Relaxation times and transport properties of moderately supercooled liquids exhibit a universal dependence on temperature. When the glass transition temperature Tg is approached, the relaxation of organic glass formers usually splits into two processes. A slower process, structural a relaxation, corresponds to cooperative molecular dynamics, which is practically frozen at Tg. Another process, slow b relaxation, occurs on a shorter time scale and exists also below Tg. The microscopic origin of this process is still disputed.

In order to understand the relaxational processes on a microscopic level and to distinguish between translational and rotational modes of motion we have applied Nuclear Forward Scattering (NFS) of synchrotron radiation and Synchrotron Radiation based Perturbed Angular Correlation (SRPAC). Both methods are sensitive to nuclear spin dynamics, which is coupled to the molecular rotations. In addition, NFS is also sensitive to translational motions on an atomic length scale. Thus the combination of NFS and SRPAC allows one to separate rotational and translational dynamics.

We applied NFS and SRPAC to investigate the dynamics of the glass former dibutyl phthalate (DBP) (Tg = 178 K) using probe molecules of ferrocene enriched with 57Fe. The experiment was carried out at beamline ID18. SRPAC was measured up to 330 K and NFS up to 210 K, where the Lamb-Mössbauer factor, and therefore also the NFS intensity, vanish. Typical time spectra are shown in Figure 10. At the lower temperatures, the SRPAC intensity follows an exponential decay modulated by quantum beats. In the regime of slow relaxation, the beats are damped at a rate proportional to rotational relaxation. Similar quantum beats modulate the decay of the NFS intensity, where the damping depends on the sum of rotational and translational relaxation. At higher temperatures, in the regime of fast relaxation, only SRPAC spectra can be measured, which exhibit a characteristic approach of the natural decay (Abragam-Pound limit).


Fig. 10: Time evolution of SRPAC and NFS intensities for several temperatures. The solid lines show the fit according to the full theory.


In Figure 11a, we compare the rotational relaxation rate obtained by SRPAC to the sum of the rotational and translational relaxation rates obtained by NFS and Mössbauer Spectroscopy (MS) [1]. Below 190 K the data sets coincide, which means that translational dynamics is absent in the experimental time window. Above 190 K the NFS and MS data begin to deviate from the SRPAC data because translational dynamics is activated. From these results the pure translational relaxation rate can be derived.


Fig. 11: a) Comparison of the rotational relaxation rate of the ferrocene molecule derived from SRPAC () to the relaxation rate derived from NFS () and from Mössbauer spectroscopy (); b) Comparison of the rotational () and translational (,) relaxation rates of the FC molecules to the DS data for pure DBP (red line, ).



In Figure 11b we compare the rotational and translational relaxation rates of the ferrocene molecules in DBP with each other and with data of pure DBP as obtained from dielectric spectroscopy (DS) [2,3]. At low temperatures, the DS data split into two branches. The branch of slow ß relaxation follows our data of rotational dynamics, whereas the branch of relaxation decreases in parallel with our data of translational dynamics. This correlation suggests an interpretation of the ferrocene data also in terms of decoupling, where one branch corresponds to rotational, the other to translational relaxation. In particular, the coincidence of the slow ß relaxation branches for probe and solvent suggests that also in pure DBP the slow ß relaxation is connected with rotational dynamics.

[1] S.L. Ruby, B.J. Zabransky, P.A. Flinn, J. Physique 37, C6-745 (1976).
[2] S.A. Dzyuba and Yu.D. Tsvetkov, J. Struct. Chem. 28, 343 (1987).
[3] P.K. Dixon et al., Phys. Rev. Lett. 65, 1108 (1990).

Principal Publication and Authors
I. Sergueev (a,b), U. van Bürck (a), A.I. Chumakov (b,c), T. Asthalter (d), G.V. Smirnov (c), H. Franz (e), R. Rüffer (b), W. Petry (a), submitted to Phys. Rev. Lett.
(a) TUM Physik E13, Garching (Germany)
(b) ESRF
(c) RRC, Moscow (Russia)
(d) Univ. Stuttgart (Germany)
(e) DESY, HASYLAB, Hamburg (Germany)