DYNAMICAL MEAN FIELD THEORY OF THE ANTIFERROMAGNETIC METAL TO ANTIFERROMAGNETIC INSULATOR TRANSITION

The correlation driven metal insulator transition (MIT) or Mott transition is one of the central problems of condensed matter physics. Recently, a great deal of progress has been made in understanding the MIT using the dynamical mean field approach (DMFT), a method which becomes exact in the well-defined limit of infinite lattice coordination [1,2]. However, all studies so far have been confined to the paramagnetic metal (PM) to paramagnetic insulator (PI) transition. In this Letter, we use DMFT to study the transition from the antiferromagnetic metal (AM) to an antiferromagnetic insulator (AI) at zero temperature. The motivation for this work is twofold. Experimentally, the interaction or pressure driven MIT in V2O3 [3,4] and NiS22xSex [4,5] takes place between magnetically ordered states. The Neel temperatures are much smaller than the respective characteristic electronic energy scales. We interpret this as a sign of reduced effective magnetic correlations as compared to estimates obtained from a simple one band Hubbard model. Close to the MIT, the behavior of physical quantities like the specific heat coefficient is, however, different in these two materials. Furthermore, measurements of the magnetic moment seem to indicate that the magnetism in V2O3 is much weaker than in NiS22xSex [6]. This suggests that the strength of the magnetic correlations influences the MIT. We would therefore like to understand how magnetic correlations, which control the scale at which the spin entropy is quenched, affect the MIT and hence, various physical quantities. While we are still far from a realistic modeling of these materials (in this work we consider only commensurate magnetic order and ignore orbital degeneracy and realistic band structure), we present a simple model which, we believe, captures the generic effects of the magnetic correlations on the MIT. We consider a simple one band Hamiltonian,