fix GCMC System is not charge neutral

Hi all,
I want to simulate the adsorption of methane in silica pore. To control the temperature, according to Andrew, I used the ‘fix nve/limit’ and ‘fix langevin’ command. As a result, the temperature of methane in output was fluctuant acutely, (0~50,000K). I think it is unnormal and the high temperature should not appear. And worst of all, in middle of simulation, about 131,000 timesteps, the warning ‘System is not charge neutral, net charge = 0.526125 (…/kspace.cpp:302)’ was raised. There are 255 Si atoms, 460 bonded O atoms, 100 nonbonded O atoms, 100 H atoms connected to nonbonded O and 10 TraPPE-U methane, and their charge are 1.2805, -0.64025,-0.526125,0.2060, 0 respectively. Obviously, my initial system is charge neutral, and my silica pore is fixed using ‘fix pore pore setforce 0.0 0.0 0.0’ command. Two question, first, Is the fluctuate temperature in output normal? Second, from the warning information, I think one nonbonded O atom was out of system, but my system was fixed. It is contradictory. Could you give me some suggestion about solving these problems? I attach my input script , data file and out file.
Looking forward to your response. Thanks to your help.

in.puregcmc (1.39 KB)

pore1nmsorption.lammps05 (75.3 KB)

puregcmc.out (27.3 KB)