Screening Cu-Zeolites for Methane Activation Using Curriculum-Based Training.

Machine learning (ML), when used synergistically with atomistic simulations, has recently emerged as a powerful tool for accelerated catalyst discovery. However, the application of these techniques has been limited by the lack of interpretable and transferable ML models. In this work, we propose a curriculum-based training (CBT) philosophy to systematically develop reactive machine learning potentials (rMLPs) for high-throughput screening of zeolite catalysts. Our CBT approach combines several different types of calculations to gradually teach the ML model about the relevant regions of the reactive potential energy surface. The resulting rMLPs are accurate, transferable, and interpretable. We further demonstrate the effectiveness of this approach by exhaustively screening thousands of [CuOCu]2+ sites across hundreds of Cu-zeolites for the industrially relevant methane activation reaction. Specifically, this large-scale analysis of the entire International Zeolite Association (IZA) database identifies a set of previously unexplored zeolites (i.e., MEI, ATN, EWO, and CAS) that show the highest ensemble-averaged rates for [CuOCu]2+-catalyzed methane activation. We believe that this CBT philosophy can be generally applied to other zeolite-catalyzed reactions and, subsequently, to other types of heterogeneous catalysts. Thus, this represents an important step toward overcoming the long-standing barriers within the computational heterogeneous catalysis community.

Predicting Structural Properties of Pure Silica Zeolites Using Deep Neural Network Potentials.

Machine learning potentials (MLPs) capable of accurately describing complex ab initio potential energy surfaces (PESs) have revolutionized the field of multiscale atomistic modeling. In this work, using an extensive density functional theory (DFT) data set (denoted as Si-ZEO22) consisting of 219 unique zeolite topologies (350,000 unique DFT calculations) found in the International Zeolite Association (IZA) database, we have trained a DeePMD-kit MLP to model the dynamics of silica frameworks. The performance of our model is evaluated by calculating various properties that probe the accuracy of the energy and force predictions. This MLP demonstrates impressive agreement with DFT for predicting zeolite structural properties, energy-volume trends, and phonon density of states. Furthermore, our model achieves reasonable predictions for stress-strain relationships without including DFT stress data during training. These results highlight the ability of MLPs to capture the flexibility of zeolite frameworks and motivate further MLP development for nanoporous materials with near-ab initio accuracy.

Elucidating correlated defects in metal organic frameworks using theory-guided inelastic neutron scattering spectroscopy.

Metal organic frameworks (MOFs) that incorporate metal oxide cluster nodes, exemplified by UiO-66, have been widely studied, especially in terms of their deviations from the ideal, defect-free crystalline structures. Although defects such as missing linkers, missing nodes, and the presence of adventitious synthesis-derived node ligands (such as acetates and formates) have been proposed, their exact structures remain unknown. Previously, it was demonstrated that defects are correlated and span multiple unit cells. The highly specialized techniques used in these studies are not easily applicable to other MOFs. Thus, there is a need to develop new experimental and computational approaches to understand the structure and properties of defects in a wider variety of MOFs. Here, we show how low-frequency phonon modes measured by inelastic neutron scattering (INS) spectroscopy can be combined with density functional theory (DFT) simulations to provide unprecedented insights into the defect structure of UiO-66. We are able to identify and assign peaks in the fingerprint region (<100 cm-1) which correspond to phonon modes only present in certain defective topologies. Specifically, this analysis suggests that our sample of UiO-66 consists of predominantly defect-free fcu regions with smaller domains corresponding to a defective bcu topology with 4 and 2 acetate ligands bound to the Zr6O8 nodes. Importantly, the INS/DFT approach provides detailed structural insights (e.g., relative positions and numbers of acetate ligands) that are not accessible with microscopy-based techniques. The quantitative agreement between DFT simulations and the experimental INS spectrum combined with the relative simplicity of sample preparation, suggests that this methodology may become part of the standard and preferred protocol for the characterization of MOFs, and, in particular, for elucidating the structure defects in these materials.

Atomically Dispersed Platinum in Surface and Subsurface Sites on MgO Have Contrasting Catalytic Properties for CO Oxidation.

