Sharp Increase in Catalytic Selectivity in Acetylene Semihydrogenation on Pd Achieved by a Machine Learning Simulation-Guided Experiment

Discovery of Alloy Catalysts Beyond Pd for Selective Hydrogenation of Reformate via First-Principle Screening with Consideration of H-Coverage

The highly selective hydrogenation to remove olefins is a significant refining approach for the reformate. Herein, a library of transition metal for reformate hydrogenation is tested experimentally to validate the predictive level of catalytic activity from our theoretical framework, which combines ab initio calculations and microkinetic modeling, with consideration of surface H-coverage effect on hydrogenation kinetics. The favorable H coverage of specific alloy surface under relevant hydrogenation condition, is found to be determined by its corresponding alloy composition. Besides, olefin hydrogenation rate is determined as a function of two descriptors, i.e. H coverage and binding energies of atomic hydrogen, paving the way to computationally screen on metal component in the periodic table. Evaluation of 172 bimetallic alloys based on the activity volcano map, as well as benzene hydrogenation rate, identifies prospective superior candidates and experimentally confirms that Zn3Ir1 outperforms pure Pd catalysts for the selective hydrogenation refining of reformate. The insights into H-coverage-related microkinetic modelling have enabled us to both theoretically understand experimental findings and identify novel catalysts, thus, bridging the gap between first-principle simulations and industrial applications. This work provides useful guidance for experimental catalyst design, which can be easily extended to other hydrogenation reaction.

Iterative multiscale and multi-physics computations for operando catalyst nanostructure elucidation and kinetic modeling

Modern heterogeneous catalysis has benefitted immensely from computational predictions of catalyst structure and its evolution under reaction conditions, first-principles mechanistic investigations, and detailed kinetic modeling, which are rungs on a multiscale workflow. Establishing connections across these rungs and integration with experiments have been challenging. Here, operando catalyst structure prediction techniques using density functional theory simulations and ab initio thermodynamics calculations, molecular dynamics, and machine learning techniques are presented. Surface structure characterization by computational spectroscopic and machine learning techniques is then discussed. Hierarchical approaches in kinetic parameter estimation involving semi-empirical, data-driven, and first-principles calculations and detailed kinetic modeling via mean-field microkinetic modeling and kinetic Monte Carlo simulations are discussed along with methods and the need for uncertainty quantification. With these as the background, this article proposes a bottom-up hierarchical and closed loop modeling framework incorporating consistency checks and iterative refinements at each level and across levels.

Machine Learning Descriptors for Data-Driven Catalysis Study

Traditional trial-and-error experiments and theoretical simulations have difficulty optimizing catalytic processes and developing new, better-performing catalysts. Machine learning (ML) provides a promising approach for accelerating catalysis research due to its powerful learning and predictive abilities. The selection of appropriate input features (descriptors) plays a decisive role in improving the predictive accuracy of ML models and uncovering the key factors that influence catalytic activity and selectivity. This review introduces tactics for the utilization and extraction of catalytic descriptors in ML-assisted experimental and theoretical research. In addition to the effectiveness and advantages of various descriptors, their limitations are also discussed. Highlighted are both 1) newly developed spectral descriptors for catalytic performance prediction and 2) a novel research paradigm combining computational and experimental ML models through suitable intermediate descriptors. Current challenges and future perspectives on the application of descriptors and ML techniques to catalysis are also presented.

Increasing the Distance of Adjacent Palladium Atoms for Configuration Matching in Selective Hydrogenation

Precisely tailoring the distance between adjacent metal sites to match adsorption configurations of key species for the targeted reaction pathway is a great challenge in heterogeneous catalysis. Here, we report a proof-of-concept study on the atomically sites-tailored pathway in Pd-catalyzed acetylene hydrogenation, i.e., increasing the distance of adjacent Pd atoms (d_Pd-a-Pd ) for configuration matching in acetylene semi-hydrogenation against coupling. d_Pd-a-Pd is identified as a structural descriptor for describing the competitiveness for reaction pathways, and the increased d_Pd-a-Pd prefers the semi-hydrogenation pathway due to simultaneously promoted C₂ H₄ desorption and the destabilized transition state of the C₂ H₃ * coupling. Spectroscopic, kinetics and electronic structure studies reveal that increasing d_Pd-a-Pd to 3.31 Å delivers superior selectivity and stability due to energy matching and appropriate hybridization of Pd 4d with In 2s and, especially, 2p orbitals.

