Perturbation theory for an Anderson quantum dot asymmetrically attached to two superconducting leads

TL;DR

This paper develops a second-order perturbation theory for an Anderson quantum dot coupled to superconducting leads, accurately predicting phase transitions and Josephson currents, offering a simpler alternative to complex numerical methods.

Contribution

It introduces a charge-conserving, thermodynamically consistent second-order perturbation approach for quantum dots with superconducting leads, providing analytical formulas and practical estimates for phase boundaries.

Findings

Accurately predicts the 0-$ ext{pi}$ phase boundary and Josephson current.

Validates the perturbation method against Numerical Renormalization Group results.

Offers a quick estimate tool for experimental data interpretation.

Abstract

Self-consistent perturbation expansion up to the second order in the interaction strength is used to study a single-level quantum dot with local Coulomb repulsion attached asymmetrically to two generally different superconducting leads. At zero temperature and wide range of other parameters the spin-symmetric version of the expansion yields excellent results for the position of the $0-\pi$ impurity quantum phase transition boundary and Josephson current together with the energy of Andreev bound states in the $0$-phase as confirmed by numerical calculations using the Numerical Renormalisation Group method. We analytically prove that the method is charge-conserving as well as thermodynamically consistent. Explicit formulas for the position of the $0-\pi$ phase boundary are presented for the Hartree-Fock approximation as well as for its variant called Generalized Atomic Limit. It is shown that the Generalized Atomic Limit can be used as a quick estimate for the position of the phase boundary at half-filling in a broad range of parameters. We apply our second order perturbation method to the interpretation of the existing experimental data on the phase boundary with very satisfactory outcome suggesting that the so far employed heavy numerical tools such as Numerical Renormalization Group and/or Quantum Monte Carlo are not necessary in a class of generic situations and can be safely replaced by a perturbative approach.