# Questions about MD model for studying the material behavior under temperature gradient

Hi, everyone!

I am attempt to simulate the material behaviors under the influence of surface heating. I have read some of the previous threads and came up with an approach but I am not sure whether my model is correct or not. Here is my MD model and setups:

#--------------------------------#
##################### → high temperature surface (5 Ångstrom)
#--------------------------------#
#####################
##################### → free layers (NVE ensemble, 45 Å)
#####################
#--------------------------------#
##################### → room temperature layer (5 Å)
#--------------------------------#
##################### → immobile layer (5 Å)
#--------------------------------#

Can you give me some advice on my MD model, really appreciate for your help and time!

1. First, I equilibrate my whole MD system at room temperature. In this procedure, I used NVT ensemble. Should I use NVE ensemble to further equilibrate my MD system (Ready for the subsequent surface heating).

2. In the first procedure, the boundary conditions are periodic in XY directions. The Z dimension is fixed. Should I add two vacuum space in z directions surrounding the MD model and use fix oneway in these vacuum spaces?

3. During the surface heating procedure, the upper surface is heat up through fix nve+langevin method. And the bottom room temperature layer is thermostat through the nve+langevin. I am not sure whether my modelling method is correct.
And compared with fix nvt, which one is a better choice? I have learned that a dissipative thermostat like langevin is perhaps suitable for my case.

4. About the thermostat, there is a parameter called time constant. Should I set the time constant to a large value like 100 timesteps of to a small value. I am not sure how to appropriately set this parameter to make my model physically reasonable because I have no deep understanding about this parameter. I am attempting to avoid unphysical modeling problems. Please give me some advice, thanks a lot!

5. Because the material surface is thermostat under a quite high temperature, the surface atoms would have large velocities. And I guess that some surface atoms perhaps fly away from the bulk material. Should I also add a rigid immobile layer on the material surface to make a barrier (just like the bottom immobile layer)? Or, should I only add a vaccum layer on the top of surface and use fix oneway?

6. After the heating up procedure, I want to dissipate and consume the system energy gradually, and make the system equilibrate at room temperature finally. In order to achieve this, I try to perform NVE ensemble to the surface layer and the langevin thermostat is removed for the surface layer. The rest of layers have no change in setting. I am not sure whether this can work or not.

I appreciate it if anyone could take a look at this and give me some advice. Thanks again for your kindly help!

Best regards,
Shusen Guo

And you expect that somebody here can tell you just from a quick look?
Moreover, how will you justify your choices to a reviewer when you publish your results?
With “Some dude on the internet said it was OK”?

What you are asking for is a discussion that you would have to have with an adviser or a collaborator.
In addition, what you are asking about is not really about LAMMPS but about how to design and plan your simulations in general. That makes it off-topic here. So you need to find a more suitable location to ask, but still keep in mind that sources you cannot quote and where you have no way to determine the credibility are of limited value.

Hi, akohlmey