As always, keep the LAMMPS list in on the communications, in case people such as yourself find this useful in the future.
Dear Eric,
Sorry to bother you.
I am Wenjuan, a graduate student in Civil Engineering in Virginia Tech. I am working on a conatct model between discrete particles and atoms to consider van der Waals interactions just like whatc colloid package does in LAMMPS.
The thing of using colloid package in LAMMPS is that computation will stop once a discrete particle or an atom approach (or run into) another particle, due to infinite value of van der Waals potential energy. Such a problem happens a lot while the radius ® of particle is great, such as R = 5 nm. I tried to solve this problem by replacing the contact model of van der Waals for overlapping particles with a new contact model that considers Hooke contact and adhesion part of van der Waals interaction for particles that are overlapping with each other or very close to each other, when van der Waals potential is infinite. However, I have a problem to smoothly match van der Waals long-range interaction with adhesive part of VDW + Hooke repulsion. I have problems in setting force and energy in the computation.
So, its important here to distinguish what we’re talking about. The van der Waals potential (in all but a few exceptional entropically dominated systems) is always adhesive/attractive. It does occur in one form in the colloid package, and there are many forms it can take depending on the geometry of your particles. For spherical particles in the micron size range things get simple (I suggest picking up Israelachvili’s book if you haven’t; there’s a length scale here you should become familiar with). Also I’m glad you’re aware DMT is your only hope. 
You might think on if a granular model is appropriate for particles with R~5nm, my guess would be that it is not. If its really 5nm I would thing the colloidal repulsive core would do fine.
I am going to talk about a -1/r attractive potential with no 1/r^Higher power repulsion, which reaches a minimum at a length-scale often denoted sigma in the lammps literature. Note that this is a meaningless separation distance for macroscopic particles, so its more of a parameter (a separation distance particles effectively see, when they are really ‘in contact’. Fortunately the 1/r goes to 0 very close to particles, so smoothness at the edge of the well isn’t a real worry. The simplest way to get this to work is by doing some alterations to the colloid package, you’ll have to redefine some force/energy calcs for consistency and removes some flags. Fortunately, you can get a few simple analytic results to validate if your potential looks like it should.
I saw on lammps mail-list that you discussed with Jeremy and Steve on this problem three years ago, and I am wonder whether you’ve got some progress in solving this problem. I would be really appreciated if you have any suggestions on it. Thank you so much!
And now for the user beware. I’ve done these alterations, but there are MANY minutiae (too many for me to feel comfortable giving it to someone else). First, and most severely, you will find that there will be a HUGE force discontinuity at contact due to the dashpot in the Hooke model. This will put severe limitations on speed due to the fact you will need to find a region in parameter space where the discontinuity is not too large (if it is too large you will not be able to reproduce analytic results for binary collisions, also important for things like rheology). Using the Hertzian (aka Kuwabara-Kono model) alleviates this problem if you want to go this route. Second, a new time-scale emerges (in the sense that it wasn’t in the granular problem) that is due to the fact that you need to resolve the particles crossing the well. Since the well is MUCH smaller than the particle for granular problems this is another source of slowness in the simulations.
Hope this helps some, its not too much work but you’ve got to be careful (there’s a lot of not so thorough plug and chug work out there esp. in this macro/micro cross-over area). I would think on whether you really need granular interactions before crawling into this can of worms. Good luck.
Best regards,
Wenjuan