Looks pretty bad to me. I am not used to GGA+U since in many cases the adjusted U parameter is unphysical. You can check in the literature for other GGA+U attempts in peroveskites. Another possibility is to do GGA-1/2 correcting only the oxygen self-interaction energy. It gives quite accurate results if the top of the valence band and the bottom of the conduction band are not made of the same orbital (p in the case of O). If you want to give it a try you can generate the GGA-1/2 PP for O here http://www.eedevice.com/dft-half/
Em qui, 4 de abr de 2019 às 17:01, doaa haleem <doaa.halee...@gmail.com> escreveu: > Thank you for your reply > > I understand your point, but The experimental band gap for my structure > is 4.8 eV , after using GGA+U it becomes 2.8 eV...so is the 2 eV > difference acceptable ? > > > > > On Wed, 3 Apr 2019, 10:02 p.m. Leonardo Fonseca, <fonsecal...@gmail.com> > wrote: > >> There is no such a thing as the exact band gap if you employ DFT within >> the various approximations for the exchange-correlation. For most cases DFT >> underestimates it. GGA+U is an approximation which in many cases improves >> the band gap considerably over plain GGA, but it is not exact. In fact not >> even with GW one can be sure to have an exact band gap. The underlining >> many-body problem is too difficult to solve, thus approximations to it >> yield approximate results for the band gap. Sometimes quite decent, most >> often disastrous. >> >> Leo >> >> Em ter, 2 de abr de 2019 às 17:02, doaa haleem <doaa.halee...@gmail.com> >> escreveu: >> >>> Dear all >>> >>> Is Siesta could find the exact band gap for peroveskite structures by >>> using any of GGA only or GGA+U ? >>> Or it will still underestimate ? >>> >>> >>>
