Suppose we have a layered crystal structure (for example Ti4C3). I save the structure as the pdb file format. Now Ti atoms labeled as the Ti1, Ti2, …, Ti4, representing 4 new Ti and 3 new carbon atoms.
My question is regarding the LAMMPS data file:
How many ‘atom types’ do we have here? 7 or 2?
What about the 2nd column of the data file; Are all of the Ti and C atoms labeled as the first molecule?
If all Ti atoms have the same interaction parameters, you can assign the same type to all four. However, you could also assign different types to each atom for easier manipulation.
If the Ti atoms have different interaction parameters, use different types for each.
Would you mind writing more about these dependencies and conditions? For example, when Ti4Cl3 is supposed as the substrate, are there all Ti and Cl labeled as one molecule?
LAMMPS normally does not use molecule IDs to control dynamics. Some “fixes” can use molecule IDs – most commonly fix rigid has input settings enabling each molecule to be integrated as a rigid body, and “compute chunk” allows particles to be “chunked” by molecule. Whether those are appropriate for your use case is of course up to you.
That would seem reasonable, yes. But as @srtee said, it won’t make any difference on the dynamic (unless you use some specific command), so its only a matter of choice.