Getting rid of negative pressure

Two problems I found from looking at the configuration.
    - "high bond coefficients" on C-H, means your timestep is
      too large for that high frequency vibration, you'll
      "loose" the bond
    - NPT simulation? What you got here is a vacuum simulation,
      you have two small molecules in an relatively large empty box.
      Where is the "P" (pressure) expected to come from?

I would assume you originally planned a simulation with a large enough
box full of your ions. Then, a NPT ensemble surely would make sense.
As Axel already said, even in this case you should select appropriate
barostat algorithms after intense testing and doubting :wink:



P.S. with some minor modifications (nvt, dt), your input runs
fine on my machine (v 2018.08.02):
=> (693 Bytes)

inputdata.lmp (8.97 KB)