I am performing MD nanoindentation using a virtual spherical indenter in LAMMPS.
The same script works perfectly at 10 K (smooth loading–unloading curve + correct penetration visualization),
but fails at 300 K, although the force–displacement curve sometimes looks smooth.
I am trying to understand what physical /numerical mechanism breaks at higher temperature. What is the correct, physically consistent nanoindentation protocol at finite temperature (300 K)?
Nobody will take any simulation seriously that uses fix temp/rescale. This has been shown to produce very bad results and is only acceptable for the initial part of the equilibration of a high potential energy system. But even then, a dissipative thermostat is usually more effective and creates fewer artifacts.
That is not really a LAMMPS question and thus off-topic for this forum. If you need advice on how to correctly do your research, you need to discuss with your adviser, supervisor, tutor, or search the published literature or text books for information about applicable simulation protocols and quality of results that can be achieved with them.
I was already aware of that. That’s why I tried approaching the problem from different mechanical perspectives. When I ran the indentation simulation at 10 K, everything worked and showed proper results. At that time, my boundary condition was p p s. During equilibration, I used only a thermostat, and for the Newtonian layer I applied NVE integration. Before starting the indentation, I did not need to apply or remove any fixes. However, at 300 K, I tried many different approaches, but the issue persisted. I followed the methodology of a recent paper, although their study was based on a Lennard-Jones potential, whereas mine uses a different Potential.
This tells me exactly NOTHING about what you are doing other than that you don’t like the results. It looks like you seriously need some proper in-person advising.
This is all too vague and thus there is no way to give you meaningful advice from remote.
The idea of procedures exists only in the realm of engineering. In physics it does not. If you want to do modelling nanoindentation, you need first to understand well the process of penetration of forces into the material with time. I do not know a single paper on nanoindentation where authors have demonstrated any understanding of the process of penetration of forces into the material with time. I am sorry to sat that.
Young researches have a tendency to use the simulation box LAMMPS as a black magic box that will give them answers. No. First, one needs to understand the physical process undergoing in there behind.
My latest article deals with the problem of force propagation. It will rather not help you in solving your nanoindentation problem, but perhaps will show, how the problem is difficult: