1: \section{Vibrational Dynamics [36]} %\protect\cite{ToZF00}}
2: \label{Vibrational_dynamics}
3: When the dynamic anomalies of the glass transition are discussed
4: it is particularly important for molecular glasses, which have a large
5: number of internal degrees of freedom, to study at least
6: qualitatively the vibrational mode structure.
7: This seems to be even more important in view of the fact that low--frequency
8: vibrational features -- for obscure reasons called Boson peak --
9: persist well into the supercooled phase.
10:
11: This can be done by comparison of the absolute frequency scales
12: of the vibrational density of states between the
13: ordered and disordered state.
14: From single crystal dispersion measurements -- if available --
15: information about low--energy internal modes and hybridisation of
16: external and internal modes can be obtained.
17:
18: \subsection{Single Crystal: Phonon Dispersions}
19: The phonon dispersions of single crystal OTP at 200\,K
20: along the three main symmetry directions [100], [010], and [001]
21: are presented in figure~\ref{Dis-relat}.
22: Some optic--like phonons are detected as well,
23: as an example, the lowest optic branch in [010] direction is
24: included in figure~\ref{Dis-relat}.
25: However, for a detailed study of them thermal neutrons instead of
26: cold ones would be needed.
27: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
28: \input{Bilder/Vibration/Dis-relat.tex}
29: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
30: In the small $Q$ limit the sound velocities of acoustic excitations
31: obtained from the initial slope are summarized in table~\ref{soundns}.
32: In all lattice directions,
33: longitudinal sound modes are almost twice as fast as transverse modes.
34: The single crystal data may also be compared with
35: sound velocities in the glass.
36: Results from Brillouin light scattering
37: are summarized in table~\ref{TsoundG}.
38: Taking the simple arithmetic average over the three crystal axes,
39: the mean longitudinal and transverse sound velocities
40: $\langle v_{\rm L,T} \rangle$
41: exceed those of the glass by about 20\,\% and 27\,\%, respectively.
42:
43: % table soundns %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
44: \input{Tables/soundns.tex}
45: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
46: % table ATsoundG %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
47: \input{Tables/TsoundG.tex}
48: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
49:
50: \subsection{Glass versus Polycrystal: Density of Vibrational States}
51: Without crystalline order, it is no longer possible
52: to measure selected phonon modes with well--defined polarization and
53: propagation vector.
54: For polycrystalline or amorphous samples,
55: the distribution of vibrational modes can be conveyed only
56: in form of a spectral density of states (DOS).
57:
58: The DOS of glassy and crystalline OTP at 100\,K are shown in
59: figure~\ref{DOS_100K}.
60: They were calculated directly from the spectra at the highest scattering
61: angles $S(2\theta,\omega)$ without interpolation to constant $Q$.
62: Multiphonon contributions were calculated be repeated convolution of
63: $g(\omega)$ with itself and subtraction from $S(2\theta,\omega)$ in
64: an interative procedure described in detail in \cite{WuKB93}.
65: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
66: \input{Bilder/Vibration/DOS_100K.tex}
67: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
68: Comparing our results with model calculations \cite{Bus82},
69: we assign all vibrations below $\nu_{\rm g}$ to
70: the 16 degrees of freedom needed to describe the crystal structure.
71: For 16 low--lying modes in a molecule with 32 atoms,
72: we expect an integrated DOS
73: \begin{equation}
74: \int_0^{\nu_{\rm g}}\!{\rm d}\nu\, g(\nu)={16\over32}=0.5\,.
75: \end{equation}
76:
77: Rather broad distributions are found for both phases showing a pronounced
78: gap above $\nu_{\rm g}\simeq5$\,THz
79: (1\,THz = 33.36\,cm$^{-1}$ = 4.1357\,meV).
80: In the glass a first shoulder around 1.5\,THz is followed by a second at
81: 3.5\,THz in accordance with results from Raman studies \cite{CrBA94,KiVP99}.
82: As expected, the crystal DOS is more structured, in particular in the
83: low energy region.
84: Distinct peaks at 0.6, 0.8, 1.1 and 1.5\,THz become apparent.
85: They are due to strong contributions from zone--boundary modes
86: (figure~\ref{Dis-relat}).
87: Compared to the glass, significant density is missing in the low
88: energy region and in the range from 1.5 to 3\,THz,
89: and reappears at higher frequencies around 3.5\,THz.
90: The gap for $\nu > 5$\,THz common for both states and indicates
91: a separation from high energy intramolecular modes.
