self consistent c parameter for MBJ

[psriv]

Dear Admin and VASP users,

Please suggest :

1) Is there any way to calculate c-parameter in MBJ self-consistently?
2) Is it possible to perform MBJ calculation for slab or nanowire?

Thanks in advance.

Best regards,
PS

[admin]

In the parameter-free MBJLDA calculation, the c-value is chosen to depend
linearly on the square root of the average of |∇ρ|/ρ over the unit cell volume
and is self-consistently determined. One can also keep the parameter c fixed
and compare the calculated band gap with the experimental value. By varying
c and comparying the calculated band gaps of many materials with experiment,
an optimized value c opt of 1.1-1.3 and 1.4-1.7 for solids with small and large
band gaps, respectively, has been proposed. [1] Note, for an exact reproduction
of the experimental band gap a fit of c is required.
N.B. The MBJLDA is just an XC potential, not an XC energy functional.
Thus E XC is taken from LSDA and the forces cannot be used from such calcu-
lations.
We recommend the following steps to perform the calculation:
1. usual DFT run including LASPH = .TRUE. in the INCAR file
2. MBJLDA calculation including the following flags in the INCAR file:
METAGGA = MBJ
CMBJ = 1.3 ! for MBJLDA calculation with fixed c value only
LASPH = .TRUE.
If the CMBJ flag is set in the INCAR file, the c value is kept constant and not
self-consistently determined.

Known issues
• The MBJLDA method was not meant for surface calculations, where the
electron density ρ and correspondingly the kinetic energy density τ can
become close to zero. Thus surface or slab calculations will diverge.
• MBJLDA calculations require POTCAR files including the kinetic energy
density term for the valence electrons (potpaw rc or potpaw PBE rc).
• For fast convergence it is suggested to use
ALGO = N
IMIX = 1 ! Kerker mixing

References
[1] F. Tran and P. Blaha, Phys. Rev. Lett. 102, 226401 (2009)

[psriv]

Dear Admin,

Thank you so much. Recently, I have found few works on using MBJ-LDA for surface calculation. They used Abinit-code.

P. V. Smith, et al. RSC Adv, 4, 48245 (2014).

Best regards,
PS