No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Speciation of Ru Molecular Complexes in a Homogeneous Catalytic System: Fingerprint XANES Analysis Guided by Machine Learning
ResearchGate has not been able to resolve any citations for this publication.
X-ray absorption spectroscopy (XAS) produces a wealth of information about the local structure of materials, but interpretation of spectra often relies on easily accessible trends and prior assumptions about the structure. Recently, researchers have demonstrated that machine learning models can automate this process to predict the coordinating environments of absorbing atoms from their XAS spectra. However, machine learning models are often difficult to interpret, making it challenging to determine when they are valid and whether they are consistent with physical theories. In this work, we present three main advances to the data-driven analysis of XAS spectra: we demonstrate the efficacy of random forests in solving two new property determination tasks (predicting Bader charge and mean nearest neighbor distance), we address how choices in data representation affect model interpretability and accuracy, and we show that multiscale featurization can elucidate the regions and trends in spectra that encode various local properties. The multiscale featurization transforms the spectrum into a vector of polynomial-fit features, and is contrasted with the commonly-used “pointwise” featurization that directly uses the entire spectrum as input. We find that across thousands of transition metal oxide spectra, the relative importance of features describing the curvature of the spectrum can be localized to individual energy ranges, and we can separate the importance of constant, linear, quadratic, and cubic trends, as well as the white line energy. This work has the potential to assist rigorous theoretical interpretations, expedite experimental data collection, and automate analysis of XAS spectra, thus accelerating the discovery of new functional materials.
X-ray spectroscopy delivers strong impact across the physical and biological sciences by providing end-users with highly-detailed information about the electronic and geometric structure of matter. To decode this information in challenging cases, e.g. in operando catalysts, batteries, and temporally-evolving systems, advanced theoretical calculations are necessary. The complexity and resource requirements often render these out of reach for end-users, and therefore data are often not interpreted exhaustively, leaving a wealth of valuable information unexploited. In this paper, we introduce supervised machine learning of X-ray absorption spectra, by developing a deep neural network (DNN) that is able to estimate Fe K-edge X-ray absorption near-edge structure spectra in less than a second with no input beyond geometric information about the local environment of the absorption site. We predict peak positions with sub-eV accuracy and peak intensities with errors over an order of magnitude smaller than the spectral variations that the model is engineered to capture. The performance of the DNN is promising, as illustrated by its application to the structural refinement of iron(II)tris(bipyridine) and nitrosylmyoglobin, but also highlights areas for which future developments should focus.
Analyzing coordination environments using X-ray absorption spectroscopy has broad applications in solid-state physics and material chemistry. Here, we show that random forest models trained on 190,000 K-edge X-ray absorption near-edge structure (XANES) spectra can identify the main atomic coordination environment with a high accuracy of 85.4% and all associated coordination environments with a high Jaccard score of 81.8% for 33 cation elements in oxides, significantly outperforming other machine-learning models. In a departure from prior works, the coordination environment is described as a distribution over 25 distinct coordination motifs with coordination numbers ranging from 1 to 12. More importantly, we show that the random forest models can be used to predict coordination environments from experimental K-edge XANES with minimal loss in accuracy. A drop-variable feature importance analysis highlights the key roles that the pre-edge and main-peak regions play in coordination environment identification.
Element selectivity and possibilities for in situ and operando applications make X-ray absorption spectroscopy a powerful tool for structural characterization of catalysts. While determination of coordination numbers and interatomic distances from extended spectral region is rather straightforward, analysis of X-ray absorption near-edge structure (XANES) spectra remains a highly debated and topical problem. The latter region of spectra is shaped depending on the local 3D geometry and electronic structure. However, there is no straightforward procedure for the unambiguous extraction of these parameters. This work gives a critical vision on the amount of information that can be practically extracted from Pd K-edge XANES spectra measured under in situ and operando conditions, in which adsorption of reactive molecules at the surface of palladium with further formation of subsurface and bulk palladium carbides are expected. We investigate how particle size, concentration of carbon impurities, and their distribution in the bulk and at the surface of palladium particles affect Pd K-edge XANES features and to which extend they should be implemented in the theoretical model to adequately reproduce experimental data. Then, we show how the step-by-step increasing the complexity of the theoretical model improves the agreement with experiment. Finally, we suggest a set of formal descriptors relevant to possible structural diversity and construct a library of theoretical spectra for machine-learning-based analysis of the experimental data.
A brief review of studies using X-ray Absorption Spectroscopy (XAS) to investigate homogeneous catalytic reactions in fine chemical and pharmaceutical context since 2010 is presented. The advantages of the techniques over traditional lab-based analytical tools, particularly when NMR spectroscopy fails to deliver mechanistic insights, are summarised using these examples. A discussion on the current limitations of the techniques and challenges in the near future is also included.
Graphical abstract
A minireview of recent developments in application of X-ray Absorption Spectroscopy as an effective mechanistic tool to synthetic catalytic reactions relevant to fine chemical and pharmaceutical syntheses.
This paper proposes a new tree-based ensemble method for supervised classification and regression problems. It essentially consists of randomizing strongly both attribute and cut-point choice while splitting a tree node. In the extreme case, it builds totally randomized trees whose structures are independent of the output values of the learning sample. The strength of the randomization can be tuned to problem specifics by the appropriate choice of a parameter. We evaluate the robustness of the default choice of this parameter, and we also provide insight on how to adjust it in particular situations. Besides accuracy, the main strength of the resulting algorithm is computational efficiency. A bias/variance analysis of the Extra-Trees algorithm is also provided as well as a geometrical and a kernel characterization of the models induced.
A software package for the analysis of X-ray absorption spectroscopy (XAS) data is presented. This package is based on the IFEFFIT library of numerical and XAS algorithms and is written in the Perl programming language using the Perl/Tk graphics toolkit. The programs described here are: (i) ATHENA, a program for XAS data processing, (ii) ARTEMIS, a program for EXAFS data analysis using theoretical standards from FEFF and (iii) HEPHAESTUS, a collection of beamline utilities based on tables of atomic absorption data. These programs enable high-quality data analysis that is accessible to novices while still powerful enough to meet the demands of an expert practitioner. The programs run on all major computer platforms and are freely available under the terms of a free software license.
The present study describes the synthesis of primary amines from long-chain fatty alcohols and ammonia using supported ruthenium catalysts over different acid supports, including a variety of zeolites with different topologies and Si/Al ratios. The morphology, acidity and location of ruthenium in the catalysts was studied in detail by combining XRD, BET, HR-TEM, NH3-TPD, octylamine-TPD, H2-TPR, XPS, EXAFS / XANES, ²⁷Al MAS NMR and TGA. In particular, Ru/HBEA (Si/Al = 25) with 5 wt% Ru afforded more than 90% conversion and 90% selectivity to 1-octylamine in the liquid-phase amination reaction of 1-octanol with ammonia at 180 °C in a batch reactor. The high selectivity of Ru/HBEA (Si/Al = 25) can be explained by the presence of Brønsted / Lewis acid centers with medium strength in the proximity of ruthenium nanoparticles. The catalyst was further tested in a pre-pilot continuous stirred-tank reactor (2 L) with flash separation of 1-octylamine. In this configuration, a steady 92% selectivity of octylamine was obtained at 87% 1-octanol conversion during 120 h on steam. The catalyst kept its integrity after the reaction.
Hydrodeoxygenation (HDO) of sugar alcohols could be a valuable reaction for the renewable production of alkenes. However, reports on deoxygenation of longer polyols (e.g. erythritol, xylitol, sorbitol) to mono-alkenes are scarce, especially when using H2 as reducing agent. Here we design a conceptually different catalyst system for polyol conversion to olefins, containing a selective Ru hydrogenation function (e.g. RuBr3) in an ionic liquid that dehydrates alcohols (tetrabutylphosphonium bromide, Bu4PBr). This system performs HDO of erythritol to butenes in yields up to 69%. Strikingly, overreduction to less valuable butane is effectively suppressed. We show that in situ formed CO is crucial in the formation and stabilization of the active and selective catalyst. The actual hydrogenation catalyst is identified as a Ru carbonyl bromide species, as demonstrated by FTIR and XANES. Finally, we show that this HDO also works for more economically relevant polyols: glycerol, xylitol and sorbitol, leading to propene, pentenes and hexenes as main products respectively.
