Mind the gap but also the spin: why the Heyd-Scuseria-Ernzerhof hybrid functional description of VO2 phases is not correct

Abstract: In contrast with recent claims that the Heyd-Scuseria-Ernzerhof (HSE) screened hybrid functional can provide a good description of the electronic and magnetic structure of VO2 phases [V. Eyert, Phys. Rev. Lett. 107, 016401 (2011)], we show here that the HSE lowest-energy solutions for both the low-temperature monoclinic (M1) phase and the high-temperature rutile (R) phase, which are obtained upon inclusion of spin polarization, are at odds with experimental observations. For the M1 phase the groundstate is (but should not be) magnetic, while the groundstate of the R phase, which is also spin-polarized, is not (but should be) metallic. The energy difference between the low-temperature and high-temperature phases is also in strong discrepancy with the experimental latent heat.