[Wien] Negative strain energy due to isoelectronic substitution

[Oleg Rubel]

Dear Wien2k users,

would you please exchange your experience on calculation of the strain
energy (sometimes also called 'formation enthalpy') of an alloy due to
isoelectronic substitutional defects.

I calculate a 64-atom GaP:B zinc-blend supercell. GaP:B means that one Ga
atom is replaced my B atom. I let the supercell keep the equilibrium
volume of GaP and relax internal degrees of freedom (atomic positions).
Obtained structure I call "relaxed structure". I would like to figure out:
What is the strain energy caused by substitutional B atom? For the stain
energy I use the following definition [see e.g., Phys. Rev. B 63, 085207
(2001)]:

Estr = Etot(GaP:B) - (1-x)*Etot(GaP) - x*Etot(BP),

where Etot(GaP:B) is the total energy of the relaxed structure, x is the
concentration of substitutional atoms (in my case it is 1/32), Etot(GaP)
and Etot(BP) are the total energies of 64-atom supercells of binary
compounds at their equilibrium lattice constants.

At the end I have:
Etot(GaP:B) = -142238.0807 Ry
Etot(GaP)   = -146071.2169 Ry
Etot(BP)    =  -23402.7274 Ry
which gives
Estr = -0.254 Ry.

This result surprises me due to its negative (!) sign. The strain energy
should be at least positive. Note: The same approach works fine on PW
codes (FHI98md). Maybe the problem is that the reference energy, with
respect which the total energy is measured, in Wien2k changes itself from
one structure to another? If such, would you please give me a suggestion
how to overcome this problem.

Thank you in advance.

Yours,

Oleg Rubel


ADDITIONAL INFORMATION: Wien2k version: WIEN2k_06.2 (Release 9/03/2006).
Run command ~>min -i 100 -s 10 -j 'run_lapw -p -I -i 40 -fc 1.0 ';
minimization method: PORT; LDA; Spin polarization - NO; Rmt(Ga)=1.92 Bohr;
Rmt(P)=1.85;  Rmt(B)=1.75; Kmax*Rmt_min = 7; 4x4x4 k-point mesh (shifted);
Lmax = 10 (in *.in1c); for the rest parameters I rely on default values.


===========================
Dr. Oleg Rubel
Faculty of Physics
Philipps University Marburg
Renthof 5, 35032 Marburg, Germany
E-mail: Oleg.Rubel at physik.uni-marburg.de