[Pw_forum] how to determine the mass of virtual-atom generated by virtual.x

[Stefano de Gironcoli]

A virtual atom is a ficticious object that you introduce with some 
purpouse...

Which mass (if any) is more appropriate depends on what is your purpouse.
If you want to relax the system any mass is good since it is not used.

If you plan to calculate phonons of a material then you should use the 
mass appropriate for it. In certain cases (GaAlAs alloys or 
superlattices for instance) it is a very good approximation to calculate 
the interatomic force constants with ana average atom and then study the 
big systems introducing the disorder by assigning the "true" Ga and Al 
masses to electronically equivalent virtual atoms (this approximation is 
called Mass Approximation)
[see for instance
S.~Baroni, S.~de~Gironcoli, and P.~Giannozzi,
{\sl Phonon dispersions in Ga$_{x}$Al$_{1-x}$As alloys},
Phys.\ Rev.\ Lett.\ {\bf 65}, 84 (1990) ;
E.~Molinari, S.~Baroni, P.~Giannozzi, and S.~de~Gironcoli,
{\sl Effects of disorder on the Raman spectra of GaAs/AlAs superlattices},
Phys.\ Rev.\ B {\bf 45}, 4280 (1992);
G.~Scamarcio, L.~Tapfer, W.~K\"onig, A.~Fisher, K.~Ploog, E.~Molinari,
S.~Baroni, P.~Giannozzi, and S.~de~Gironcoli,
{\sl Infrared reflectivity by TO phonons in (GaAs)$_{m}$(AlAs)$_{n}$
ultrathin superlattices},
Phys.\ Rev.\ B (RC) {\bf 43}, 14754 (1991);
C.~Bungaro, and S.~de~Gironcoli,
{\sl Ab initio study of phonons in Al$_x$Ga$_{1-x}$N alloys},
Appl.\ Phys.\ Lett.\ {\bf 76}, 2101 (2000). ]

I'm sure that there are systems where no choice is reasonable.

stefano

wangjunjie1981_0 wrote:
> Dear Stefano
> Thank you for your answer.
> I have another question that how to determine the mass of virtual-atom potential generated by virtual.x. it is equal to the mass of the main atom, or (1-x) M1+x M2, M1 is mass of potential1 and M2 is mass of potential2.
>  Thank you for your advice 
> Best regards
> Junjie Wang