and/or to the anharmonicity of the interatomic potential.                 % 
76: 
77: In this Letter we report a Molecular Dynamics (MD) simulation 
78: study of the $Q$ dependence of the sound velocity in a model 
79: monatomic Lennard-Jones glass in the {\it harmonic} approximation. 
80: We show that even in a {\it harmonic} glass, by increasing $Q$
81: there is a positive dispersion of the sound velocity, thus proving
82: one of the prediction of the MCT theory \cite{Goetze}, and 
83: relating this phenomenon to the topological disorder. Since 
84: this dispersion is similar to that found in presence of a relaxation 
85: process, we attempt to apply a generalized Langevin equation with an 
86: effective memory function approach to describe the density fluctuations
87: dynamics. This formalism allows to account for the ubiquitous 
88: $Q^2$-dependence of the high frequency sound absorption
89: observed in many glasses by experimenys \cite{Q2g} and by MD 
90: simulations \cite{Q2MD,dellanna2}.
91: These results suggest to identify this process with the one
92: referred to as {\it microscopic} or {\it instantaneous} 
93: relaxation in {\it real} systems.
94: 
95: The investigated systems consist of $N$=2048, 10976 and 32000 
96: argon atoms interacting via a (6-12) Lennard-Jones potential 
97: ($\epsilon$=125.2 K, $\sigma$=3.405 A). A standard microcanonical 
98: MD simulation, performed at decreasing temperatures in the normal 
99: liquid phase, is followed by a fast quench ($\dot T$$\approx$ 
100: 10$^{12}$ K/s) of the slightly supercooled liquid down to 
101: $\approx$5 K \cite{Mazz}. Starting from the glass configuration 
102: at $T$=5 K, the atomic trajectories $\bar r_i(t)$ 
103: (here $i$=$1...N$ is the particle label) are followed and 
104: stored for subsequent analysis, and the "inherent" 
105: configuration  $\{\bar x_i\}_{i=1..N}$ \cite{StWe} at $T$=0 K 
106: is calculated by the steepest descent method. Two different 
107: procedures have been used to derive the $S(Q,\omega)$:
108: (i) The trajectories calculated from MD in the glass 
109: are used to compute the corresponding time correlation 
110: function and (ii) The Normal Mode Analysis (NMA) is applied 
111: to the inherent configuration. This last procedure works in the 
112: harmonic approximation, which is obtained by retaining only the 
113: quadratic term of the interaction potential. From the 
114: calculation of the dynamical matrix ${\bf D}$, one computes 
115: the eigenvalues, $\omega_p$ ($p$=$1..3N$ is the mode index) 
116: and the eigenvectors ($\bar e_p(i)$). For large size samples 
117: (where a direct diagonalization of ${\bf D}$ is not feasible) 
118: the $S(Q,\omega)$ is obtained by the method of moments \cite{mom}.
119: 
120: In the MD runs, the dynamics structure factor is
121: calculated from the numerical time-Fourier transform
122: of the intermediate scattering function $F(Q,t)$,
123: defined as:
124: \begin{equation}
125: \label{fqt}
126: F(Q,t)= 1/N \langle 
127: \Sigma_{ij} \exp{(i \bar Q \cdot \bar r_i(t) )}
128: \exp{(-i \bar Q \cdot \bar r_j(0))}
129: \rangle .
130: \end{equation}
131: In the harmonic framework and in the classical limit, 
132: the one excitation approximation of the dynamic structure 
133: factor $S^{^{(1)}}\!(Q,\omega)$ is obtained
134: by the normal mode expansion of the atomic displacements:
135: \begin{equation}
136: \label{udt}
137: \bar r_i(t)=\bar x_i+
138: \sqrt{K_BT/M} \Sigma_p \bar e_p(i) A_p(t)/\sqrt{\omega_p}.
139: \end{equation}
140: $A_p(t)$ is the amplitude of the $p$-th normal mode 
141: and is characterized by $\langle|A_p(t)|^2\rangle=1$. 
