Dear all

I want to relax a periodic mineral bulk. It's a structure of hundreds of atoms, I tried to perform  geometry optimization +  NVT MD, but I got stuck during the NVT MD process with the following error:

Dynamics calculation - Aborting dynamics.

An unphysical integration step (>100 Angstroms) has been detected.

Try reducing the time step or relaxing the structure first.

I already had 1 fs as the time step. I used CLAYFF. After the geometry optimization, I tried to first equilibrate the system with Berendsen or Velocity Scale thermostat, then proceed to Nose production, but I failed in both case.

With Berendsen, I set the decay constant at 0.005ps, temperature at 300K, but NVT process stopped with the above mentioned error, and the temperature went up over 1000K. I attached the temperature and energy figures of Berendsen NVT as following:

             Temperature curve of NVT MD with Berendsen thermostat

                     Energy curve of NVT MD with Berendsen thermostat

With Velocity Scale,  I also set the temperature at 300K, with a temperature difference of 50K, I attached temperature and energy figures as following:

             Temperature curve of NVT MD with velocity scale thermostat

                  Energy curve of NVT MD with velocity scale thermostat

After 500 ps, I proceed to Nose NVT with Q-ratio of 0.01, but after several pecosecond, it failed with the above mentioned error, and the temperature went up to several thousands K. I attached the temperature and energy figures:

                     Temperature curve of NVT MD with Nose thermostat

                       Energy curve of NVT MD with Nose thermostat

I also tried to start with Andersen thermostat, but I failed too.

Any suggestion would be appreciated.

Thank you.

Jingjing Bu