GROMACS version:4.5.5
GROMACS modification: Yes/No NO
Here post your question
Hello,
does anyone here as experience in simulation liquid crystal through highly idealized
model? Typically, consider a fused soft sphere linear molecule. This would be a
rigid rod-like molecule, with fixed sphere-sphere distance, say of half diameter.
I managed to setup such system, and simulate with Gromacs 4.5.5. However,
I am unable to stabilize an ordered nematic phase, even when starting from a
perfect ordered system. The system ends up in a disordered bunch of small
ordered aggregates. I tried with 5 and 7 spheres, with very similar results.
This sounds like a force field issue to me. What force field are you using and has it been parameterised for the liquid crystal you are simulating?
Thanks Magnus for your reply.
My guideline is the fact that, in theory, these models should show a nematic
phase at high densiy, and possibly a smectic phase at even higher ones.
I use liquid state theories (integral equations) to compute various structural
quantities, and I would like to check how accurate these are against “exact”
results as provided by simulations. I have a Monte Carlo code that finds
the nematic phase, but it is not as fast as Gromacs, and I cannot use it for
large systems where other interesting phenomena appear.
The force field is very simple: I use a 1/r**12 repulsion for each pair of spherical
sites. This should in principle mimic a hard rod-like particle.
I have already checked many things using my slow Monte Carlo code. I never found this strange situation where all the rods pack in small parallel aggregates which themselves pack in disordered configuration. It sounds like some metastable phase, and I wonder why Gromacs would go straight to this phase?
I’m afraid it’s very difficult to say what is going wrong. I would not expect any internal GROMACS errors to cause this difference. I still suspect a force field issue (are the units correct?), system setup errors or errors in the simulation parameters. But it is very difficult (at least for me) to say from the given information. It is of course possible that there are users on the forum who have experienced similar problems, but I am afraid you might have to wait long for a good answer.
I guess you are using tabulated interaction potentials, so I’m afraid I cannot recommend you to test it in a modern and supported GROMACS version, to rule out (or find) potential errors in the code.