14
$\begingroup$

I know of several general molecular mechanics methods suitable for metals, including Dreiding and UFF.

I know it's difficult to get many things right for such classical treatment of inorganic / organometallic molecules, because oxidation state, spin state and other factors can greatly influence geometries and energetics.

But I'm surprised that so few attempts have been made to improve on these methods. Are there no other methods published since 1992? I'm looking for force fields that cover a substantial part of the periodic table (or at least most metal-ligand interactions) and are completely published - proprietary methods don't count.

Improve this question
$\endgroup$
2
  • $\begingroup$ Paging @F'x - I believe you have some expertise here.. $\endgroup$
    – Geoff Hutchison
    Commented Dec 29, 2015 at 20:16
  • $\begingroup$ I think the cynical answer is that it's probably very hard to get funding to do something like this now. I'm looking for more technical answers. Thanks! $\endgroup$
    – Geoff Hutchison
    Commented Dec 30, 2015 at 14:28

2 Answers 2

Reset to default
5
$\begingroup$

The modelling of metals using a 'simple' potential is relatively difficult. I have a suggestion for a force field, though it does not fit exactly all your criteria, but I would like to offer it anyway.

Adri van Duin has developed a force field called ReaxFF. This force field is non-proprietary and is readily available in the open-source package LAMMPS. This force field gives pretty good results and can be applied to any kind of elements, it can describe chemical reactions and is also valid for hetero-atomic interactions. That being said, the biggest challenge when using the ReaxFF force field is that you have to fit / train it to your system. This requires, for instance, a database of ab initio calculations that are representative for your system.

If you cannot find any force field, than this might be a interesting alternative to developing a force field completely from scratch.

Improve this answer
$\endgroup$
3
  • $\begingroup$ I know of ReaxFF, but my understanding is that there isn't a generally-parameterized version like UFF. People end up tailoring the parameters for reactions of interest. So I think of ReaxFF as a "family" of force fields, rather than a general one. Is there a "base" parameterization for general use? $\endgroup$
    – Geoff Hutchison
    Commented Dec 29, 2015 at 18:38
  • $\begingroup$ Your understanding is correct. :-) That is why I mentioned in my answer that it does not exactly fit all your criteria. The advantage of fitting a force field though, if you base it on ab initio calculations, is that you tend to get better results than using a 'general' force field. $\endgroup$
    – Ivo Filot
    Commented Dec 29, 2015 at 18:40
  • $\begingroup$ I think most new parameterization uses high-quality calculations with schemes like ForceBalance. My point is rather that this seemingly productive area of research (e.g., "fast geometry cleanup") has quietly, while other areas of MM have gained significant advances. $\endgroup$
    – Geoff Hutchison
    Commented Dec 29, 2015 at 20:08
4
$\begingroup$

You could have a look at VALBOND from Clark Landis, don't know how generally it was parameterized. LFMM from Rob Deeth is very good, but not many metals. I think old PCModel is still around, one of the better for metals. Allinger did a metal extension in MM3(94), with generalized parameters for the periodic table, but not much used. I've created a number of specific metal force fields based on MM3* in MacroModel, but nothing general. The easiest is usually a POS model with bonds from covalent radii, but no directionality.

Improve this answer
$\endgroup$
1
  • $\begingroup$ Wow, great to get an answer from you! I know that UFF incorporates VALBOND concepts, as does ReaxFF. I'm aware of LFMM and similar ligand-field schemes, but as you say, they don't seem very general. $\endgroup$
    – Geoff Hutchison
    Commented Dec 30, 2015 at 14:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.