9
$\begingroup$

Most quantum chemistry packages are capable of computing ZPE and thermal corrections for a given temperature. But how big is the influence of the basis set chosen and the functional used?

Is it sufficient to calculate this at a lower level than the corresponding final high-level single point calculation? What about differences between the energies - if there is an error, is it systematic so that energy differences are not affected?

$\endgroup$
  • 6
    $\begingroup$ The short answer is without benchmarking you will not know. The slightly longer answer is, that these corrections only make sense for well converged optimised structures. It is not too uncommon, to use these corrective terms and add it to a single point calculation of higher level. Computing frequencies at the SP level should give you (almost) the same corrections and is a waste of resources. $\endgroup$ – Martin - マーチン Feb 12 '16 at 20:37
  • $\begingroup$ I agree on Martin's comment - that's what I usually read in (benchmark) papers. $\endgroup$ – pH13 - Yet another Philipp Feb 14 '16 at 12:14
  • 1
    $\begingroup$ I agree with all the comments, but I want to add: Please also consider what kind of method you are using. A DFT calculation on a transition metal complexes has error bars where I wouldn't even consider ZPE corrections at all without a very good reason. A small molecule where you want the heat of formation as accurately as possible? Do whatever you can afford. I am surprised that you did not mention a method or system to be honest. $\endgroup$ – AMT Feb 23 '17 at 15:04
1
$\begingroup$

This is more of a comment, apologies.

Consider this paper (Ruth L. Jacobsen, Russell D. Johnson III, Karl K. Irikura, Raghu N. Kacker: Anharmonic Vibrational Frequency Calculations Are Not Worthwhile for Small Basis Sets, J. Chem. Theory Comput., 2013, 9 (2), pp 951-954). The magnitude of the computational-level-dependent scaling factors for the frequencies reported therein should give you a sense of the magnitude (I would also think that at least some other studies of the topic are cited).

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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