Topic Options
#26785 - 03/16/11 11:09 AM van der Waals Mixing rules
DanSandbergUCONN Offline
Forum Member

Registered: 09/17/07
Posts: 74
Loc: United States
Hi All - I am reading a paper "Comparison of AMBER, CHARMM, COMPASS, GROMOS, OPLS, TraPPE, and UFF forcefield for prediction of vapor-liquid coexistence curves and liquid densities." By M.G. Martin, 2006, in Fluid Phase Equilibria.

He discusses "Geometric mean mixing rules" versus "Lorentz-Berthelot" mixing rules.

For the latter is shown,

e(ij)=k(ij)*sqrt(e(ii)*e(jj)), where k(ij) is an empirical constant and
a(ij)=l(ij)*sqrt(a(ii)*a(jj)), where l(ij) is an empirical constant.


Does this mean that Lorentz-Berthelot mixing rules require every pair of atoms' van der waals interactions to be empirically fit to experiment? I thought the entire point of "mixing rules" was to take an atom pair parameter, e.g. e(ij) and make it dependent on two single atom parameters, e.g. e(ii) and e(jj).
_________________________
Daniel J. Sandberg University of Connecticut Department of Chemistry Birge Research Group

Top
#26786 - 03/16/11 02:22 PM Re: van der Waals Mixing rules [Re: DanSandbergUCONN]
Kenno Online   content
Forum Member

Registered: 12/19/05
Posts: 1393
Loc: Baltimore, MD
You are absolutely correct. We're not using the empirical constants k(ij) and l(ij) in CHARMM. As far as I know, they are not part of the original Geometric or Lorentz-Berthelot mixing rules and their presence indeed negates the very purpose of mixing rules.

Edit: I believe such constants might be used by other force fields to systematically scale 1-4 interactions; however, in this case, the notation k(ij) and l(ij) is somewhat misleading because the constant would not depend on the atom types i and j but on the connectivity of the corresponding atoms. At any rate, the CHARMM force field uses an entirely different mechanism to handle 1-4 interactions.


Edited by Kenno (03/16/11 02:32 PM)

Top
#26858 - 03/21/11 07:10 PM Re: van der Waals Mixing rules [Re: Kenno]
DanSandbergUCONN Offline
Forum Member

Registered: 09/17/07
Posts: 74
Loc: United States
Thank you for the information Kenno. I guess I'm just trying to understand different mixing rules. As far as I can tell almost all mixing rules for non-similar atoms simply average, in some way, the like atom terms.
_________________________
Daniel J. Sandberg University of Connecticut Department of Chemistry Birge Research Group

Top
#26861 - 03/22/11 11:35 AM Re: van der Waals Mixing rules [Re: DanSandbergUCONN]
Kenno Online   content
Forum Member

Registered: 12/19/05
Posts: 1393
Loc: Baltimore, MD
Just for the record, CHARMM is parameterized for use with the Lorentz-Berthelot mixing rules (and NBXMOD 5), and is not expected to perform well with other mixing rules (or other nonbond exclusion schemes).

Top

Moderator:  alex, chmgr