PolyODENet: Deriving mass-action rate equations from incomplete transient kinetics data.

Kinetics of a reaction network that follows mass-action rate laws can be described with a system of ordinary differential equations (ODEs) with polynomial right-hand side. However, it is challenging to derive such kinetic differential equations from transient kinetic data without knowing the reaction network, especially when the data are incomplete due to experimental limitations. We introduce a program, PolyODENet, toward this goal. Based on the machine-learning method Neural ODE, PolyODENet defines a generative model and predicts concentrations at arbitrary time. As such, it is possible to include unmeasurable intermediate species in the kinetic equations. Importantly, we have implemented various measures to apply physical constraints and chemical knowledge in the training to regularize the solution space. Using simple catalytic reaction models, we demonstrate that PolyODENet can predict reaction profiles of unknown species and doing so even reveal hidden parts of reaction mechanisms.

Quantitative and Atomic-Scale View of CO-Induced Pt Nanoparticle Surface Reconstruction at Saturation Coverage via DFT Calculations Coupled with in Situ TEM and IR.

Atomic-scale insights into how supported metal nanoparticles catalyze chemical reactions are critical for the optimization of chemical conversion processes. It is well-known that different geometric configurations of surface atoms on supported metal nanoparticles have different catalytic reactivity and that the adsorption of reactive species can cause reconstruction of metal surfaces. Thus, characterizing metallic surface structures under reaction conditions at atomic scale is critical for understanding reactivity. Elucidation of such insights on high surface area oxide supported metal nanoparticles has been limited by less than atomic resolution typically achieved by environmental transmission electron microscopy (TEM) when operated under realistic conditions and a lack of correlated experimental measurements providing quantitative information about the distribution of exposed surface atoms under relevant reaction conditions. We overcome these limitations by correlating density functional theory predictions of adsorbate-induced surface reconstruction visually with atom-resolved imaging by in situ TEM and quantitatively with sample-averaged measurements of surface atom configurations by in situ infrared spectroscopy all at identical saturation adsorbate coverage. This is demonstrated for platinum (Pt) nanoparticle surface reconstruction induced by CO adsorption at saturation coverage and elevated (>400 K) temperature, which is relevant for the CO oxidation reaction under cold-start conditions in the catalytic convertor. Through our correlated approach, it is observed that the truncated octahedron shape adopted by bare Pt nanoparticles undergoes a reversible, facet selective reconstruction due to saturation CO coverage, where {100} facets roughen into vicinal stepped high Miller index facets, while {111} facets remain intact.

Controlling catalytic selectivity on metal nanoparticles by direct photoexcitation of adsorbate-metal bonds.

Engineering heterogeneous metal catalysts for high selectivity in thermal driven reactions typically involves the synthesis of nanostructures with well-controlled geometries and compositions. However, inherent relationships between the energetics of elementary steps limit the control of catalytic selectivity through these approaches. Photon excitation of metal catalysts can induce chemical reactivity channels that cannot be accessed using thermal energy, although the potential for targeted activation of adsorbate-metal bonds is limited because the processes of photon absorption and adsorbate-metal bond photoexcitation are typically separated spatially. Here, we show that the use of sub-5-nanometer metal particles as photocatalysts enables direct photoexcitation of hybridized adsorbate-metal states as the dominant mechanism driving photochemistry. Activation of targeted adsorbate-metal bonds through direct photoexcitation of hybridized electronic states enabled selectivity control in preferential CO oxidation in H2 rich streams. This mechanism opens new avenues to drive selective catalytic reactions that cannot be achieved using thermal energy.