I am building a crystal lattice of organic molecules. For all of the constitutent atoms OPLS-AA parameters are available. I felt that re-evaluation of partial-charges using GAMESS software with HF-6/31G* method would provide us with more accurate results, however, I later realized that OPLS-AA being a complete force-field it would be improper to modify the parameters. Am I correct or should I use re-evaulated charges ?
The short answer is yes, since there is no right or more accurate answer when it comes to point charge calculations. However, this depends on how this will affect the overall goal of your calculation. You must have some way of justifying your use of a different set of parameters. The Mulliken point charge calculations are many times (but not always) a reasonable approximation, but there are other methods out there for calculating point charges.
Notice that there are numerous ways of evaluating atomic charges, such as Mulliken, CHELPG, RESP, ADCH, Hirshfeld, CM5 and so on. Irrespective of the choice of the calculation method, deriving the charges based on the very poor HF/6-31G* wavefunction should always be avoided.
In J. Phys. Chem. B, 121, 3864 (2017), it was proven that the "1.2*CM5" atomic charge is well compatible with OPLS-AA forcefield, it is simply 1.2 times of CM5 atomic charges. CM5 charge can be easily evaluated using subfunction 16 of main function 7 of Multiwfn (http://sobereva.com/multiwfn) based on the .wfn/wfx/fch/molden... file produced by mainstream quantum chemistry codes (GAMESS-US output file is also supported). In addition, an atomic charge model very popular in MD field is RESP, which can also be very conveniently evaluated by Multiwfn (see Section 4.7.7 of Multiwfn manual for examples), but the compatibility between RESP charge and OPLS-AA is not quite clear.
Dear Hrushikesh,
as it was pointed out before, the question shouldn't be about the "correctness" or not of the modification of partial charges of a generic force field in order to be used in a specific case. In generic force fields (like OPLS, TraPPE, AMBER, DREIDING etc...) the suggested atomic point charges for specific atom types have been obtained using ab initio calculations of a wide range of representative molecules from several categories (eg alcohols, ethers, ketones, amides, amines etc...) and have been presented then in a generic way, combined with proposed 'optimum' parameters to describe Van der Waals interactions (most commonly via Lennard-Jones potentials). These parameter sets have been tested for a wide range of compounds and the goal was to provide a reasonable description of mainly thermodynamic, transport and dynamic properties of all these compounds using these generic force field parameters. Therefore, in each specific case someone has to test first if these parameters describe properly the properties of interest for each specific system under investigation. If not, then you could start modifying the force field parameters, like the charges that you mentioned in your questions. However, this modification is not trivial at all. First of all, you should use the most appropriate level of theory (eg MP2, DFT etc...) taking into account the size (unfortunately this is still something you should take into account if you need to study very big molecules) and chemical composition of your system and then the appropriate basis sets (and functionals as well in the case that you use DFT) to describe better your system. This is a quite complicated task and you should check each time the literature, to see which combination is proposed for similar systems to the one you want to investigate. Then, as it was mentioned, you should be aware that there are many methods to extract atomic charge distributions and each one can give quite different results. A very popular one is the CHELPG method, which is based on the reproduction of the electrostatic field around a molecule, and charges extracted by this method are quite reliable for condensed phase calculations. But, as I've said, nothing is trivial and you should check which method provides you the best description of the properties of interest for your system with available experimental data.
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Atomic, Molecular and Optical Physics
- Atoms