I am trying to investigate superconductivity in a noncentrosymmetric compound which likely has a d-wave gap. However, I struggle to converge Tc, especially with respect to the number of bands. For example:
lbee=1 and fbee=nbnd=30 -> Tc=4.32 K
lbee=1 and fbee=nbnd=35 -> Tc=3.52 K
lbee=1 and fbee=nbnd=40 -> Tc=52.13 K
In all cases a nodal line is present. Fermi surface is built from bands 19-20. The first 6 bands are nondispersive and lay over 3 Ry below EF, yet including them affected Tc by 0.1 K.
The shifted q grid is 4x4x4 and has 12 irreducible q points. SCF calculations were done on 12x12x12 grid and nscf on 24x24x24. The unshifted 4x4x4 q grid has 21 irreducible points and is converged for phonons and electron-phonon interaction with interpolation method with smearing in Quantum Espresso. Electron-phonon coupling constant on the shifted grid plotted on the Fermi surface looks reasonable, with no jagged pieces, so I doubt the grid is too sparse for the tetrahedra method. Ecutfock also stronly affects Tc, but is converged for 80 Ry. nci/nmf above the default 5/10 changed the results only by 0.05 K.
Can SCTK deal with a d-wave gap? Few months ago I wrote about a bug that prevented calculations for noncentrosymmetric compounds. In a fixed version they complete, but maybe there is another subtle bug?
Dear developers,
I am trying to investigate superconductivity in a noncentrosymmetric compound which likely has a d-wave gap. However, I struggle to converge Tc, especially with respect to the number of bands. For example:
In all cases a nodal line is present. Fermi surface is built from bands 19-20. The first 6 bands are nondispersive and lay over 3 Ry below EF, yet including them affected Tc by 0.1 K.
The shifted q grid is 4x4x4 and has 12 irreducible q points. SCF calculations were done on 12x12x12 grid and nscf on 24x24x24. The unshifted 4x4x4 q grid has 21 irreducible points and is converged for phonons and electron-phonon interaction with interpolation method with smearing in Quantum Espresso. Electron-phonon coupling constant on the shifted grid plotted on the Fermi surface looks reasonable, with no jagged pieces, so I doubt the grid is too sparse for the tetrahedra method. Ecutfock also stronly affects Tc, but is converged for 80 Ry. nci/nmf above the default 5/10 changed the results only by 0.05 K.
Can SCTK deal with a d-wave gap? Few months ago I wrote about a bug that prevented calculations for noncentrosymmetric compounds. In a fixed version they complete, but maybe there is another subtle bug?
Best regards, Gabriel Kuderowicz