Entropy and Nonlinear Nonequilibrium Thermodynamic Relation for Heat Conducting Steady States

TL;DR

This paper develops a new thermodynamic relation for heat conducting steady states, incorporating nonlinear nonequilibrium effects, and clarifies its connection with linear response theory, advancing the understanding of nonequilibrium thermodynamics.

Contribution

It introduces a simplified derivation of the extended Clausius relation and derives an improved relation with a nonlinear contribution based on correlation between work and heat.

Findings

Derived a new simpler derivation of the extended Clausius relation.

Established the relation between the extended Clausius relation and linear response theory.

Identified a nonlinear nonequilibrium contribution that is essential and uniquely determined.

Abstract

Among various possible routes to extend entropy and thermodynamics to nonequilibrium steady states (NESS), we take the one which is guided by operational thermodynamics and the Clausius relation. In our previous study, we derived the extended Clausius relation for NESS, where the heat in the original relation is replaced by its "renormalized" counterpart called the excess heat, and the Gibbs-Shannon expression for the entropy by a new symmetrized Gibbs-Shannon-like expression. Here we concentrate on Markov processes describing heat conducting systems, and develop a new method for deriving thermodynamic relations. We first present a new simpler derivation of the extended Clausius relation, and clarify its close relation with the linear response theory. We then derive a new improved extended Clausius relation with a "nonlinear nonequilibrium" contribution which is written as a correlation between work and heat. We argue that the "nonlinear nonequilibrium" contribution is unavoidable, and is determined uniquely once we accept the (very natural) definition of the excess heat. Moreover it turns out that to operationally determine the difference in the nonequilibrium entropy to the second order in the temperature difference, one may only use the previous Clausius relation without a nonlinear term or must use the new relation, depending on the operation (i.e., the path in the parameter space). This peculiar "twist" may be a clue to a better understanding of thermodynamics and statistical mechanics of NESS.