Could Someone Give me Advice on Optimizing Basis Set Selection for Calculating Reaction Energies?

Hello there,

I am working on a research project where I need to calculate reaction energies for a series of chemical reactions using Psi4.

I have encountered some challenges in selecting the appropriate basis set for my calculations and would greatly appreciate some advice from those experienced with Psi4’s capabilities.

My goal is to ensure that the basis set I choose provides accurate reaction energies while also being computationally feasible for larger systems.

I have been considering basis sets like cc-pVDZ and cc-pVTZ; but I am unsure about their suitability for my specific reactions; which involve both organic and inorganic molecules.

What factors should I consider when selecting a basis set for calculating reaction energies in Psi4? :thinking:

Are there recommended basis sets known to perform well for a wide range of chemical reactions; particularly those involving transition metals?

How can I assess the balance between accuracy and computational cost when choosing between different basis sets? :thinking:

Also; I have gone through this post; https://forum.psicode.org/t/how-to-user-define-a-df-basis-dcft-basis-in-psi4-uipath/ which definitely helped me out a lot

Additionally; if there are any best practices or resources you could recommend for understanding basis set selection in Psi4; that would be incredibly helpful.

Thank you in advance for your help and assistance. :innocent:

I’m confused. What about that post was helpful? That post is about a bug which is both very specific to dct methods and resolved years ago.

…And yes, I do need an answer for that before I can answer your actual question.

The Karlsruhe def2 basis sets are a good place to start. They are available for the whole periodic table, they have effective core potentials for the heavy elements, and they are widely used also in industry.

The biggest factor to consider is the level of theory: density functional theory converges much faster to the basis set limit than post-Hartree-Fock theories like coupled-cluster do. Generally, DFT calculations employ a triple-zeta basis like def2-TZVP (or the larger def2-TZVPP), as this is considered to yield the best tradeoff between cost and accuracy.

As Susi said, it really depends on the level of theory. I think this article will probably be a good start: “Best-Practice DFT Protocols for Basic Molecular Computational Chemistry” by Stefan Grimme via https://doi.org/10.1002/anie.202205735 or just google the title.

I’d stay away from the old Pople basis sets such as 6-31G/6-311G*, as there are many more modern basis sets that are better. Also, a good rule of thumb for beginners regarding basis set and level of theory combinations is: DFT and the def2 basis sets are great together, while for post-HF CCSD(T) use the correlation-consistent sets like aug-cc-pVTZ.