Questions about dosanalyze.pl

[jianmin]

Dear guys,

I tried to get the d-band center of Pd(111) surface and Pt(111) surface. I found that in the dosanalyze.pl script, "cut=0.25*$maxDOS", which means that the width of the quarter-height of maximum DOS times a value of "w" (the default value of "w" is 2.5) is used as the integrated range to calculate the band center. If I use this default, the calculated d-band center is oscilating with different NBANDS values. But if I change it to "cut=0.10*$maxDOS", the calcuated d-band center becomes less sensitive to NBANDS.

My question are: Is there a special reason to choose cut=0.25*$maxDOS? Can I change it to some other value, like 0.10 for example?

Thank you for your time.

[Wenjie]

NBANDS changes the number of empty bands in your calculation. It shouldn't change the d-band center so much when you used same cutoff. However, there is no special reason to choose cut=0.25*$maxDOS. You can choose a proper integration range for your own system.

We've updated the dosanalyze.pl script so that the default is to integrate the whole band. You can also indicate an integration range with this new script. We'll put the new script online as soon as possible so you can download it.

[jianmin]

Hi Wenjie, thank you very much for your answer.
Though, I still found that the values of NBANDS has a big effect on the calculated d-band center by the dosanalyze.pl.

[graeme]

We have recently been fiddling with this script. Our attempt at defining an automatic integration window seemed like a good idea, but we have basically decided that it is not robust enough to use as the default.

Integrating over the entire energy range in the DOS is at least robust, but as you are saying, the occupied bands can significantly effect the center of the d-band. I am leaning towards a strategy where you specify a fixed range which contains most of the DOS. With a fixed window, the number of states should not significantly effect the d-band center, since a higher value of nbands will increase the number of unoccupied states at the high energy range of the DOS.

[jianmin]

Hi Prof. Graeme, thanks a lot for your note.
Is the new version of dosanalyze.pl script available for us at the moment?