Simulations of excited state properties, such as spectral functions, are often computationally expensive and therefore not suitable for high-throughput modeling. As a proof of principle, we demonstrate that graph-based neural networks can be used to predict the x-ray absorption near-edge structure spectra of molecules to quantitative accuracy. Specifically, the predicted spectra reproduce nearly all prominent peaks, with 90% of the predicted peak locations within 1 eV of the ground truth. Besides its own utility in spectral analysis and structure inference, our method can be combined with structure search algorithms to enable high-throughput spectrum sampling of the vast material configuration space, which opens up new pathways to material design and discovery.
Palladium-based catalysts are exploited in the industrial scale for selective hydrogenation of hydrocarbons. Formation of palladium carbide and hydride phases under reaction conditions changes the catalytic properties of the material, which points to the importance of operando characterization for determining a relation between the relative fraction of the two phases and the catalyst performance. We present a combined time-resolved characterization by X-ray absorption spectroscopy (in both near-edge and extended regions) and X-ray diffraction of a working palladium-based catalyst during hydrogenation of ethylene in a wide range of partial pressures of ethylene and hydrogen. Synergistic coupling of multiple techniques allowed us to follow the structural evolution of palladium lattice as well as the transitions between metallic, hydride and carbide phases of palladium. Nanometric dimensions of the particles resulted in the considerable contribution of both surface and bulk carbides to the absorption spectra. During reaction, palladium carbide is formed, which does not lead to a loss of activity. Unusual contraction of the unit cell parameter of palladium lattice in a spent catalyst was observed upon increasing hydrogen partial pressure.
X-ray absorption near-edge spectroscopy (XANES) is becoming an extremely popular tool for material science thanks to the development of new synchrotron radiation light sources. It provides information about charge state and local geometry around atoms of interest in operando and extreme conditions. However, in contrast to X-ray diffraction, a quantitative analysis of XANES spectra is rarely performed in the research papers. The reason must be found in the larger amount of time required for the calculation of a single spectrum compared to a diffractogram. For such time-consuming calculations, in the space of several structural parameters, we developed an interpolation approach proposed originally by Smolentsev and Soldatov (2007). The current version of this software, named PyFitIt, is a major upgrade version of FitIt and it is based on machine learning algorithms. We have chosen Jupyter Notebook framework to be friendly for users and at the same time being available for remastering. The analytical work is divided into two steps. First, the series of experimental spectra are analyzed statistically and decomposed into principal components. Second, pure spectral profiles, recovered by principal components, are fitted by theoretical interpolated spectra. We implemented different schemes of choice of nodes for approximation and learning algorithms including Gradient Boosting of Random Trees, Radial Basis Functions and Neural Networks. The fitting procedure can be performed both for a XANES spectrum or for a difference spectrum, thus minimizing the systematic errors of theoretical simulations. The problem of several local minima is addressed in the framework of direct and indirect approaches.
Program summary
Program title: PyFitIt.
Program Files doi: http://dx.doi.org/10.17632/ydkgfdc38t.1
Licensing provisions: GNU General Public License 3.
Programming language: Python, Jupyter Notebook framework.
Nature of problem: Quantitative structural refinements of the X-ray absorption near-edge structure spectra (XANES). Identification of the pure spectral and concentration profiles associated with an experimental XANES dataset.
Solution method: The fitting procedure of the experimental XANES spectra or of their differences is realized by means of the inverse and direct approaches based on the training set and approximation machine learning algorithms. The spectral resolution method is based on the PCA technique involving the usage of a target transformation matrix.
Additional comments including restrictions and unusual features: The current version is compatible with the free FDMNES program package for XANES simulations. However, users can prepare their own matrices of spectra calculated by an arbitrary software and the corresponding structural parameters to perform the fitting procedure in PyFitIt. The complete set of examples is distributed along with the program.
