Some remarks on 'superradiant' phase transitions in light-matter systems

TL;DR

This paper examines superradiant phase transitions in various light-matter models, clarifying their mean-field nature, negligible quantum fluctuations, and the impact of photon losses on their critical behavior.

Contribution

It provides a detailed analysis showing these transitions are mean-field, not quantum phase transitions, and discusses how photon losses affect criticality in these models.

Findings

Transitions are of the mean-field type with negligible quantum fluctuations.

Quantum fluctuations are strictly zero in Tavis-Cummings and Jaynes-Cummings models.

Photon losses eliminate criticality in open Tavis-Cummings and Jaynes-Cummings models.

Abstract

In this paper we analyze properties of the phase transition that appears in a set of quantum optical models; Dicke, Tavis-Cummings, quantum Rabi, and finally the Jaynes-Cummings model. As the light-matter coupling is increased into the deep strong coupling regime, the ground state turns from vacuum to become a superradiant state characterized by both atomic and photonic excitations. It is pointed out that all four transitions are of the mean-field type, that quantum fluctuations are negligible, and hence these fluctuations cannot be responsible for the corresponding vacuum instability. In this respect, these are not quantum phase transitions. In the case of the Tavis-Cummings and Jaynes-Cummings models, the continuous symmetry of these models implies that quantum fluctuations are not only negligible, but strictly zero. However, all models possess a non-analyticity in the ground state in agreement with a continuous quantum phase transition. As such, it is a matter of taste whether the transitions should be termed quantum or not. In addition, we also consider the modifications of the transitions when photon losses are present. For the Dicke and Rabi models these non-equilibrium steady states remain critical, while the criticality for the open Tavis-Cummings and Jaynes-Cummings models is completely lost, i.e. in realistic settings one cannot expect a true critical behaviour for the two last models.