Application of a new informatics tool in heterogeneous catalysis: Analysis of methanol dehydrogenation on transition metal catalysts for the production of anhydrous formaldehyde
Explore the source record for details and available documents.
SEARCH · Engineering Papers
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Explore the source record for details and available documents.
Not Available
We have developed a multi-scale approach (Burke, M. P.; Klippenstein, S. J.; Harding, L. B. Proc. Combust. Inst. 2013, 34, 547–555.) to kinetic model formulation that directly incorporates elementary kinetic theories as a means to provide reliable, physics-based extrapolation to unexplored conditions. Here, we extend and generalize the multi-scale modeling strategy to treat systems of considerable complexity – involving multi-well reactions, potentially missing reactions, non-statistical product branching ratios, and non-Boltzmann (i.e. non-thermal) reactant distributions. The methodology is demonstrated here for a subsystem of low-temperature propane oxidation, as a representative system for low-temperature fuel oxidation. A multi-scale model is assembled and informed by a wide variety of targets that include ab initio calculations of molecular properties, rate constant measurements of isolated reactions, and complex systems measurements. Active model parameters are chosen to accommodate both “parametric” and “structural” uncertainties. Theoretical parameters (e.g. barrier heights) are included as active model parameters to account for parametric uncertainties in the theoretical treatment; experimental parameters (e.g. initial temperatures) are included to account for parametric uncertainties in the physical models of the experiments. RMG software is used to assess potential structural uncertainties due to missing reactions. Additionally, branching ratios among product channels are included as active model parameters to account for structural uncertainties related to difficulties in modeling sequences of multiple chemically activated steps. The approach is demonstrated here for interpreting time-resolved measurements of OH, HO2, n-propyl, i-propyl, propene, oxetane, and methyloxirane from photolysis-initiated low-temperature oxidation of propane at pressures from 4 to 60 Torr and temperatures from 300 to 700 K. In particular, the multi-scale informed model provides a consistent quantitative explanation of both ab initio calculations and time-resolved species measurements. The present results show that interpretations of OH measurements are significantly more complicated than previously thought – in addition to barrier heights for key transition states considered previously, OH profiles also depend on additional theoretical parameters for R + O2 reactions, secondary reactions, QOOH + O2 reactions, and treatment of non-Boltzmann reaction sequences. Extraction of physically rigorous information from those measurements may require more sophisticated treatment of all of those model aspects, as well as additional experimental data under more conditions, to discriminate among possible interpretations and ensure model reliability. Keywords: Optimization, Uncertainty quantification, Chemical mechanism, Low-Temperature Oxidation, Non-Boltzmann
Redox proteomics has yielded molecular insight into diseases of protein dysfunction attributable to oxidative stress, underscoring the need for robust detection of protein oxidation products. Additionally, oxidative protein surface mapping techniques utilize hydroxyl radicals to gain structural insight about solvent exposure. Interpretation of tandem mass spectral data is a critical challenge for such investigations, because reactive oxygen species target a wide breadth of amino acids. Additionally, oxidized peptides may be generated in a wide range of abundances since the reactivity of hydroxyl radicals with different amino acids spans three orders of magnitude. Taken together, these attributes of oxidative footprinting pose both experimental and computational challenges to detecting oxidized peptides that are naturally less abundant than their unoxidized counterparts. In this study, three model proteins were oxidized electrochemically and analyzed at both the intact protein and peptide levels. A multidimensional chromatographic strategy was utilized to expand the dynamic range of oxidized peptides measurements. Peptide mass spectral data were searched by the hybrid software packages Inspect and Byonic, which incorporate de novo elements of spectral interpretation into a database search. This dynamic search capacity accommodates the challenge of searching for more than forty oxidative mass shifts that can occur in a staggering variety of possible combinatorial occurrences. A prevailing set of oxidized residues was identified with this comparative approach, and evaluation of these sites was informed by solvent accessible surface area gleaned through molecular dynamics simulations. Along with increased levels of oxidation around highly reactive hotspot sites as expected, the enhanced sensitivity of these measurements uncovered a surprising level of oxidation on less reactive residues.
