Direct entropy determination and application to artificial spin ice

TL;DR

This paper introduces a new method for directly measuring configurational entropy in microscopic systems, demonstrated on artificial spin ice, revealing how local correlations influence longer-range order.

Contribution

It presents a conditional-probability technique to calculate entropy densities from limited data, applicable to various disordered systems including artificial spin ice.

Findings

Nearest-neighbour correlations influence longer-range order.

Method effectively assesses models of disordered systems.

Applied to artificial spin ice, revealing correlation effects.

Abstract

From thermodynamic origins, the concept of entropy has expanded to a range of statistical measures of uncertainty, which may still be thermodynamically significant. However, laboratory measurements of entropy continue to rely on direct measurements of heat. New technologies that can map out myriads of microscopic degrees of freedom suggest direct determination of configurational entropy by counting in systems where it is thermodynamically inaccessible, such as granular and colloidal materials, proteins and lithographically fabricated nanometre-scale arrays. Here, we demonstrate a conditional-probability technique to calculate entropy densities of translation-invariant states on lattices using limited configuration data on small clusters, and apply it to arrays of interacting nanometre-scale magnetic islands (artificial spin ice). Models for statistically disordered systems can be assessed by applying the method to relative entropy densities. For artificial spin ice, this analysis shows that nearest-neighbour correlations drive longer-range ones.