[lammps-users] msst with reaxff on N2

Hello LAMMPS mailing list,

I have a question about using the ReaxFF in combination with msst (shock-compression). Such setups have been reported in the literature and I’m trying it out on a simple system of pure liquid nitrogen where solid experimental data exists, e.g. Nellis et al., JCP, 1991. Up to a certain shock strength (Us=8.5 km/s), the agreement (in Us-up or P(V) space) is excellent. This limit seems to coincide with the expected dissociative phase transition, where a softening in the Us-up curve appears in the experimental data. My model doesn’t capture this and I’m wondering if you have some clue about this?

Upon compression, the density is increased from 0.807 to almost 2.2 g/cm3 so the system is very dense. The force field seems to capture molecular dissociation at about the right temperature. I’m trying to figure out if this is a limitation of the model or if this type of model could capture this softening effect by further refining?

Attached are logfile from run and plot against experimental data.

Best Regards,

Magnus (FOI, Sweden)

MSST-REAXFF-N2-Us10500.txt (20.2 KB)

This is not so much a question about LAMMPS but whether the ReaxFF parameterization you use (or ReaxFF in general) is capable of representing the property you are trying to compute. Please keep in mind that classical forcefield parameterizations (and ReaxFF is effectively that) have a limited transferability, i.e. they may only be able to represent a limited range of temperature/pressure conditions well. often you may be able to represent certain phenomena, but only in a relative sense, i.e. condition 1 will happen at a higher temperature than condition 2, but the transition temperature may be off, or the influence of a second parameter may be visible, but stronger or weaker.

For more details you need to look through the published literature to see whether people have already investigated the property of interest under similar conditions, or consult with people that have intimate knowledge of the model and/or its parameterizations about whether the property you are interested in can be modeled with sufficient accuracy at all.

axel.