Importance of subleading corrections for the Mott critical point

Abstract

The interaction-induced metal-insulator transition should be in the Ising universality class. Experiments on layered organic superconductors suggest that the observed critical endpoint of the first-order Mott transition belongs instead to a different universality class. To address this question, we use dynamical mean-field theory and a cluster generalization that is necessary to account for short-range spatial correlations in two dimensions. Such calculations can give information on crossover effects, in particular quantum ones, that are not included in the simplest mean-field. In the cluster calculation, a canonical transformation that minimizes the sign problem in continuous-time quantum Monte Carlo calculations allows us to obtain very accurate results for double occupancy. These results show that there are important subleading corrections that can lead to apparent exponents that are different from mean-field. Experiments on optical lattices could verify our predictions.