# Why can't I reproduce the behavior of an H-saturated graphene flake?

I'm trying to simulate a graphene flake with its edge C atoms saturated by H atoms, in a temperature ramp from 300 K to 1600 K, using LAMMPS and the AIREBO potential proposed by Stuart et al. in 2000.

The problem is when the temperature is close to 500 K the C-H bonds start breaking and at 1600 K all of the H atoms are unbonded. It also occurs if I use a bilayer graphene flake even if it is two times bigger than the here shown.

What do you think about that?

My input card is below. It looks very similar to the example included in the LAMMPS folder, so I don't know why is this happening.

Pic1: simulation at 300 K Pic2: simulation at 1600 K (H atoms are randomly distributed in the box)

################
# graphene n14_l1
################

units           metal
boundary        p p p
atom_style      atomic

mass        1 12.01
mass            2 1.008

pair_style      airebo 3.0 1 1
pair_coeff      * * CH.airebo C H

neighbor        0.3 bin
neigh_modify    delay 10

#---------------------------------------- relax
timestep        0.0005  # 0.5 fs
thermo          20
dump            1 all custom 2 min.lammpstrj id element x y z
minimize        1.0e-10 1.0e-11 10000 100000

reset_timestep  0
undump          1

#---------------------------------------- MD
timestep        0.002   # 2 fs
velocity        all create 273.00 32345 rot yes mom yes dist gaussian
fix     1 all nvt temp 273.0 1600.0 0.01
thermo          1000
thermo_style    custom step temp pe ke etotal
dump            1 all custom 200 md.lammpstrj id element x y z
dump_modify     1 element C H
restart         20000 restart1 restart2

#---------------------------------------- Run
run             2500000    # 5 ns


• +1. Thank you for this question, and welcome to our site!!! We hope to see much more of you here !!! It might be helpful if you add a link to the Stuart paper. Also we have a LAMMPS chat room here, if you think that discussing it with other LAMMPS users will help converge on a solution faster: chat.stackexchange.com/rooms/109805/lammps Jun 29, 2020 at 2:53
• Hi biddies, thanks a lot for your suggestions. As Magic_Numer said, the problem was solved decreasing the T_{damp}. I Also will take into account your suggestion to change the neighbor skin It is also a mistake. The correct date is 3 A. Thanks!! Jul 1, 2020 at 18:07
• @GabrielOlguín now that your question has been solved, please write an answer with the solution and your working file(s). We need to clear our unanswered queue, and your answer will help anyone in the future with the same problem, and you will be rewarded for your answer. Jul 1, 2020 at 20:04
• @NikeDattani Ok I will turn my comment to an answer since it solved Gabriel's problem. Jul 2, 2020 at 12:25
• @SaiSmaranSBPES1201701189PES, if you are speaking about the picture that Gabriel posted, that is really due to enhanced atomic vibrations (atoms on the edge definitely have higher degree of freedom). Jul 2, 2020 at 12:27

From the available information, I think it is caused by the small value of $$T_{damp}$$. This causes the $$T$$ to fluctuate wildly, which can induce unwanted bond-breaking. Best practise is to keep $$T_{damp}$$ value around $$100 \times \text{timestep}$$.
Also, I will suggest you use a much larger neighbour bin size ($$\approx 3 Å$$).