Search NASA⌕ Search

Engineering topics

Chen, Benjamin W. J.

Publications and source records attributed to Chen, Benjamin W. J..

Enhancing the Quality and Reliability of Machine Learning Interatomic Potentials through Better Reporting Practices

Recent developments in machine learning interatomic potentials (MLIPs) have empowered even nonexperts in machine learning to train MLIPs for accelerating materials simulations. However, reproducibility and independent evaluation of presented MLIP results is hindered by a lack of clear standards in current literature. In this Perspective, we aim to provide guidance on best practices for documenting MLIP use while walking the reader through the development and deployment of MLIPs including hardware and software requirements, generating training data, training models, validating predictions, and MLIP inference. We also suggest useful plotting practices and analyses to validate and boost confidence in the deployed models. Finally, we provide a step-by-step checklist for practitioners to use directly before publication to standardize the information to be reported. Altogether, we hope that our work will encourage the reliable and reproducible use of these MLIPs, which will accelerate their ability to make a positive impact in various disciplines including materials science, chemistry, and biology, among others.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Computational Methods in Heterogeneous Catalysis

The unprecedented ability of computations to probe atomic-level details of catalytic systems holds immense promise for the fundamentals-based bottom-up design of novel heterogeneous catalysts, which are at the heart of the chemical and energy sectors of industry. Here, we critically analyze recent advances in computational heterogeneous catalysis. First, we will survey the progress in electronic structure methods and atomistic catalyst models employed, which have enabled the catalysis community to build increasingly intricate, realistic, and accurate models of the active sites of supported transition-metal catalysts. We then review developments in microkinetic modeling, specifically mean-field microkinetic models and kinetic Monte Carlo simulations, which bridge the gap between nanoscale computational insights and macroscale experimental kinetics data with increasing fidelity. Here, we finally review the advancements in theoretical methods for accelerating catalyst design and discovery. Throughout the review, we provide ample examples of applications, discuss remaining challenges, and provide our outlook for the near future.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

An automated cluster surface scanning method for exploring reaction paths on metal-cluster surfaces

Metal-cluster surfaces present a wide variety of unique coordination environments. This complexity makes it difficult to manually probe the surface reactivity of such clusters. Here, we present a simple and automated method to systematically discover reaction pathways on cluster surfaces, based on the automated cluster surface scanning (ACSS) technique for mapping out potential energy surfaces. We showcase our method on 55-atom icosahedral Cu and Ag clusters, where we determine the activation energies of four elementary steps common in heterogeneous catalysis – hydrogen recombination (H* + H* → H 2 * + *), oxygen recombination (O* + O* → O 2 * + *), water formation (OH* + H* → H 2 O* + *), and CO oxidation (CO* + O* → CO 2 * + *) – with density functional theory calculations (DFT-PBE + D3). We show that the ACSS method requires significantly less human effort than the established manually performed climbing-image nudged elastic band (MP + CI-NEB) technique and locates transition states with comparable accuracy (root-mean-squared error of 0.10 eV) and similar computational cost. Rigorous sampling of the potential energy surface with the ACSS method allows one to locate all lowest-energy reaction pathways obtained via the MP+CI-NEB approach, as well as alternative pathways that one may have missed with the MP+CI-NEB approach due to the many possible pathways available on these clusters. The accuracy and efficiency afforded by the ACSS method could enable high-throughput exploration of the diverse reactivity of metal clusters.

