[Wien] dSCF vs Slater's transition state for XPS shifts

[pavel.ondracka at email.cz]

Dear Wien2k mailing list,

recently I've become interested in calculating XPS shifts for some TiO2-
based compounds. For some initial testing I've took plain rutile and 
anatase, made a supercells around 100 atoms and calculated shifts of core O 
1s state to get familiar with the process. I've tried two approaches: 
deltaSCF and the Slater's transition state as suggested for example here: 
http://www.mail-archive.com/wien%40zeus.theochem.tuwien.ac.at/msg01128.html 
.
The Slater's transition state approach seems to work fine and the rutile O 1
s state is predicted 0.38eV deeper than in the anatase, in accordance with 
experimental value (0.4eV [1]), however for deltaSCF approach I get 
completely different trend and the rutile O 1s state is predicted 1eV above 
anatase O 1s.
Since in all papers and tutorials I've read the both approaches should give 
very similar results, it is quite bothering me, why I'm getting this 
difference. Can anyone suggest a reason for this discrepancy? The cases were
converged with respect to number of k-points as well as Rkm and the 
supercell size should also hopefully be enough... I've also tried LDA vs 
PBE, however the result were the same. 
Any ideas?

Best regards
Pavel Ondračka

[1] Pfeifer, V., Erhart, P., Li, S., Rachut, K., Morasch, J., Brötz, J., … 
Klein, A. (2013). Energy Band Alignment between Anatase and Rutile TiO 2. 
The Journal of Physical Chemistry Letters, 4(23), 4182–4187. http://doi.org/
10.1021/jz402165b
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20151008/2227d9e9/attachment.html>