potpaw_PBE for Zr

[butterfly]

Hello,
I have two questions about the PAW_PBE psudopotentials for Zr element.
1. We know that Zr atom don't have 5p electrons. Why the valence electrons for PAW_PBE_Zr are chosed as 5s4d5p?
2. Are there anyone know the differece of PAW_PBE for the standard version and Zr_sv version? Or are there anyone know the author of the two PAW_PBE psudopotential for Zr element?
Thanks in advance!

[admin]

5s4d5p are not electrons but atomic energy elvels (orbitals). Number of electrons is 4 (ZVAL in POTCAR).
Zr_sv pseudopotential comprises electrons of the previous sphere 4s4p. Total number of electrons is therefore 12.
Author of pseudopotentials is Georg Kresse.

[butterfly]

Thanks for your pateint reply.
I'm still confuse about in which condition it's OK to choose Zr(5s4d5p) and in which condition we must choose Zr_sv. Because the latter may increase the computation cost. Are there some reference articles about the different suitable situation of the two psudopotentials for Zr?
Thanks for your kindly help.

<span class='smallblacktext'>[ Edited Fri Nov 18 2011, 11:18AM ]</span>

[admin]

Hard potentiasl are cheaper, potentials with semicore states are more expensive. Use potentials with semicore states when these states can play a role. In very simple terms: ionic properties are well described by hard potentials - covalent properties are better described by potentials with semicore states (including of semicore states increases the basis set).

[law5son]

I do not see a 4 electron potpaw_PBE for Zr on the current distribution (potpaw_PBE.54.tar.gz). I only see Zr_sv (12 electrons) and Zr_sv_GW (12 electrons). Is a 4 electron version available for Zr? This would speed up our calculations considerably.