Using already existing potentials with GROMOS or OPLS simultaneously

Hello everyone!
I have a question about how to use already existing potentials and other force fields like GROMOS for example. Let’s say I want to study adsorbtion of long alkanes on the hBN surface for example. I am going to use Tersoff potential for hBN, GROMOS or OPLS-AA for alkanes and LJ parameters for intermolecular interactions. My question is that whether it is technically possible to use them at the same time?
If it is so, generally is it good idea to use these two different force fields simultaneously or do you suggest me other possible options?

[I am using latest version of LAMMPS (23 Jun 2022)]

There is no simple “yes or no” answer to this, but - like with many issues in science - the answer is “it depends”.

Generally speaking - and as your question implies you are already suspecting - it is not a good idea to mix and match different force fields. Each have their own protocol of deriving parameters and with classical models, you have too few parameters to match the real behavior, so you have to make compromises and balance different components of the force fields. Since those choices are different for different force fields, you are introducing additional - and sometimes quite drastic - errors to your model.

Technically speaking, it is possible to assign different potential functions to subsets of your system/model using hybrid styles, but these have to be used with care. The most important requirement for those to yield acceptable results, is that you have spatially separated groups of atoms where you apply the different force fields. Your case of having molecules interacting with a surface is a case where this applies. However, there are a few kinks that need to be worked out. The weak part of the model is where you have the two parts interact. So you have to choose how accurate this could be and whether the approximations you need to do are acceptable.

For example, if your surface is a metal and you have non-polar molecules, then some Lennard-Jones or Morse potential might be sufficient for combining an EAM model and some polymer. If the molecular part would be water, the situation is far more complicated, since the quite polar water will polarize the metal and thus change the interaction strength between surface and adsorbent. Since the molecules will be ordered at the surface, this effect is increased, so your model will be quite bad when only using a simple Morse potential to couple the two subsystems. On the other hand, if you plan to do some nanomachining of a metal workpiece with a diamond, that may be very sufficient and yield good results (mostly because of the pushing from the tool, the dominant interaction is the (strong) repulsion and that is orders of magnitude stronger than the absorbent force).

For your specific system, using pair style hybrid is worth considering, but you have to consider (again) how well you want (and need!) to model the interaction between the two subsystems. If you have a diamond-like structure as surface, this would be easier, but you still have to keep in mind that BN is polar and thus you would have at the very least assign suitable partial charges so that the interaction with the molecules is correct or make some tests with models confirming that those do not have a significant impact. With a graphene-like or graphite-like surface, this becomes more problematic, since you have those large delocated electron clouds that make this a bit like a 2d metal and then - similar to the metal surface example - you need to consider the polarization as well.

At any rate, the best way to go about doing these kind of studies is to first to an exhaustive search of the published literature and see whether others have studies similar systems and what kind of models they used. Please keep in mind that the kind of journal and who its reviewers are has a significant impact on what level of detail and accuracy is required to get some study published: reviewers in a journal focused on physics will be far more critical of a simulation setup, than a journal focused on materials for engineering.

1 Like