How good is DFT for no-pairs ferromagnetic bonds?
Luis Rincón
Universidad San Francisco de Quito (USFQ), Grupo de Química Computacional y Teórica (QCT-USFQ), Departamento de Ingeniería Química, Colegio Politécnico de Ciencias e Ingeniería, Diego de Robles y Via Interoceanica, Quito 17-1200-841, Ecuador
The electron pairing constitutes the principal bonding mechanism in almost any molecule, in contrast, triplet pairs bonds is an elusive bonding mechanism due to the Pauli repulsion of identical spins. Nevertheless, nowadays it is accepted that high-spin clusters of univalent atoms (Li, Na, Cu, Ag, Au), even without having a single electron pair in the valence shell, should be weakly bonded.1,2 In these systems, the triplet repulsion need to be compensated by an attractive interaction which is a consequence of the electron correlation. This bonding mechanism was called no-pair ferromagnetic bond. Because real no-pair ferromagnetic bonds are typically investigated using modern DFT calculations, it seems important to check the accuracy of common exchange-correlation functional in these high-spin unusual situations. In this work, we will explore the performance of 20 functionals, compared to CCSD(T), for the triplet dimers of Li, Na and Cu, and some of their small clusters. The tested functional involved pure-Local, pure-GGA, meta-GGA, hybrids and double-hybrids. In all cases, DFT functionals show very small distances, for example, in the best hybrid functionals a shortening of 20-30% in the bond distance compared with CCSD(T) is observed. To understand this behavior, we study the electron localization due to the influence of the Coulomb hole using the information content of the conditional pair density.3
1. D. Danovich and S. Shaik “Bonding with Parallel Spins: High-Spin Clusters of Monovalent Metal Atoms” Acc. Chem. Res. 47, 417-426 (2014)
2. D. Danovich and S. Shaik “On the nature of bonding in Parallel Spins in monovalent metal clusters” Annu. Rev. Phys. Chem. 67, 419-439 (2016)
3. A.S. Urbina, F.J. Torres and L. Rincón “The electron localization as the information content of the conditional pair density” J. Chem. Phys. 144, 244104 (2016)