This paper describes the analysis of methods and basis sets for theoretical calculations on oxorhenium(V) complexes. A variety of ligand systems with different coordinating atoms (N, O, P, S, Cl, and Se) were incorporated into this study. Two types of geometries viz. ReOX3 and ReOX41- were considered for model compounds M1 – M12 while structures of some complexes (L1 – L5) related to rhenium radiopharmaceuticals were collected from CCDC (section 2.1). The authors’ effort at standardizing a procedure for DFT calculations for oxorhenium complexes is a good one and the results indicate reliably accurate output geometries at low computational costs and the authors comment that “Single-point energy calculations at the PBE0/6-311+G*, LANL2TZ level of theory are recommended for energetics”.
A variety of combinations were used to identify the correct combination of functional and basis set. The basis set and ECP for Re were LANL2DZ, SDD, LANL2TZ, LANL2TZ(f), aug-cc-pVTZ-pp. Furthermore, 6-31G*, 6-31G**, 6-31+G**, 6-31++G*, 6-31++G**, 6-311G*, 6-311+G*, LANL2DZ, SDD, Midi!, aug-cc-pVTZ all were tested for other elements. For Se, the sets LANL2DZ, LANL08 and LANL08d were used.
Goodness of fit (GOF) calculations was carried out by comparing the model structures with the MP2/ACPTZ optimized geometries. “This choice is supported by numerous previous studies that have employed MP2 optimized or reoptimized structures as a reference for transition metal complexes, especially for (but not limited to) closed-shell, singlet species [24–33]”. For compounds L1 – L5, the optimized structures were compared directly with the X-ray crystal structures found in CCDC.
Complexation energies for ReOX3 and ReOX41- were put to GOF calculations “as the absolute value of the difference between the CCSD(T)/ACPTZ//MP2/ACPTZ energy and that from the method/basis set employed (UE complexation )”.
The inclusion of compounds L1 – L5 is indeed a good step, as it benchmarks the theoretical methods for predicting properties of rhenium radiopharmaceuticals.
The findings are consistent with the methods, but I found some other systems like pbe1pbe/ STMIDI to be more close to benchmarks in case of ReO(SH)3, ReOCl3 and ReOCl4-.
The non-inclusion of Se in autoscaling calculations or the overall rank for Table 4 renders the work somewhat incomplete, but it is understandable, as basis sets other than LANL2DZ, LANL08 and LANL08d could not be employed for it. The “Stuttgart RLC ECP” basis set can perform well for Se and it can be tested.
The authors should also have included ReS4 systems in line with the findings reported by Safi et. al in J. Phys. Chem. A 2005, 109, 1944-1951.