I simulated a silicon nanofilm with a central crack under tensile loading. The boundary condition I used is psp and the interatomic potential is MEAM. The system was simulated with the NVE ensemble at 100k for 10000 steps to equilibrate the energy and then with the NPT ensemble to equilibrate the pressure. After obtaining the equilibrium state, the tensile displacement was applied to the boundary layers (the topmost and bottommost three layers of atoms) with an increment of 0.002a (lattice parameter). The whole system was then relaxed by running NVT for 1000 steps with the boundary layers being fixed. After applying the tensile loading the crack did not propagate and the boundary atoms were in disorder, as shown by the attached figure, which is not what I expected. I cannot figure out the reason. Did anyone have this problem before? Any comment will be helpful. Thank you.
Most likely because you time integrated your boundary layers by using fix 2 all nvt. If you want the boundary layers immobilized, you should exclude them in nvt.
Your fix nvt tdamp is too large. It is wrong to use fix nvt with fix temp/rescale. There may be more errors in your input script.
Thank you very much for your reply. I revised my input file according to your suggestions. The problem of disordered atoms is solved. I have a new problem. I run 50000 steps with NVE ensemble, followed by 50000 steps with NPT ensemble to equilibrate the system. But, the pressure of the system did not converge to zero and the stress field is not uniform as shown by the attached figure which is colored according to the stress component in y direction (loading direction). Do you have any suggestions?
Step Temp PotEng Pxx Pyy Pzz Lx Ly Lz
100000 99.741256 -32690.421 -435.72545 1571.2609 293.89165 119.72081 109.28372 10.877614
50000 steps is too short to obtain meaningful equilibration. Your NPT pdamp may be inappropriate. A small system has large fluctuations in stresses. If you have psp boundary conditions, your Pyy is not equilibrated - so color coding Pyy has no meaning. There could be more reasons.
Thank you very much for your reply. Pxx and Pzz are not uniform either. I will do more tests.