 Re: Description of keywords in DFT-D3 ( No.1 ) |
- Date: 2022/12/01 18:48
- Name: T. Ozaki
- Hi,
Such a specification for the keywords corresponds to the latter, i.e.,
Atom 1: GGA-PBE + DFT-D3 and Vdw interaction with the surrounding cells of the unit cell is also taken into account.
Atoms 2,3,4: GGA-PBE + DFT-D3 without considering Vdw interaction with surrounding cells of unit cell
I wonder that only the use of the keyword:
DFTD.IntDirection 1 1 1
might be better than specifying the interaction of each atom.
If you do not specify the keyword; DFTD.periodicity, the defalut value of 1 is set to all the atoms.
Regards,
TO
|
| Re: Description of keywords in DFT-D3 ( No.2 ) |
- Date: 2022/12/01 19:30
- Name: Masanobu Miyata <m-miyata@jaist.ac.jp>
- Dear Ozaki-sensei,
This is Miyata.
Thank you very much for prompt reply.
I performed structural optimization for phonon calculation of Sb2Te3 using DFT-D3, and found that the Max force is about 1/100 smaller in condition B than in condition A.
Condition A
Basis set: Standard
Conventional cell
MD.type OptC4
IntDirection 1 1 1
Condition B
Basis set: Standard
Conventional cell
MD.type OptC4
DFTD.IntDirection 1 1 1
<DFTD.periodicity
1 0
2 0
3 0
4 0
5 0
6 0
7 1
8 1
9 1
10 1
11 1
12 1
13 0
14 0
15 0
DFTD.periodicity>
# 1-6: Sb 6c site
#7-12: Te 6c site
#13-15: Te 3a site
In Sb2Te3, the Te-Te bond at the 6c site is Vdw coupled, and the covalent bond between Sb at the 6c site and Te at the 3a site in the plane is dominant.
I think that If the long-range interactions outside the cell was included for Sb at 6c and Te at 3a, this is double counting the original covalent bonding and the long-range false Vdw interactions.
Therefore, I excluded Sb and Te at site 3a for the long-range interactions outside the cell.
Is the calculation condition B not suitable?
Best regareds,
Miyata,
 |
| Re: Description of keywords in DFT-D3 ( No.3 ) |
- Date: 2022/12/02 21:19
- Name: T. Ozaki
- Dear Miyata-san,
I am not sure which treatment is better for your specific case.
But, I wonder that the original DFTD-D3 has been developed so that the structures and energetics
under interactions from the periodic cells in the bulk can be well compared to the corresponding
results calulated by more accurate methods. As for the issue, please check the original paper by Grimme et al.,
https://doi.org/10.1063/1.3382344
Recently I also performed the geometry optimization with variable cell for bulk MoS2 within DFT-D3
using deverloper's version of OpenMX, and reproduced the published computational result reported in
https://doi.org/10.1088/0953-8984/26/30/305502
It turned out that the cell paremeters I obtained are well compared to the experimental ones.
The calculation was performed by the condition A you mentioned.
(Note that the geometry optimization with variable cell is not supported for the DFT-D3 in the public
version of OpenMX.)
So, this implies that the DFTD-D3 has been developed so that the structures and energetics
under interactions from the periodic cells in the bulk can be reproduced.
Of course, we need to eliminate the spurious vdW interaction if the system of interest is
intrinsically 0D, 1D, or 2D.
I hope that my experience may help you.
Regards,
TO
|
| Re: Description of keywords in DFT-D3 ( No.4 ) |
- Date: 2022/12/02 19:17
- Name: Masanobu Miyata <m-miyata@jaist.ac.jp>
- Dear Ozaki-sensei,
This is Miyata.
Thank you very much for your detailed response.
This calculation was performed with OpenMX 3.9.9. The patch mentioned that cell variable structure optimization was implemented in DFT-D3.
Related to DFTDvdW_init.c and DFT3DvdW_init.c:
Variable cell optimizations are supported.
Since DFT-D3 is designed to reproduce the energy and structure of periodic cells, condition A is more in line with the principle.
In order to perform phonon calculations, I needed to reduce the forces on the cell as much as possible while maintaining crystal symmetry. I don't know why, but condition B reduced the force on the cell to about 1/100 of that of condition A.
I thought that by adjusting the value of DFTD.rcut_dftD, I might be able to get good convergence of forces in Condition A.
Since we were dealing with Sb2Te3 with 3R type structure, I had taken 100 Bohr as the cutoff for my calculation to account for the interaction of all three layers.
Thank you very much for your valuable discussion.
Best regards,
Miyata
|
| Re: Description of keywords in DFT-D3 ( No.5 ) |
- Date: 2022/12/02 21:20
- Name: T. Ozaki
- Dear Miyata-san,
> This calculation was performed with OpenMX 3.9.9. The patch mentioned that cell variable structure optimization was implemented in DFT-D3.
Yes, you are right. I did not make an official announcement except in README.txt, and thought that nobody recognized it.
Anyway, the patch 3.9.9 supports the variable cell optimization.
> In order to perform phonon calculations, I needed to reduce the forces on the cell as much as possible while maintaining crystal symmetry.
> I don't know why, but condition B reduced the force on the cell to about 1/100 of that of condition A.
From the computational point of view, the forces on atoms and stress tensor should be miminized by the variable cell optimization
for either cases, and the final force and stress must be small enough after achieving the convergence, though it must be a system-dependent
issue how the preservation of the crystal symmetry is related to how the vdW interaction is included.
> I thought that by adjusting the value of DFTD.rcut_dftD, I might be able to get good convergence of forces in Condition A.
In principle, we should not regard the cutoff radius as an adjustable parameter, but take a large value which guarantee the convergence
of the energy as a function of the cutoff radius.
Regards,
TO
|
| Re: Description of keywords in DFT-D3 ( No.6 ) |
- Date: 2022/12/02 22:09
- Name: Masanobu Miyata <m-miyata@jaist.ac.jp>
- Dear Ozaki-sensei,
Thank you very much for your kind response.
I will examine the conditions for structural optimization more carefully based on condition A. I will also try to make the the value of DFTD.rcut_dftD for DFT-D3 large enough.
Thank you very much for your valuable discussion.
Best regards,
Miyata
|
|
This thread is locked.Only browsing is available.
Description of keywords in DFT-D3