Temperature rise in tensile test

Dear lammps users,

I try to simulate the process of a tensile test of a grapheme nanoribbon by applying fix deform under nvt ensemble.
But I find a temperature rise from 300 to 400 or higher near the fracture point. Why is that happen? Is my strain rate too fast? Or should I perform the deformation every 500 or more steps?
Thank you


What do you think? After all none here has a clue about what your input script looks like.

Thermostatting, even when done properly, does not provide instantaneous action, i.e., usually takes a bit to set the temperature of a system that is out of thermal equilibrium.

Bond breaking can also result in transit temperature gradients and fluctuations. But then again, who else really knows whats going on with your simulation but you…


Thank you Carlos. Just notice that my input file wasn’t uploaded successfully. Here’s part of it.

variable T equal 300
units metal
boundary f f f
atom_style full
read_data data.line7
neighbor 2.0 bin
neigh_modify delay 5
#Potential function
pair_style airebo 2.0
pair_coeff * * E:/Gao/strgnr/CH.airebo C C C C

mass 1 12.081
mass 2 12.081
mass 3 12.081
mass 4 12.081

#groups & heat layers
group HOT molecule 1
group COLD molecule 10
group A1 molecule 2
group A2 molecule 3
group A3 molecule 4
group A4 molecule 5
group A5 molecule 6
group A6 molecule 7
group A7 molecule 8
group A8 molecule 9
group Lbound molecule 11
group Rbound molecule 12
group total subtract all Lbound Rbound
fix Lbound Lbound setforce 0.0 0.0 0.0
fix Rbound Rbound setforce 0.0 0.0 0.0

#Compute Temperature & variables definition
compute Ttotal total temp
timestep 0.0005
#Initial condition: System is at 300K
velocity HOT create $T 798456 dist gaussian
velocity COLD create $T 788456 dist gaussian
velocity A1 create $T 720414 dist gaussian
velocity A2 create $T 720416 dist gaussian
velocity A3 create $T 720418 dist gaussian
velocity A4 create $T 720420 dist gaussian
velocity A5 create $T 120422 dist gaussian
velocity A6 create $T 220424 dist gaussian
velocity A7 create $T 420414 dist gaussian
velocity A8 create $T 520416 dist gaussian

#1st Equilibrium, NH
fix 1 total nvt temp $T $T 50

thermo_style custom step c_Ttotal etotal
thermo 1000
dump trajectory all atom 5000 3c60tra.dump
run 1000000
fix Delta_out all ave/time 1 500 500 f_Lbound[1] f_Rbound[1] &
file stress.dat

fix 2 all deform 1000 x erate 0.00025 remap x
run 20000000

Any suggestions will be appreciated. Thank you.


Actually the problem is I find a ‘hardening’ in my stress strain curve near the fracture point and I notice the temperature rise up to larger than 400K. I suspect the latter leads to the former since some paper says Young’s modulus positively depends on temperature.

Not sure I am your guy for this example. Is this how people handle such simulations in the lit? I mean, if I was to strain a finite size system I would have anchored one end of the system and applied a constant velocity (equivalent to the desired strain rate) to the other end. As to the thermostatting, I would have placed a Langevin one only in the plane perpendicular to the straining direction. Your combination of “f” boundaries, fix deform and

fix nvt…hmmm. My suggestion (buyer beware) feels to me closer to what an experimentalist could do in real life with some fancy AFM tip and some chemical functionalization to achieve the anchoring. This is as far as my comment goes cause I don’t want to shot myself in the foot due to my limited knowledge of this particular topic.