Dear Dr. Axel and all,
Yes you are totally right. The problem comes from the Born potential parameters. When I changed the parameters so that the Born potential has a proper shape, the simulation could went up to 2000K. But I am sure I used correct parameters with correct units. I attached the literature for your perusal. The author did not mentioned what parameters units they used. I emailed him, and he said they used KJ/mol energy units. After conversion from KJ/mol to Kcal/mol (real units in lammps), I could not reproduce the right lattice constant with respect to experiment and also the literature (experimental lattice constant = 3.9051A at r.t., literature lattice cosnt. = 3.89A at r.t.). So I assume the potential parameters reported in the literature were in Kcal/mol, and the lattice const. calculated (3.907A) was in closed agreement within a reasonable error. If I use potential parameters in KJ/mol unit, I get lattice const. ~ 5.0A, which is not right w.r.t. the experiment. I calculate the lattice constant by averaging the simulation box boundary lx over large equilibrium time steps at a desired temperature.
The first parameter in Born style is 4.184 KJ/mol/A times rho(A), so it is
are you sure about the “times rho(A)”?
Yes I am sure. As you can see from the literature, the Born potential depth is f_0*(b_i+b_j), which is equals to A in lammps born style and b_i+b_j=rho_i,j in lammps born style. F_0 = 4.184 KJ/mol/A = 1Kcal/mol/A.
As a conclusion, the potential parameters reported in the literature were failed, I wonder how come they still can be published?
Thank you for your time.