36 MATERIALS SCIENCE↗

Formic Acid: A Hydrogen-Bonding Cocatalyst for Formate Decomposition

Hydrogen bonding accelerates many catalytic reactions by orienting intermediates, stabilizing transition states, and even opening reaction pathways. However, most mechanistic studies regarding the decomposition of formic acid (FA), a promising hydrogen storage material, neglect hydrogen-bonding interactions even though FA is a strong hydrogen-bond donor and acceptor. Here, we probe the formation of bimolecular hydrogen-bonded complexes between FA and formate (FA–HCOO complexes) adsorbed on metal surfaces and how these complexes affect HCOO* decomposition. Using first-principles density functional theory (DFT) calculations on 12 close-packed (111)/(0001) and 8 open (100) surfaces of 12 transition metals—Ag, Au, Co, Cu, Ir, Ni, Os, Re, Pd, Pt, Rh, and Ru, we—show that FA–HCOO complexes are generally thermodynamically stable, even at elevated temperatures and pressures. We then illustrate that these complexes produce infrared spectroscopic signatures consistent with as yet unassigned experimental peaks. We last demonstrate that by stabilizing the dangling bond of monodentate HCOO*, these complexes significantly lower the barriers for rotation of HCOO* from a bidentate to a monodentate configuration, the rate-limiting step for HCOO* decomposition on many surfaces. FA thus acts as a cocatalyst for HCOO* decomposition. Our results may guide the community toward improved catalysts for reactions involving HCOO* such as FA decomposition, methanol steam reforming, and the water gas shift reaction. More broadly, our work highlights the ability of hydrogen bonding to modify the adsorbed structures of intermediates and lower the barriers for their reaction on heterogeneous catalysts. Lastly, this phenomenon can be relevant for other reactions involving ammonia, alcohols, and carboxylic acids.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Exploring driving forces for length growth in graphene nanoribbons during chemical vapor deposition of hydrocarbons on Ge(0 0 1) via kinetic Monte Carlo simulations

Graphene grown slowly on Ge(0 0 1) using chemical vapor deposition of hydrocarbons leads to high-aspect ratio graphene nanoribbons with smooth edges and a technologically relevant band gap at room temperature; however, the driving forces leading to one-dimensional growth of such graphene crystals are not well understood. Here, we combine a lattice kinetic Monte Carlo approach based on steps in graphene growth and experimental measurements to study the growth of graphene nanoribbons via chemical vapor deposition on Ge(0 0 1). To identify potential reasons for growth of graphene as anisotropic ribbons, we study the impact of anisotropy in various growth parameters on the resulting graphene crystals. Comparing our model with experimental measurements indicates that anisotropy in the stabilization of a graphene precursor species bound to the graphene edge is the most likely reason why high aspect ratio graphene ribbons with smooth edges grow on Ge(0 0 1). Using the growth model developed here, we reproduce experimental trends in the synthesis of graphene nanoribbons on Ge(001) and arrive at an intuitive picture for their growth. These insights shed light on the driving forces governing this highly anisotropic regime of crystal growth.

36 MATERIALS SCIENCE↗

Identification of stable adsorption sites and diffusion paths on nanocluster surfaces: an automated scanning algorithm

The diverse coordination environments on the surfaces of discrete, three-dimensional (3D) nanoclusters contribute significantly to their unique catalytic properties. Identifying the numerous adsorption sites and diffusion paths on these clusters is however tedious and time-consuming, especially for large, asymmetric nanoclusters. Here, we present a simple, automated method for constructing approximate 2D potential energy surfaces for the adsorption of atomic species on the surfaces of 3D nanoclusters with minimal human intervention. These potential energy surfaces fully characterize the important adsorption sites and diffusion paths on the nanocluster surfaces with accuracies similar to current approaches and at comparable computational cost. Our method can treat complex nanoclusters, such as alloy nanoclusters, and accounts for cluster relaxation and adsorbate-induced reconstruction, important for obtaining accurate energetics. Moreover, its highly parallelizable nature is ideal for modern supercomputer architectures. We showcase our method using two clusters: Au 18 and Pt 55 . For Au 18 , diffusion of atomic hydrogen between the most stable sites occurs via non-intuitive paths, underlining the necessity of exploring the complete potential energy surface. By enabling the rapid and unbiased assessment of adsorption and diffusion on large, complex nanoclusters, which are particularly difficult to handle manually, our method will help advance materials discovery and the rational design of catalysts.

36 MATERIALS SCIENCE↗

Kinetic Isolation between Turnovers on Au 18 Nanoclusters: Formic Acid Decomposition One Molecule at a Time

Formic acid (HCOOH or FA) is a clean, safe, and renewable hydrogen storage material. Although Au catalysts decompose vapor-phase FA with high activity and selectivity toward hydrogen, the active site and reaction mechanism remain unclear. Here, we show that the subnanometric Au 18 cluster (0.8 nm in diameter) is likely the active species for FA decomposition. We performed coverage self-consistent, density functional theory-based kinetic Monte Carlo simulations of FA decomposition on gas-phase Au18 clusters, predicting 100% selectivity toward hydrogen and turnover frequencies close to experimentally determined values. The active site is made up of a triangular ensemble of three atoms each possessing a coordination number of 5. Although there are two active site ensembles on the Au18 cluster, their occupations are strongly correlated because of strong, stabilizing interactions between pairs of open-shell adsorbates mediated by the superatomic nature of the cluster. Because the occupation of the active sites blocks the dissociation of additional HCOOH molecules, there is kinetic isolation between turnovers: only one HCOOH molecule can dissociate on the cluster at a time. This explains the extraordinary, experimentally observed selectivity of Au catalysts toward HD during decomposition of HCOOD and DCOOH. Our work offers nanoscale insights into the reaction mechanisms of FA decomposition over Au. This represents a rare example of heterogeneous catalysis by a cluster that catalyzes reactants one molecule at a time. Our work on Au18 thus sheds light on how the unique electronic properties of subnanometric clusters can be used to design quasi-molecular catalysts with high activity and selectivity.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Effects of composition and morphology on the hydrogen storage properties of transition metal hydrides: Insights from PtPd nanoclusters