142: This gives:
143: \begin{equation}
144: \label{sqw}
145: S^{^{(1)}}\!(Q,\omega)=
146: (K_BTQ^2/M\omega^2) \; \Sigma_p E_p(Q) \delta(\omega-\omega_p),
147: \end{equation}
148: where we have introduced the spatial power spectrum of the 
149: longitudinal component of the eigenvectors, $E_p(Q)$:
150: \begin{equation}
151: \label{eqw}
152: E_p(Q)=|\Sigma_i (\hat Q \cdot \bar e_p(i))
153: \exp{{(i \bar Q \cdot \bar x_i )} |^2}.
154: \end{equation}
155: Here, $\hat Q$=$\bar Q/|Q|$ and the Debye-Waller factor
156: has been neglected. Differently from the second 
157: ($\int \omega^2 S(Q,\omega) d\omega$=
158: $\int \omega^2 S^{^{(1)}}\!(Q,\omega) d\omega$=%
159: $K_BTQ^2/M$) and higher moments sum rules,
160: the zeroth moment sum rule for $S(Q,\omega)$, 
161: $\int S(Q,\omega) d\omega$=$S(Q)$, does not hold for
162: $S^{^{(1)}}\!(Q,\omega)$, as in this function the 
163: elastic intensity is missing. Rather, it is useful 
164: to define the "inelastic" contribution to $S(Q)$:
165: $S_{_{i}}(Q)$=$\int S^{^{(1)}}\!(Q,\omega) d\omega$.
166: 
167: Selected examples of the $S(Q,\omega)$ calculated for
168: different size systems, and with different methods are
169: reported in Fig.~1. As there is a trivial dependence
170: of the $S^{^{(1)}}\!(Q,\omega)$ on $T$, here we report
171: this quantity multiplied by the factor $(M/K_BT)$. 
172: Figure~1a shows the $S(Q,\omega)$ 
173: in the low $Q$ range, as calculated from the MD
174: runs for the $N$=32000 particles system, while Fig.~1b
175: shows the intermediate $Q$ range, in the case of the
176: harmonic approximation for the $N$=2048 system.
177: As a check of consistency, the 
178: inset of Fig.~1b  shows the $Q$ dependence of the 
179: second moment of $(M/K_BT)S(Q,\omega)$ (+) which, 
180: according to the sum rule, should be $Q^2$ (full line). 
181: The inset of Fig.~1a shows, at two $Q$ values, the
182: comparison of the $S(Q,\omega)$ ($N$=2048) calculated
183: either from the MD runs (full line) or in the harmonic
184: approximation ($\circ$). The $S(Q,\omega)$, as derived 
185: with the two different methods, are equivalent in the 
186: whole considered $Q$ range. This indicates that the 
187: Newtonian dynamics at $T$=5 K is truly harmonic, and 
188: that the results obtained by MD at this $T$ and NMA 
189: can be interchanged among each other. 
190: 
191: The general features of the $S(Q,\omega)$ reported in 
192: Fig.~1 are: {\it i)} A Brillouin peak, dispersing and 
193: becoming broader with increasing $Q$, dominates the
194: spectrum up to $Q$$\approx$10 nm$^{-1}$ 
195: ({$Q/Q_o$$\approx$0.45). {\it ii)} At larger $Q$ 
196: values a second peak, observed at frequencies below
197: the Brillouin peak, starts to dominate the data. 
198: This secondary peak has been already detected 
199: in many systems: 1) in liquid water, both experimentally 
200: \cite{wat97} and by MD \cite{watmix}, 
201: where it has been interpreted as a signature of the 
202: transverse dynamics; 2) in vitreous silica, by MD,
203: where it has been interpreted either 
204: in terms of a transverse dynamics \cite{dellanna2} or
205: as an evidence of the Boson peak \cite{kob}; and 3) in
206: a hard sphere glass, where it has been theoretically 
207: predicted \cite{Goetze}. In this respect, it is
208: worth to underline the striking similarity 
209: between the $S(Q,\omega)$ reported here and those
210: of Fig.~5 of Ref.~\cite{Goetze}. Finally,
211: {\it iii)}, the $S(Q,\omega)$ up to 
212: $Q$$\approx$5 nm$^{-1}$ shows a nearly constant 
213: and $Q$-independent intensity below the Brillouin 
214: peak frequency. This plateau, whose value is 
215: $\lim_{\omega\rightarrow 0} (M/K_BT)S(Q,\omega)$=
216: $A_0$$\approx$0.03~10$^{-18}$ s$^3$/m$^2$, 
217: can be identified with the plateau observed in the 
218: constant $\omega$ cuts of the $S(Q,\omega)$
219: at $Q$ larger than the Brillouin peak in simulated
220: \cite{and_v}, calculated (Fig~18 in Ref. \cite{Goetze})
221: and measured \cite{nh3} systems.
