Decoding reactive structures in dilute alloy catalysts

• Published in: Nature communications Vol. 13; no. 1; pp. 832 - 9
• Main Authors: Marcella, Nicholas, Lim, Jin Soo, Płonka, Anna M., Yan, George, Owen, Cameron J., van der Hoeven, Jessi E. S., Foucher, Alexandre C., Ngan, Hio Tong, Torrisi, Steven B., Marinkovic, Nebojsa S., Stach, Eric A., Weaver, Jason F., Aizenberg, Joanna, Sautet, Philippe, Kozinsky, Boris, Frenkel, Anatoly I.
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 11.02.2022; Nature Publishing Group; Nature Portfolio
• Subjects: 119/118; 140/58; 147/28; 639/301/299/1013; 639/638/77/885; 639/638/77/887; Absorption spectroscopy; Bimetals; Catalysts; Catalytic activity; Catalytic mechanisms; Deuterium; Dilution; Energy efficiency; First principles; Gold; Heterogeneous catalysis; Humanities and Social Sciences; Hydrogen-deuterium exchange; INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CHEMISTRY; Learning algorithms; Machine learning; MATERIALS SCIENCE; Modelling; multidisciplinary; Nanoalloys; Nanoparticles; Palladium; Porous materials; Renewable energy; Science; Science (multidisciplinary); Spectrum analysis; X ray absorption; X ray spectra; X-ray absorption spectroscopy
• DOI: 10.1038/s41467-022-28366-w
• ISSN: 2041-1723

Rational catalyst design is crucial toward achieving more energy-efficient and sustainable catalytic processes. Understanding and modeling catalytic reaction pathways and kinetics require atomic level knowledge of the active sites. These structures often change dynamically during reactions and are difficult to decipher. A prototypical example is the hydrogen-deuterium exchange reaction catalyzed by dilute Pd-in-Au alloy nanoparticles. From a combination of catalytic activity measurements, machine learning-enabled spectroscopic analysis, and first-principles based kinetic modeling, we demonstrate that the active species are surface Pd ensembles containing only a few (from 1 to 3) Pd atoms. These species simultaneously explain the observed X-ray spectra and equate the experimental and theoretical values of the apparent activation energy. Remarkably, we find that the catalytic activity can be tuned on demand by controlling the size of the Pd ensembles through catalyst pretreatment. Our data-driven multimodal approach enables decoding of reactive structures in complex and dynamic alloy catalysts. Rational catalyst design is crucial toward achieving more energy-efficient and sustainable catalytic processes. Here the authors report a data-driven approach for understanding catalytic reactions mechanisms in dilute bimetallic catalysts by combining X-ray absorption spectroscopy with activity studies and kinetic modeling.