I am attempting to simulate water molecules between two hydrocarbon substrates (Fig. 1). The substrates comprise carbon and hydrogen atoms. At present, the reaxff potential is a suitable choice for this model, but it is very computationally demanding. Therefore, I want to use pair_style hybrid as an alternative method. As shown in Fig. 1, the airebo potential is used to describe interactions within hydrocarbon substrates and the charge equilibration is performed on hydrocarbon substrates (There are no Coulomb interactions within substrates). The lj/coul is used to describe interactions within water molecules. The interactions between substrates and water molecules are LJ and Coulomb. To obtain the Coulomb interactions, the qeq is performed on substrates. My question is whether my method involves double counting on interactions or whether my results are unreasonable and artificial.
Fig. 1
#############
hydrocarbon ---------- airebo (interactions within hydrocarbons) + qeq (to obtain atomic charge)
#############
water ---- LJ/coul (interactions within waters) + LJ/coul (interactions between substrates and waters)
#############
hydrocarbon ---------- airebo (interactions within hydrocarbons) + qeq (to obtain atomic charge)
#############
Could you please help me understand what might be going wrong in this model?
The use of charge equilibration is qeq/reaxff. I want to use qeq/reaxff to obtain the atomic charge in hydrocarbon substrates. The atomic charges in water molecules are kept constant.
The atomic charges in substrates are used to determine the coulomb interactions between substrates and water molecules.
Could you please give me some explanation as to why the use of equilibration is wrong?
Based on the manual, it seems that the charge equilibration determines the atomic charge. And there would be coulomb interactions between these substrate atoms and water molecules. I know that the airebo have already considered the coulomb interactions implicitly. And the coulomb interactions within substrates have thus been excluded.
The problems is that charge equilibration computes partial charges for compatible force field. That is why there are so many different qeq fixes. Each computes “some” charges, but their meaning outside the specific force field is very limited.
For your system, I would not even bother with this, and instead use a force field like OPLS/AA for everything. There the (partial) charges are fixed and determined through an increment system after assigning atom types following then description in the publications describing OPLS/AA. Also the water potential is set by the definition of the force field, because the force field has been parameterized to give consistent solvation behavior.
If you want a more specific and detailed analysis of the charge distribution, I don’t think that any classical model is sufficient, rather you need to do ab initio calculations.
At any rate, I don’t recall you explaining what those charges are to be used for. So please elaborate.
Do you mean that each qeq fix is used for the corresponding force field? For example, the qeq/reaxff determines atomic charges that will be used by the reaxff force field only.
Thanks for your advice, I will adopt OPLS/AA for interactions between water molecules. However, since my substrate is a carbon material incorporated with hydrogen atoms, the airebo (C,H) or ReaxFF (C,H,O) is more appropriate here based on some similar research papers. Sorry for mistakenly describing the substrate as hydrocarbons. For sure, the OPLS/AA is a suitable choice for some hydrocarbon materials. And is this model more suitable here?
Fig. 1
#############
hydrocarbon ---------- airebo (interactions within the substrate)
#############
water ---- OPLS/AA (interactions within waters) + LJ (interactions between substrates and waters)
#############
hydrocarbon ---------- airebo (interactions within the substrate)
#############
As you mentioned, charge equilibration is wrong during the model. So, I will not use qeq to determine the atomic charge anymore, since the qeq fix is designed for a compatible force field. However, there will be only vdw interactions between substrate and water molecules. Initially, I think the water molecules between two substrates will influence the atomic charge in substrates. The Coulomb interactions will affect the interactions between substrate and water molecules. So I want to use qeq initially, which perhaps leads to some artificial or wrong results. If I want to obtain the atomic charge in substrate, whether the ab initio calculations is the only method?
For the most part, yes. But the logic is the other way around: if you add add any QEq fix to a force field that is not set up for charge equilibration, you create a new force field and it will likely be bogus. The same is true, when you use a different QEq fix for a force field than what it expects. Normal ReaxFF parameter files require using pair style reaxff and fix qeq/reax or fix qeq/shielded which both implement the same QEq algorithm. Using a different QEq fix will produce charges that are not consistent with those ReaxFF parameters.
Instead, you must use fix acks2/reaxff if you use a ReaxFF parameter set generated for ACKS2. Using fix qeq/reaxff or fix qeq/shielded would be an error then.
I cannot or do not want to answer the remaining questions because they are not LAMMPS related any more (already the discussion about charge equilibration above is only peripherally a LAMMPS question), but are questions about how to do your research. This is a topic for a discussion with an adviser, tutor, or other person that has a vested interest in your research and the success of your research. I am not any of those and am not really interested in the details of your research. I have given you my perspective on this and what I would do, what you actually will be doing is your choice and something that you will have to defend and argue when you present your research or publish the results.