Nonequilibrium thermodynamics of interacting tunneling transport: variational grand potential, density-functional formulation, and nature of steady-state forces

Abstract

The standard formulation of tunneling transport rests on an open-boundary modeling. There, conserving approximations to nonequilibrium Green function or quantum-statistical mechanics provide consistent but computational costly approaches; alternatively, use of density-dependent ballistic-transport calculations [e.g., Phys. Rev. B 52, 5335 (1995)], here denoted `DBT', provide computationally efficient (approximate) atomistic characterizations of the electron behavior but has until now lacked a formal justification. This paper presents an exact, variational nonequilibrium thermodynamic theory for fully interacting tunneling and provides a rigorous foundation for frozen-nuclei DBT calculations as a lowest order approximation to an exact nonequilibrium thermodynamics density functional evaluation. The theory starts from the complete electron nonequilibrium quantum statistical mechanics and I identify the operator for the nonequilibrium Gibbs free energy. I demonstrate a minimal property of a functional for the nonequilibrium thermodynamic grand potential which thus uniquely identifies the solution as the exact nonequilibrium density matrix. I also show that a uniqueness-of-density proof from a closely related study [Phys. Rev. B 78, 165109 (2008)] makes it possible to provide a single-particle formulation based on universal electron-density functionals. I illustrate a formal evaluation of the thermodynamics grand potential value which is closely related to the variation in scattering phase shifts and hence to Friedel density oscillations. This paper also discusses the difference between the here-presented exact thermodynamics forces and the often-used electrostatic forces. Finally the paper documents an inherent adiabatic nature of the thermodynamics forces and observes that these are suited for a nonequilibrium implementation of the Born-Oppenheimer approximation.