Toward Molecular Catalysts by Computer

• Published in: Accounts of chemical research Vol. 48; no. 2
• Main Authors: Raugei, Simone, DuBois, Daniel L., Rousseau, Roger J., Chen, Shentan, Ho, Ming-Hsun, Bullock, Ronald Morris, Dupuis, Michel
• Format: Journal Article
• Language: English
• Published: United States American Chemical Society 09.01.2015
• Subjects: acidity; carbon dioxide reduction; catalyst optimization; computer-aided design; density functional theory; hydrogen oxidation and production; INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CHEMISTRY; molecular electrocatalysis; nitrogen reduction; oxygen reduction; redox potentials; thermodynamics
• ISSN: 0001-4842; 1520-4898

Rational design of molecular catalysts requires a systematic approach to designing ligands with specific functionality and precisely tailored electronic and steric properties. It then becomes possible to devise computer protocols to design catalysts by computer. In this Account, we first review how thermodynamic properties such as redox potentials (E°), acidity constants (pKa), and hydride donor abilities (ΔGH–) form the basis for a framework for the systematic design of molecular catalysts for reactions that are critical for a secure energy future. Next, we illustrate this for hydrogen evolution and oxidation, oxygen reduction, and CO conversion, and we give references to other instances where it has been successfully applied. The framework is amenable to quantum-chemical calculations and conducive to predictions by computer. We review how density functional theory allows the determination and prediction of these thermodynamic properties within an accuracy relevant to experimentalists (~0.06 eV for redox potentials, ~1 pKa unit for pKa values, and 1–2 kcal/mol for hydricities). Computation yielded correlations among thermodynamic properties as they reflect the electron population in the d shell of the metal center, thus substantiating empirical correlations used by experimentalists. These correlations point to the key role of redox potentials and other properties (pKa of the parent aminium for the proton-relay-based catalysts designed in our laboratory) that are easily accessible experimentally or computationally in reducing the parameter space for design. These properties suffice to fully determine free energies maps and profiles associated with catalytic cycles, i.e., the relative energies of intermediates. Their prediction puts us in a position to distinguish a priori between desirable and undesirable pathways and mechanisms. Efficient catalysts have flat free energy profiles that avoid high activation barriers due to low- and high-energy intermediates. The criterion of a flat energy profile can be mathematically resolved in a functional in the reduced parameter space that can be efficaciously calculated by means of the correlation expressions. Optimization of the functional permits the prediction by computer of design points for optimum catalysts. Specifically, the optimization yields the values of the thermodynamic properties for efficient (high rate and low overpotential) catalysts. We are on the verge of design of molecular electrocatalysts by computer. Future efforts must focus on identifying actual ligands that possess these properties. We believe that this can also be achieved through computation, using Taft-like relationships linking molecular composition and structure with electron-donating ability and steric effects. We note also that the approach adopted here of using free energy maps to decipher catalytic pathways and mechanisms does not account for kinetic barriers associated with elementary steps along the catalytic pathway, which may make thermodynamically accessible intermediates kinetically inaccessible. Such an extension of the approach will require further computations that, however, can take advantage of Polanyi-like linear free energy relationships linking activation barriers and reaction free energies.