Hydrogenation of CO2 to Formic Acid with a Highly Active Ruthenium Acriphos Complex in DMSO and DMSO/Water.

The novel [Ru(Acriphos)(PPh3 )(Cl)(PhCO2 )] [1; Acriphos=4,5-bis(diphenylphosphino)acridine] is an excellent precatalyst for the hydrogenation of CO2 to give formic acid in dimethyl sulfoxide (DMSO) and DMSO/H2 O without the need for amine bases as co-reagents. Turnover numbers (TONs) of up to 4200 and turnover frequencies (TOFs) of up to 260 h(-1) were achieved, thus rendering 1 one of the most active catalysts for CO2 hydrogenations under additive-free conditions reported to date. The thermodynamic stabilization of the reaction product by the reaction medium, through hydrogen bonds between formic acid and clusters of solvent or water, were rationalized by DFT calculations. The relatively low final concentration of formic acid obtained experimentally under catalytic conditions (0.33 mol L(-1) ) was shown to be limited by product-dependent catalyst inhibition rather than thermodynamic limits, and could be overcome by addition of small amounts of acetate buffer, thus leading to a maximum concentration of free formic acid of 1.27 mol L(-1) , which corresponds to optimized values of TON=16×10(3) and TOFavg ≈10(3)  h(-1) .

Can Contemporary Density Functional Theory Predict Energy Spans in Molecular Catalysis Accurately Enough To Be Applicable for in Silico Catalyst Design? A Computational/Experimental Case Study for the Ruthenium-Catalyzed Hydrogenation of Olefins.

The catalytic hydrogenation of cyclohexene and 1-methylcyclohexene is investigated experimentally and by means of density functional theory (DFT) computations using novel ruthenium Xantphos(Ph) (4,5-bis(diphenylphosphino)-9,9-dimethylxanthene) and Xantphos(Cy) (4,5-bis(dicyclohexylphosphino)-9,9-dimethylxanthene) precatalysts [Ru(Xantphos(Ph))(PhCO2)(Cl)] (1) and [Ru(Xantphos(Cy))(PhCO2)(Cl)] (2), the synthesis, characterization, and crystal structures of which are reported. The intention of this work is to (i) understand the reaction mechanisms on the microscopic level and (ii) compare experimentally observed activation barriers with computed barriers. The Gibbs free activation energy ΔG(⧧) was obtained experimentally with precatalyst 1 from Eyring plots for the hydrogenation of cyclohexene (ΔG(⧧) = 17.2 ± 1.0 kcal/mol) and 1-methylcyclohexene (ΔG(⧧) = 18.8 ± 2.4 kcal/mol), while the Gibbs free activation energy ΔG(⧧) for the hydrogenation of cyclohexene with precatalyst 2 was determined to be 21.1 ± 2.3 kcal/mol. Plausible activation pathways and catalytic cycles were computed in the gas phase (M06-L/def2-SVP). A variety of popular density functionals (ωB97X-D, LC-ωPBE, CAM-B3LYP, B3LYP, B97-D3BJ, B3LYP-D3, BP86-D3, PBE0-D3, M06-L, MN12-L) were used to reoptimize the turnover determining states in the solvent phase (DF/def2-TZVP; IEF-PCM and/or SMD) to investigate how well the experimentally obtained activation barriers can be reproduced by the calculations. The density functionals B97-D3BJ, MN12-L, M06-L, B3LYP-D3, and CAM-B3LYP reproduce the experimentally observed activation barriers for both olefins very well with very small (0.1 kcal/mol) to moderate (3.0 kcal/mol) mean deviations from the experimental values indicating for the field of hydrogenation catalysis most of these functionals to be useful for in silico catalyst design prior to experimental work.