Hello, I am conducting simulations of single crystal melting and calculating the RMSD based on these simulations. I aim to run the simulations in the NPT ensemble. The issue I am facing is separating thermal expansion from the RMSD caused by atomic vibrations. Currently, instead of calculating RMSD in all three directions, I am only calculating RMSD for motion in the Z direction. This is because I fix the box size in the Z direction using the command: fix 4 all npt temp 300.0 2500 100.0 x 0.0 0.0 100 y 0.0 0.0 100 couple xy.
My question is: Is it possible to fix the box size in two directions, for example, in the XY directions, while running the NPT ensemble?
First off, please note that there is a difference between running a simulation in NPT ensemble and using fix npt. If you don’t have all box dimensions mobile, then you are not in an NPT ensemble.
That said, you can easily couple only the z-direction to the barostat in the same way you had coupled x- and y-direction in the example you have quoted in your post.
In general, I don’t think you can separate “thermal expansion” from “atomic vibrations” because the former is caused by growth of the latter.
Please also note that (spontaneous) melting is an activated process (requiring nucleation) and for atomic scale simulation associated with very large finite size effects and hysteresis due to the comparatively small size of the system. Studies of phase transitions are thus usually performed by performing and analyzing coexistance simulations where both phases are present.
Thank you for your response. I didn’t clarify, but indeed in NPT, I meant the ability to change the size in all three directions. The finite-size effect is indeed significant, and we observe that only systems with over 250k atoms correctly replicate the systems we measure (in a timescale similar to that of the simulation!). And we are studying such large systems. Expanding on my question – when using the TTM (two-temperature model) method, can I also freely couple the directions with the barostat (in this case, we are using a Langevin thermostat-barostat)?