The role of the chemical potential in the BCS theory

TL;DR

This paper investigates how the chemical potential's deviation from the system's chemical potential affects BCS superconductivity, revealing significant changes in energy gap, phase transition order, and solution multiplicity.

Contribution

It introduces a generalized BCS model with a separate chemical potential parameter, showing how deviations influence superconducting properties and phase transitions.

Findings

Energy gap decreases when $\mu eq \mu_R$

First-order phase transition occurs with $\mu eq \\mu_R$

Multiple solutions for the gap equation can exist at the same temperature

Abstract

We study the effect of the chemical potential on the results of the BCS theory of superconductivity. We assume that the pairing interaction is manifested between electrons of single-particle energies in an interval $[\mu - \hbar\omega_c, \mu + \hbar\omega_c]$, where $\mu$ and $\omega_c$ are parameters of the model--$\mu$ needs not be equal to the chemical potential of the system, denoted here by $\mu_R$. The BCS results are recovered if $\mu = \mu_R$. If $\mu \ne \mu_R$ the physical properties change significantly: the energy gap $\Delta$ is smaller than the BCS gap, a population imbalance appears, and the superconductor-normal metal phase transition is of the first order. The quasiparticle imbalance is an equilibrium property that appears due to the asymmetry with respect to $\mu_R$ of the single-particle energy interval in which the pairing potential is manifested. For $\mu_R - \mu$ taking values in some ranges, the equation for $\Delta$ may have more than one solution at the same temperature, forming branches of solutions when $\Delta$ is plotted vs $\mu_R-\mu$ at fixed $T$. The solution with the highest energy gap, which corresponds to the BCS solution when $\mu = \mu_R$, cease to exist if $|\mu-\mu_R| \ge 2\Delta_0$ ($\Delta_0$ is the BCS gap at zero temperature). Therefore the superconductivity is conditioned by the existence of the pairing interaction and also by the value of $\mu_R - \mu$.