shankar1729
commented
3 months ago I could reproduce your issue with the PSlib pseudopotential in the attached results. I did adjust the plane wave cutoffs to half of what you had since the recommendations in the UPF are in Ry, while JDFTx uses Eh. I also used U = 0.075 Eh that we had fit to ARPES in an old work, but that's not too far from the values you tried. All these don't make a difference though, and the bandstructures are bad.
Seems like that PSlib pseudopotential is bad. I also attach results for the PBEsol FR NC pseudopotential from DOJO, with everything else the same, and it gives a reasonable looking bandstructure.
Best, Shankar
wmatter
I have been trying to do a DFT+U calculation for solid Au using the pseudopotential https://pseudopotentials.quantum-espresso.org/upf_files/Au.rel-pbesol-spn-rrkjus_psl.1.0.0.UPF
I have tried using a variety of Hubbard U parameters, including U=-0.0583 Hartrees from linear response and U=2.5 eV from optimizing the energy of the d-band to match literature. I am running into an issue where the band structure (attached) intersects the Fermi level to the right of the L symmetry point, meaning that the Fermi surface doesn’t intersect the Bragg plane as would be expected for the noble metals. This issue doesn’t seem to resolve no matter what I have been trying (input file attached).
Ultimately we are aiming to produce an accurate DOS with the correct d-band energy and with a relativistic pseudopotential and XC functional. Any suggestions would be greatly appreciated. Many thanks.
Au.in.txt