Wrong gap on 1H-MoS2

Concerns issues with computing quasiparticle corrections to the DFT eigenvalues - i.e., the self-energy within the GW approximation (-g n), or considering the Hartree-Fock exchange only (-x)

Moderators: Davide Sangalli, andrea.ferretti, myrta gruning, andrea marini, Daniele Varsano

Post Reply
Nhat Minh Pham
Posts: 14
Joined: Tue Feb 15, 2022 12:07 am
Location: 01 Mac Dinh Chi, Ben Nghe, District 1, Ho Chi Minh City, Viet Nam

Wrong gap on 1H-MoS2

Post by Nhat Minh Pham » Mon Feb 21, 2022 7:23 am

Dear developers,

I'm calculating the band structure of 1H-MoS2 monolayer using G0W0 corrections (Yambo-5.0.4) and parameters from PhysRevB.88.045412, but it results in a large gap of about 3 eV, compared to 2,41 eV from the paper.
To me, I think I used good parameters and PP (SG15 ONCV), the only parameter I'm concerned about is the cut box size, from discussions in the forum I use 39 a.u. (my system is 21 Angstrom in Z direction).
I realize the gap would be significantly reduced by reducing the cut box to about 8 a.u. I don't know what should I do.
I wonder if you'd mind taking a look at my QE inputs and Yambo reports?

Best,
Minh
You do not have the required permissions to view the files attached to this post.
Pham Nhat Minh
MS student, Ho Chi Minh University of Science,
Ho Chi Minh City Institute of Physics
https://www.hcmus.edu.vn/

User avatar
Daniele Varsano
Posts: 3773
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: Wrong gap on 1H-MoS2

Post by Daniele Varsano » Mon Feb 21, 2022 9:56 am

Dear Minh,

the cutoff size is correct. Looking at your input:
VXCRLvcs is a very small value, you should use all the g vectors of the density (77001 RL).
50 bands also seem to me it is not at convergence as well as NGsBlkXp
Aso the k point sampling should be carefully checked.

Anyway, someone expert on the system can probably provide you with more details.

Best,
Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

User avatar
palful
Posts: 56
Joined: Tue Jan 26, 2016 11:23 am
Location: Modena and Milan

Re: Wrong gap on 1H-MoS2

Post by palful » Mon Feb 21, 2022 11:36 am

Dear Minh,

I agree with Daniele.

In order to get reasonably converged results, I think that the kpoint grid should be between 39x39x1 and 42x42x1.
In addition, for NGsBlkXp you should put at least 100 bands, probably more.

The band gap dependence on the cutoff that you observed usually happens when the kgrid is not well converged. Indeed, the various convergence parameters are not exactly independent from one another but have a degree of interdependence.

Cheers,
Fulvio
Dr. Fulvio Paleari
S3-CNR Institute of Nanoscience and MaX Center
Modena, Italy

User avatar
Daniele Varsano
Posts: 3773
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: Wrong gap on 1H-MoS2

Post by Daniele Varsano » Mon Feb 21, 2022 11:45 am

Dear Minh,

please have also a look to FIg.4 of Phys. Rev. B 94, 155406.
We have recently implemented a very efficinet scheme to accelerate convergence wrt k points for 2D materials that will released very soon.

Best,
Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

Nhat Minh Pham
Posts: 14
Joined: Tue Feb 15, 2022 12:07 am
Location: 01 Mac Dinh Chi, Ben Nghe, District 1, Ho Chi Minh City, Viet Nam

Re: Wrong gap on 1H-MoS2

Post by Nhat Minh Pham » Fri Mar 04, 2022 5:34 pm

Dear Fulvio and Daniele,

Thank you for helping, after increasing K-grid to 36x36 and changing lattice parameters to relaxed values I get a gap of about 2.65 eV.

I want to include excitonic effects in nonlinear optical calculations for the material, but as I've read monolayer MoS2 has a large exciton binding energy.
So I don't know if it is sufficient when I only include the exchange term <cvk|V|c'v'k'> and ignore the screened Coulomb <cvk|W|c'v'k'>?

In Yambo to Wannier90 procedure, I see there is an update in the overlap matrices from DFT to GW.
I don't understand how it's done because GW uses KS orbitals as the basis.
With this misunderstanding, I'm wondering should there be an update for Coulomb matrix elements?
Because I think <cvk|V|c'v'k'> is calculated with KS states and if the overlap matrices are changed using some QP states, Coulomb matrices should also use these QP states.
I'm looking for your help.

Best,
Minh
Pham Nhat Minh
MS student, Ho Chi Minh University of Science,
Ho Chi Minh City Institute of Physics
https://www.hcmus.edu.vn/

User avatar
Daniele Varsano
Posts: 3773
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: Wrong gap on 1H-MoS2

Post by Daniele Varsano » Sat Mar 05, 2022 7:02 am

Dear Minh,
So I don't know if it is sufficient when I only include the exchange term <cvk|V|c'v'k'> and ignore the screened Coulomb <cvk|W|c'v'k'>?
Of course in this way you are missing the excitonic effects.
Because I think <cvk|V|c'v'k'> is calculated with KS states and if the overlap matrices are changed using some QP states, Coulomb matrices should also use these QP states.
In GoWo procedure wavefunctions are not changed, there is the assumption that Psi_QP ~ Psi_KS and in general the approximation is valid.

Best,
Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

Nhat Minh Pham
Posts: 14
Joined: Tue Feb 15, 2022 12:07 am
Location: 01 Mac Dinh Chi, Ben Nghe, District 1, Ho Chi Minh City, Viet Nam

Re: Wrong gap on 1H-MoS2

Post by Nhat Minh Pham » Sun Mar 06, 2022 4:22 pm

Dear Daniele,

I still don't understand why overlap matrices are updated if we assume that Psi_QP ~ Psi_KS.
I think if we have the same k-grid and band numbering, overlap matrices will stay the same as DFT overlap matrices. Is that correct?

Best,
Minh
Pham Nhat Minh
MS student, Ho Chi Minh University of Science,
Ho Chi Minh City Institute of Physics
https://www.hcmus.edu.vn/

AntimoMarrazzo
Posts: 19
Joined: Fri Jan 22, 2016 2:46 pm

Re: Wrong gap on 1H-MoS2

Post by AntimoMarrazzo » Mon Mar 07, 2022 11:46 am

I still don't understand why overlap matrices are updated if we assume that Psi_QP ~ Psi_KS.
I think if we have the same k-grid and band numbering, overlap matrices will stay the same as DFT overlap matrices. Is that correct?
G0W0 can change the ordering of the eigenvalues, that means that the eigenvalues (and so the eigenstates) might not be ordered in energy anymore.
This is a problem for Wannier90, that assumes energy ordering in many steps, for instance when the energy windows are used.
Hence we reorder the indices of the KS orbitals: the orbitals themselves are unchanged, it's just the ordering that is updated.

HTH
Antimo Marrazzo
University of Trieste
Physics Department
Strada Costiera 11, 34151 Trieste (Italy)

Post Reply