And because the H's VdW sphere is entirely inside the O's VdW sphere, the L-J parameters on the H have a negligible effect on bulk phase properties. They're there only for practical purposes. Specifically, without them (and if SHAKE is not in use), it's not unthinkable that during a simulation with an anionic solute, a hydrogen atom would get so close to the solute's negatively charged center X that the H-X electrostatic attraction would become stronger than the O-H bond stretching potentials, causing the simulation to crash because of electrostatic catastrophe.Edit 2
: this is somewhat of a moot point, however, because TIP3P was originally conceived as a rigid water model, so SHAKE should
be in use, which rules out the scenario outlined above. In theory, it's still possible for a negative center to penetrate the oxygen's L-J sphere far enough to make the electrostatic energy go to minus infinity, thereby overriding the L-J repulsion, but I wonder what the odds are that this would happen in practice.Edit 1
: A similar question has been answered before
. I'm starting to thing we should have a "CHARMM FAQ" somewhere...