I am working on classical molecular dynamics simulations of hydride perovskite using LAMMPS, particularly to study hydrogen diffusion (MSD and diffusion coefficient).
My main difficulty is finding an appropriate interatomic potential for the ternary Li–Mg–H system.
I searched the NIST Interatomic Potentials Repository and found potentials for some binary systems, such as Li–Mg and Mg–H, but I could not find a potential specifically parameterized for Li–Mg–H or LiMgH₃.
Could you please advise me on the following points?
Is there an existing Li–Mg–H potential compatible with LAMMPS that could be used for LiMgH₃?
If no ternary Li–Mg–H potential exists, can binary potentials such as Li–Mg and Mg–H be combined, or would this be physically inappropriate?
In the absence of a reliable classical potential, would you recommend developing a new potential or using a machine-learning interatomic potential trained on DFT data?
Which databases or literature sources would you recommend for searching for potentials for hydride/perovskite systems?
That is a question for a domain expert. You should either ask your adviser or supervisor or senior colleagues working in the same field. Otherwise, you need to identify a suitably competent collaborator, which is again something that you need to discuss with your adviser/supervisor.
This is almost never a good idea unless you are using a pair-wise additive potential, and then you would also need a Li-H potential. If you cannot find a suitable potential in the NIST database, you have to survey the published literature.
This is both essentially the same thing and requires about the same kind of work and neither is a trivial task, but requires experience and guidance from an experienced person. On top of using DFT calculation data you can also use experimental data for fitting or training the potential.
This is - again - a question for a domain expert. You are not likely to find one by posting in an online forum.