Not Available
Progress in mass spectrometry lipidomics has led to a rapid proliferation of studies across biology and biomedicine. These generate extremely large raw datasets requiring sophisticated solutions to support automated data processing. To address this, numerous software tools have been developed and tailored for specific tasks. However, for researchers, deciding which approach best suits their application relies on ad hoc testing, which is inefficient and time consuming. Here we first review the data processing pipeline, summarizing the scope of available tools. Next, to support researchers, LIPID MAPS provides an interactive online portal listing open-access tools with a graphical user interface. This guides users towards appropriate solutions within major areas in data processing, including (1) lipid-oriented databases, (2) mass spectrometry data repositories, (3) analysis of targeted lipidomics datasets, (4) lipid identification and (5) quantification from untargeted lipidomics datasets, (6) statistical analysis and visualization, and (7) data integration solutions. Detailed descriptions of functions and requirements are provided to guide customized data analysis workflows.
The classification of miscible and immiscible systems of binary alloys plays a critical role in the design of multicomponent alloys. By mining data from hundreds of experimental phase diagrams, and thousands of thermodynamic data sets from experiments and high-throughput first-principles (HTFP) calculations, we have obtained a comprehensive classification of alloying behavior for 813 binary alloy systems consisting of transition and lanthanide metals. Among several physics-based descriptors, the slightly modified Pettifor chemical scale provides a unique two-dimensional map that divides the miscible and immiscible systems into distinctly clustered regions. Based on an artificial neural network algorithm and elemental similarity, the miscibility of the unknown systems is further predicted and a complete miscibility map is thus obtained. Impressively, the classification by the miscibility map yields a robust validation on the capability of the well-known Miedema’s theory (95% agreement) and shows good agreement with the HTFP method (90% agreement). Our results demonstrate that a state-of-the-art physics-guided data mining can provide an efficient pathway for knowledge discovery in the next generation of materials design.
Abstract The application of machine learning (ML) techniques in materials science has attracted significant attention in recent years, due to their impressive ability to efficiently extract data-driven linkages from various input materials representations to their output properties. While the application of traditional ML techniques has become quite ubiquitous, there have been limited applications of more advanced deep learning (DL) techniques, primarily because big materials datasets are relatively rare. Given the demonstrated potential and advantages of DL and the increasing availability of big materials datasets, it is attractive to go for deeper neural networks in a bid to boost model performance, but in reality, it leads to performance degradation due to the vanishing gradient problem. In this paper, we address the question of how to enable deeper learning for cases where big materials data is available. Here, we present a general deep learning framework based on Individual Residual learning (IRNet) composed of very deep neural networks that can work with any vector-based materials representation as input to build accurate property prediction models. We find that the proposed IRNet models can not only successfully alleviate the vanishing gradient problem and enable deeper learning, but also lead to significantly (up to 47%) better model accuracy as compared to plain deep neural networks and traditional ML techniques for a given input materials representation in the presence of big data.
Abstract Modern machine learning (ML) and deep learning (DL) techniques using high-dimensional data representations have helped accelerate the materials discovery process by efficiently detecting hidden patterns in existing datasets and linking input representations to output properties for a better understanding of the scientific phenomenon. While a deep neural network comprised of fully connected layers has been widely used for materials property prediction, simply creating a deeper model with a large number of layers often faces with vanishing gradient problem, causing a degradation in the performance, thereby limiting usage. In this paper, we study and propose architectural principles to address the question of improving the performance of model training and inference under fixed parametric constraints. Here, we present a general deep-learning framework based on branched residual learning (BRNet) with fully connected layers that can work with any numerical vector-based representation as input to build accurate models to predict materials properties. We perform model training for materials properties using numerical vectors representing different composition-based attributes of the respective materials and compare the performance of the proposed models against traditional ML and existing DL architectures. We find that the proposed models are significantly more accurate than the ML/DL models for all data sizes by using different composition-based attributes as input. Further, branched learning requires fewer parameters and results in faster model training due to better convergence during the training phase than existing neural networks, thereby efficiently building accurate models for predicting materials properties.