222: 
223: Figure~2 shows the position of the maxima 
224: ($\Omega_{_C}(Q)$) of the current spectra $C_L(Q,\omega)$ 
225: (=$\omega^2/Q^2S(Q,\omega)$) as a function of $Q$ in the 
226: investigated $Q$ range. 
227: The inset of Fig.~2 shows the peak position $\Omega(Q)$ and 
228: the broadening $\Gamma(Q)$ of the $S(Q,\omega)$ in the low $Q$ 
229: region, as measured directly from the spectra of Fig.~1a. The $Q$ 
230: dependence of these parameters agrees with the behavior 
231: observed in all the glasses investigated so far \cite{Q2g}:
232: a linear (square) dependece of $\Omega(Q)$ ($\Gamma(Q)$).
233: The behavior of $\Omega_{_C}(Q)$ closely resembles that of 
234: the acoustic phonon branches in crystals, i.~e.
235: the almost linear behavior in the small $Q$ region, a 
236: maximum around $Q_o/2$, and a minimum around $Q_o$. Most
237: importantly one can clearly detect a positive dispersion 
238: of the sound velocity, as highlighted by the dashed line 
239: in Fig.~2. This dispersion is better seen 
240: in Fig.~3, where the apparent sound velocity, 
241: $v(Q)$=$\Omega_{_C}(Q)/Q$, is reported. The quantity $v(Q)$
242: (full dots) undergoes a transition from the "low frequency" 
243: \cite{nota} sound velocity $v_o(Q)$ towards the infinite frequency 
244: sound velocity $v_{\infty}(Q)$.
245: Here $v_o(Q)$ is calculated as 
246: $v_o(Q)$=$\sqrt{K_BT/MS_i(Q)}$ (dashed line),
247: and $v_\infty(Q)$ is calculated either as the
248: fourth moment of the calculated $S(Q,\omega)$ (open points)
249: or from the expression of the fourth moment in terms of both
250: the pair correlation function and the interaction potential 
251: (full line) \cite{baluc}.
252: The velocity dispersion is observed up to $Q$$\approx$5 nm$^{-1}$, 
253: i.~e. $\Omega_{_C}$$\approx$35 cm$^{-1}$, and an opposite
254: dispersion is observed in the region approaching $Q_o$.
255: These observations recall of a typicall relaxation scenario.
256: When a relaxation process with characteristic time $\tau$ is 
257: active in the system, the transition from $v_o$ to $v_{\infty}$ 
258: takes place when the condition $\omega\tau$=1 is fulfilled. In 
259: the present case, considering that the first transition is at at 
260: $Q$$\approx$2 nm$^{-1}$  where $\Omega_{_C}$$\approx$15 cm$^{-1}$, 
261: the value of $\tau$ results to be around 0.3 ps. The second 
262: and third transitions between $v_o$ and $v_\infty$ are observed
263: just below and above $Q_o$ as a consequence of the slowing down 
264: of the dynamics (deGennes narrowing) around $Q_o$.
265: Consequently the whole behavior of $v(Q)$, as reported in Fig.~3, 
266: can be qualitatively understood in terms of a relaxation process 
267: with a characteristic time of $\tau$$\approx$0.3 ps. 
