Multiwfn official website: http://sobereva.com/multiwfn. Multiwfn forum in Chinese: http://bbs.keinsci.com/wfn
You are not logged in.
Dear all,
I am studying the mechanism of an organic reaction involving a hydride transfer.
Looking into the various basis sets on BasisSetExchange, I looked into the basis functions and coefficients for hydrogen atoms and am puzzled to notice this:
- the difference between 6-311+G** and 6-311++G** is an 's' function
- the difference between def2-TZVP and def2-TZVPD is a 'p' function
What is the reason for the differences in Pople vs. Karlsruhe basis sets?
Why is the diffuse function an s orbital in one case, and a p orbital in the other ?
How important is that difference and what impact will it have on the accuracy of the calculation?
Thanks in advance!
Offline
The design idea of def2-D series of basis set is very different to common basis sets (such as 6-311G series), which are suitable for general calculation and the exponents are usually variationally optimized for representing electronic energy. The diffuse functions of def2-D basis set were specifically optimized for polarizabilty calculation, for hydrogen it is found that adding a 'p' diffuse function plays a much more significant role than adding a 's' diffuse function, therefore only 'p' diffuse function is supplemented to the original def2 basis set.
See J. Chem. Phys. 133, 134105 (2010) for more detail, which is the original paper of def2-D.
Since def2-D basis set doesn't aim at general purpose calculation, and the 's' diffuse function important for representing H- is not available, using it for studying hydride transfer is not recommended.
Offline
Thank you for you reply. The hydride I'm studying is transferred from a boron to a carbon atom, so it is not a 'free' hydride, and the highest transition state (the H- transfer) has C-H-B atoms quite linear.
The whole energy surface (geometry optimization and thermochemistry) has been calculated with 6-311++G** for the whole molecule bearing 'light atoms' (heaviest is nitrogen), but now I am looking for a higher basis set for single point calculations.
Would it be recommended to use a mixed basis set of def2-TZVPD for all non-hydrogen atoms, and 6-311++G** for hydrogen atoms only ? Or is it best to not do single points for electronic energies with def2 sets at all?
Thanks in advance.
Last edited by lmangin (2021-01-24 18:14:35)
Offline
In your case, diffuse function of hydrogen is not absolutely needed. You can compare the result of 6-311+G** and 6-311++G**, I believe the calculated barrier should be very similar. It is expected that def2-TZVP should work much better than 6-311++G**.
If you indeed hope to use a high-quality basis set with general-purpose diffuse function for hydrogen, I suggest considering aug-pcseg-2. pcseg-2 has comparable quality with def2-TZVP, and the "aug-" adds diffuse functions for all atoms, and hydrogen has 's','p','d' diffuse functions. If you are using double-hybrid functional or post-HF method, you can also consider aug-cc-pVTZ.
Offline
Thanks again for your time and your answer. The study is done with B3LYP and M06-2X functionals.
I forgot to mention that in the system, there is a counter anion and some intermediates in the whole process bear fluorine atoms like (F3C-SO2)2N- anion. This is the main reason I had added diffuse functions in Pople's basis set for the geometry optimizaiton and thermochemistry.
You mentioned that def2-TZVP would work better than Pople's and I totally agree but was rather aiming at using Karlruhe's for single points only (so the calculation is less exepensive in time). But now, considering the presence of anion and fluorine atoms, I need to keep diffuse functions, so I guess def2-TZVPD would be more appropriate that def2-TZVP... but again there will only be a 'p' diffuse function for the hydrogens.. that's why I'm wondering if it was reasonable do the single points with def2-TZVPD for non-H and 6-311++G** for H; or if using def2-TZVPD for all atoms (including H) would cause a problem/errors in the electronic energy calculation.
Otherwise, I will read over the pcseg series and consider it for the electronic energies.
Thanks in advance for your input!
Last edited by lmangin (2021-01-24 23:38:30)
Offline