Density functional dependence of molecular geometries in lanthanide(III) complexes relevant to bioanalytical and biomedical applications

• Published in: Computational and theoretical chemistry 2012-11, Vol.999, p.93-104
• Main Authors: Roca-Sabio, Adrián, Regueiro-Figueroa, Martín, Esteban-Gómez, David, de Blas, Andrés, Rodríguez-Blas, Teresa, Platas-Iglesias, Carlos
• Format: Article
• Language: English
• Subjects: Density functional calculations; Effective core potentials; f-Elements; Lanthanides; MRI contrast agents; Solvent effects
• ISSN: 2210-271X
• DOI: 10.1016/j.comptc.2012.08.020

[Display omitted] ► The performance of different density functionals to predict the geometries of LnIII complexes has been evaluated. ► Meta-GGA functionals and hybrid meta-GGA functional TPSSh perform better than hybrid GGA and GGA functionals. ► The use of large-core RECPs provides somewhat longer LnIII–donor distances than the SCRECP approach. ► Including solvent effects is important to obtain a more accurate description of the structures of LnIII complexes in solution. A set of 15 lanthanide-containing model systems was used to evaluate the performance of 15 commonly available density functionals (SVWN, SPL, BLYP, G96LYP, mPWLYP, B3LYP, BH&HLYP, B3PW91, BB95, mPWB95, TPSS, TPSSh, M06, CAM-B3LYP and wB97XD) in geometry determination, benchmarked against MP2 calculations. The best agreement between DFT optimized geometries and those obtained from MP2 calculations is provided by meta-GGA and hybrid meta-GGA functionals. The use of hybrid-GGA functionals such as BH&HLYP and B3PW91 also provide reasonably good results, while B3LYP provides an important overestimation of the metal–ligand bonds. The performance of different basis sets to describe the ligand(s) atoms, as well as the use of large-core (LC) RECPs and small-core (SC) RECPs, has been also assessed. Our calculations show that SCRECP calculations provide somewhat shorter GdIII–donor distances than the LCRECP approach, the average contraction of bond distances for the systems investigated amounting to 0.033Å. However, geometry optimizations with the SCRECP (in combination with the mPWB95 functional and the 6-31G(d) basis set for the ligand atoms) take about 15 times longer than the LC counterparts, and about four times longer than MP2/LCRECP/6-31G(d) calculations. The 6-31G(d), 6-311G(d), 6-311G(d,p) or cc-pVDZ basis sets, in combination with LCRECPs, appear to offer an adequate balance between accuracy and computational cost for the description of molecular geometries of LnIII complexes. Electronic energies calculated with the the cc-pVxZ family (x=D-6) indicate a relative fast convergence to the complete basis set (CBS) limit with basis set size. The inclusion of bulk solvent effects (IEFPCM) was shown to provoke an important impact on the calculated geometries, particularly on the metal–nitrogen distances. Calculations performed on lanthanide complexes relevant for practical applications confirmed the important effect of the solvent on the calculated geometries.