The role of dynamical (or Born effective) charges in classification of octet AB-type binary compounds between four-fold (zincblende/wurtzite crystal structures) and six-fold (rocksalt crystal structure) coordinated systems is discussed. We show that the difference in the dynamical charges of the fourfold and sixfold coordinated structures, in combination with Harrison’s polarity, serves as an excellent feature to classify the coordination of 82 sp–bonded binary octet compounds. We use a support vector machine classifier to estimate the average classification accuracy and the associated variance in our model where a decision boundary is learned in a supervised manner. Lastly, we compare the out-of-sample classification accuracy achieved by our feature pair with those reported previously.
Explore the source record for details and available documents.
Not Available
This paper addresses cognitive implications and research needs surrounding the problem of cyber friendly fire (FF). We define cyber FF as intentional offensive or defensive cyber/electronic actions intended to protect cyber systems against enemy forces or to attack enemy cyber systems, which unintention-ally harms the mission effectiveness of friendly or neutral forces. Just as with combat friendly fire, maintaining situation awareness (SA) is paramount to avoiding cyber FF incidents. Cyber SA concerns knowledge of a system’s topology (connectedness and relationships of the nodes in a system), and critical knowledge elements such as the characteristics and vulnerabilities of the components that comprise the system and its nodes, the nature of the activities or work performed, and the available defensive and offensive countermeasures that may be applied to thwart network attacks. Mitigation strategies to combat cyber FF— including both training concepts and suggestions for decision aids and visualization approaches—are discussed.
Not provided.
Glass corrosion data from the ALTGLASS™ database were used to determine if gel compositions, which evolve as glass systems corrode, are correlated with the generation of zeolites and subsequent increase in the glass dissolution rate at long times. The gel compositions were estimated based on the difference between the elemental glass starting compositions and the measured elemental leachate concentrations from the long-term product consistency tests (ASTM C1285) at various stages of dissolution, ie, reaction progress. A well-characterized subset of high level waste glasses from the database was selected: these glasses had been leached for 15-20 years at reaction progresses up to ~80%. The gel composition data, at various reaction progresses, were subjected to a step-wise regression, which demonstrated that hydrogel compositions with Si*/Al* ratios of <1.0 did not generate zeolites and maintained low dissolution rates for the duration of the experiments. Glasses that formed hydrogel compositions with Si^*/Al^* ratios ≥1, generated zeolites accompanied by a resumption in the glass dissolution rate. Finally, the role of the gel Si/Al ratio, and the interactions with the leachate, provides the fundamental understanding needed to predict if and when the glass dissolution rate will increase due to zeolitization.
Purpose: Recent legislative and accreditation requirements have driven rapid development and implementation of CT radiation dose monitoring solutions. Institutions must determine how to improve quality, safety, and consistency of their clinical performance. The purpose of this work was to design a strategy and meaningful characterization of results from an in-house, clinically-deployed dose monitoring solution. Methods: A dose monitoring platform was designed by our imaging physics group that focused on extracting protocol parameters, dose metrics, and patient demographics and size. Compared to most commercial solutions, which focus on individual exam alerts and global thresholds, the program sought to characterize overall consistency and targeted thresholds based on eight analytic interrogations. Those were based on explicit questions related to protocol application, national benchmarks, protocol and size-specific dose targets, operational consistency, outliers, temporal trends, intra-system variability, and consistent use of electronic protocols. Using historical data since the start of 2013, 95% and 99% intervals were used to establish yellow and amber parameterized dose alert thresholds, respectively, as a function of protocol, scanner, and size. Results: Quarterly reports have been generated for three hospitals for 3 quarters of 2015 totaling 27880, 28502, 30631 exams, respectively. Four adult and two pediatric protocols were higher than external institutional benchmarks. Four protocol dose levels were being inconsistently applied as a function of patient size. For the three hospitals, the minimum and maximum amber outlier percentages were [1.53%,2.28%], [0.76%,1.8%], [0.94%,1.17%], respectively. Compared with the electronic protocols, 10 protocols were found to be used with some inconsistency. Conclusion: Dose monitoring can satisfy requirements with global alert thresholds and patient dose records, but the real value is in optimizing patient-specific protocols, balancing image quality trade-offs that dose-reduction strategies promise, and improving the performance and consistency of a clinical operation. Data plots that capture patient demographics and scanner performance demonstrate that value.