92:
93: The DOS of the polycrystalline sample
94: is in accord with the measured dispersion of the single crystal:
95: A small shoulder at 0.4\,THz can be attributed
96: to the transverse acoustic zone--boundary phonons in $[100]$ direction,
97: and the main peak at 0.6\,THz corresponds
98: to the transverse acoustic zone--boundary phonons in the other two lattice
99: directions.
100: The peak at 0.8\,THz reflects the longitudinal acoustic zone--boundary
101: phonon in $[001]$ direction and the transverse optic in $[010]$ direction.
102:
103: In order to show the low energy modes on an enhanced scale,
104: we plot in figure~\ref{DOS_100K} $g(\nu)/\nu^2$
105: which in the one--phonon approximation is proportional to $S(Q,\nu)$ itself.
106: In this representation,
107: the excess of the glass over the crystal becomes evident.
108: A well defined frequency peak appears around 0.35\,THz which is down shifted
109: with respect to the first peak of the crystal at 0.6\,THz
110: and superposed to a long tail which is similar for both, glass and crystal.
111: Note that the maximum of the boson peak at 0.35\,THz is located
112: below the lowest acoustic zone--boundary phonons in the crystal.
113:
114: Another dynamic observable which can be obtained from neutron scattering
115: is the atomic mean square displacement $\langle r^2(T) \rangle$.
116: Roughly speaking, $\langle r^2(T) \rangle^{3/2}$ measures the volume
117: to which an atom remains confined in the limit $t\rightarrow\infty$.
118: For a large class of model situations
119: (harmonic solid, Markovian diffusion, \ldots)
120: it can be obtained directly
121: from the Gaussian $Q$ dependence of the elastic scattering intensity
122: \begin{equation} \label{Sq0}
123: S(Q,\nu\!=\!0)=\exp(-Q^2 \langle r(T)^2 \rangle)\,.
124: \end{equation}
125: For a harmonic solid,
126: \begin{equation} \label{msd}
127: \langle r(T)^2\rangle =
128: {\hbar^2 \over 6 M k_{\rm B} T } \int_0^\infty\! {\rm d}\nu\,
129: \frac{g(\nu)}{\beta} {\rm coth}(\frac{\beta}{2})
130: \end{equation}
131: (with $\beta=h\nu/k_{\rm B}T$)
132: crosses over from zero--point oscillations
133: $\langle r^2(0) \rangle$
134: to a linear regime
135: $\langle r^2(T) \rangle \propto T$. % \cite{x43}.
136: Note, for a meaningful application of (\protect\ref{msd}), we have to
137: assume the rigid--body limit (Sect.~3.1, Ref.~\protect\cite{WuKB93}).
138: Therefore, $M$ has to be taken as the average atomic mass which in the case
139: of OTP is 7.2.
140: In any real experiment,
141: which integrates over the elastic line with a resolution $\Delta\nu$,
142: one actually measures atomic displacements within finite times
143: $t_\Delta\simeq 2\pi/\Delta\nu$.
144:
145: Figure~\ref{MSD-MSD_w} shows mean square displacements of OTP,
146: determined according to (\ref{Sq0})
147: from elastic back--scattering
148: (fixed window scans on IN13, with $t_\Delta\simeq100-200\,$psec)
149: and from Fourier--deconvoluted time-of-flight spectra
150: (taking the plateau $S(Q,t_\Delta)$ with $t_\Delta\simeq5-10\,$psec
151: from Fourier transformed IN6 data).
152: For the glassy sample, a direct comparison can be made and shows
153: good agreement between IN6 and IN13.
154: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
155: \input{Bilder/Vibration/MSD-MSD_w.tex}
156: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
157:
158: The lines in figure~\ref{MSD-MSD_w}
159: are calculated through (\ref{msd}) from the DOS at 100\,K.
160: For low temperatures, the $\langle r^2(T) \rangle$
161: are in full accord with the values determined through (\ref{Sq0}).
162: This comparison can be seen as a cross--check between
163: the analysis of elastic and inelastic neutron scattering data.
164:
165: Equation (\ref{msd}) not only gives the absolute value
166: of~$\langle r^2(T) \rangle$,
167: but enables us also to read off which modes contribute most
168: to the atomic displacement.
169: To this end, we restrict the integration (\ref{msd}) to
170: modes with $0<\nu<\nu'$.
171: The inset in figure~\ref{MSD-MSD_w}
172: shows the relative value
173: $\langle r^2(\nu';T) \rangle^{1/2} / \langle r^2(\nu;T) \rangle^{1/2}$
174: for $T=100\,$K as function of~$\nu'$.
