Hi everyone,
I am doing a calculation for the elastic constants of hexagonal AlN. Basically, I used both energy-strain and stress-strain methods. I run into trouble of getting inaccurate results for C12, C13. I got C12 = 100 and C13 = 60 GPa, however, the accurate values for these components should be C12 = 150 and C13 = 100 GPa. The results I got for C11, C33, and C44 are quite close to literature (within 5% error). I followed the procedure of this paper "Applied Physics 82, 2833 (1997)" for the calculation of elastic constants (via energy-strain method).
I think that this problem is not due to the unrelaxed structure since after doing relaxation the pressure on my structure is low enough (0.037 GPa). I try increasing ecutwfc up to 100 (ecutrho = 1000), however, it did not help.
I attached the input and output files of scf simulation (with zero strain applied). If anyone gets any idea of why I got the wrong values for C12 and C13, please give me your comment. Any help from you will be appreciated.
Thanks!
Hi, have you tried by setting ibrav=4 and the respective celldm(1), celldm(3)?
Hi again,
from the input file, I guess you are using the pbe functional. Have you tried pbesol? I have never used it myself (I use LDA on Siesta), but I have read that pbesol improves structural /elastic properties.
It is always good to try different potential types. I have actually tried PBE and LDA functionals, however, both of them gave the same result. Now I did try the PBESOL, it also did not help. I forgot to mention some other ways I have tried like increasing the size of the cell to (2x2x2) with a reasonable q-point mesh of (6x6x2), improving the convergence threshold (conv_thr =1d-10), increasing q mesh up to (20x20x20), and finally using different Quantum Espresso versions.
Checking for elastic constants and pwscf, I found this
https://people.sissa.it/~dalcorso/thermo_pw/user_guide/user_guide.html
check entry 4.12.
Maybe it is worth a try.
Have you tried asking on the pwscf mailing list?
Thank Giovanni for spending your time on this discussion. It took me some time to run the thermo_pw for the elastic constant. When I input the same input file as I used before (ibrav=0), the program shows some errors related to the symmetry. However, by using ibrav=4, everything was going in the right way, I got all the elastic constants close to the literature (smaller by 6-7% -> acceptable). The bad point of using the thermo_pw is that I could not trace back the theory behind as the manual is very limited.
I then do the same for GaN instead of AlN (using thermo_pw package), the same functional type of pseudopotential was used. This time I got all the elastic constants smaller than the literature by approximately 20%. Definitely, this might be linked to the use of pseudopotential.
Anyway, your comments are very helpful and I really appreciate. Thanks!
Hi, maybe for the theory behind this program, you can contact the author directly. I think his name is Dal Corso
https://dalcorso.github.io/thermo_pw/
Dear Dat Tran
You mentioned that you used thermo_pw to calculate the elastic constant and get the not bad result. But I faced with some problem in elastic constants through thermo_pw.
Could you please take a look of my file?
Thanks a lot if you can help me!
Best wishes!
- Badges
- Science topic
- Similar topics
- Condensed Matter Physics
- Zero