How to make geometry optimization calculation faster


Clicked A Few Times
I am trying to calculate optimized structure of a series of clusters of metal-organic framework (MOF). Then use this structure as the starting conformation to run the following molecular dynamics simulation. The number of atom varies from 500 to 700. MOF is a type of microporous material and is well-organized. To make the result reasonable in appropriate calculation time, what is the better way to rationalize the calculation? Is Hartree-Fock with STO-3G or 3-21G enough? Or I should use DFT methods?

Thanks.
Chun-Hung

Gets Around
What metals are involved? HF and minimal basis sets will probably not do well with transition metals. Both HF and DFT are going to be rather slow with hundreds of atoms.

I would actually recommend that you try modern semi-empirical methods PM6 or PM7. They will be much faster than HF or DFT calculation and offer geometries that rival DFT and surpass HF with small basis sets in most cases. If you are an academic not-for-profit user you can get MOPAC2012, which implements these methods, for free after registration: http://openmopac.net/download-c.html

Clicked A Few Times
Thanks for your reply. The metals I study are Fe, Cr, and Zn. I am from academy and would consider using semi-empirical methods by MOPAC2012.

Quote:Mernst Jul 3rd 7:29 pm
What metals are involved? HF and minimal basis sets will probably not do well with transition metals. Both HF and DFT are going to be rather slow with hundreds of atoms.

I would actually recommend that you try modern semi-empirical methods PM6 or PM7. They will be much faster than HF or DFT calculation and offer geometries that rival DFT and surpass HF with small basis sets in most cases. If you are an academic not-for-profit user you can get MOPAC2012, which implements these methods, for free after registration: http://openmopac.net/download-c.html


Forum >> NWChem's corner >> General Topics