Bad stress-strain plot

Hello lammps users,

I am performing MD simulation on BN using reaxff at 1K but the shape of stress-strain plot I am getting is very bad (see attached pic). Please suggest what could be the reason for the same?? I am also attaching my logfile.



Stress-strain plot.png

log.lammps (462 KB)

There are many possible reasons. The most common one is that the potential that you have defined in LAMMPS is not the same as the one used in the previously published work. To test this, you should check that you can accurately reproduce published results.

Thank you Aidan. These parameters were obtained directly from Adri van. I mentioned that we are working on boron nitride and he provided latest parameters for the same. The same parameters are working fine at 100K. This problem is coming at 1K.Can box size, npt fix parameters, or control file affect results in such an extent??


Thanks for explaining that you have gotten good results at 100K. That is an important piece of information, which eliminates a lot of questions about the potential. I looked at your log file, and noticed this:

fix 2 all npt temp 1.0 1.0 0.29999 x 0 0 199

fix 3 all deform 1 y erate 1e-06 units box remap x

If you read the doc page for fix deform, you will see this:

“Dimensions corresponding to unspecified parameters can also be controlled by a fix npt or fix nph command.”

Clearly, your simulation violates the rule implied by this statement (i.e. you should only apply npt to dimensions not controlled by fix deform). You were asking fix npt to maintain an average scalar pressure of zero by changing the box dimensions isotropically, while at the same time you were deforming the box anisotropically. Such a combination of conflicting operations is ill-defined, at any temperature.

Start by running fix deform with fix nve. After that, if you really feel you need it, try running fix deform y with fix npt x, but be very careful.


I am asking fix npt command only to maintain an average scalar pressure of zero in X-direction not in z. I tried “fix 1 all npt temp 1.0 1.0 0.29999 x 0 0 199 z 0 0 199” with tersoff potential but this was not giving correct fracture stress values because box dimensions were changing abruptly. And fix 2 all npt temp 1.0 1.0 0.29999 x 0 0 199 is required in situations where we want zero stress in other direction to represent a pure mode loading.

Good. So I think you understand that combining fix deform y with fix npt with any keywords that include changes in the y dimension (iso, aniso, tri, y) are completely wrong. Using x and or z might work. However, please do not ignore my original advice of replacing npt with nve. Once you get that working, you can try doing more complicated stuff with npt.