Metal-oxo moieties are important catalytic intermediates in the selective partial oxidation of hydrocarbons and in water splitting. Stable metal-oxo species have reactive properties that vary depending on the spin state of the metal, complicating the development of structure-property relationships. To overcome these challenges, we train machine learning (ML) models capable of predicting metal-oxo formation energies across a range of first-row metals, oxidation states, and spin states. Using connectivity-only features tailored for inorganic chemistry as inputs to kernel ridge regression or artificial neural network ML models, we achieve good mean absolute errors (4-5 kcal/mol) on set-aside test data across a range of ligand orientations. Analysis of feature importance for oxo formation energy prediction reveals the dominance of non-local, electronic ligand properties in contrast to other transition metal complex properties (e.g., spin-state or ionization potential). We enumerate the theoretical catalyst space with an ANN, revealing both expected trends in oxo formation energetics, such as destabilization of the metal-oxo species with increasing d-filling, as well as exceptions, such as weak correlations with indicators of oxidative stability of the metal in the resting state or unexpected spin-state dependence in reactivity. We carry out uncertainty aware evolutionary optimization using the ANN to explore a > 37,000 candidate catalyst space. New metal and oxidation state combinations are uncovered and validated with density functional theory (DFT), including counter-intuitive oxo-formation energies for oxidatively stable complexes. This approach doubles the density of confirmed DFT leads in originally sparsely populated regions of property space, highlighting the potential of ML-model-driven discovery to uncover catalyst design rules and exceptions.

Bioinspired strategies have been given extensive attention for the recovery of rare earth elements (REEs) from waste streams because of their high selectivity, regeneration potential, and sustainability as well as low cost. Lanmodulin protein is an emerging biotechnology that is highly selective for REE binding. Mimicking lanmodulin with shorter peptides is advantageous because they are simpler and potentially easier to manipulate and optimize. Lanmodulin-derived peptides have been found to bind REEs, but their properties have not been explored when immobilized on solid substrates, which is required for many advanced separation technologies. Here, two peptides, LanM1 and scrambled LanM1, are designed from the EF-hand loop 1 of lanmodulin and investigated for their binding affinity toward different REEs when surface-bound. First, the ability of LanM1 to bind REEs was confirmed and characterized in solution using circular dichroism (CD), nuclear magnetic resonance (NMR), and molecular dynamics (MD) simulations for Ce(III) ions. Isothermal titration calorimetry (ITC) was used to further analyze the binding of the LanM1 to Ce(III), Nd(III), Eu(III), and Y(III) ions and in low-pH conditions. The performance of the immobilized peptides on a model gold surface was examined using a quartz crystal microbalance with dissipation (QCM-D). The studies show that the LanM1 peptide has a stronger REE binding affinity than that of scrambled LanM1 when in solution and when immobilized on a gold surface. QCM-D data were fit to the Langmuir adsorption model to estimate the surface-bound dissociation constant (K_d) of LanM1 with Ce(III) and Nd(III). The results indicate that LanM1 peptides maintain a high affinity for REEs when immobilized, and surface-bound LanM1 has no affinity for potential competitor calcium and copper ions. The utility of surface-bound LanM1 peptides was further demonstrated by immobilizing them to gold nanoparticles (GNPs) and capturing REEs from solution in experiments utilizing an Arsenazo III-based colorimetric dye displacement assay and ultraviolet-visible (UV-vis) spectrophotometry. The saturated adsorption capacity of GNPs was estimated to be around 3.5 μmol REE/g for Ce(III), Nd(III), Eu(III), and Y(III) ions, with no binding of non-REE Ca(II) ions observed.

Aqueous phase reforming (APR) of sugar alcohol molecules derived from biomass, e.g., C_xH_(2x+2)O_x (aq) + xH₂O → xCO₂ (g) + (2x + 1)H₂ (g), creates hydrogen gas sustainably, making it an important component of future bio-refineries; however, problems with the cost, activity, and selectivity of present precious metal based catalysts impede its broader adoption. Ideally, new catalysts would be designed to optimize activity and selectivity; however, a comprehensive understanding of the APR mechanism is lacking. This is complicated by the fact that the primary biomass-derived sugar alcohols are large molecules (meaning that their reaction networks are large) and because of the presence of liquid water. Water influences catalytic phenomena in multiple ways, including altering the thermodynamics of catalytic surface species and participating in catalytic reactions. Understanding the mechanism of APR requires understanding these various effects; however, computational strategies based solely on density functional theory (DFT) are computationally prohibitive for such large and complicated reaction networks. In this work, we investigate the mechanism of APR reactions in the context of glycerol reforming. To calculate the reaction network, we combine DFT calculations, force-field molecular dynamics (MD) simulations, linear scaling relations (LSRs), transition state scaling (TSS) relationships, and data from the literature into a microkinetic model. The microkinetic model is run under vacuum and aqueous phases in order to learn about the roles of water molecules on the mechanism of glycerol APR. We identify four such roles: providing surface hydroxyl groups, which promote oxidation of surface CO formed in glycerol decomposition; promoting C–H scissions; promoting O–H scissions; and inhibiting the thermodynamics of decarbonylation of C3 intermediates.