[Q] band structure calculation with spin-orbit interaction
Posted: Tue Sep 07, 2010 4:21 pm
Dear VASPers,
I am trying to calculate the band structure of Bi2Se3
with Spin-orbit coupling/interaction (SOC) included. What I have done for that is:
i) Optimization & SCF without SOC
ii) Generation of CHGCAR by SCF as usual, with and without SOC
iii) From the CHGCAR, band structure calculation with line mode, with SOC
But I have some troubles with the result...
What I would like to know is
[1] Correct way of including SOC for non-magnetic system
I just add the following lines to my usual INCAR:
ISPIN = 1 # this will be ignored, so I have left this line
LSORBIT = .TRUE.
LORBMOM = .TRUE.
LMAXMIX=4
GGA_COMPAT=.FALSE.
NBANDS=XXX # double the number of non-SOC calculation
I hope that this will set
SAXIS = 0+ 0 1, and MAGMOM = 0 0 0 (or some other default values),
and the result will not be sensitive to the setting of SAXIS.
Is this setting legitimate?
Should I test the SAXIS dependence?
And,
I think the kpoint convergence with SOC might be quite demanding wrt the calculation without SOC.
In this point in mind, if I use the same set-up except LSORBIT options above, which one would be the most reasonable set-up,
i. Use CHGCAR from SCF with ISPIN=1 without SOC, and turn on the SOC option for band calculation only
ii. Use CHGCAR from SCF with ISPIN=1 with SOC, and turn on the SOC option for band calculation too
iii. Use CHGCAR from SCF with ISPIN=2 without SOC, and turn on the SOC option for band calculation only
iv. Use CHGCAR from SCF with ISPIN=2 with SOC, and turn on the SOC option for band calculation too
[2] Discrepancy between TDOS and band structure
After the SCF is done with the above additional lines, I have TDOS.
But the consistency bewteen the band structure with the CHGCAR from the SCF and the TDOS is not so satisfactory.
I believe it is correct to use Fermi-level from my SCF calculation,
but the band structure is more reasonable when I set the Fermi-level as the value from the band structure calculation.
Which Fermi level should I use, E_f from the SCF of from the band structure calculation?
I know the DOSCAR is different when I include the SOC,
because total charge and the magnetization is accompanied.
But I just could not find any significant difference in EIGENVAL.
Is there anything to take care of during the band structure calculation with SOC? or during drawing the band structure?
Many thanks in advance for your advice (-_-) (_ _)
<span class='smallblacktext'>[ Edited ]</span>
I am trying to calculate the band structure of Bi2Se3
with Spin-orbit coupling/interaction (SOC) included. What I have done for that is:
i) Optimization & SCF without SOC
ii) Generation of CHGCAR by SCF as usual, with and without SOC
iii) From the CHGCAR, band structure calculation with line mode, with SOC
But I have some troubles with the result...
What I would like to know is
[1] Correct way of including SOC for non-magnetic system
I just add the following lines to my usual INCAR:
ISPIN = 1 # this will be ignored, so I have left this line
LSORBIT = .TRUE.
LORBMOM = .TRUE.
LMAXMIX=4
GGA_COMPAT=.FALSE.
NBANDS=XXX # double the number of non-SOC calculation
I hope that this will set
SAXIS = 0+ 0 1, and MAGMOM = 0 0 0 (or some other default values),
and the result will not be sensitive to the setting of SAXIS.
Is this setting legitimate?
Should I test the SAXIS dependence?
And,
I think the kpoint convergence with SOC might be quite demanding wrt the calculation without SOC.
In this point in mind, if I use the same set-up except LSORBIT options above, which one would be the most reasonable set-up,
i. Use CHGCAR from SCF with ISPIN=1 without SOC, and turn on the SOC option for band calculation only
ii. Use CHGCAR from SCF with ISPIN=1 with SOC, and turn on the SOC option for band calculation too
iii. Use CHGCAR from SCF with ISPIN=2 without SOC, and turn on the SOC option for band calculation only
iv. Use CHGCAR from SCF with ISPIN=2 with SOC, and turn on the SOC option for band calculation too
[2] Discrepancy between TDOS and band structure
After the SCF is done with the above additional lines, I have TDOS.
But the consistency bewteen the band structure with the CHGCAR from the SCF and the TDOS is not so satisfactory.
I believe it is correct to use Fermi-level from my SCF calculation,
but the band structure is more reasonable when I set the Fermi-level as the value from the band structure calculation.
Which Fermi level should I use, E_f from the SCF of from the band structure calculation?
I know the DOSCAR is different when I include the SOC,
because total charge and the magnetization is accompanied.
But I just could not find any significant difference in EIGENVAL.
Is there anything to take care of during the band structure calculation with SOC? or during drawing the band structure?
Many thanks in advance for your advice (-_-) (_ _)
<span class='smallblacktext'>[ Edited ]</span>