Better potential than Lennard-Jones for Carbon Carbon interactions?

[Signifier]

Hello! I was wondering if anyone could provide me with any advice on a better interatomic potential than the Lennard-Jones (Mie 6-12) potential for carbon-carbon interactions. I know that L-J fails for metals, but I am not sure how good it would be for carbon-carbon interactions. Are there any much better potentials to use for carbon-carbon?

Thank you for any help.

 

[quetzalcoatl9]

that really depends upon the kind of interactions that you are talking about..

Obvious, the LJ potential is only good for intermolecular interactions. It may still be a good potential to use if the many-bodied nature of your problem is "effectively" handled in a mean-field way by the force field.

for example: for carbon nanotubes systems, some people find good results using an intramolecular force-field of harmonic potentials and then C-C LJ for the long-range potential. There are also more complicated potentials used for nanotubes that are more accurate.

tell us more about your system. To speak of just carbon is too generic.

 

[Signifier]

Sorry! Yeah, that was too generic. Specifically, my system is a carbon nanotube wrapped into a torus. The carbons I am dealing with are all bonded, so I guess the L-J potential really wouldn't apply. Anyway, I am not interested in PRECISE data, so I don't need an extremely complex potential. Just wondering what might be a decent fit.

Thank you.

 

[quetzalcoatl9]

your lennard-jones forces will still apply past the 1-4 interactions

i assume you are doing MD? or just energy minimizing the structure?

i know of people using the CHARMM force-field for nanotubes and C60...i don't know about their results though.

 

[Signifier]

I am doing molecular dynamics. I am thinking about using some other program to energy minimize the structure, but such programs seem to be hard to find without paying.

Perhaps the best potential would be a modified L-J, with a quadratic term multiplied by an exponential decay to include the bonding length (IE, so small displacements / inputs of energy just cause the C atoms to oscillate, large displacements / inputs of energy can break the bond)...

 

[quetzalcoatl9]

Classical MD won't allow you to break any bonds accurately.

These sorts of studies have been done on nanotubes, why don't you search the literature?

I would suggest doing an energy minimization before MD, and then slowly heating. GROMACS, NAMD, DLPOLY, etc. are all good codes.

 

[Signifier]

I've decided to use a Buckingham potential combined with the stretch potential (and anharmonic term) from the MM2 force field, just because it's simple and at least decent. I don't know if I'm going to consider angular terms yet.

quetzalcoatl9, you seem to have good knowledge of MD stuff. I was wondering if you could perhaps recommend a free energy minimization program, one that would help me get rough cartesian coordinates for the structures of compounds (with around IE 350 atoms).

Thank you for all your help and any more help you might provide!

 

[quetzalcoatl9]

Indeed I do MD and MC for a living :) I have written MD and MC code for biomolecular and material science studies, including a current PIMC project (that is currently working - in the true Trotter fermionic case for both electron and proton - in simulating a hydrogen atom!).

well, if your structure is around 350 atoms then you could do a semi-empirical geometry optimization using something like GAMESS. An SE method like PM3 or AM1 will probably work to get a decent energy minimized structure. If you have symmetry exploitation and a supercomputer available, you _might_ be able to crank out a few optimization steps using DFT and a minimal basis set (for 350 atoms i think this is pushing it - but then again if your structure is highly symmetric it might not be so bad actually. Don't single walled tubes have Dnh symmetry?). I would be nearly certain that others have done things like this, and you could probably just grab their coordinates from the literature and focus your efforts on something else, or just use structures from X-ray or INS studies (any C-13 NMR experiments?) if the resolution is decent (i.e. ~1.0 A).

also, if I may ask - why 350 atoms? i assume you have more than just a standard nanotube (i.e. one that is contorted in some way and breaking symmetry).

are you studying normal modes by chance?

 

[chuckschuldiner]

Have you considered using the Tersoff potential. Unlike, LJ it is not a pair potential so it should be more accurate. You can do some literature search to see if people have used tersoff like potentials for nanotubes. As far as i know, http://www.ivec.org/GULP/" is one of the MD programs that implements tersoff. Using GULP is fairly simple. However, GULP is not paralellized...but considering the system sizes you are talking about, it should be OK . Also, if i can remember,a MS Windows version of GULP also exists. Also Brenner has his own set of Fortran codes which you can d/l by googling for them.