175: Modes below 0.6\,THz in the crystal, or 0.4\,THz in the glass
176: contribute about 55\% to the total displacement;
177: 90\% are reached only at about 2\,THz.
178: This means that the modes which are responsible for
179: the mean square displacement and the Debye--Waller factor
180: are not rigid body motions alone,
181: but contain a significant contribution from
182: other modes.
183: %intramolecular degrees of freedom.
184:
185: Anyway, one immediately recognizes that for all temperatures the
186: mean square displacement in the glass is larger than in the crystal.
187: For higher temperatures the measured mean square displacements
188: are clearly larger than expected from harmonic theory.
189: These anharmonic contributions to $\langle r^2(T) \rangle$ start
190: around 140--150\,K.
191: Around the same temperature deviations from the proportionality
192: $\langle r^2(T) \rangle \propto T$ can be observed in
193: coherent elastic scans and in the \Mob\ experiment \cite{VaFl79}
194: (figure~\ref{DWF-VaFl79} in section~\ref{t-squarerootsingularity}).
195: In addition, around 140--150\,K the temperature dependence of the
196: experimental specific heat $c_{\rm P}(T)$ \cite{ChBe72} shows a
197: slight change for both glass and crystal.
198:
199: Anharmonicities can directly be observed via the temperature dependence
200: of the density of states.
201: With increasing temperature the first moment, which provides a measure for
202: the center of gravity, clearly shifts to lower frequencies
203: for the glass and even more for the crystal.
204: %Exploiting the first logarithmic moment yields an
205: %explicit anharmonic contribution to
206: %the specific heat at constant pressure of about 1.9\% for the crystal
207: %and only 1.2\% for the glass.
208: Note however, the strong increase above 240\,K in the glass is connected
209: with the glass transition.
210:
211: Our vibrational density of states can also be checked againts
212: the heat capacity, which for a harmonic solid is given by the integral
213: \begin{equation}\label{cv}
214: c_{\rm P}(T) \simeq c_{\rm V}(T) = N_{\rm at}R\int_0^\infty\!{\rm d}\nu\, g(\nu)
215: \frac{(\beta/2)^2}{\sinh^2(\beta/2)}\,.
216: \end{equation}
217: With a Debye DOS, this yields the well--known $c_{\rm P}\propto T^3$.
218: Therefore, in figure~\ref{cpT} experimental data \cite{ChBe72}
219: for the specific heat of glassy and polycrystalline OTP
220: are plotted as $c_{\rm P}/T^3$.
221: In this representation,
222: a boson peak at 0.35\,THz is expected to lead to a maximum at about 4\,K.
223: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
224: \input{Bilder/Vibration/cpT.tex}
225: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
226:
227: The lines are calculated through (\ref{cv}) from the neutron DOS.
228: They agree with the measured data for both glass and crystal in
229: absolute units and over a broad temperature range.
230: Similar accord has been reported for a number of other systems
231: \cite{CrBA94,BeCA96,TaRV98,WuPC95,BuPN86}.
232:
233: At higher temperatures,
234: the heat capacities of crystalline and glassy OTP differ only little
235: ({\it e.g.} at 200\,K: $c_{\rm P}^{\rm cryst}= 182.8$\,J\,mol$^{-1}$\,K$^{-1}$
236: and $c_{\rm P}^{\rm glass}= 186.1$\,J\,mol$^{-1}$\,K$^{-1}$ \cite{ChBe72}).
237: Towards higher frequencies, the DOS becomes less sensitive
238: to the presence or absence of crystalline order
239: and remaining differences (clearly visible in figure~\ref{DOS_100K})
240: are largely averaged out by the integral (\ref{cv}).
241:
242: Altogether, these observations suggest that the mode structure in the glass
243: and the crystal is quite similar except for an overall softening in the
244: glass.
245: Concerning the above presented analysis of quasielastic
246: scattering in the supercooled liquid we can state that
247: above 0.6\,THz a very large portion of inelastic
248: intensity in $S(Q,\nu)$ of glassy OTP must arise from optic modes.
249: Hybridisation and coupling between inter- and intramolecular modes
250: plays an important role for frequencies higher than 0.6\,THz
251: \cite{CrBG93,CrBA94}.
252: As a consequence, the quasielastic scattering, which is confined to
253: below 0.25\,THz (1\,meV), is clearly dominated by rigid--body motions
254: and the analysis in terms of relaxations remains reasonable.
255:
256:
257: