Inverse Meyer-Neldel behavior for activated processes in model glasses

The activation-relaxation technique is used to explore the distribution of escape times from metastable minima of a model metallic glass, estimated using the harmonic transition state theory. We investigate in particular the distribution of attempt frequencies and find that these prefactors are distributed over a range of almost ten orders of magnitude, in stark contrast with the common assumption that they are given by a typical vibrational frequency. Moreover, the typical attempt frequencies appear to decrease with increasing barrier energy, a behavior inverse to the so-called Meyer-Neldel rule observed in many processes with apparent Arrhenius behavior. When combined with a multiple-hop analysis that takes into account the multiplicity of possible transitions, a direct Meyer-Neldel behavior is recovered, albeit with a reduced characteristic prefactor.