Graphene sheet at finite temprature

Dear Lammps Users,
I am trying to simulate a graphene sheet in yz plane ( 720 atoms) at finite temprature 320k using fix npt command. I am using airebo potential. The simulation result shows graphene sheet contracts (as supposed to be) and then expands and this contraction and expansion cycles repeats up to end of simulation and it never become stable. I tried many different dump value, drag value, time steps and simulation time but none of them make considerable changes. The initial atom coordinates are read from a data file. I visualized the initial atom configurations and it shows a very perfect graphene sheet.
I appriciate if you can help me solving this problem.
Here is my script:

units metal
boundary p p p
atom_style atomic
newton on
read_data data1.rtf
pair_style airebo 2.5 1 1
pair_coeff * * CH.airebo C
neighbor 3.0 nsq
neigh_modify once no every 1 delay 0 check yes
timestep 0.0001
thermo 1000
thermo_style custom step temp lx ly lz
thermo_modify flush yes
velocity all create 640 5812576 dist gaussian
fix 1 all npt temp 320 320 0.5 y 0.01 0.01 5 z 0.01 0.01 5 drag 1.5 pchain 0 mtk yes nreset 0

run 250000

Thanks!
Parnia Mohammadi

Just because you start with a perfect graphene sheet does
not mean it is in equilibrium. Is the initial pressure equal
to your target pressure (0.0?). If not, then it's not
in equil, for an NPT simultaion. Other than that, pressures
oscillate. I assume you also tried different damping values,
and that they are values that make sense for your time units.

Steve