Please keep the list in the loop.
I suggest you run the vapor by itself
or without interactions with the surface,
or with the interactions dialed down
and see if it conserves energy. Or run
with a single vapor particle and track
its energy as it bounces off the surface
and see what happens. In other words,
debug what is going on. My guess is
that it is not an issue with LAMMPS, but
with what your understanding is of
what is going on.
Steve