268: 
269: The identification of a such a {\it relaxation process} in an 
270: {\it harmonic system}, suggests to use the formalism that 
271: describes the density correlators $\phi(Q,t)$=$F(Q,t)/S(Q)$ 
272: through its generalized Langevin equation \cite{baluc}:
273: \begin{equation}
274: \label{lang}
275: \ddot \phi(Q,t)+\omega^2_o \phi(Q,t)+
276: \mbox{$\int$}_o^t m(Q,t-t') \dot \phi(Q,t') dt =0
277: \end{equation}
278: where $\omega^2_o$=$K_BTQ^2/MS(Q)$ and $m(Q,t)$ is the 
279: "memory function". This 
280: equation has been rigorously derived for ergodic 
281: systems, but, as recently shown in the framework of the MCT
282: \cite{Goetze}, 
283: it can be still applied in the non-ergodic glassy phase 
284: making the the following substitutions:
285: $\phi(Q,t)\rightarrow \phi'(Q,t)$=$(\phi(Q,t)-f_{_Q})/(1-f_{_Q})$, 
286: $\omega_o^2(Q)\rightarrow \omega^2_{o \mu}$
287: $\equiv$$\omega^2_{\infty \alpha}$=$K_BTQ^2/MS_i(Q)$
288: and $m(Q,t)\rightarrow m_\mu(Q,t)$=$m(Q,t)\!-\!\omega^2_o 
289: f_{_Q}/(1\!-\!f_{_Q})$. Therefore, the "vibrational" dynamics
290: of interest is now described by the correlators $\phi'(Q,t)$, 
291: which is obtained subtracting from $\phi(Q,t)$ the long time 
292: plateau level, whose value is the non-ergodicity parameter 
293: $f_{_Q}$. It is worth to note the subtraction of the constant 
294: term $S(Q)f_{_Q}$ from F(Q,t) is equivalent to neglect the 
295: elastic contribution in the $S(Q,\omega)$, and that 
296: $F'(Q,t)$=$S_i(Q)\phi'(Q,t)$ is the Fourier transform 
297: of the $S^{^{(1)}}\!(Q,\omega)$ as defined in Eq.~(\ref{sqw}).
298: The whole dynamic behavior of $\phi'(Q,t)$ is now contained in the
299: {\it microscopic} contribution to the memory function $m_\mu(Q,t)$.
300: In the case of a harmonic system, the function $m_\mu(Q,t)$ can be 
301: explicitly calculated from the eigenvalues and eigenvectors of the 
302: system. Indeed, the Laplace transform (indicated by hats) of 
303: Eq.~(\ref{lang}) (after the previously indicated substitutions) 
304: and a straightforward algebra gives:
305: \begin{equation}
306: \label{langs}
307: \hat m_\mu(Q,\!s)\!=\!\left[ 
308: \hat \phi'(Q,s)[s^2+\omega^2_{o \mu}]\!-\!s \right]
309: \! \left[ 1\!-\!s\hat \phi'(Q,s)\right ]^{-1}.
310: \end{equation}
311: Then, from an inverse Fourier and a subsequent Laplace transform of 
312: Eq.~(\ref{sqw}), it is easy to get an explicit expression for 
313: $\hat \phi'(Q,s)$ to be inserted in Eq.~(\ref{langs}). This gives:
314: \begin{equation}
315: \label{ms}
316: \hat m_\mu(Q,\!s)\!=\!\!\left[ {\Sigma} \frac{E_p(Q)}
317: {\omega_p^2} \frac{s}{s^2\!\!+\!\omega^2_p}\right ] \!
318: \left[ {\Sigma} \frac{E_p(Q)}{\omega_p^2} {\Sigma}
319: \frac{E_p(Q)}{s^2\!\!+\!\omega^2_p}\right ]^{-1}
320: \!\!\!\!\!\!\!-\!s
321: \end{equation}
322: This equation provides an explicit expression of the memory function 
323: in terms of the system eigenstates.
324: %Besides a detailed analysis of the time-dependence of $m_\mu(Q,\!t)$,
325: %which is beyond the aim of this work, one can access its main
326: %features from its $s$$\rightarrow$0 and $s$$\rightarrow$$\infty$ limits.
327: Considering that $m_\mu(Q,\! t)$ is mainly characterized by parameters 
328: as its initial value $\Delta_{_Q}^2$ and total area $\Gamma_{_Q}$ and 
329: therefore by 
330: a decaying time-scale $\tau_{_Q}$$\approx$$\Gamma_{_Q}/\Delta_{_Q}^2$
331: \cite{notatau}, one can show that
332: $\hat m(Q,s$$\rightarrow$$\infty)$=$\Delta_{_Q}^2/s$ and
333: $\hat m(Q,s$$\rightarrow$0)=$\Gamma_{_Q}$  \cite{baluc}. 
