I’m trying to create amorphous silica(a-SiO2) by melting cristobalite SiO2 using tersoff potential. I followed the method mentioned in this  publication for the melting and quenching. The steps are briefly as follows, (Timestep is 0.5fs and pressure in all the npt ensembles were 1bar)
1: Create cristobalite at 1800K
2: Equilibrate the system for 0.5ns at 1800K under npt ensemble
3: Take the system into nve ensemble and check the stability
4: Heat the system to 5000K in npt with 1K/ps rate.
5: Equilibrate the system at 5000K in npt for 10ns.
6: Take the system into nve and check the stability.**
7: Quench the system to 300K at a rate of 0.1K/ps in npt.
8: Equilibrate the system for 1ns at 300K in npt.
9: Take the system into nve and check the stability.**
My problems are,
- At step NO 6, when I check the stability of the system at 5000K in nve, temperature of the system was increasing.
- At step NO 9, when I was checking the stability of the system at 300K in nve, temperature of the system was increasing.
I was able to solve both of the problems by reducing the timestep to 0.1fs. As my understanding, it should be possible to have energy conservation with a 0.5fs timestep in a-SiO2. But here I’am with a unstable system. Going back to 0.1fs timestep will cost me a lot of time in my calculations and so I really want to keep the stimestep at 0.5fs.
I would like to have some advice on the issue from someone who is familiar with a-SiO2. And is there any way of creating a-SiO2 by placing Si and O atoms randomly rather than heating and quenching method?
I have attached the input file of my simulation.
 - http://dx.doi.org/10.1063/1.4983753
I wish you well during this challenging time. Hope everyone keeps safe!
Thank you very much,
SIO2.in (2.92 KB)