Dear Wenjie Tang and Graeme Henkelman,
I plan on including a density-of-states analysis based on the Bader volumes in one of my upcoming papers. Therefore, I wonder if there is any more references you would like me to include for the projection scheme in addition to the three references for the Bader analysis given at http://theory.cm.utexas.edu/vtsttools/bader/ ? At the moment there is no reference information specified at http://theory.cm.utexas.edu/vtsttools/dos/ .
Best regards,
/Dan Fors
DOS projection in the Bader volumes: References?
Moderator: moderators
Re: DOS projection in the Bader volumes: References?
We don't actually have a publication about this. But thank you for asking!
Did you find any significant differences using the Bader volumes for the DOS calculations instead of (for example) spheres around the atoms? We have not tested this extensively, but in the metal surfaces that we have looked at so far, there does not seem to be much difference. I'm curious if you found a case where the Bader partitioning improved the DOS analysis, or alternatively, caused problems.
Did you find any significant differences using the Bader volumes for the DOS calculations instead of (for example) spheres around the atoms? We have not tested this extensively, but in the metal surfaces that we have looked at so far, there does not seem to be much difference. I'm curious if you found a case where the Bader partitioning improved the DOS analysis, or alternatively, caused problems.
Re: DOS projection in the Bader volumes: References?
Dear Graeme Henkelman,
Thank you very much for your response. I will then settle by referring to your implementation in the text, and give a reference to the homepage for further information.
Regarding your questions:
I have compared two approaches:
i) A DOS analysis based on the Bader volumes,
ii) A DOS analysis where the ratios between the volumes of the projection spheres were determined from an analysis of the ratios between the Bader volumes. The spheres were made as large as possible without overlap.
The two approaches have been applied to binary bulk systems and metal/ceramic interfaces. From the analyzes I find that the two approaches in most cases give the same qualitative results, where 1st approach only affects the absolute values of the DOS. I can therefore draw the same conclusions about the bond mechanisms for both approaches.
However, for certain interface geometries where the interface distances are rather different than in the bulk phases, I have noticed an advantage with the Bader partitioning. In the 2nd approach the trade-off between keeping the projections ratios similar to the bulk values while minimizing the overlap between the projections spheres (especially across the interface) as well as being consistent for different translation states, occasionally lead to that a large portion of the interstitial regions and an important part of the interface region are being left out. This trade-off leads to a less clear picture of the bond mechanisms at the interfaces. Usually, I can circumvent this by multiple modifications of the projection radii in the 2nd approach and doing some systematic comparisons, but there is an arbitrariness in the analysis. For my interface systems, the Bader partitioning therefore seems to provide a more consistent picture and have in particular simplified the comparison between different translation states where the interphase distances have large variations. At least this is my impression.
I have not noticed any case where the Bader partitioning have caused problems for DOS analysis.
Best regards,
/Dan Fors
Thank you very much for your response. I will then settle by referring to your implementation in the text, and give a reference to the homepage for further information.
Regarding your questions:
I have compared two approaches:
i) A DOS analysis based on the Bader volumes,
ii) A DOS analysis where the ratios between the volumes of the projection spheres were determined from an analysis of the ratios between the Bader volumes. The spheres were made as large as possible without overlap.
The two approaches have been applied to binary bulk systems and metal/ceramic interfaces. From the analyzes I find that the two approaches in most cases give the same qualitative results, where 1st approach only affects the absolute values of the DOS. I can therefore draw the same conclusions about the bond mechanisms for both approaches.
However, for certain interface geometries where the interface distances are rather different than in the bulk phases, I have noticed an advantage with the Bader partitioning. In the 2nd approach the trade-off between keeping the projections ratios similar to the bulk values while minimizing the overlap between the projections spheres (especially across the interface) as well as being consistent for different translation states, occasionally lead to that a large portion of the interstitial regions and an important part of the interface region are being left out. This trade-off leads to a less clear picture of the bond mechanisms at the interfaces. Usually, I can circumvent this by multiple modifications of the projection radii in the 2nd approach and doing some systematic comparisons, but there is an arbitrariness in the analysis. For my interface systems, the Bader partitioning therefore seems to provide a more consistent picture and have in particular simplified the comparison between different translation states where the interphase distances have large variations. At least this is my impression.
I have not noticed any case where the Bader partitioning have caused problems for DOS analysis.
Best regards,
/Dan Fors