# Weird uniaxial tension test of graphene

Hi all,

Just for fun and learning, I’m testing the mechanical properties of graphene sheets. The sheets are 100 x 100 Angstrom squared (non periodic system). So far, what I do is equilibrate the system under NPT @ 300 K, and P = 0 bar. Then, I change to NVT ensemble and stress the system by pulling one edge at a velocity of 1 m/s (strain rate 1E10 1/s) and fixing the opposite edge. What I had done so far was track the coordinate of the pulling edge to get the strain, and then calculate the stress by using the pressure in the x direction (Pxx component). However, I get a HUGELY underestimated modulus (4 MPa versus the theoretical 1 TPa). I’m using the AIREBO force field.

This brings me to my question: what is the common practice for stress calculations? Global pressure? Pressure component in the pulling direction? I’m curious because I didn’t expect such a underestimate when people in literature have used similar dimensions and gotten good agreement. Also, other sources I have seen use the pulling direction stress (or pressure if you want the whole system scalar value) but in my case that doesn’t seem to be the case at all.

Thanks,
Rafael

Hi all,

Just for fun and learning, I'm testing the mechanical properties of graphene
sheets. The sheets are 100 x 100 Angstrom squared (non periodic system). So
far, what I do is equilibrate the system under NPT @ 300 K, and P = 0 bar.

i am confused, why use NPT when you have a non-periodic system?

Then, I change to NVT ensemble and stress the system by pulling one edge at
a velocity of 1 m/s (strain rate 1E10 1/s) and fixing the opposite edge.
What I had done so far was track the coordinate of the pulling edge to get
the strain, and then calculate the stress by using the pressure in the x
direction (Pxx component). However, I get a HUGELY underestimated modulus (4
MPa versus the theoretical 1 TPa). I'm using the AIREBO force field.

This brings me to my question: what is the common practice for stress
calculations? Global pressure? Pressure component in the pulling direction?

global pressure is not that useful a property in a non-periodic system.

I'm curious because I didn't expect such a underestimate when people in
literature have used similar dimensions and gotten good agreement. Also,
other sources I have seen use the pulling direction stress (or pressure if
you want the whole system scalar value) but in my case that doesn't seem to
be the case at all.

there should be lots of discussions on the subject in the mailing list archives.
you can readily access the per atom virial stress.

HTH,
axel.

Hi Axel,

I’d like to share my understanding so far after going through some documentation and the mailing list, if you would not mind corroborating my idea!

Thanks for your reply, first of all. Going through the documentation of the compute stress/atom, and the virial formulation, it makes sense that the pressure in the system might not be a global metric for the stress. Nonetheless, for the pressure calculation, the code simply takes the hydrostatic stress components and divides by the volume of the box. For the stress in the atoms, I could simply take the compute stress/atom and reduce the sum to one scalar for each component. However, in the end, I guess I must divide this stress by the “volume” of the graphene sheet as the the compute stress/atom gives a stress-volume formulation.

Does this make sense?

Thanks a lot,
Rafael