Different roles of similar aromatic rings in half-sandwich lanthanide single-ion magnets unravelled by ab initio electronic structure calculations

Abstract

Herein, a systematic ab initio study was carried out on a group of half-sandwich lanthanide (Ln) single-ion magnets (SIM) composed of one Ln^III ion, one aromatic ring and several non-aromatic ligands. Our theoretical predictions are generally consistent with experimental results. Although not perfect from an electrostatic viewpoint, the cyclopentadienyl (Cp⁻) ring can become the crucial ligand in Dy^III-SIM due to its capability to generate a certain metal–ligand orbital interaction, i.e., covalency, favouring SIM performance. However this case is subject to the condition that other non-aromatic ligands are weak enough. If this condition was not fulfilled, the Cp⁻ ring will not be the crucial ligand in Dy^III-SIM. Then the orbital interaction between the Cp⁻ ring and Dy^III becomes detrimental to the SIM performance. The neutral benzene (Bz) ring behaves in a way similar to Cp⁻ but requires the non-aromatic ligands to be even weaker. From the viewpoints of both electrostatics and covalency, the cyclooctatetraenyl (COT²⁻) ring is not suitable for Dy^III-SIM and thus it never becomes the crucial ligand there. The crucial ligating atom in those Dy^III-SIMs involving COT²⁻ are actually phenolic O atoms with a formal charge of −1. For Er^III-based half-sandwich SIMs, neither the Cp⁻ nor the Bz ring is a good choice. However, the half-sandwich Er^III-SIMs, involving COT²⁻ rings, show a SIM performance inferior to that of the corresponding half-sandwich Dy^III-SIMs. This is due to two points. Firstly, the orbital interaction pattern between COT²⁻ and Er^III is less favourable to SIM performance. Secondly, there exist strong ligating O atoms, not from COT²⁻, in the corresponding Dy^III-SIMs. Starting from 1, we designed a new structure 1a wherein a phenoxy group gives its formally negatively charged O atom as the trans ligating atom to Dy^III. Compared to 1, 1a is predicted to have significantly longer τ_QTM (around 14 000 second) and higher U_eff (around 2000 Kelvin). 1a holds the possibility of a T_B of 30 Kelvin, which is more than twice of that of 1.