@mkanski ok this was what I had been trying. I agree this is less than ideal, as I want the binding energies to be accurate (key for emission energy as you already know).
However, I may have found a work around or slight fix. I noticed that during impacts I reduce my timestep by 5-10x the typical reax time step. As such, I thought perhaps it is just doing charge equilibration too often (every 1 time step in my case).
When I switched to every 10 time steps I got a much cleaner behavior, without a major spike and no huge emission energies. In this case I am equilibrating charges every 0.5-1 fs, as would be if I had a timestep of 0.5 fs. Would this make sense as to why a large spike was being seen?