Band structure SOC and optB88-vdW of Klimeš et al.

Message

karthik_hj · #1 Post by **karthik_hj** » Wed Nov 22, 2023 1:07 pm

I was plotting the Band structure and DOS with and without SOC for an SnS2 hetero-structure. I already know from literature that they are semi-conducting. But when I use vdw tags also for DOS and BS calculations, it shows metallic nature. If I remove the tags, it turns semiconducting. My question is ,

1. Do we include vdw tags for Non-scf runs also i.e. DOS and BS calculations including SOC interactions? Or do we only give it initially for scf runs and then remove it when doing DOS and BS including SOC interactions
2. Why does vdw tags change the electronic properties like this?
3. Why doesn't it change when I do not give SOC tags

#2 Post by **fabien_tran1** » Wed Nov 22, 2023 1:52 pm

In order to help you, we need to know more details about the procedure that you used for the calculations. So, please describe precisely the steps of the calculations, and also show us the INCAR files of the SCF and non-SCF runs.

karthik_hj · #3 Post by **karthik_hj** » Thu Nov 23, 2023 7:12 am

Procedure1 - INCAR1 attached
Step1: Optimisation of structure with vdw (No SOC)
Step2: DOS and BS calculation with vdw and SOC

Procedure2 - INCAR2 attached
Step1: Optimisation of structure with vdw (No SOC)
Step2: DOS and BS calculation without vdw and SOC

Procedure 1 is showing metallic nature and 2 is showing semiconducting. Band diagrams attached as pdf's.

#4 Post by **fabien_tran1** » Thu Nov 23, 2023 9:02 am

This is still not completely clear what procedures you used for the calculations. In total, there are 4 calculations, right? Please, provide the INCAR for each of these 4 calculations, such that I can really understand what you did.

For the moment I can just say the following:
Step 2 of procedure 1 should lead to a crash, because starting a calculation with LSORBIT=.TRUE. from a WAVECAR file that was generated from a previous calculation with LSORBIT=.FALSE. is not possible.

karthik_hj · #5 Post by **karthik_hj** » Thu Nov 23, 2023 10:02 am

WAVECAR was not generated from the initial calculation. It ran as usual. The INCAR for the initial optimisations is attached. It is same for Step1 in procedure 1 and Step 1 in procedure 2

#6 Post by **fabien_tran1** » Thu Nov 23, 2023 2:50 pm

I still can not understand the procedures that you followed. Your explanations and naming of the INCAR files are confusing. Provide again all INCAR files with a proper naming and a clear description of the steps (and specify CLEARLY which INCAR is associated with which step).

karthik_hj · #7 Post by **karthik_hj** » Fri Nov 24, 2023 5:28 am

My question is; do we include vdw tags also for the band structure and dos calculations?

karthik_hj · #8 Post by **karthik_hj** » Fri Nov 24, 2023 5:33 am

Procedure1 - BS attached
Step1: Optimisation of structure with vdw (No SOC) INCAR1
Step2: DOS and BS calculation with vdw and SOC included INCAR2

Procedure2 - BS attached
Step1: Optimisation of structure with vdw (No SOC) INCAR3
Step2: DOS and BS calculation without vdw and SOC included INCAR4

#9 Post by **fabien_tran1** » Fri Nov 24, 2023 9:46 am

A few points:

1) The calculation that does not include vdW (INCAR4) uses the functional PBE. Is it really PBE that you want to use for calculating BS and DOS?

2) PBE and optB88-vdW should lead to very similar BS. What is more important for the BS is SOC.

3) I do not know why you get a metal from BS, but a semiconductor from procedure 2.

4) As explained for the MAGMOM tag:
-One value per atom has to be specified when LSORBIT=.FALSE.
-Three values per atom have to be specified when LSORBIT=.TRUE.
What is specified for MAGMOM is the same in all your INCAR files, and I guess that what you specified in INCAR2 and INCAR4 is wrong and should be (with magnetic moment oriented along z):
MAGMOM = 0.0 0.0 2.0 0.0 0.0 2.0 ...

Anyway, I suggest to follow these steps:

1) Do a geometry optimization with the optB88-vdW functional. If including SOC leads to a lengthy calculation, then do not include it. I don't think that SOC is so important for the geometry.

2) At the geometry obtained at step 1 (do "cp CONTCAR POSCAR"), do a SCF calculation (but no geometry optimization) with optB88-vdW and including SOC (and correct specification for MAGMOM). Include "LORBIT=11" in INCAR, but not "ICHARG=11".

karthik_hj · #10 Post by **karthik_hj** » Fri Nov 24, 2023 10:04 am

1. Yes I wish to use PBE only
I shall change the MAGMOM tag to what you have said. Could you please let me know the IBRION and ISIF to be put if no geometry optimisation is needed?

#11 Post by **fabien_tran1** » Fri Nov 24, 2023 10:15 am

As you did in INCAR2 and INCAR4: IBRION=-1 and NSW=0. It does not matter for ISIF. This is what is explained at IBRION, NSW and ISIF.

My Community

Band structure SOC and optB88-vdW of Klimeš et al.

Band structure SOC and optB88-vdW of Klimeš et al.

Re: Band structure SOC and optB88-vdW of Klimeš et al.

Re: Band structure SOC and optB88-vdW of Klimeš et al.

Re: Band structure SOC and optB88-vdW of Klimeš et al.

Re: Band structure SOC and optB88-vdW of Klimeš et al.

Re: Band structure SOC and optB88-vdW of Klimeš et al.

Re: Band structure SOC and optB88-vdW of Klimeš et al.

Re: Band structure SOC and optB88-vdW of Klimeš et al.

Re: Band structure SOC and optB88-vdW of Klimeš et al.

Re: Band structure SOC and optB88-vdW of Klimeš et al.

Re: Band structure SOC and optB88-vdW of Klimeš et al.