Engineering topics
Medford, Andrew J.
Publications and source records attributed to Medford, Andrew J..
Gaussian approximation of dispersion potentials for efficient featurization and machine-learning predictions of metal–organic frameworks
Energy-related descriptors in machine learning are a promising strategy to predict adsorption properties of metal–organic frameworks (MOFs) in the low-pressure regime. Interactions between hosts and guests in these systems are typically expressed as a sum of dispersion and electrostatic potentials. The energy landscape of dispersion potentials plays a crucial role in defining Henry’s constants for simple probe molecules in MOFs. To incorporate more information about this energy landscape, we introduce the Gaussian-approximated Lennard-Jones (GALJ) potential, which fits pairwise Lennard-Jones potentials with multiple Gaussians by varying their heights and widths. The GALJ approach is capable of replicating information that can be obtained from the original LJ potentials and enables efficient development of Gaussian integral (GI) descriptors that account for spatial correlations in the dispersion energy environment. GI descriptors would be computationally inconvenient to compute using the usual direct evaluation of the dispersion potential energy surface. Here, we show that these new GI descriptors lead to improvement in ML predictions of Henry’s constants for a diverse set of adsorbates in MOFs compared to previous approaches to this task.
Quantifying the impact of temporal analysis of products reactor initial state uncertainties on kinetic parameters
The temporal analysis of products (TAP) reactor, a transient kinetic tool, provides users with information as the catalyst state evolves. However, the state of the art for TAP uncertainty quantification only considers the experimental noise present in the outlet flow signal. Additional sources of uncertainty, including, but not limited to, initial surface coverages, the catalyst zone location, the inert void fraction, and the gas pulse intensity and pulse delay, are not included. For this reason, a framework for quantifying all uncertainty sources present in TAP experiments is presented and applied to a carbon monoxide oxidation case study. Herein, two methods for quantifying these sources of uncertainty are introduced. The first utilizes initial state sensitivities to approximate the parameter variances, as well as to provide insights into the structural certainty of the model. The second generates parameter confidence distributions through an ensemble-based sampling algorithm. This initial state covariance matrix can ultimately be merged with the experimental noise covariance matrix, providing a unified description of the parameter uncertainties for a single TAP experiment.
Kinetics-Informed Neural Networks
Chemical kinetics and reaction engineering consists of the phenomenological framework for the disentanglement of reaction mechanisms, optimization of reaction performance and the rational design of chemical processes. Here, we utilize feed-forward artificial neural networks as basis functions to solve ordinary differential equations (ODEs) constrained by differential algebraic equations (DAEs) that describe microkinetic models (MKMs). We present an algebraic framework for the mathematical description and classification of reaction networks, types of elementary reaction, and chemical species. Under this framework, we demonstrate that the simultaneous training of neural nets and kinetic model parameters in a regularized multi-objective optimization setting leads to the solution of the inverse problem through the estimation of kinetic parameters from synthetic experimental data. We analyze a set of scenarios to establish the extent to which kinetic parameters can be retrieved from transient kinetic data, and assess the robustness of the methodology with respect to statistical noise. Furthermore, this approach to inverse kinetic ODEs can assist in the elucidation of reaction mechanisms based on transient data.
Supported molybdenum oxides for the aldol condensation reaction of acetaldehyde
The (retro-)aldol condensation reaction is an important chemical transformation in the upgrading of biomass-derived compounds into fuels and valuable specialty chemicals. In this study, we found that supported molybdenum oxide (MoO x ) catalysts were active and selective for the aldol condensation of acetaldehyde to crotonaldehyde under steady-state reactor conditions. Through a combination of transmission electron microscopy (TEM), ultraviolet–visible (UV–VIS) diffuse reflectance spectroscopy, Fourier transform infrared (FTIR) spectroscopy of adsorbed pyridine, and steady-state reactor testing, we determined that highly dispersed MoO x has a strong interaction with a γ-Al 2 O 3 support resulting in optimal catalyst performance at low weight loadings. In contrast, MoO x particles supported on SiO 2 have a weaker interaction with the support, resulting in a monotonic relationship between Mo loading and aldol condensation activity. The Lewis acid site density and strength are important parameters for predicting aldol condensation activity across all samples. The concentration of weak acid sites had a poor correlation with aldol condensation activity, most likely because these sites are too weak to activate acetaldehyde for the reaction. Medium and strong acid sites both had good correlations to aldol condensation activity. Results from X-ray absorption near edge structure (XANES) and acetaldehyde temperature programmed desorption (TPD) indicated that partially reduced MoO x was more active for aldol condensation, but pretreatment in reducing or oxidizing environments had no significant effect on steady-state catalytic activity. Finally, characterization of spent catalyst samples through temperature programmed oxidation (TPO) and thermogravimetric analysis (TGA) revealed that catalysts with high densities of strong acid sites tended to form more carbonaceous deposits on the surface over the course of the reaction.
