[Paper Review] Unraveling the Catalytic Effect of Hydrogen Adsorption on Pt Nanoparticle Shape-Change

Cameron J. Owen, Nicholas Marcella+5 more|arXiv (Cornell University)|Jun 1, 2023

Machine Learning in Materials Science4 citations

TL;DR

This study combines first-principles machine-learned molecular dynamics and X-ray absorption spectroscopy to reveal how hydrogen adsorption catalyzes shape transformations in sub-2 nm Pt nanoparticles. It identifies a previously unknown quasi-icosahedral intermediate and surface rosette formation, with hydrogen lowering transition temperatures and accelerating structural reorganization, providing atomic-level insight into dynamic catalyst behavior during operation.

ABSTRACT

The activity of metal catalysts depends sensitively on dynamic structural changes that occur during operating conditions. The mechanistic understanding underlying such transformations in small Pt nanoparticles (NPs) of $\sim1-5$ nm in diameter, commonly used in hydrogenation reactions, is currently far from complete. In this study, we investigate the structural evolution of Pt NPs in the presence of hydrogen using reactive molecular dynamics (MD) simulations and X-ray spectroscopy measurements. To gain atomistic insights into adsorbate-induced structural transformation phenomena, we employ a combination of MD based on first-principles machine-learned force fields with extended X-ray absorption fine structure (EXAFS) measurements. Simulations and experiments provide complementary information, mutual validation, and interpretation. We obtain atomic-level mechanistic insights into `order-disorder' structural transformations exhibited by highly dispersed heterogeneous Pt catalysts upon exposure to hydrogen. We report the emergence of previously unknown candidate structures in the small Pt NP limit, where exposure to hydrogen leads to the appearance of a `quasi-icosahedral' intermediate symmetry, followed by the formation of `rosettes' on the NP surface. Hydrogen adsorption is found to catalyze these shape transitions by lowering their temperatures and increasing the apparent rates, revealing the dual catalytic and dynamic nature of interaction between nanoparticle and adsorbate. Our study also offers a new pathway for deciphering the reversible evolution of catalyst structure resulting from the chemisorption of reactive species, enabling the determination of active sites and improved interpretation of experimental results with atomic resolution.

Motivation & Objective

• To understand the dynamic structural evolution of sub-2 nm Pt nanoparticles under hydrogen exposure, a key condition in catalytic hydrogenation.
• To overcome limitations of conventional intuition-based models in predicting active site geometries in highly dynamic, heterogeneous nanocatalysts.
• To establish a synergistic framework combining reactive machine-learned force fields and experimental EXAFS for unbiased, atomically resolved mechanistic insight.
• To identify previously unknown intermediate structures and transformation pathways in Pt NPs during adsorbate-induced restructuring.
• To enable improved interpretation of experimental EXAFS data by linking local atomic structures to spectroscopic signatures with high fidelity.

Proposed method

• Employed first-principles machine-learned force fields (MLFFs) trained on ab initio data to simulate reactive molecular dynamics of Pt nanoparticles over experimentally relevant time and length scales.
• Used a deep neural network surrogate model to predict EXAFS spectra from radial distribution functions (RDFs), with input derived from MLFF trajectories and output validated against FEFF10 calculations.
• Applied temperature matching via linear regression between MD-simulated mean square radial deviation (MSRD) and experimental Einstein or RMC-derived MSRD to align MD temperatures with experimental conditions.
• Performed trajectory analysis using OVITO and custom Python code to extract Pt-Pt pairwise distances, bin them into RDFs (0–6 Å, Δr = 0.025 Å), and compute the first peak’s mean (R) and variance (MSRD) as structural descriptors.
• Calculated EXAFS spectra from simulated atomic configurations and compared them directly to experimental EXAFS data to validate the simulated structures and interpret experimental observations.
• Used active learning via FLARE to expand the initial H/Pt data set to include nanoparticle configurations, improving MLFF generalization and training efficiency.

Experimental results

Research questions

• RQ1 What atomic-scale structural transformations occur in Pt nanoparticles during hydrogen exposure, and how do they differ from static or bulk-like behavior?
• RQ2 How does hydrogen adsorption influence the thermodynamics and kinetics of shape changes in sub-2 nm Pt NPs?
• RQ3 What intermediate structures emerge during the order-disorder transition in Pt NPs under H2 exposure, and how do they relate to catalytic activity?
• RQ4 To what extent can machine-learned force fields and EXAFS spectral predictions jointly validate and interpret experimental observations in dynamic nanocatalysts?
• RQ5 Can the catalytic effect of hydrogen on structural reorganization be quantified in terms of reduced transition temperatures and increased apparent rates?

Key findings

• Hydrogen adsorption induces a previously unreported 'quasi-icosahedral' intermediate symmetry in small Pt nanoparticles (1–5 nm), observed during shape evolution under H2 exposure.
• Hydrogen adsorption catalyzes structural transitions by lowering the apparent transition temperature and increasing the rate of shape change, demonstrating a dual catalytic and dynamic role.
• Surface 'rosette' structures form on Pt nanoparticle surfaces during the transformation, indicating a non-uniform, anisotropic restructuring process driven by adsorbate interactions.
• The combination of MLFF-based MD simulations and EXAFS spectral prediction enables mutual validation: simulated structures reproduce experimental EXAFS, and experimental data constrain the simulated pathways.
• The mean square radial deviation (MSRD) of the first Pt-Pt peak (2–3.34 Å) serves as a sensitive metric for tracking structural disorder, with MD and experimental MSRD values showing linear correlation for temperature matching.
• The neural network surrogate model accurately predicts EXAFS from RDFs with high fidelity, achieving strong agreement between simulated and experimental spectra, even on unseen 309-atom NP configurations.