Trouble Validating Stress-Strain Response of 2D MoS2 with Reaxff Potential


I am trying to simulate Stress Vs Strain plot for Uniaxial Tensile Test of Pristine 2D monolayer MoS2 through LAMMPS. I used Reaxff potential developed by Professor Dr. Adri Van Duin. The Boundary Conditions (Nose Hoover Thermostatting) in the input script is set up following the paper Ostadhossein et al ( DOI: 10.1021/acs.jpclett.6b02902) & Bohayra et al ( DOI: 10.1039/c6cp03612k). But neither the ultimate tensile strength nor the Young’s modulus matches with the output they had obtained. In fact those parameters in my output is much lower than what they had got. I have attached the LAMMPS input script along with the output data files and reference plot in this Google Drive Link - LAMMPS Input Script for uniaxial stress-strain responses of pristine 2H MoS2 - Google Drive .

I appreciate you taking time to read this. It would be very helpful if you can help me by pointing out the mistakes I have made.

N:B: Even after nose hoover thermostatting and barostatting (npt ensemble) Pressure and temperature still fluctuates a lot during the straining process . That might lead to the result not being same as the reference. Would like to know expert opinion on how to solve it.


1 Like

Your request is too specific and can only be properly addressed by repeating your work, which is far too much effort for most people reading here. You need to get extremely lucky that somebody downloads and looks at your files and detects an obvious error.

Until then you can make tests with different system sizes and also repeat the same run with different initial random seeds to get better phase space sampling.

1 Like

Professor Dr. Axel Kohlmeyer,

Thank you for taking an interest in this problem. Till now I have run hundreds of tests with different system sizes as well as different conditions. I am also attaching output files from two more simulations I ran for this purpose as tials. I am new to working with Reaxff potential. That’s why I am expecting someone with expertise in this field to share his/her take on this.

To acquire better phase space sampling, I’ll keep performing trials with various system sizes and repeat the same run with different initial random seeds as you have suggested. Hopefully, someone experienced will kindly look into the files in the meantime and help me to figure it out.


dump_Trial 1.compression (2.4 MB)
log.PE_nc10_clK_Trial 1.txt (139.9 KB)

log_Trial 2.lammps (112.7 KB)