Machine Learning: A New Paradigm in Computational Electrocatalysis

Designing and screening novel electrocatalysts, understanding electrocatalytic mechanisms at an atomic level, and uncovering scientific insights lie at the center of the development of electrocatalysis. Despite certain success in experiments and computations, it is still difficult to achieve the above objectives due to the complexity of electrocatalytic systems and the vastness of the chemical space for candidate electrocatalysts. With the advantage of machine learning (ML) and increasing interest in electrocatalysis for energy conversion and storage, data-driven scientific research motivated by artificial intelligence (AI) has provided new opportunities to discover promising electrocatalysts, investigate dynamic reaction processes, and extract knowledge from huge data. In this Perspective, we summarize the recent applications of ML in electrocatalysis, including the screening of electrocatalysts and simulation of electrocatalytic processes. Furthermore, interpretable machine learning methods for electrocatalysis are discussed to accelerate knowledge generation. Finally, the blueprint of machine learning is envisaged for future development of electrocatalysis.

Tuning crystal-phase of bimetallic single-nanoparticle for catalytic hydrogenation

Bimetallic nanoparticles afford geometric variation and electron redistribution via strong metal-metal interactions that substantially promote the activity and selectivity in catalysis. Quantitatively describing the atomic configuration of the catalytically active sites, however, is experimentally challenged by the averaging ensemble effect that is caused by the interplay between particle size and crystal-phase at elevated temperatures and under reactive gases. Here, we report that the intrinsic activity of the body-centered cubic PdCu nanoparticle, for acetylene hydrogenation, is one order of magnitude greater than that of the face-centered cubic one. This finding is based on precisely identifying the atomic structures of the active sites over the same-sized but crystal-phase-varied single-particles. The densely-populated Pd-Cu bond on the chemically ordered nanoparticle possesses isolated Pd site with a lower coordination number and a high-lying valence d-band center, and thus greatly expedites the dissociation of H₂ over Pd atom and efficiently accommodates the activated H atoms on the particle top/subsurfaces.

Achieving Theory-Experiment Parity for Activity and Selectivity in Heterogeneous Catalysis Using Microkinetic Modeling

Microkinetic modeling based on density functional theory (DFT) energies plays an essential role in heterogeneous catalysis because it reveals the fundamental chemistry for catalytic reactions and bridges the microscopic understanding from theoretical calculations and experimental observations. Microkinetic modeling requires building a set of ordinary differential equations (ODEs) based on the calculation results of thermodynamic properties of adsorbates and kinetic parameters for the reaction elementary steps. Solving a microkinetic model can extract information on catalytic chemistry, including critical reaction intermediates, reaction pathways, the surface species distribution, activity, and selectivity, thus providing vital guidelines for altering catalysts.

However, the quantitative reliability of traditional microkinetic models is often insufficient to conclusively extrapolate the mechanistic details of complex reaction systems. This can be attributed to several factors, the most important of which is the limitation of obtaining an accurate estimation of the energy inputs via traditional calculation methods. These limitations include the difficulty of using static DFT methods to calculate reaction energies of adsorption/desorption processes, often rate-controlling or selectivity-determining steps, and the inadequate consideration of surface coverage effects. In addition, the robust microkinetic software is rare, which also complicates the resolution of complex catalytic systems.

In this Account, we review our recent works toward refining the predictions of microkinetic modeling in heterogeneous catalysis and achieving theory–experiment parity for activity and selectivity. First, we introduce CATKINAS, a microkinetic software developed in our group, and show how it disentangles the problem that traditional microkinetic software has and how it can now be applied to obtain kinetic results for more sophisticated reaction systems. Second, we describe a molecular dynamics method developed recently to obtain the free-energy changes for the adsorption/desorption process to fill in the missing energy inputs. Third, we show that a rigorous consideration of surface coverage effects is pivotal for building more realistic models and obtaining accurate kinetic results. Following a series of studies on acetylene hydrogenation reactions on Pd catalysts, we demonstrate how this new approach can provide an improved quantitative understanding of the mechanism, active site, and intrinsic structural sensitivity. Finally, we conclude with a brief outlook and the remaining challenges in this field.

Dynamics of Heterogeneous Catalytic Processes at Operando Conditions

The rational design of high-performance catalysts is hindered by the lack of knowledge of the structures of active sites and the reaction pathways under reaction conditions, which can be ideally addressed by an in situ/operando characterization. Besides the experimental insights, a theoretical investigation that simulates reaction conditions-so-called operando modeling-is necessary for a plausible understanding of a working catalyst system at the atomic scale. However, there is still a huge gap between the current widely used computational model and the concept of operando modeling, which should be achieved through multiscale computational modeling. This Perspective describes various modeling approaches and machine learning techniques that step toward operando modeling, followed by selected experimental examples that present an operando understanding in the thermo- and electrocatalytic processes. At last, the remaining challenges in this area are outlined.

