Forums
I am thinking about incorporating some calculation exercises into my inorganic course in the fall (possibly something like the Using Computational Chemistry to discuss backbonding in CO). We currently have a license for Spartan (which can certainly do a number of different types of levels of calculations, including DFT). Has anyone else used Spartan for such things? If so, what basis sets did you use for your calculations?
Hilary,
It depends!
I used to use Spartan for a lots of stuff - a full lab in a second-year structure/bonding course, and various complements to experimental labs in organometallics, main group, and so on. I found that it's worth your while to play around with different levels of theory and different basis sets, and see what least amount of computation gives you the results you're looking for. You could run everything at a DFT level with extra polarizations and diffuse functionals, but there's no reason to do so if a simpler method gets you answers that teaches your students what they need to figure out. You want this to be an exercise in which computational results are used to help understand the experiments, not one in which students sit there for 30 minutes (or hours) for their geometry optimization to finish.
Most of the time you don't need the high-level quantitiative accuracy that a DFT with big basis set will give you. For simple, qualitative calculations (atomic orbital energies, VSEPR shapes, even MOs if you just care about what they look like and not so much about their energies), a semi-empirical PM3 is fine. For small main group molecules, HF/3-21G is good for most stuff, 6-31G for greater accuracy of geometries. We did studies where we were calculating geometries, total energies (and then reaction energies), bond strengths, dipole moments, charge distributions and spin density, vibrational frequencies, etc etc.
One nice thing about Spartan is that they have some home-brewed PM3 parameters for transition metals, so small coordination compounds are much faster than they would be otherwise. Just at the PM3 level you can get useful information about M-CO orbital overlaps and stretching frequencies. We did stuff like that: looking at IR spectra in isomers of L2Mo(CO)4 and L2Fe(CO)3, for example.
I've since moved away from Spartan, and we're using WebMO interfaced with Gaussian. It offers a number of advantages, but you lose Spartan's d-block PM3 paramterizations.
Let me know if you want any specifics about the sorts of things we've done at UBC Okanagan; I can just send you lab manuals.
WSM
Hello, Stephen.
I also have been using SPARTAN for six-to-eight years. Two questions about WebMO:
1. What are those advantages?
2. What is the cost?
Thank you.
Craig Davis
WebMO isn't a computational package as much as an easy to use interface that is accessible from the web. As far as I know, it doesn't interface directly with Spartan, but there are free packages with similar capabilities that it does work with (Gamess, NWChem). WebMO is free, as long as you don't need some of the advanced features like integration into cluster management systems like Torque or SGE.
Really, if you and your students are comfortable and happy using Spartan on just a few machines, there's probably no need to switch. If you think you're going to start adding a lot more users who want to be able to access their calculations from anywhere, then WebMO would be a good choice.
Hello, Gerard.
Thank you for your response.
Craig.