References: PyFitIt web page: http://hpc.nano.sfedu.ru/pyfitit/
Rapid growth of methods emerging in the last decade for synthesis of “designer” catalysts – ranging from the size and shape-selected nanoparticles to mass-selected clusters, to precisely engineered bimetallic surfaces, to single site and pair site catalysts - opened the opportunities for tailoring the catalyst structure for the desired activity and selectivity. It also sharpened the need for developing approaches to the operando characterization, ones that identify the catalytic active sites and follow their evolutions in reaction conditions. Commonly used methods for determination of the activity descriptors in the nanocatalysts, based on the correlation between the changes in catalyst performance and evolution of its structural and electronic properties, are hampered by the paucity of experimental techniques that can detect such properties with high accuracy and in reaction conditions. Out of many such techniques, X-ray absorption spectroscopy (XAS) stands out as an element-specific method that is very sensitive to the local geometric and electronic properties of the metal atoms and their surroundings and, therefore, is able to track catalyst structure modifications in operando conditions. Despite the vast amount of structure-specific information (such as, e.g., the charge states and radial distribution function of neighbors of selected atomic species) stored in the XAS data of catalysts, extracting it from the spectra is challenging, especially in the conditions of low metal weight loading, nanoscale dimensions, heterogeneous size and composition distributions, and harsh reaction environment. In this Perspective, we discuss the recent developments in XAS data analysis achieved by employing supervised and unsupervised machine learning (ML) methods for structural characterization of catalysts. By benefiting from the sensitivity of ML methods to subtle variations in experimental data, a previously “hidden” relationship between the X-ray absorption spectrum and descriptors of material’s structure and/or composition can be found, as illustrated on representative examples of mono-, hetero- and non-metallic catalysts. In the case of supervised ML (SML) the experimental spectra can be rapidly “inverted” and structure of the catalyst can be tracked in real time and in reaction conditions. Emerging opportunities for catalysis research that the ML methods enable, such as high throughput data analysis, and their applications to other experimental probes of catalyst structure, are discussed.
In this work we have applied machine learning methods (Extra Trees, Ridge Regression and Neural Networks) to predict structural parameters of the system based on its XANES spectrum. We used two ML approaches: direct one, i.e. when ML model is trained to predict the structural parameters directly from the XANES spectrum and inverse one when ML model is used to approximate spectrum as a function of structural parameters. We show the applicability of several ML approaches to predict the geometry of CO2 molecule adsorbed on Ni²⁺ surface sites hosted in the channels of CPO-27-Ni metal-organic framework. Quantitative fitting is based on difference XANES spectra and we discuss advantages and disadvantages of the two ML approaches and critically examine the overfitting phenomenon, caused by systematic differences of experimental data and learning dataset.
Palladium nanoparticles (NPs) find numerous applications as catalysts for hydrogenation of hydrocarbons. Under reaction conditions, formation of palladium hydrides and carbides may occur, which affects the catalytic properties of the catalyst. Unlike pure hydride phase, whose phase diagram is well-studied, conditions of formation of palladium carbides and mixed hydride/carbide structures in Pd NPs is poorly investigated. We present an operando study of a supported Pd/C catalyst during ethylene hydrogenation reaction in a variable C 2 H 4 /H 2 ratio. By simultaneous analysis of the extended X-ray absorption fine structure (EXAFS) and X-ray absorption near-edge structure (XANES) spectra, we demonstrated an irreversible behavior in the catalyst structure explained by carbide phase formation even in the instant excess of molecular hydrogen with respect to molecular ethylene in the gas feed. The relevant role of the carbide phase was demonstrated by correlations between the carbide phase amount and ethylene conversion. An interesting behavior of the atomic structure of the Pd NPs has been observed upon periodic variation of the amount of ethylene in the reaction mixture. The obtained results provide significant information about the structural evolution of Pd NPs during ethylene hydrogenation reaction and highlight the importance of operando XANES spectroscopy for characterization of the working catalysts.
