Are there any semiempirical methods that work with heavy alkaline earth metals?

Question

I did some research. Found some semiempirical methods that work even with heavy lanthanide compounds (RM1). But could not find any references to semiempirical methods compatible with the heavier members of the alkali earth group: Sr, Ba and Ra. If there is such method, does any open source / free software implements it? I'm learning to use Gamess-US, but the only semiempirical methods implemented in this package are MNDO, AM1, PM3 and RM1 (refs file).

Answer 1

The implementations of MNDO, AM1, and PM3 along with PM6 and PM7 in MOPAC include Sr and Ba, but not Ra. MOPAC is free for academic use, but modern versions of it are not open source. Most of these models have an open-source implementation in SPARROW (GitHub). An entirely different class of semiempirical models (GFNx) is available in the open-source XTB code (GitHub), which also includes Sr and Ba, but not Ra. The lower portions of the periodic table (lanthanides and beyond) are of less general interest to scientists and have less reference data for fitting purposes, so there is lower demand and higher difficulty in covering this portion of the periodic table. The only parameterization that I know of that covers Ra is the QUASINANO2013 parameter set (J. Chem. Theory Comput. 2013, 9 (9), 4006–4017) in DFTB, which is proprietary and only available in the commercial implementation of DFTB by SCM.

If you only need Ra to behave like a +2 cation with a slightly larger ionic radius than Ba, then you probably could modify the parameter set of one of the aforementioned models and tune the repulsion parameters of Ba to increase its ionic radius in a few reference ionic systems. That would of course take a bit of effort, in both learning what the parameters of the model mean, where they are stored in a particular simulation code, identifying some suitable reference data, and tuning one or more parameters to fit the data.

I'm sorry that there isn't a simple, affirmative answer to your question, but I hope that this helps anyways.