Questions about MD reservoir temperature and its analysis

[Lim changmin]

Hello.

I am currently simulating amorphous structure of silicon using MD, and also have some questions regarding reservoir temperature.

1. What is exactly reservoir temperature? Can it be understood as starting point?

2. When melting the silicon, is it the right procedure to set the reservoir temperature to 300 or to set reservoir temperature 3000 or more?

3. When checking MD trajectory, the some part of temperature and reservoir temperature are not aligned. For example, in the graph, reservoir temperature shows 300 to 4000K line, while the temperature line only stayed at 300K. How can I interpret this part?

Thank you.

 

[AsifShah]

I will try to answer:
1. Reservoir temperature is like the temperature you set on the AC in your room, i.e., the reference temperature you want. Then there is instantaneous temperature- which is the actual temperature calculated from kinetic energy of atoms. Insta. temp. oscillates around the reservoir temperature.
2. I would suggest to heat the system slowly from 300 to 3000K. (The term slowly is generic here- it depends on how much computational resources you have, type of study you want to conduct etc. Also, more slow means more time taking process).
3. Can you please the graph? I don't understand.

[Lim changmin]

Thank you for your reply on question 1,2.

These are the pictures from my MD simulator.
As seen in the pictures, the reservoir temperature and temperature don't aligned together.
I used NVT-nose hoover both, each step of 752(time step : 1.33 fs), 3500(time step : 1.33 fs).

[AsifShah]

I think this is the equilibration period since the simulation has only progressed a few femtoseconds. Can you show behavior after 10 ps?

[Lim changmin]

This is the results after 10ps. And the timestep is 1.33fs 200steps.

[AsifShah]

yes you can see after 10ps both are coming close to each other..
If you are using DFT then 2-3 ps is enough for equilibration however for classical forcefield, do use longer timescales maybe up to 100 ps.

[Lim changmin]

Thank you for your help.
And would there be some problems if the reservoir temperature and temperature graph in the figures are not close together?
Or is it just an minor issue?

[AsifShah]

The instan. temperature will always fluctuate around reservoir temperature and it is normal.
May I know:
1. What type of forcefield you are using?
2.  If you can provide a longer time scale graph?
3. What time do you set for thermostat timescale and barostate timescale?