Resolving the chicken-and-egg problem in VO₂

A new paradigm for the Mott transition

Óscar Nájera¹, M. Civelli¹,
V. Dobrosavljević², M. J. Rozenberg¹

najera.oscar@gmail.com / Titan-C

¹ Laboratoire de Physique des Solides, CNRS-UMR8502, Université Paris-Sud, Orsay 91405, France
² Department of Physics and National High Magnetic Field Laboratory, Florida State University, Tallahassee, FL 32306, USA

March 13, 2017 - New Orleans

The transition in VO₂

Peierls vs Mott

VO₂ is a very interesting transition metal oxide because it displays a Metal insulator transition close to room temperature. Making it attractive for novel electronic devices. Explaining this transition remains a challenging endeavor, between our hand waving arguments to our most sophisticated bandstructure calculations great progress has been made to describe this material but not enough to settle the debate about the origins of its transition. VO₂ is a correlated metal at high temperature and one would argue for Mott transition to justify the insulator, where electrons localize due to the strong Coulomb repulsion they experience between eachother. Nevertheless at the transition temperature the crystal structure changes from rutile to Monoclinic and dimers form and inspire an alternative justification where the insulating behavior is attributed to this spin-Peierls structural instability.

Outline

Solve using DMFT a reference model: The Dimer Hubbard Model
Characterize the correlated Insulator to correlated Metal transition
Dimer-Mott Transition explains key experiments in the transition of VO₂

Dynamical Mean Field Theory (DMFT)

Same cluster impurity problem as other LDA+DMFT studies for VO₂
Solved Exactly in the Bethe Lattice
DMFT Exact by construction in the large coordination limit
DMFT yields generic behavior of a high-dimensional lattice

For our study we trade the complications of the real crystal structure and the orbital degeneracy of VO₂ for a model Hamiltonian where we focus on just few things. Take the fact that there is a preferred direction in the unit cell which gives the raise to dimer and build a single orbital Hubbard model out of it, the Dimer Hubbard Model. It will capture the key competition between Mott localization due to coulomb repulsion and the singlet dimerization. We will explore if a purely electronic transition takes place in this model, for which in this study the lattice will stay fixed. We will address the question if this purely electronic transition is of first order as a function of temperature within a relevant parameter region, having a bearing on the physics of VO₂. We will explore the physical nature of the different electronic states and relate our results to key available experiments.

Emphasize that you are using the same cluster-impurity model as in the LDA+DMFT studies. We simplify the geometry and the multi-orbital nature, which is known to be qualitatively OK from extensive single site studies (ie LDA+DMFT in single site always give the 3-peak structure or the incoherent Mott insulator)

DMFT Phase diagram at \(T=0\)

Reference Mott transition: V\(_2\)O₃

The case \(t_\perp/D=0\) no dimers

The Dimer-Mott transition: VO₂

The dimerized \(t_\perp/D=0.3\) lattice

The Effect of correlation is to enhance dimerization

Electronic Structure

Mid Infrared Peak appears in Correlated Metal

Optical conductivity at the transition

Metallic \(\sigma(w)\) is only from the metallic islands. The high \(T\) rutile metal does not have a MIR peak
Qazilbash et al., Science (2007), 318(5857), 1750–1753
O. Nájera, et al. (2017). PRB, 95(3), 035113

Conclusions

The Dimer Hubbard Model describes a temperature driven insulator to metal correlated transition
These results explain key observations in the transition in VO₂: the Mid-Infrared features in the optical conductivity before the structural transition takes place
The existence of a metastable correlated dimer metal phase is the key ingredient that shows that the VO₂ Insulator Metal transition is electronically driven(Dimer Mott Mechanism)
Read more about our work in: O. Nájera, et al. (2017). PRB, 95(3), 035113

Optics Support