334: Inserting these limiting values into Eq.~(\ref{ms}) one obtains:
335: \begin{eqnarray}
336: \label{limit1}
337: \Delta_{_Q}^2&=&\left [ \Sigma_p E_p(Q)\omega^2_p \right ]-
338: \left [ \Sigma_p E_p(Q)\omega^{-2}_p \right ]^{-1}\\
339: \label{limit2}
340: \Gamma_{_Q}&=&\left[\Sigma_p \frac{E_p(Q)}
341: {\omega^2_p} \right]^{-2} \!\!\!
342: \lim_{s\rightarrow 0}  
343: \left[ \Sigma_p \frac{E_p(Q)}{\omega_p^2}
344: \frac{s}{s^2\!+\!\omega^2_p}\right ]
345: \end{eqnarray}
346: It is now easy to identify, through the explicit expression of the 
347: zeroth and fourth moments of Eq.~(\ref{sqw}) that 
348: $\Sigma_p E_p(Q)\omega^2_p$=$v^2_\infty Q^2$ and
349: $(\Sigma_p E_p(Q)\omega^{-2}_p)^{-1}$=$v^2_o Q^2$, confirming
350: that $\Delta_{_Q}^2$=$(v_\infty^2\!-\!v_o^2)Q^2$. 
351: The determination of $\Gamma_{_Q}$ which, being 
352: the area of the memory function, coincides with the Brillouin 
353: broadening in the $\omega\tau$$<<$1 limit is slightly more
354: involved. Using the representation 
355: $\delta(x)=1/\pi \; \lim_{s\rightarrow 0} s/(s^2+x^2)$,
356: in Eq.~(\ref{sqw}) one sees that:
357: \begin{equation}
358: \label{sss}
359: S^{^{(1)}}\!(Q,\omega)\!=\!\frac{K_BTQ^2}{\pi M} 
360: \lim_{s\rightarrow 0} \Sigma_p \frac{E_p(Q)}
361: {\omega_p^2} \frac{s}{s^2\!+\!(\omega\!-\!\omega_p)^2}, 
362: \end{equation}
363: and comparing of Eqs.~(\ref{sss}) and (\ref{limit2})
364: one gets \cite{nota2}:
365: \begin{equation}
366: \label{gamma}
367: \Gamma_{_Q}=v_o^4 Q^2 \frac{\pi M}{K_BT} 
368: \lim_{\omega\rightarrow 0} S(Q,\omega).
369: \end{equation}
370: Similarly:
371: \begin{equation}
372: \label{tau}
373: \tau_{_Q} \approx \frac{v_o^4}{(v_\infty^2-v_o^2)}  
374: \frac{\pi M}{K_BT} \lim_{\omega\rightarrow 0} S(Q,\omega).
375: \end{equation}
376: Considering that $\lim_{\omega\rightarrow 0} S(Q,\omega)$ 
377: is $Q$-independent (see Fig.~1), these expressions give
378: account of the observation in the low $Q$ region -where
379: the $Q$ dependence of $v_o$ and $v_\infty$ is negligible-
380: that the Brillouin peak broadening is proportional to $Q^2$
381: and the $\tau$ is $Q$-independent. As a check of internal
382: consistency, using the value of $A_o$ previously reported,
383: one finds $\tau$$\approx$0.6 ps and $\Gamma_{_Q}$[cm$^{-1}$]=
384: 1.35$Q^2$[nm$^{-1}$]. The value of $\tau_{_Q}$ 
385: overestimates the one deduced from
386: Fig.~3, and this probably due to the rough expression
387: of $\tau$ as $\Gamma_{_Q}/\Delta_{_Q}^2$. On the contrary,
388: as shown in the inset of Fig.~2,
389: $\Gamma_{_Q}$ from Eq.~(\ref{gamma}) is in excellent
390: agreement with the one directly derived from the width
391: of the $S(Q,\omega)$.
392: 
393: It is worth to discuss the microscopic origin 
394: of the observed relaxation process. A relaxation process 
395: can be pictured as the macroscopic manifestation of 
396: microscopic phenomena associated with the existence of 
397: channels by which the energy stored in a given "mode" relaxes 
398: towards other degrees of freedom.
