Entanglement entropy and quantum phase transitions in quantum dots coupled to Luttinger liquid wires

TL;DR

This paper investigates a quantum phase transition in a system of two quantum dots coupled to Luttinger-liquid leads, revealing an entanglement entropy drop at the transition and validating analytical predictions with numerical simulations.

Contribution

It analytically and numerically demonstrates a Berezinskii-Kosterlitz-Thouless transition in a two-impurity Luttinger-liquid system using entanglement entropy as a key indicator.

Findings

Entanglement entropy drops from ln(2) to zero at the transition.

Finite size scaling accurately predicts the transition point.

Numerical results agree with analytical predictions.

Abstract

We study a quantum phase transition which occurs in a system composed of two impurities (or quantum dots) each coupled to a different interacting (Luttinger-liquid) lead. While the impurities are coupled electrostatically, there is no tunneling between them. Using a mapping of this system onto a Kondo model, we show analytically that the system undergoes a Berezinskii-Kosterlitz-Thouless quantum phase transition as function of the Luttinger liquid parameter in the leads and the dot-lead interaction. The phase with low values of the Luttinger-liquid parameter is characterized by an abrupt switch of the population between the impurities as function of a common applied gate voltage. However, this behavior is hard to verify numerically since one would have to study extremely long systems. Interestingly though, at the transition the entanglement entropy drops from a finite value of $\ln(2)$ to zero. The drop becomes sharp for infinite systems. One can employ finite size scaling to extrapolate the transition point and the behavior in its vicinity from the behavior of the entanglement entropy in moderate size samples. We employ the density matrix renormalization group numerical procedure to calculate the entanglement entropy of systems with lead lengths of up to 480 sites. Using finite size scaling we extract the transition value and show it to be in good agreement with the analytical prediction.