In computational chemistry of transition metals, there are many popular basis sets, from pople+LANL2DZ mixed basis sets, to Ahlrich and Karlsruhe basis sets like def-SVPD, def2-TZVPPD. Often the ECP versions of these basis sets were used to minimise computational expenses. However there are molecular properties such as topology analysis and energy decomposition analysis where a full electron basis set is not only recommended but also necessary.

One of the full electron basis sets commonly used in the literature is the ANO-RCC basis set. However, such a basis set is so large that it computationally infeasible in applications like the EDA2 calculation in serial qchem (where there are no parallelisation implemented) and barely feasible in generating wfx files within hours for analysis. Thus a slightly smaller full electron basis set (if any) is needed to be selected.

Say I start with some ECP basis set X, which I can read the list of all exponents and angular momentum functions used, what is the recommended approach to upgrade it into a full electron version of it so it retains most of the properties of the original basis set while making it full electron. Do I need to perform a benchmarking again and compare single point energies at optimised geometries to be sure?

  • $\begingroup$ Shouldn’t you benchmarking to the property which you want to study? It is not clear what you want to study. $\endgroup$ – Greg Sep 25 '19 at 12:07
  • $\begingroup$ I want to perform a energy decomposition analysis to elucidate steric effects. The problem is that ECP basis disallow the decomposition of the frozen interaction energy term into the pauli and dispersion term thus I cannot directly benchmark that. The only thing I can benchmark is the SCF energy but I am not sure if it is enough $\endgroup$ – Secret Sep 25 '19 at 12:10
  • $\begingroup$ How SCF energy is relevant in this case? Also, what is the discrepancy you would like to check in a benchmark? $\endgroup$ – Greg Sep 25 '19 at 12:13
  • $\begingroup$ I want to e.g. check whether e.g. def-SVPD-Rifit (the full electron counterpart of def-SVPD) is a good approximate for LANL2DZ or other double zeta ECP basis sets such that I will not get qualitatively different results thus allowing the def-SPVD-Rifit result to be comparable to the energy, TS barriers and geometry of the LANL2DZ basis. However, I am not sure what benchmark parameters I need to ensure the energies calculated using the two methods will be comparable to analyse together (e.g. SCF and Gibbs free energies for the calculations with def-SVPD and LANL2DZ vs EDA2 $\endgroup$ – Secret Sep 25 '19 at 12:19
  • $\begingroup$ energies using def-SVPD-Rifit. That is, if the basis sets differ too much, I cannot draw any meaningful correlations between the EDA2 energies and the Gibbs free energy unlesss I recalculate the whole library of 1000+ molecules with def-SVPD-Rifit (all optimisations and vibrations), which is infeasibale in my project $\endgroup$ – Secret Sep 25 '19 at 12:19

Your Answer

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

Browse other questions tagged or ask your own question.