In the last decade the appearance of progressively more sophisticated codes, together with the increased computational capabilities, has made XANES a spectroscopic technique able to quantitatively confirm (or discard) a structural model, thus becoming a new fundamental diagnostic tool in catalysis, where the active species are often diluted metal centers supported on a matrix. After providing a brief historical introduction and the basic insights on the technique, in this review article, we provide a selection of four examples where operando XANES technique has been able to provide capital information on the structure of the active site in catalysts of industrial relevance: (i) Phillips catalyst for ethylene polymerization reaction; (ii) TS-1 catalyst for selective hydrogenation reactions; (iii) carbon supported Pd nanoparticles for hydrogenation reactions; (iv) Cu-CHA zeolite for NH3-assisted selective reduction of NOx and for partial oxidation of methane to methanol. The last example testifies how the multivariate curve resolution supported by the alternating least-squares algorithm applied to a high number of XANES spectra collected under operando conditions allows to quantitatively determine different species in mutual transformation. This approach is particularly powerful in the analysis of experiments where a large number of spectra has been collected, typical of time- or space-resolved experiments. Finally, machine learning approaches (both indirect and direct) have been applied to determine, from the XANES spectra, the structure of CO, CO2 and NO adsorbed on Ni²⁺ sites of activated CPO-27-Ni metal-organic framework.
Formation of palladium hydride and carbide phases in palladium-based catalysts is a critical process which changes the catalytic performance and selectivity of the catalysts in important industrial reactions, such as selective hydrogenation of alkynes or alkadienes. We present a comprehensive study of a 5 wt.% carbon supported Pd nanoparticles (NPs) catalyst under various environments by in situ and operando X-ray absorption spectroscopy and diffraction, to determine structure and evolution of palladium hydride and carbide phases, and their distribution over the volume of the NPs. We demonstrate how the simultaneous analysis of extended X-ray absorption fine structure (EXAFS) spectra and X-ray powder diffraction (XRPD) patterns allows discriminating between the inner “core” and outer “shell” regions of the nanoparticle during hydride phase formation at different temperatures and under different hydrogen pressures, indicating that the amount of hydrogen in the shell region of the nanoparticle is lower than that in the core. For palladium carbide, advanced analysis of X-ray absorption near-edge structure (XANES) spectra allows detecting Pd-C bonds with carbon-containing molecules adsorbed at the surface of the nanoparticles. In addition, H/Pd and C/Pd stoichiometries of PdHx and PdCy phases were obtained by theoretical modelling and fitting of XANES spectra. Finally, the collection of operando time-resolved XRPD patterns (with a time resolution of 5 s) allowed highlighting, during ethylene hydrogenation reaction, periodic oscillations in the NPs core lattice parameter, which are in phase with the MS signal of the ethane (product) and in antiphase with the MS signal of the H2, highlighting an interesting direct structural-reactivity relationship. The presented studies show how a careful combination of X-ray absorption and diffraction can highlight the structure of core, shell and surface of the palladium nanoparticles and prove their relevant roles in catalysis.
During the last two decades, remarkable and often spectacular progress has been made in the methodological and instrumental aspects of x-ray absorption and emission spectroscopy. This progress includes considerable technological improvements in the design and production of detectors especially with the development and expansion of large-scale synchrotron reactors All this has resulted in improved analytical performance and new applications, as well as in the perspective of a dramatic enhancement in the potential of x-ray based analysis techniques for the near future. This comprehensive two-volume treatise features articles that explain the phenomena and describe examples of X-ray absorption and emission applications in several fields, including chemistry, biochemistry, catalysis, amorphous and liquid systems, synchrotron radiation, and surface phenomena. Contributors explain the underlying theory, how to set up X-ray absorption experiments, and how to analyze the details of the resulting spectra. X-Ray Absorption and X-ray Emission Spectroscopy: Theory and Applications: Combines the theory, instrumentation and applications of x-ray absorption and emission spectroscopies which offer unique diagnostics to study almost any object in the Universe. Is the go-to reference book in the subject for all researchers across multi-disciplines since intense beams from modern sources have revolutionized x-ray science in recent years. Is relevant to students, postdocurates and researchers working on x-rays and related synchrotron sources and applications in materials, physics, medicine, environment/geology, and biomedical materials.
