(147w) Rational Design of Low-Dimensional Catalysts Using First Principles Methods for CO2 Reduction to C1 Products

The Intergovernmental Panel on Climate Change (IPCC) report emphasizes the urgent need for immediate and ambitious actions to reduce greenhouse gas emissions and transition to a low-carbon economy. Carbon dioxide (CO₂) can be used as a feedstock for manufacturing chemicals and fuels, thus contributing to the circular economy by closing the carbon cycle and lowering net greenhouse gas emissions. Bifunctional catalysts containing two distinct catalytic sites in proximity to each other are needed to activate relatively inert CO₂ for these chemical conversions. Examples of such catalysts include active sites at the interface formed between precious metal nanoparticles like Au and earth-abundant catalyst supports like transition metal carbides and nitrides. While such bifunctional catalysts can activate and hydrogenate CO₂, these catalysts are currently designed through trial-and-error methods. My research interests are focused on using Density Functional Theory (DFT) in conjunction with machine learning methods to establish rational design principles for reactive, selective, and stable bifunctional catalysts. We explore computational methods to design highly efficient catalysts that can help mitigate the effects of greenhouse gas emissions by converting CO₂ to valuable chemicals. I believe that my research has significant implications for the industry, especially for companies that are committed to reducing their carbon footprint and mitigating the effects of climate change. By developing highly efficient and selective catalysts, we can convert CO₂ into valuable chemicals and fuels, reducing our dependence on fossil fuels and creating a more sustainable future.

The reactivity of reaction intermediates involved in a reaction is linked through scaling relationships, which are limited by the bond-order conservation principle. These limitations restrict the number of materials that can be screened for catalytic activity, as all tested metals exhibit strong scaling. To overcome this limitation, we introduce bifunctionality in heterostructures by creating metal/support interfaces, which break the observed constraints in scaling relations. This results in a synergy between metallic nanoparticles and supports that enhances catalytic turnovers. The potential of such interfaces displaying bifunctional gain has already been reported (Choksi T. et al., Angew. Chemie Int. Ed. (2018)). Our aim is to use 2D/3D transition metal carbides and nitrides as supports for important catalytic processes involving C- and O-based intermediates. These materials, including MXenes have sparked growing interest in the field of catalysis because of its ultrathin nature, high surface tunability, and well-defined structure.

Breaking scaling relations enables the discovery of optimal catalysts by modification of reactive sites. In our study, we investigate bifunctional metal/support interfaces with site-by-site resolution to predict catalytic reactivity, selectivity, and stability metrics. We use a data-driven approach to present a generalized design principle that predicts catalytic rates at every site of the nanoparticle. We utilize DFT to obtain ground state energies of reaction intermediates and transition states and convert them to free energies under reaction conditions using statistical thermodynamics treatments. DFT calculations are performed via Quantum Espresso within the Atomic Simulation Environment. Total energies of reaction intermediates are determined using the PBE-D3 functional. The computational hydrogen electrode is employed to determine thermodynamic barriers as a function of the applied potential. To simplify complex reaction mechanisms for high-throughput screening, we leverage natural energy correlations between structurally similar reaction intermediates and reduce them to one or two descriptors (such as binding energies of CO* and OH*). Our material space includes low-strain (< 5%) epitaxial films and moiré patterns of Au, Ag, and Cu monolayers supported on metal/carbon/nitrogen/oxygen-terminated supports. The functional groups on 2D/3D carbides and nitrides induce different charge states on the Au, Ag, and Cu sites, which affect selectivity for CO₂ reduction.