Transition metal hydride nanoparticles are promising hydrogen storage materials as they store and release hydrogen at near-ambient conditions. Their storage capacities can be modulated by adjusting their size, shape, morphology, and composition. It is however unclear why these factors influence hydrogen storage properties. Here, we present a first-principles study on six 55-atom PtPd nanoclusters to elucidate the effects of nanoparticle morphology and composition on the hydrogen storage characteristics of small nanoclusters. We employ the cluster expansion methodology to analyze how the spatial distribution and energetics of hydrogen adsorption and absorption evolve as a function of hydrogen loading. Our results reveal that at near-ambient pressures and above, hydrogen is actively stored in both the surface—top sites on Pt, and bridge sites on Pd—and the subsurface. The hydrogen storage capacity is controlled by the binding strength of hydrogen, itself influenced by two factors: i) the interaction energy between hydrogen atoms, which dictates how binding strength varies with increasing loading; and ii) the intrinsic binding affinity of a metal cluster for hydrogen, which dictates the magnitude of the nearly-uniform offsets in binding strength across all hydrogen loadings. Bimetallic clusters with alloyed shells and core-shell morphologies modulate these factors by tuning the distribution of hydrogen binding strengths on the surface and subsurface respectively. We shed light on the factors that affect hydrogen storage in clusters and provide insights for the rational design of metal clusters with maximal hydrogen storage capacities.

08 HYDROGEN↗

Density functional theory study of thermodynamic and kinetic isotope effects of H 2 /D 2 dissociative adsorption on transition metals

We studied here the thermodynamic isotope effects (TIEs) and kinetic isotope effects (KIEs) for H 2 /D 2 dissociative adsorption using periodic, density functional theory (DFT)-based calculations. We examined the TIEs on the close-packed, open, and stepped surfaces, of twelve transition metals (Fe, Co, Ni, Cu, Ru, Rh, Pd, Ag, Re, Ir, Pt, and Au), and the KIEs on the surfaces of three noble metals (Cu, Ag, and Au). Both TIEs and KIEs were evaluated at 1/9 ML coverage. We find distinct TIEs on different adsorption sites, indicating that TIEs could be used in conjunction with binding energies to determine the dominant adsorption sites for hydrogen. Additionally, we find that while H 2 dissociative adsorption may traditionally be considered structure insensitive in terms of reaction rates, it can exhibit structure sensitivity in terms of its KIEs. Complementarily to TIEs, KIEs might therefore be useful for identifying active sites for H 2 dissociative adsorption on the three noble metal transition metal catalysts studied.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Atomic and Molecular Adsorption on Ag(111)

The adsorption of atomic (H, C, N, O, S) and molecular (OH, CH x , NH x , CO, NO, CN, N 2 , HNO, NOH, HCN, x = 1–3) species at 1/4 monolayer coverage on an extended Ag(111) surface was studied here using periodic density functional theory. Geometries and energies were calculated self-consistently using the PW91 functional; nonself-consistent energies using the RPBE functional are also provided. We analyze the binding energies, binding geometries, estimated diffusion barriers, harmonic vibrational frequencies, and energetic and geometric deformation parameters of these adsorbates, comparing them to experimental and theoretical results whenever possible. PW91 gives binding energies that match experimental binding energies more closely than RPBE, which consistently predicts weaker binding than PW91. The data were then used to construct and analyze thermochemistry-only potential energy pathways for the hydrocarbon-assisted and hydrogen-assisted selective catalytic reduction (SCR) of nitric oxide (NO). These analyses provide preliminary insights into the possible mechanistic paths of the SCR of NO on Ag(111). Specifically, we show that deep dehydrogenation leading to the formation of atomic intermediates is not favored on Ag(111).

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