399: The $S(Q^*,\omega)$, through the fluctuation-dissipation 
400: relation, reflects the time evolution of the energy initially 
401: stored ($t$=$t_o$) in a Plane Wave (PW) of wavelength $2\pi/Q^*$. 
402: As the PW is not an eigenstate of the disordered system, 
403: at $t\!>\!t_o$ there will be a transfer of amplitude from this 
404: PW towards other PWs of different $Q$ values. 
405: This process is controlled by the difference bewteen
406: the considered PW and the normal modes of the topologically 
407: disordered glassy structure. This energy flow takes
408: place on the time scale $\tau$ as derived from Eq.~(\ref{tau}),
409: and gives rise to the observed relaxation process phenomenology.
410: Consequently, one can speculate that this process is the
411: {\it instantaneous} or {\it microscopic} process 
412: empirically introduced to explain the
413: $S(Q,\omega)$ measured in real glasses and liquids
414: by Brillouin light and x-ray Scattering \cite{cumm,IXS}. 
415: 
416: We thanks W.~G\"otze for helpful discussions and 
417: critical readings of the manuscript.
418: 
419: \begin{references}
420: 
421: \bibitem{Goetze}
422: W.~G\"otze and M.~R.~Mayr, 
423: Phys. Rev. {\bf E61}, 0000 (2000);
424: {\it ibidem} cond-mat/9910054.
425: 
426: \bibitem{MCT}
427: W.~G\"otze,
428: J. of Phys.: Condens. Matt. {\bf 11}, A1 (1999)
429: 
430: \bibitem{Q2g}
431: F.~Sette, M.~Krisch, C.~Masciovecchio, G.~Ruocco 
432: and G.~Monaco,  Science {\bf 280}, 1550 (1998),
433: and references therein.
434: 
435: \bibitem{Q2MD}
436: S.N.~Taraskin and S.R.~Elliot, 
437: Phys. Rev. {\bf B56}, 8605 (1997);
438: {\it ibidem} Europhys. Lett. {\bf B39}, 37 (1997);
439: M.C.C.~Ribeiro, M.~Wilson, and P.A.~Madden
440: J. Chem. Phys. {\bf 108}, 9027 (1998);
441: J.L.~Feldman, P.B.~Allen, and S.R.~Bickham,
442: Phys. Rev. {\bf B59}, 3551 (1999).
443: 
444: \bibitem{dellanna2}
445: R.~Dell'Anna, G.~Ruocco, M.~Sampoli, G.~Viliani.
446: Phys. Rev. Lett. {\bf 80}, 1236 (1998).
447: 
448: \bibitem{Mazz}
449: V.~Mazzacurati, G.~Ruocco, M.~Sampoli,
450: Europhys. Lett. {\bf 34}, 681 (1996).
451: 
452: \bibitem{StWe}
453: F.H.~Stillinger, and T.A.~Weber, 
454: Phys. Rev. {\bf A28}, 2408 (1983).
455: 
456: \bibitem{mom}
457: C.~Benoit, E.~Royer, and G.~Poussigue G., 
458: J. Phys.: Condens. Matt. {\bf 4}, 3125 (1992);
459: G.~Viliani, R.~Dell'Anna, O.~Pilla, M.~Montagna, G.~Ruocco,
460: G.~Signorelli, and V.~Mazzacurati,
461: Phys. Rev. {\bf B52}, 3346 (1995).
462: 
463: \bibitem{wat97}
464: F.~Sette, G.~Ruocco, M.~Krisch, C.~Masciovecchio,
465: R.~Verbeni, U.~Bergmann. Phys. Rev. Lett. 
466: {\bf 77}, 83 (1996). G.~Ruocco and F.~Sette, 
467: J. Phys:Cond. Matt. {\bf 11}, R259 (1999).
468: 
469: \bibitem{watmix}
470: M.~Sampoli, G.~Ruocco, F.~Sette.
471: Phys. Rev. Lett. {\bf 79}, 1678 (1997).
472: 
473: \bibitem{kob}
474: J.~Horbach, W.~Kob, and K.~Binder,
475: cond-mat/9901162.
