Thermopower of an SU(4) Kondo resonance under an SU(2) symmetry-breaking field

TL;DR

This paper investigates the thermopower behavior of an SU(4) Kondo quantum dot under symmetry-breaking fields, revealing its sensitivity to electronic structure and energy scales through advanced theoretical methods.

Contribution

It provides a detailed analysis of thermopower in SU(4) Kondo systems with symmetry-breaking, comparing multiple computational approaches for accuracy.

Findings

Thermopower exhibits two dips and a peak related to Kondo temperature and level splitting.

Thermopower is more sensitive than conductance to electronic structure at intermediate energies.

SBMFA is inadequate for large splitting; NCA combined with Fermi liquid relations yields reliable results.

Abstract

We calculate the thermopower of a quantum dot described by two doublets hybridized with two degenerate bands of two conducting leads, conserving orbital (band) and spin quantum numbers, as a function of the temperature $T$ and a splitting $\delta$ of the quantum dot levels which breaks the SU(4) symmetry. The splitting can be regarded as a Zeeman (spin) or valley (orbital) splitting. We use the non-crossing approximation (NCA), the slave bosons in the mean-field approximation (SBMFA) and also the numerical renormalization group (NRG) for large $\delta$. The model describes transport through clean C nanotubes %with weak disorder and in Si fin-type field effect transistors, under an applied magnetic field. The thermopower as a function of temperature $S(T)$ displays two dips that correspond to the energy scales given by the Kondo temperature $T_K$ and $\delta$ and one peak when $k_BT$ reaches the charge-transfer energy. These features are much more pronounced than the corresponding ones in the conductance, indicating that the thermopower is a more sensitive probe of the electronic structure at intermediate or high energies. At low temperatures ($T \ll T_K$) $T_K S(T)/T$ is a constant that increases strongly near the degeneracy point $\delta=0$. We find that the SBMFA fails to provide an accurate description of the thermopower for large $\delta$. Instead, a combination of Fermi liquid relations with the quantum-dot occupations calculated within the NCA gives reliable results for $T \ll T_K$.