I believe that the parameters written in library.meam for K are wrong. The rest of this post explains my reasoning. I run the following code, which creates a box of K atoms in a bcc lattice, assigns the meam K potential to it, minimizes energy, sets starting velocities, and uses fix npt. I use the parameters for K in library.meam with no additional specified parameters.
units metal
dimension 3
boundary p p p
atom_style atomic
variable latparam equal 5.344
variable Tstart equal 300
variable Patm equal 1.01325
timestep 0.001
lattice bcc ${latparam}
region potato block 0 5 0 5 0 5
create_box 1 potato
create_atoms 1 region potato
pair_style meam
pair_coeff * * library.meam K NULL K
minimize 1.0e-5 1.0e-4 100 1000
velocity all create ${Tstart} 83051 dist gaussian mom yes rot yes
compute pe_id all pe pair
dump dumpid all atom 100 dumpfile
thermo_style custom step temp press etotal vol pe
thermo 50
fix id1 all npt temp ${Tstart} ${Tstart} $(100*dt) iso ${Patm} ${Patm} $(100*dt)
run 1000
The energy values printed are all zero during minimization. During the npt run, the box expands very quickly with atoms spreading out really far.
When I instead replace K with Na in the pair_coeff line and update the lattice parameter, the simulation works normally (bcc crystal structure maintained and energy nonzero during minimization). This proves that the format of the meam potential assignment is okay.
Furthermore, when I replace the potential with an available K eam/fs potential, I get a normal, working simulation. Therefore, it seems that the error lies in the K meam parameters printed in library.meam.
If I am missing something here and have an error in my reasoning or simulation code, please let me know. If not, hopefully this can help others in the future using simulations with potassium. Thank you!