Kent Bower, Chemistry
The use of Molecular Mechanics has taken a strong foothold in the world of computational chemistry today. Molecular Mechanics methods offer quick solutions to complex molecular conformation calculations that could take up to days or even months to perform on supercomputers using more exact methods, such as ab initio calculations. For this speed the price of accuracy is often paid when using Molecular Mechanics methods, particularly when the system under inspection is very specialized. For example, much current research centers on the conformations of crown-ether systems with metal-cation ligands. The C-C-O-C-C pattern in these ethers exhibits specific properties which make Molecular Mechanics often inadequate for them.
Our research centered on improving the van der Waals term used in nearly all Molecular Mechanics force-fields. The goal was to incorporate the work of Waltman, et al. into a force field tailored for use with crown-ether systems.
After applying Waltman’s “New Combination Rules” for van der Waals parameters to several force-fields, we found ambiguous results. There appeared to be no general improvement in the MM results when compared to ab initio calculations on the same systems.
We decided to take a step back to see whether the error produced by the method of van der Waals calculation, Lennard-Jones 12-6 in our case, was small enough to allow for improvements in combination rules to do any good. In other words, we asked, “Can any Lennard-Jones 12-6 parameters accurately model the true van der Waals potential between two atoms?”
To start, we chose hydrogen atoms. We used the work of Wind & Roeggen 2, who accurately modeled the hydrogen dimer-hydrogen dimer van der Waals potential surface for comparison.
Using a least-squares fit strategy we found several candidate parameters for the Lennard-Jones 12-6 van der Waals potential between two hydrogen atoms. With these parameters, we generated a large potential energy surface for many conformations of two hydrogen dimers.
What we found was that the Lennard-Jones 12-6 strategy apparently doesn’t give enough freedom to consistently model even major trends in the potential energy surface of two hydrogen dimers. The same type of research for methane molecules is currently underway.