Ab-Initio Investigation of Finite Size Effects in Rutile Titania Nanoparticles with Semilocal and Nonlocal Density Functionals
In this work, we employ hybrid and generalized gradient approximation (GGA) level density functional theory (DFT) calculations to investigate the convergence of surface properties and electronic gap of rutile titania nanoparticles with particle size. The surface energies and electronic gaps are calculated for cuboidal particles with minimum dimension ranging from 3.7 Angstrom (24 atoms) to 10.3 Angstrom (384 atoms) using a highly-parallel real-space DFT code to enable hybrid level DFT calculations of larger nanoparticles than are typically practical. We deconvolute the geometric and electronic finite size effects in surface energy, and evaluate the influence of defects on electronic gap and density of states (DOS). The electronic finite size effects in surface energy vanish when the minimum length scale of the nanoparticles becomes greater than 10 Angstrom. We show that this length scale is consistent with a computationally efficient numerical analysis of the characteristic length scale of electronic interactions. The surface energy of nanoparticles having minimum dimension beyond this characteristic length can be approximated using slab calculations that account for the geometric defects. In contrast, the finite size effects on the electronic gap and DOS is highly dependent on the shape and size of these particles. Furthermore, the DOS for cuboidal particles and more realistic particles constructed using the Wulff algorithm reveal that defect states within the electronic gap play a key role in determining the eigen value distribution of nanoparticles and the electronic gap does not converge to the bulk limit for the particle sizes investigated.
TAPsolver: A Python package for the simulation and analysis of TAP reactor experiments
Not Available
Direct aromatization of CO 2 via combined CO 2 hydrogenation and zeolite-based acid catalysis
Aromatics, including benzene, toluene, and xylenes (BTX), are essential chemical building blocks and are widely used as solvents, fuel additives, and polymers. With the recent development in CO 2 capture technologies and the progress made in producing H 2 using renewable energy, direct hydrogenation of CO 2 to aromatics via heterogeneous catalysis has emerged as a promising pathway to accomplish the production of aromatics with simultaneous utilization of waste CO 2 . In this review, we focus on recent advances in the nascent field of direct CO 2 aromatization, whereby tandem catalysts composed of CO 2 hydrogenation and aromatization functionalities are designed and deployed. We review two categories of tandem catalysts: catalysts integrating Fe-based/H-ZSM-5 components following RWGS (reverse water-gas shift of CO 2 to CO)-FT (Fischer-Tropsch synthesis of lower olefins)-aromatization pathways, and catalysts combining metal oxide/H-ZSM-5 domains following CO 2 to methanolaromatization pathways. The key parameters that determine the catalytic performance, such as the composition and structure of the Fe-based or metal oxide-based CO 2 conversion catalysts, the properties of H-ZSM-5, and the synergy between the two components, are analyzed to provide insights for the design of efficient tandem catalysts for CO 2 aromatization. In parallel, thermodynamic analyses, mechanistic studies, and density functional theory (DFT) computations for the relevant reaction routes and pathways are discussed to offer improved understanding of CO 2 activation, reaction intermediates, and product formation. In closing, the challenges and prospects for these tandem reactions are addressed to provide suggested paths forward for future research.
Mechanocatalytic Ammonia Synthesis over TiN in Transient Microenvironments
Environmentally friendly and energy-efficient ways to produce ammonia are essential to meet global food demands. In this study, a new approach for ammonia production at nominally ambient conditions is introduced. As proof of concept, ammonia is synthesized mechanocatalytically by ball milling titanium in a continuous gas flow. The ammonia synthesis reaction is proposed to follow a transient Mars–van Krevelen mechanism under mechanically activated conditions, where molecular nitrogen incorporation into the titanium lattice and titanium nitride hydrogenation occur in thermodynamically distinct environments. X-ray powder diffraction and X-ray absorption spectroscopy confirm the formation of titanium nitride from titanium and N2. The reactivity of nitrided titanium supports that lattice nitrogen plays a role in ammonia formation. The in situ formed titanium nitride is catalytically active, and the nitride regeneration reaction is determined to be the rate-limiting step. A preliminary technoeconomic analysis shows that this approach could be feasible for distributed ammonia production.
Pretreatment Effects on the Surface Chemistry of Small Oxygenates on Molybdenum Trioxide
Understanding surface reactions of biomass-derived oxygenates on metal oxides is important for designing catalysts for valorization of biomass. As such, This work elucidated the effect of different pretreatments on molybdenum trioxide (MoO 3 ) to understand how surface reactivity is controlled by the surface oxidation state. The catalyst was pretreated in oxidative, inert, and reducing environments. The inert and reducing pretreatments created oxygen vacancies on the catalyst surface that acted as active sites for the adsorption of oxygenated molecules, with the reducing pretreatment yielding a higher density of these active sites. Exposing the catalyst to an alcoholic solvent such as methanol also led to a partial reduction similar to the inert pretreatment. After pretreatment, the catalyst was exposed to ethanol, acetaldehyde, and crotonaldehyde with subsequent characterization by diffuse reflectance infrared spectroscopy (DRIFTS), temperature-programmed desorption (TPD), X-ray absorption near edge spectroscopy (XANES), and X-ray photoelectron spectroscopy (XPS). Density functional theory (DFT) was also used to determine adsorption configurations and energies of ethanol, acetaldehyde, and crotonaldehyde. Reduced surfaces were shown to have a stronger affinity for carbonyls, leading to a higher activity for the aldol condensation of acetaldehyde and ethanol to C 4 molecules. Catalysts pretreated in an oxidative environment were completely inactive toward chemisorption and reaction of acetaldehyde.