Pressure/Stress in VASP

[pengx]

Dear VASP master,

I am calculating free-standing Ge nanowire with diameter 2.5 nm (surface is terminated by hydrogen atoms). I found that the relaxed wire has pressure parameters such as external pressure = -0.62 kB Pullay stress = 0.00 kB, and stress in x, y, and z directions are (in kB) -1.21, -0.35, -0.29, respectively,. I wonder how these stress/pressure are calculated by the program. Since the Ge nanowire is free-standing (energetically relaxed), I should expect a zero stress in Z direction while in the lateral direction only tangential stress exists. I have struggled a lot for this inconsistency. I really appreciate any suggestion from you.
Thank you very much.

[pkroll]

Now I don't know, if you used ISIF = 2 or ISIF=3, and how large the parameters of your simulation cell are.
Assuming you work with ISIF=2 and have somehow adjusted your z-axis (or manipulated the stress tensor in the code), the values are pretty nice. They still depend on your convergence criterion EDIFFG, but with such a huge model this can't be so low as to yield less than 0.1 kbar.
I have a cluster (3nm diameter, cell 3.6 nm dimensions, ISIF=2) and stresses are -4 kbar.
Even if you have used ISIF=3, which may be debatable, since your wires will get somewhat close (>5 Ã… distance from H to H to be "free", > 8 Ã… to get a "converged" dielectric function), these values are "OK" given the size of your model.

[panda]

Yes, I have been doing some one dimensional structures, where the stresses on some of the ions is high or negative, which bothered me at first as I am so accustomed to metal calculations, but if you have varied the parameters and the cell a bit and the forces are still high try running a single point (no relaxation) and see how the forces compare. I agree with above poster in the movement of ions during relaxation and the magnitude of your forces, I think it is fine.