Atomically dispersed metals on metal oxide supports are a rapidly growing class of catalysts. Developing an understanding of where and how the metals are bonded to the supports is challenging because support surfaces are heterogeneous, and most reports lack a detailed consideration of these points. Herein, we report two atomically dispersed CO oxidation catalysts having markedly different metal-support interactions: platinum in the first layer of crystalline MgO powder and platinum in the second layer of this support. Structural models have been determined on the basis of data and computations, including those determined by extended X-ray absorption fine structure and X-ray absorption near edge structure spectroscopies, infrared spectroscopy of adsorbed CO, and scanning transmission electron microscopy. The data demonstrate the transformation of surface to subsurface platinum as the temperature of sample calcination increased. Catalyst performance data demonstrate the lower activity but greater stability of the subsurface platinum than of the surface platinum.

A Theory-Guided X-ray Absorption Spectroscopy Approach for Identifying Active Sites in Atomically Dispersed Transition-Metal Catalysts.

Atomically dispersed supported metal catalysts offer new properties and the benefits of maximized metal accessibility and utilization. The characterization of these materials, however, remains challenging. Using atomically dispersed platinum supported on crystalline MgO (chosen for its well-defined bonding sites) as a prototypical example, we demonstrate how systematic density functional theory calculations for assessing all the potentially stable platinum sites, combined with automated analysis of extended X-ray absorption fine structure (EXAFS) spectra, leads to unbiased identification of isolated, surface-enveloped platinum cations as the catalytic species for CO oxidation. The catalyst has been characterized by atomic-resolution imaging and EXAFS and high-energy resolution fluorescence detection X-ray absorption near edge spectroscopy. The proposed platinum sites are in agreement with experiment. This theory-guided workflow leads to rigorously determined structural models and provides a more detailed picture of the structure of the catalytically active site than what is currently possible with conventional EXAFS analyses. As this approach is efficient and agnostic to the metal, support, and catalytic reaction, we posit that it will be of broad interest to the materials characterization and catalysis communities.

Circumventing Scaling Relations in Oxygen Electrochemistry Using Metal-Organic Frameworks.

It has been well-established that unfavorable scaling relationships between *OOH, *OH, and *O are responsible for the high overpotentials associated with oxygen electrochemistry. A number of strategies have been proposed for breaking these linear constraints for traditional electrocatalysts (e.g., metals, alloys, metal-doped carbons); such approaches have not yet been validated experimentally for heterogeneous catalysts. Development of a new class of catalysts capable of circumventing such scaling relations remains an ongoing challenge in the field. In this work, we use density functional theory (DFT) calculations to demonstrate that bimetallic porphyrin-based MOFs (PMOFs) are an ideal materials platform for rationally designing the 3-D active site environments for oxygen reduction reaction (ORR). Specifically, we show that the *OOH binding energy and the theoretical limiting potential can be optimized by appropriately tuning the transition metal active site, the oxophilic spectator, and the MOF topology. Our calculations predict theoretical limiting potentials as high as 1.07 V for Fe/Cr-PMOF-Al, which exceeds the Pt/C benchmark for 4e ORR. More broadly, by highlighting their unique characteristics, this work aims to establish bimetallic porphyrin-based MOFs as a viable materials platform for future experimental and theoretical ORR studies.

Structural determinants in a glucose-containing lipopolysaccharide from Mycobacterium tuberculosis critical for inducing a subset of protective T cells.

Mycobacteria synthesize intracellular, 6-O-methylglucose-containing lipopolysaccharides (mGLPs) proposed to modulate bacterial fatty acid metabolism. Recently, it has been shown that Mycobacterium tuberculosis mGLP specifically induces a specific subset of protective γ9δ2 T cells. Mild base treatment, which removes all the base-labile groups, reduces the specific activity of mGLP required for induction of these T cells, suggesting that acylation of the saccharide moieties is required for γ9δ2 T-cell activation. On the basis of this premise, we used analytical LC/MS and NMR methods to identify and locate the acyl functions on the mGLP saccharides. We found that mGLP is heterogeneous with respect to acyl functions and contains acetyl, isobutyryl, succinyl, and octanoyl groups and that all acylations in mGLP, except for succinyl and octanoyl residues, reside on the glucosyl residues immediately following the terminal 3-O-methylglucose. Our analyses also indicated that the octanoyl residue resides at position 2 of an internal glucose toward the reducing end. LC/MS analysis of the residual product obtained by digesting the mGLP with pancreatic α-amylase revealed that the product is an oligosaccharide terminated by α-(1→4)-linked 6-O-methyl-d-glucosyl residues. This oligosaccharide retained none of the acyl groups, except for the octanoyl group, and was unable to induce protective γ9δ2 T cells. This observation confirmed that mGLP induces γ9δ2 T cells and indicated that the acylated glucosyl residues at the nonreducing terminus of mGLP are required for this activity.