Comparing dynamic correlation lengths from an approximation to the four-point dynamic susceptibility and from the picosecond vibrational dynamics

TL;DR

This study compares two methods for measuring dynamic correlation lengths in supercooled liquids, finding good correlation between them but highlighting discrepancies in Boson peak measurements above Tg.

Contribution

It introduces a comparison between vibrational dynamics-based and four-point susceptibility-based methods for determining dynamic correlation lengths.

Findings

Good correlation between the two measures for 26 materials

Pressure dependences are consistent within experimental error

Boson peak measurements above Tg show unrealistic temperature dependence

Abstract

Recently a new approach to the determination of dynamic correlation lengths, {\xi}, for supercooled liquids, based on the properties of the slow (picosecond) vibrational dynamics, was carried out [L. Hong, V.N. Novikov, and A.P. Sokolov, Phys. Rev. E 83, 061508 (2011)]. Although these vibrational measurements are typically conducted well below the glass transition temperature, Tg, the assumption is that the structure of the liquid is frozen at Tg, so that the {\xi} characterize dynamic heterogeneity in the supercooled liquid state. We compare {\xi} from this method to values calculated using an approximation to the four-point dynamic susceptibility. For 26 different materials we find good correlation between the two measures; moreover, the pressure dependences are consistent within the large experimental error. However, {\xi} from Boson peak measurements above Tg have a different, and unrealistic, temperature dependence.