Liquid-Phase Hydrogenation of 1-Phenyl-1-propyne on the Pd1Ag3/Al2O3 Single-Atom Alloy Catalyst: Kinetic Modeling and the Reaction Mechanism

This research was focused on studying the performance of the Pd₁Ag₃/Al₂O₃ single-atom alloy (SAA) in the liquid-phase hydrogenation of di-substituted alkyne (1-phenyl-1-propyne), and development of a kinetic model adequately describing the reaction kinetic being also consistent with the reaction mechanism suggested for alkyne hydrogenation on SAA catalysts. Formation of the SAA structure on the surface of PdAg₃ nanoparticles was confirmed by DRIFTS-CO, revealing the presence of single-atom Pd₁ sites surrounded by Ag atoms (characteristic symmetrical band at 2046 cm⁻¹) and almost complete absence of multiatomic Pd_n surface sites (<0.2%). The catalyst demonstrated excellent selectivity in alkyne formation (95–97%), which is essentially independent of P(H₂) and alkyne concentration. It is remarkable that selectivity remains almost constant upon variation of 1-phenyl-1-propyne (1-Ph-1-Pr) conversion from 5 to 95–98%, which indicates that a direct alkyne to alkane hydrogenation is negligible over Pd₁Ag₃ catalyst. The kinetics of 1-phenyl-1-propyne hydrogenation on Pd₁Ag₃/Al₂O₃ was adequately described by the Langmuir-Hinshelwood type of model developed on the basis of the reaction mechanism, which suggests competitive H₂ and alkyne/alkene adsorption on single atom Pd₁ centers surrounded by inactive Ag atoms. The model is capable to describe kinetic characteristics of 1-phenyl-1-propyne hydrogenation on SAA Pd₁Ag₃/Al₂O₃ catalyst with the excellent explanation degree (98.9%).

In Situ Surface Structures of PdAg Catalyst and Their Influence on Acetylene Semihydrogenation Revealed by Machine Learning and Experiment

PdAg alloy is an industrial catalyst for acetylene-selective hydrogenation in excess ethene. While significant efforts have been devoted to increase the selectivity, there has been little progress in the catalyst performance at low temperatures. Here by combining a machine-learning atomic simulation and catalysis experiment, we clarify the surface status of PdAg alloy catalyst under the reaction conditions and screen out a rutile-TiO₂ supported Pd₁Ag₃ catalyst with high performance: i.e., 85% selectivity at >96% acetylene conversion over a 100 h period in an experiment. The machine-learning global potential energy surface exploration determines the Pd-Ag-H bulk and surface phase diagrams under the reaction conditions, which reveals two key bulk compositions, Pd₁Ag₁ (R3̅m) and Pd₁Ag₃ (Pm3̅m), and quantifies the surface structures with varied Pd:Ag ratios under the reaction conditions. We show that the catalyst activity is controlled by the PdAg patterns on the (111) surface that are variable under reaction conditions, but the selectivity is largely determined by the amount of Pd exposure on the (100) surface. These insights provide the fundamental basis for the rational design of a better catalyst via three measures: (i) controlling the Pd:Ag ratio at 1:3, (ii) reducing the nanoparticle size to limit PdAg local patterns, (iii) searching for active supports to terminate the (100) facets.

Influence of surface defects on activity and selectivity: a quantitative study of structure sensitivity of Pd catalysts for acetylene hydrogenation

The structure sensitivity of Pd catalysed acetylene hydrogenation is quantitatively examined using a coverage-dependent microkinetic model. Pd(211) was found to be more active than Pd(111), but present a poorer selectivity toward ethylene.

Interstitial and substitutional light elements in transition metals for heterogeneous catalysis

The addition of foreign element dopants to monometallic nanoparticle catalysts is of great importance in industrial applications. Both substitutional and interstitial doping of pure metallic phases can give profound effects such as altering electronic and transport properties, lattice parameters, phase transitions, and consequently various physicochemical properties. For transition metal catalysts, this often leads to changes in catalytic activity and selectivity. This article provides an overview of the recent developments regarding the catalytic properties and characterisation of such systems. In particular, the structure-activity relationship for a number of important chemical reactions is summarised and the future prospects of this area are also explored.