[Pw_forum] Does VCA method give positive formation energy?

[GAO Zhe]

 Dear QE developers and users:
Hi.
I am using Quantum-ESPRESSO v4.2.1 to deal with DFPT and QHA. Because of the computational limitation of my PC, I chose VCA method to compare the phonon dispersion and thermo-properties change as a function of compositions.
The system I calculated was ZrC-ZrN. Since based on our experiments, we have known that the Zr(CN) solid-solution is possible to produce. PW91 xc-functional has been selected at 13x13x13 k-points mesh with ecutwfc=40Ry and ecutrho=400Ry. The q-mesh was 4x4x4. Structure of ZrC, ZrN and Zr(CN) is Fm-3m (NaCl Structure). The Vanderbilt ultrasoft pseudo-potential of C and N were combined by virtual.x in certain ratio.
However, I found the formation energy:
Ff[T, Zr(CNx)] = F[T, Zr(CNx)] - (1-x) * F[T, ZrC] - x * F[T, ZrN]
was always positive number.
Then, I checked ZrC, ZrN and Zr(CN0.5) carefully by both of 'vc-relax' and E-V curve for total energy calculation. Finally, I found thatE[Zr(CN0.5)] - 0.5 * E[ZrC] - 0.5 * E[ZrN] was about +0.4Ry/cell, that means+525.088kJ/mol. According to thermodynamics principle, Zr(CN0.5) cannot be prepared. However, when I tried supercell-doping method, the formation energy was negative ( about-10kJ/mol ).
Therefore, I want to know why VCA method gives positive formation energy. Is it reasonable? If it should be like this, then can we believe the phonon result from VCA as reliable one?
Any suggestion will be welcomed. Thanks.


--
GAO Zhe
CMC Lab, MSE, SNU, Seoul, S.Korea
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20110923/b7d640da/attachment.html>