Gauging the Performance of Density Functionals for Lanthanide-Containing Molecules.

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Lanthanide Chemistry

Background:

• Accurate prediction of thermochemical properties for lanthanide compounds is challenging.
• Existing computational methods require validation for lanthanide systems, especially those in unusual oxidation states.

Purpose of the Study:

• To assess the performance of various density functional approximations (DFAs) for predicting enthalpies of formation and bond dissociation energies in lanthanide-containing molecules.
• To introduce and utilize the Ln54 dataset, encompassing lanthanides in diverse oxidation states, for benchmarking computational methods.
• To propose a "lanthanide chemical accuracy" threshold of 5.0 kcal mol⁻¹ for experimental comparison.

Main Methods:

• Evaluation of a wide range of density functionals, from local density approximation (LDA) to double hybrids.
• Application of these functionals to the curated Ln54 dataset, including molecules with lanthanides in both common (3+) and exotic oxidation states.
• Statistical analysis of mean absolute deviation (MAD) and mean signed deviation (MSD) to assess functional performance.

Main Results:

• Hybrid functionals generally underperformed compared to functionals lacking exact exchange.
• The TPSS functional demonstrated the best overall performance on the Ln54 set, with a MAD of 19.2 kcal mol⁻¹ and MSD of -1.9 kcal mol⁻¹.
• Performance varied across different classes of lanthanide-containing molecules.

Conclusions:

• TPSS is identified as a reliable functional for calculating properties of lanthanide-containing molecules within the proposed "lanthanide chemical accuracy".
• The study highlights the importance of selecting appropriate DFAs for accurate lanthanide chemistry predictions.
• The Ln54 dataset serves as a valuable resource for future benchmarking studies in this field.