Self-consistent GW study of VO2*

The question of the metal-insulator transition (MIT) in VO₂ remains open. Hybrid functionals [1] and non-self-consistent G₀W₀ [2] studies state that non-local exact exchange (EEX, included in hybrids and GW) is the most important. However, the arbitrariness of the percentage of EEX (hybrids) or of the starting point (G₀W₀) makes the agreement of the results with experiment not convincing. From the LDA+DMFT studies of VO₂ [3,4] it follows that strong on-site correlations are important for MIT. However, LDA+DMFT operates with local parameters (U and J) and the argument in favor of strong correlations might be just a consequence of missing the non-local EEX in LDA+DMFT. The talk will be about the role of self-consistency in GW approach, which eliminates dependence on the starting point. It will be shown that indeed the GW level of theory provide all necessary basis for description of MIT in VO₂ quantitatively.
[1] V. Eyert, Phys. Rev. Lett. 107, 016401 (2011)
[2] M. Gatti et al., Phys. Rev. Lett. 99, 264402 (2007)
[3] S. Biermann et al., Phys. Rev. Lett. 94, 026404 (2005)
[4] W.H. Brito et al., Phys. Rev. Lett. 117, 056402 (2016)