Radical Addition Reactions: Hierarchical Ab Initio Benchmark and DFT Performance Study

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Theoretical Chemistry

Background:

• Accurate theoretical prediction of reaction energetics and mechanisms is crucial in chemistry.
• Radical addition reactions to unsaturated hydrocarbons are fundamental processes with broad implications.
• Previous studies have often relied on less rigorous computational methods, limiting predictive accuracy.

Purpose of the Study:

• To perform a hierarchical ab initio benchmark study of gas-phase radical addition reactions.
• To establish high-accuracy reference data for reactions involving methyl, amino, hydroxyl, and thiomethyl radicals with acetylene and ethylene.
• To evaluate the performance of a wide range of density functional theory (DFT) approximations against this benchmark data.

Main Methods:

• Employed a hierarchical series of ab initio methods (HF, MP2, CCSD, CCSD(T)) and Dunning basis sets ((aug)-cc-pVDZ, (aug)-cc-pVTZ, (aug)-cc-pVQZ).
• Utilized quasi-restricted orbital (QRO) reference wavefunctions to mitigate spin contamination in HF calculations.
• Extrapolated results to the Complete Basis Set (CBS) limit for high accuracy.
• Assessed 98 DFT approximations for their ability to reproduce benchmark reaction barriers, energies, and geometries.

Main Results:

• Ab initio calculations converged to within 0.0-3.4 kcal mol⁻¹ for methods and 0.0-1.0 kcal mol⁻¹ for basis sets at the QRO-CCSD(T)/CBS+ level.
• Identified several DFT functionals with Mean Absolute Errors (MAE) as low as 0.8-1.0 kcal mol⁻¹ for reaction barriers and energies.
• Best performing DFT functionals include MN12-SX, CAM-B3LYP, M06-2X, BMK, and OLYP.
• These top DFT functionals accurately reproduced key geometrical parameters, with an average deviation of 2% from the reference.

Conclusions:

• The study provides a robust benchmark dataset for radical addition reactions using high-level ab initio theory.
• Specific DFT functionals demonstrate high accuracy for predicting reaction energetics and geometries, offering reliable alternatives for computational studies.
• The findings guide the selection of appropriate DFT methods for future research on similar chemical systems.