Finite difference method (FDM) implemented in the FDMNES software [Phys. Rev. B, 2001, 63, 125120] was revised. Thorough analysis shows, that the calculated diagonal in the FDM matrix consists of about 96% zero elements. Thus a sparse solver would be more suitable for the problem instead of traditional Gaussian elimination for the diagonal neighbourhood. We have tried several iterative sparse solvers and the direct one MUMPS solver with METIS ordering turned out to be the best. Compared to the Gaussian solver present method is up to 40 times faster and allows XANES simulations for complex systems already on personal computers. We show applicability of the software for metal-organic [Fe(bpy)3]2+ complex both for low spin and high spin states populated after laser excitation.
The rising interest in XANES is due to the experimental evidence that information on coordination geometry and bonding angles, not given by EXAFS, can indeed be extracted from XANES [1,2]. In this paper I would like to point out that certain features of XANES, being sensitive to interatomic distances, can also be exploited to determine first coordination bond lengths to a few percent accuracy (5% or less) [3]. This aspect of XANES is particularly useful whenever EXAFS modulations are too weak to be measured, as in the case of low 7 backscatterers. For the sake of concreteness, I shall limit myself to K-edge absorption spectra of simple molecules containing C-C, C-N, C-O type of bonds and of two series of Mn and V oxides (see Tables I and II). These cases are general enough to serve as an illustration of other cases.
Accurate modeling of the X-ray absorption near-edge spectra (XANES) is required to unravel the local structure of metal sites in complex systems and their structural changes upon chemical or light stimuli. Two relevant examples are reported here concerning the following: (i) the effect of molecular adsorption on 3d metals hosted inside metal?organic frameworks and (ii) light induced dynamics of spin crossover in metal?organic complexes. In both cases, the amount of structural models for simulation can reach a hundred, depending on the number of structural parameters. Thus, the choice of an accurate but computationally demanding finite difference method for the ab initio X-ray absorption simulations severely restricts the range of molecular systems that can be analyzed by personal computers. Employing the FDMNES code [Phys. Rev. B, 2001, 63, 125120] we show that this problem can be handled if a proper diagonalization scheme is applied. Due to the use of dedicated solvers for sparse matrices, the calcula
The effects of hydrogen in Pd nanoparticles on the features of Pd K-edge
X-ray absorption near edge structure (XANES) are studied. Simulations
indicate a linear dependence of the intensities of XANES maxima upon the
concentration of hydrogen in the bulk of Pd clusters at low
H-concentrations and stepwise increase of the first near edge peak at
high concentrations. The effects of the diffusion of H atoms over the
interstitial sites are considered. The application of the obtained
results to the experimental spectra measured in PdH nanoparticles of
sizes from 1.3nm to 10.5nm indicates low H-concentrations and high
mobility of H-atoms in the bulk of the nanoparticles.
This paper gives a recent overview on numerical aspects of multivariate interpolation and approximation by radial basis functions. It comments on the correct choice of the basis function. It discusses the reduction of complexity by different methods as well as the problem of ill-conditioning. It is aimed to be a user’s guide to an efficient employment of radial basis functions for the reconstruction of multivariate functions. For the entire collection, see [K. Jetter (ed.), M. Buhmann (ed.), W. Haussmann (ed.), R. Schaback (ed.), J. Stoeckler (ed.), Topics in multivariate approximation and interpolation. Studies in Computational Mathematics 12. Amsterdam: Elsevier. (2005; Zbl 1205.41002)].
Starting from the late seventies, the progressively increased availability of synchrotron light sources allowed the execution of experiments requiring a high X-ray flux in a continuous interval. In particular, XAFS techniques are extremely powerful to shed light on the structure and reactivity of surface species in heterogeneous catalysts, in many steps of the catalyst life and also during the catalytic act. A quantitative EXAFS and XANES analysis, possible only in these conditions, was fundamental to reveal the local structure of carbonyl species. IR spectroscopy gave independent information on the fraction of chromium sites active in ethylene polymerization on the Phillips catalysts as a function of polymerization conditions, in well agreement with the values obtained by XANES. The second is the experimental hutch, where the experiment is performed. The third is the control room, where where scientists monitor the experiment and perform a first data analysis.