Re: atomic structure for NEGF ( No.1 ) |
- Date: 2015/02/11 23:09
- Name: Darek <dariusz.chrobak@us.edu.pl>
- Described above problem appears when NEGF solver is used:
<Check_System> The system is bulk. lattice vectors (bohr) A = 47.376525000000, -1.353615000000, -1.353615000000 B = 0.000000000000, 10.828920000000, 0.000000000000 C = 0.000000000000, 0.000000000000, 10.828920000000
Non-orthogonal supercell was detected...
Band solver produces correct atomic arrangements in orthogonal supercell.
Best regards, Darek
|
Re: atomic structure for NEGF ( No.2 ) |
- Date: 2015/02/17 20:44
- Name: Artem Pulkin <artem.pulkin@epfl.ch>
- The only thing I can imagine is that you made some mistake previously (for example, used a different atomic configuration for the leads), fixed the input file but did not restart the calculation from scratch. Try re-calculating .hks files for the leads.
Artem
|
Re: atomic structure for NEGF ( No.3 ) |
- Date: 2015/02/19 00:48
- Name: Darek <dariusz.chrobak@us.edu.pl>
- Hi Artem,
First of all thank you very much for your response. I did not suppose that some of OpenMX users would like to study all the atomic coordinates provided in my first message :)
Going back to the problem, I would like to calculate I-V characteristics of ZB|RS junction. Trying to put into my system the same leads, I can imagine that my system will be composed as follows: ...ZBZB|RS|ZBZB..., ie. two ZB|RS junctions in the central region. So, the leads should posses different crystallographic structure in my case.
A few months ago I have done 'similar' calculation but for silicon. The left lead was diamond-Si while the right lead was also diamond-Si but compressed in a-direction. Between these two different structures (cubic and tetragonal unit cells) I placed intermediate structure with tetragonality that changes from 1 to that defined for the right lead. In spite of having not the same leads I obtained reasonable results.
Concerning hks files and atomic configuration of the leads, I think they are good.
Kind regards, Darek
![メンテ](./img/mente.gif) |
Re: atomic structure for NEGF ( No.4 ) |
- Date: 2015/02/20 16:37
- Name: T. Ozaki
- Hi,
The violation of orthogonality in the unit cell in the NEGF calculation comes from a fact that Grid_Origin in the band calculation for the right lead differs from that for the left lead. In the band structure calculation, the central unit cell is placed so that the cell properly covers all the atoms, and the regular mesh for integration is generated in the unit cell. Since the atomic coordinates of the left and right leads are different from each other, the origins of the unit cell are also different. The origin of unit cell can be found from the standard output in the band structure calculation of the left lead as
Grid_Origin -1.233293666667 -1.233293666667 -1.233293666667
in the band calculation of the right leas as
Grid_Origin -2.586908666667 -2.586908666667 -2.586908666667
They are apparently different.
On the other hand, the unit cell in the NEGF calculation is generated by merging the left and right unit cells with the grid origin determined by the band calculations, and by considering the relative atomic positions. Therefore, if the grid origin of the right lead does not match with that of the left lead, the merging leads to the displaced cell in the NEGF calculation. This is the reason why you got the non-orthogonal cell. To avoid the situation, you only have to include the following keyword for the band structure calculation of the right lead:
scf.fixed.grid -1.233293666667 -1.233293666667 -1.233293666667
Then, you can keep the same grid origin for the right lead as for the left lead, and obtain the orthogonal cell in the NEGF calculation.
Regards,
TO
![メンテ](./img/mente.gif) |
Re: atomic structure for NEGF ( No.5 ) |
- Date: 2015/02/20 19:31
- Name: Darek <dariusz.chrobak@us.edu.pl>
- Dear Ozaki-sensei,
Dear Artem,
My faith in interpersonal assistance has increased significantly. It works!
Kind regards, Darek
|
|