UT theoretical chemistry code forum

Hi,

Recently, I used the dosanalyze.pl script to calculate the d-band center. I found that in the latest version of the dosanalyze.pl script, the width of the quarter-height of maximum DOS times a value of "w" (the defult value of "w" is set to 2.5) is used to obtain a new integrated range to calculate the band center. If the value of "w" is sufficently large, the integrated range will be the full range. And if the band center shifts by more than 10% compared to full integrated range, the script will print a warning on the screen.

So, my question is , why don't you simply use the full integrated range for the calculations? Is there any physical origin of the width of the quarter-height of maximum DOS and the "w" in your script ?

Thanks.

Sincerely

andyfan

We may not have the script quite right yet. The simplest solution is to integrate over the entire range. The reason we changed it is because of significant contributions to unoccupied states in d-band calculations. The current scheme aims to integrate a symmetric region around the d-band, even if the width of the window is fairly arbitrary (set by the value of w). The warning is given if the integration over the window differs from the full range, indicating that something might be wrong with the automatic window.

If you have any suggestions, we could change this. I can see the merits of going back to a simple integration over the full range with more sophisticated options available.

Hi,
Thanks for your kind reply.

Do you mean that the d-band center should be the center of the occupied states of the d band?

As I know, the d-band center should be the center of the entire d band (correct me if i am wrong). so i think the unoccupied states should not be neglected in the calculation, the integration over the full range may be more appropriate.


SIncerely,

andyfan

Right, it should be over the entire band. But there has to be some limit to the energy of the unoccupied states included in the average. Otherwise, as you add more unoccupied states in your calculation, the center of the d-band will continue to increase.

Yes. I agree with you about "there has to be some limit to the energy of the unoccupied states included in the average."
But the VASP code has already limited the energy range (e.g., the defult is from -10 to 10 eV). So I think the "w" value in your script may be unnecessary, since the "w" is an uncertain value and we don't know how to determine it.

Thanks a lot for your helpful reply.

According to the definition of the d-band center for surface science and catalysis, I think the integrated range should only include the occupied d states, not the entire d band range.