Our models predict reactivity descriptors for CO₂ hydrogenation reactions with active-site-resolution for metals dispersed on supports, including binding energies of CO*, HCOO*, HCOOH*, and H*. We demonstrate that the strongest interfacial perturbation occurs at sites closest to the support and having the lowest number of nearest neighbours. We establish systematic patterns in the reactivity of Au-heterostructures at the active site level using site-specific scaling relationships. These relationships correlate the binding energies of Au sites to the binding energies of reaction intermediates, such as CO*, as the Au coordination number is varied. The stability-reactivity relationships provide a common framework for understanding reactivity trends in Au sites across coordination environments ranging from adatoms to monolayered films. For a generic Au-heterostructure, we demonstrate that a set of site-specific scaling relations, categorized by the charged state of Au, can predict reactivity descriptors at each active site located at various proximities from the interface. For a metal/support heterostructure, the scaling relations reveal the adsorption energies of CO* to be dependent on the type of functional group and the support for the first layer of the metal. When we move away from the interface to the second layer, these adsorption energies are independent of the type of support for a fixed termination. For the third layer and beyond, reactivity only depends on the coordination number of the metal atom. Our models also quantify the structure sensitivity, which refers to the degree to which catalytic descriptors at an active site change as the local structure of that site is modified. We demonstrate that tuning the support composition or the surface termination alters the charged state of gold sites, resulting in variations in the adsorption energies of CO* by > 1 eV. These variations suggest that supported Au sites may be selective for different products (CO, C₁₊, syn-gas), thus expanding the range of potential catalysts beyond Cu and its alloys.

We classify metal/MXene heterostructures based on their selectivity towards CO, H₂, HCOOH, and C₁₊ products using a thermodynamic formalism introduced by Tang et al., Appl. Catal. B Environ. (2020). We develop linear scaling relationships between complex reaction intermediates in electrochemical CO₂ reduction and reactivity descriptors such as the adsorption energies of HCOO* and OH*, COOH* and CO*/OH*, and H* and CO* for a metal film dispersed on support. The scaling relationships are specific to a particular metal, such as Au, with different supports having diverse surface terminations. The slopes of these scaling relations (e.g., 1.13 for HCOO* vs OH* across Au heterostructures) deviate from the rigid bond-order conservation rules (slope is 1.00) that are prevalent in unsupported metallic systems used in CO₂ reduction. These scaling relationships are combined with free energy equilibrium conditions to determine the selectivity boundaries for competing pathways (e.g., HCOO*/COOH* pathways, H₂ evolution). Metal/MXenes are selective to CO, C₁₊, and H₂, depending on the MXene termination and composition. Gold and copper heterostructures are not selective to HCOOH. We have used thermodynamics-based selectivity classification to lay the groundwork for a microkinetic model that can identify metal/MXene combinations yielding high faradaic efficiencies for CO, C₁₊ products, or syn-gas with specific CO/H₂ ratios.

In addition to reactivity and selectivity considerations, we construct stability models that provide atomic level insights into surface restructuring processes. We employ a descriptor-based model based on the physicochemical properties to predict stability of bifunctional catalysts. The adhesion energy is a metal/support interaction descriptor which indicates thermodynamic stability of a catalyst. It reflects how strongly a metal binds to the support. We develop a linear model and identify important features such as strain on the metal electronegativity of the termination, work function of the terminated support and binding energy of termination on the support that can predict adhesion energy of Au and Pt films on different supports with errors of 0.03 eV/Ų. Our previous work on machine learning model to predict work function of MXenes based on handbook properties (Roy P.^#, Rekhi L.^#, et. al., JPhys Energy (2023)) could be used to predict this electronic feature for different supports. The predicted adhesion energies are correlated with adatom binding energies, which could further be used to design single atom catalysts and determine materials resistant against sintering.

We have deployed theoretical methods to design efficient bifunctional catalysts engineered with atomic-level specificity for CO₂ hydrogenation chemistries. The performance of these candidate materials will be verified through experimental collaboration. By integrating physics-based scaling relations and machine learning, we systematize reactivity, selectivity, and stability trends of supported metal catalysts for CO₂ reduction. These energy correlations will be propagated through microkinetic models to express rates as a function of catalytic descriptors. Additionally, our study seeks to correlate the catalytic properties of 2D materials with their more complex 3D counterparts, providing molecular-scale insights into the development of new catalysts. We introduce a new design principle for tailoring the selectivity of CO₂ electroreduction catalysts by manipulating interfacial charge transfer. Our model enables rapid classification of the selectivity and stability of metal/MXene heterostructures, thus expanding the pool of candidate catalysts to beyond Cu and its alloys.

These catalyst design efforts will lead to the development of new catalytic candidates that are stable, active, and selective towards important chemical reactions. I am passionate about collaborating with industry partners to work on research projects that are relevant to their needs and goals, and I believe that together, we can make significant progress in advancing our field.