Sampling Energy Landscapes in Germanium Telluride Glass Through Resistance Fluctuations

The concept of energy landscapes is very successful in explaining activated biological, chemical and physical processes. For the highly disordered systems of supercooled liquids and glasses, it has found wide support in computer simulations of model systems, which can resolve detailed characteristics of the potential energy landscape governing their structural dynamics. In experimental studies, however, any details about an underlying landscape are usually obscured due to accessible sample sizes and the averaging nature of spectroscopic techniques. Here we demonstrate that individual resistance states can be resolved in a nanoscopic volume of germanium telluride glass. The fluctuations between these states are measured over a wide temperature range, six orders of magnitude in time and modeled with a hidden Markov model. The resulting attempt frequencies and activation energies reveal a complex free energy landscape, where transitions between states are not only governed by energetic barriers but limited by entropic contributions. Beyond their practical relevance for electronic memory and neuromorphic hardware applications, these results illustrate the feasibility of the experimental approach for fundamental investigations of the energy landscape of glasses and point to the importance of entropic effects for their structural dynamics. Our approach can provide new insights into all kinds of memristive materials, where equilibrium and non-equilibrium (noisy) state dynamics can be exploited for neuromorphic hardware applications exactly because of a close connection between electrical resistance and smallest reconfiguration on the atomic scale.