Due to the constraint of the limited capacity of computer clusters, I think usually in molecular dynamics simulation, the target material is heated from room temperature to a temperature that is below the melting temperature in tens of picosecond or several picoseconds or even less simulation time.
Different materials have different melting temperatures, but it seems for metals or graphene or such material with a melting temperature on the magnitude of several thousands Kelvin, the heating rate is about 1000 K/ps. Is this heating rate on the same magnitude with that in the experiments or common life? It feels like the heating speed is much higher in the molecular dynamics simulation.
Could you please tell me what is the proper heating rate to be set in molecular dynamics simulation? What property of the material should be considered as the most important factor for the setting of the heating speed? If it varies with different kinds of materials, please talk about graphene or metals first.
Any help would be much appreciated!
Ideally it can be 5-10 degree per 1000 timesteps but again it is dependent on type of the target material
Can you elaborate more on the type of your target material ? Wenqiang Liu
I think that if the aim is to accurately compute the melting temperature, as long as you heat up to a temperature that is surely below the melting temperature does not matter the heating rate. In order to compute the melting temperature you can equilibrate the system at a temperature a little lower than the melting temperature than perform many different simulations in which the heating rate is lower and lower, if the melting temperature that you find converges at some value you could extrapolate this as the value to be compared with experimental one. Anyway you have always an overestimation of the melting temperature due to the fact that you do not sample the whole phase space at each temperature. Sometimes could be more profitable run isothermal simulations after a very brief equilibration using bisection on the temperature to find the melting one.
Hi Rajesh Pavan Pothukuchi , thanks very much for your answer!
The material I am interested in is graphene, if it is possible, please give some advice on metals like gold as well.
You talked about `5-10 degree per 1000 timesteps`, I am wondering that whether the time span a timestep represents matters?
Hi Andrea Pedrielli , thanks very much for your reply!
You provide very useful information on how to get the melting point of a material, thanks!
In my case, point defects exist in my simulation system, I want to let them evolve at a higher temperature because the time can be simulated in molecular dynamics is very limited and at higher temperature the evolution process is much faster. How should I choose the heating rate to the target temperature? Please advise me.
- Badges
- Science topic