CCL: Choice of multiplicity for NiO

 Sent to CCL by: Susi Lehtola [susi.lehtola||]
 On 04/15/2018 10:54 AM, Grigoriy Zhurko wrote:
Sent to CCL by: Grigoriy Zhurko [] Hello, I compute Ni4O4, Ni9O4 clusters and their complexes with
 carbon monoxide (me research is devoted to CO absorption on NiO).
 These systems can be computed in different multiplicities. Two
 approaches can be used for choosing the multiplicities: 1) Compute
 each multiplicity and find the one with minimal energy; 2) In the
 work of Thomas Bredow, only the multiplicity 19 was chosen for Ni9O9.
 He writes in his paper:
The first option works in simple cases, but for transition metal compounds in general - and transition metal oxides appear to be especially nasty - getting the right energy ordering for the different spin states may be highly nontrivial. Usually DFT gets it wrong, i.e. you get different orderings with different functionals, in which case the results shouldn't be trusted.
But, ab initio methods may not work very well either. In addition to significant effects due to strong correlation (for which you need something like CAS-SCF or DMRG-SCF), dynamic correlation can also play a significant part in the energy splittings between the spin states: the true spin state ordering is due to a delicate balance between static and dynamic correlation.
However, your system, Ni9O9 would appear to be well outside the computational tractability of reliable ab initio methods. So, I guess most people would just go with option 1: pick a functional that (happens to) give the right spin state ordering for a closely similar system, and then pick the spin state that has the lowest energy.
 Mr. Susi Lehtola, PhD             Junior Fellow, Adjunct Professor
 susi.lehtola]![   University of Helsinki  Finland
 Susi Lehtola, dosentti, FT        tutkijatohtori
 susi.lehtola]![   Helsingin yliopisto