Search for: All records

The electrochemical ammonia oxidation to dinitrogen as a means for energy and environmental applications is a key technology toward the realization of a sustainable nitrogen cycle. The state-of-the-art metal catalysts including Pt and its bimetallics with Ir show promising activity, albeit suffering from high overpotentials for appreciable current densities and the soaring price of precious metals. Herein, the immense design space of ternary Pt alloy nanostructures is explored by graph neural networks trained on ab initio data for concurrently predicting site reactivity, surface stability, and catalyst synthesizability descriptors. Among a few Ir-free candidates that emerge from the active learning workflow, Pt₃Ru-M (M: Fe, Co, or Ni) alloys were successfully synthesized and experimentally verified to be more active toward ammonia oxidation than Pt, Pt₃Ir, and Pt₃Ru. More importantly, feature attribution analyses using the machine-learned representation of site motifs provide fundamental insights into chemical bonding at metal surfaces and shed light on design strategies for high-performance catalytic systems beyond thed-band center metric of binding sites.

Stereoselective ring-opening polymerization catalysts are used to produce degradable stereoregular poly(lactic acids) with thermal and mechanical properties that are superior to those of atactic polymers. However, the process of discovering highly stereoselective catalysts is still largely empirical. We aim to develop an integrated computational and experimental framework for efficient, predictive catalyst selection and optimization. As a proof of principle, we have developed a Bayesian optimization workflow on a subset of literature results for stereoselective lactide ring-opening polymerization, and using the algorithm, we identify multiple new Al complexes that catalyze either isoselective or heteroselective polymerization. In addition, feature attribution analysis uncovers mechanistically meaningful ligand descriptors, such as percent buried volume (%V_bur) and the highest occupied molecular orbital energy (E_HOMO), that can access quantitative and predictive models for catalyst development.

The electrochemical nitrate reduction reaction (NO₃RR) to ammonia is an essential step toward restoring the globally disrupted nitrogen cycle. In search of highly efficient electrocatalysts, tailoring catalytic sites with ligand and strain effects in random alloys is a common approach but remains limited due to the ubiquitous energy-scaling relations. With interpretable machine learning, we unravel a mechanism of breaking adsorption-energy scaling relations through the site-specific Pauli repulsion interactions of the metald-states with adsorbate frontier orbitals. The non-scaling behavior can be realized on (100)-type sites of ordered B2 intermetallics, in which the orbital overlap between the hollow *N and subsurface metal atoms is significant while the bridge-bidentate *NO₃is not directly affected. Among those intermetallics predicted, we synthesize monodisperse ordered B2 CuPd nanocubes that demonstrate high performance for NO₃RR to ammonia with a Faradaic efficiency of 92.5% at −0.5 V_RHEand a yield rate of 6.25 mol h⁻¹g⁻¹at −0.6 V_RHE. This study provides machine-learned design rules besides thed-band center metrics, paving the path toward data-driven discovery of catalytic materials beyond linear scaling limitations.

Despite recent advances of data acquisition and algorithms development, machine learning (ML) faces tremendous challenges to being adopted in practical catalyst design, largely due to its limited generalizability and poor explainability. Herein, we develop a theory-infused neural network (TinNet) approach that integrates deep learning algorithms with the well-establishedd-band theory of chemisorption for reactivity prediction of transition-metal surfaces. With simple adsorbates (e.g., *OH, *O, and *N) at active site ensembles as representative descriptor species, we demonstrate that the TinNet is on par with purely data-driven ML methods in prediction performance while being inherently interpretable. Incorporation of scientific knowledge of physical interactions into learning from data sheds further light on the nature of chemical bonding and opens up new avenues for ML discovery of novel motifs with desired catalytic properties.

Building upon thed-band reactivity theory in surface chemistry and catalysis, we develop a Bayesian learning approach to probing chemisorption processes at atomically tailored metal sites. With representative species, e.g., *O and *OH, Bayesian models trained with ab initio adsorption properties of transition metals predict site reactivity at a diverse range of intermetallics and near-surface alloys while naturally providing uncertainty quantification from posterior sampling. More importantly, this conceptual framework sheds light on the orbitalwise nature of chemical bonding at adsorption sites withd-states characteristics ranging from bulk-like semi-elliptic bands to free-atom-like discrete energy levels, bridging the complexity of electronic descriptors for the prediction of novel catalytic materials.