476: 
477: \bibitem{and_v}
478: M.~Sampoli, P.~Benassi, R.~Dell'Anna, 
479: V.~Mazzacurati, G.~Ruocco,
480: Phil. Mag. {\bf 77}, 473 (1998).
481: 
482: \bibitem{nh3}
483: F.~Sette, G.~Ruocco, A.~Cunsolo, C.~Masciovecchio, 
484: G.~Monaco, and R.~Verbeni, preprint.
485: 
486: \bibitem{nota}
487: "Low frequency" means here the relaxed side
488: of the {\it microscopic} relaxation process, that in a
489: glass coincides with the fully unrelaxed, high frequency,
490: side of the frozen structural ($\alpha$) process.
491: 
492: \bibitem{notatau}
493: This "relaxation time" $\tau_{_Q}$ can be interpreted as 
494: the time needed by a set of harmonic oscillators of 
495: different frequencies to dephase each others.
496: 
497: \bibitem{baluc}
498: U.~Balucani and M.~Zoppi, 
499: {\it Dynamics of the Liquid State}, 
500: (Clarendon Press, Oxford, 1994). 
501: 
502: \bibitem{nota2}
503: Eq.~(\ref{gamma}) can also be derived from the Fourier
504: transform of Eq.~(\ref{lang}), i.~e. 
505: $S(Q,\omega)$= $S(Q)\tilde \phi(Q,\omega)$= 
506: $S(Q) \omega^2_{o\mu}(Q)/\pi \omega Im 
507: \left[ \omega^2-\omega_{o\mu}^2(Q)-i\omega\tilde 
508: m(Q,\omega) \right]^{-1}$, 
509: performing the $\omega$$\rightarrow$0 
510: limit, and remembering that the area of the 
511: memory function is $\tilde m'(Q,\omega=0)$.
512: 
513: \bibitem{cumm}
514: H.Z.~Cummins
515: J. of Phys.: Condens. Matt. {\bf 11}, A95 (1999)
516: and references therein.
517: 
518: \bibitem{IXS}
519: G.~Monaco et al., Phys. Rev. {\bf E60}, 5505 (1999).
520: A.~Cunsolo et al., preprint; 
521: T.~Scopigno et al., preprint.
522: 
523: \end{references}
524: 
525: \newpage
526: 
527: \begin{figure}[f]
528: \caption[short fig description]{ \footnotesize{
529: The quantity $(M/K_BT)S(Q,\omega)$ is reported in absolute
530: value at the indicated $Q$ values for a) the very low $Q$
531: region ($N$=32000, MD calculation)
532: and b) the intermediate $Q$ region ($N$=2048, NMA calculation).
533: The inset of a) shows the comparison ($N$=2048) of the 
534: MD $S(Q,\omega)$ (full line) and the NMA 
535: $S^{^{(1)}}\!(Q,\omega)$ ($\circ$).
536: The inset of b) shows the $Q$ dependence of $M^{(2)}$, the
537: second moment of $(M/K_BT)S(Q,\omega)$ (+) and its
538: theoretical prediction $M^{(2)}_{th}$=$Q^2$ (full line).
539: }}
540: \label{fig1}
541: \end{figure} 
542: \noindent
543: 
544: \begin{figure}[f]
545: \caption[short fig description]{ \footnotesize{
546: Frequency position of the maxima of the longitudinal current
547: spectra ($\Omega_M(Q)$) for different indicated size samples.
548: The inset shows in log-log scale the maxima ($\Omega(Q)$) 
549: and the broadening ($\Gamma(Q)$) of the Brillouin peaks
550: in the low $Q$ region. The full line is the prediction
551: according to Eq.~(\ref{gamma}) and the dashed line is
552: the best fit to the data.}}
553: \label{fig2}
554: \end{figure} 
555: \noindent
556: 
557: \begin{figure}[f]
558: \caption[short fig description]{ \footnotesize{
559: Sound velocity in glassy LJ Argon: apparent sound velocity ($v$,
560: full symbols), zero frequency sound velocity ($v_o$, dashed line) 
561: and ininite frequency sound velocity ($v_\infty$, full line
562: and open symbols). 
563: }}
564: \label{fig3}
565: \end{figure} 
566: \noindent
567: 
568: 
569: 
570: \end{document}
571: 
572: