The Materials Provenance Store

We present a database resulting from high throughput experimentation, primarily on metal oxide solid state materials. The central relational database, the Materials Provenance Store (MPS), manages the metadata and experimental provenance from acquisition of raw materials, through synthesis, to a broad range of materials characterization techniques. Given the primary research goal of materials discovery of solar fuels materials, many of the characterization experiments involve electrochemistry, along with optical, structural, and compositional characterizations. The MPS is populated with all information required for executing common data queries, which typically do not involve direct query of raw data. The result is a database file that can be distributed to users so that they can independently execute queries and subsequently download the data of interest. We propose this strategy as an approach to manage the highly heterogeneous and distributed data that arises from materials science experiments, as demonstrated by the management of over 30 million experiments run on over 12 million samples in the present MPS release.

Agents for sequential learning using multiple-fidelity data

Sequential learning for materials discovery is a paradigm where a computational agent solicits new data to simultaneously update a model in service of exploration (finding the largest number of materials that meet some criteria) or exploitation (finding materials with an ideal figure of merit). In real-world discovery campaigns, new data acquisition may be costly and an optimal strategy may involve using and acquiring data with different levels of fidelity, such as first-principles calculation to supplement an experiment. In this work, we introduce agents which can operate on multiple data fidelities, and benchmark their performance on an emulated discovery campaign to find materials with desired band gap values. The fidelities of data come from the results of DFT calculations as low fidelity and experimental results as high fidelity. We demonstrate performance gains of agents which incorporate multi-fidelity data in two contexts: either using a large body of low fidelity data as a prior knowledge base or acquiring low fidelity data in-tandem with experimental data. This advance provides a tool that enables materials scientists to test various acquisition and model hyperparameters to maximize the discovery rate of their own multi-fidelity sequential learning campaigns for materials discovery. This may also serve as a reference point for those who are interested in practical strategies that can be used when multiple data sources are available for active or sequential learning campaigns.

Machine learning-accelerated design and synthesis of polyelemental heterostructures

In materials discovery efforts, synthetic capabilities far outpace the ability to extract meaningful data from them. To bridge this gap, machine learning methods are necessary to reduce the search space for identifying desired materials. Here, we present a machine learning–driven, closed-loop experimental process to guide the synthesis of polyelemental nanomaterials with targeted structural properties. By leveraging data from an eight-dimensional chemical space (Au-Ag-Cu-Co-Ni-Pd-Sn-Pt) as inputs, a Bayesian optimization algorithm is used to suggest previously unidentified nanoparticle compositions that target specific interfacial motifs for synthesis, results of which are iteratively shared back with the algorithm. This feedback loop resulted in successful syntheses of 18 heterojunction nanomaterials that are too complex to discover by chemical intuition alone, including extremely chemically complex biphasic nanoparticles reported to date. Platforms like the one developed here are poised to transform materials discovery across a wide swath of applications and industries.

Benchmarking the acceleration of materials discovery by sequential learning

Sequential learning (SL) strategies, i.e. iteratively updating a machine learning model to guide experiments, have been proposed to significantly accelerate materials discovery and research. Applications on computational datasets and a handful of optimization experiments have demonstrated the promise of SL, motivating a quantitative evaluation of its ability to accelerate materials discovery, specifically in the case of physical experiments. The benchmarking effort in the present work quantifies the performance of SL algorithms with respect to a breadth of research goals: discovery of any "good" material, discovery of all "good" materials, and discovery of a model that accurately predicts the performance of new materials. To benchmark the effectiveness of different machine learning models against these goals, we use datasets in which the performance of all materials in the search space is known from high-throughput synthesis and electrochemistry experiments. Each dataset contains all pseudo-quaternary metal oxide combinations from a set of six elements (chemical space), the performance metric chosen is the electrocatalytic activity (overpotential) for the oxygen evolution reaction (OER). A diverse set of SL schemes is tested on four chemical spaces, each containing 2121 catalysts. The presented work suggests that research can be accelerated by up to a factor of 20 compared to random acquisition in specific scenarios. The results also show that certain choices of SL models are ill-suited for a given research goal resulting in substantial deceleration compared to random acquisition methods. The results provide quantitative guidance on how to tune an SL strategy for a given research goal and demonstrate the need for a new generation of materials-aware SL algorithms to further accelerate materials discovery.

Alkaline-stable nickel manganese oxides with ideal band gap for solar fuel photoanodes

Combinatorial (photo)electrochemical studies of the (Ni-Mn)Ox system reveal a range of promising materials for oxygen evolution photoanodes. X-ray diffraction, quantum efficiency, and optical spectroscopy mapping reveal stable photoactivity of NiMnO3 in alkaline conditions with photocurrent onset commensurate with its 1.9 eV direct band gap. The photoactivity increases upon mixture with 10-60% Ni6MnO8 providing an example of enhanced charge separation via heterojunction formation in mixed-phase thin film photoelectrodes. Density functional theory-based hybrid functional calculations of the band edge energies in this oxide reveal that a somewhat smaller than typical fraction of exact exchange is required to explain the favorable valence band alignment for water oxidation.

Phase Mapper: Accelerating Materials Discovery with AI

From the stone age, to the bronze, iron age, and modern silicon age, the discovery and characterization of new materials has always been instrumental to humanity's progress and development. With the current pressing need to address sustainability challenges and find alternatives to fossil fuels, we look for solutions in the development of new materials that will allow for renewable energy. To discover materials with the required properties, materials scientists can perform high-throughput materials discovery, which includes rapid synthesis and characterization via X-ray diffraction (XRD) of thousands of materials. A central problem in materials discovery, the phase map identification problem, involves the determination of the crystal structure of materials from materials composition and structural characterization data. This analysis is traditionally performed mainly by hand, which can take days for a single material system. In this work we present Phase-Mapper, a solution platform that tightly integrates XRD experimentation, AI problem solving, and human intelligence for interpreting XRD patterns and inferring the crystal structures of the underlying materials. Phase-Mapper is compatible with any spectral demixing algorithm, including our novel solver, AgileFD, which is based on convolutive non-negative matrix factorization. AgileFD allows materials scientists to rapidly interpret XRD patterns, and incorporates constraints to capture prior knowledge about the physics of the materials as well as human feedback. With our system, materials scientists have been able to interpret previously unsolvable systems of XRD data at the Department of Energy’s Joint Center for Artificial Photosynthesis, including the Nb-Mn-V oxide system, which led to the discovery of new solar light absorbers and is provided as an illustrative example of AI-enabled high throughput materials discovery

Combinatorial Discovery of Lanthanum-Tantalum Oxynitride Solar Light Absorbers with Dilute Nitrogen for Solar Fuel Applications

Oxynitrides with the photoelectrochemical stability of oxides and desirable band energetics of nitrides comprise a promising class of materials for solar photochemistry. Challenges in synthesizing a wide variety of oxynitride materials has limited exploration of this class of functional materials, which we address using a reactive cosputtering combined with rapid thermal processing method to synthesize multi-cation-multi-anion libraries. We demonstrate the synthesis of a La_xTa_1-xO_yN_z thin film composition spread library and its characterization by both traditional thin film materials characterization and custom combinatorial optical spectroscopy and X-ray absorption near edge spectroscopy (XANES) techniques, ultimately establishing structure-chemistry-property relationships. We observe that over a substantial La-Ta composition range the thin films crystallize in the same perovskite LaTaON₂ structure with significant variation of anion chemistry. The relative invariance in optical band gap demonstrates a remarkable decoupling of composition and band energetics so that the composition can be optimized while retaining the desirable 2 eV band gap energy. We also demonstrate the intercalation of diatomic nitrogen into the La₃TaO₇ structure, which gives rise to a direct-allowed optical transition at 2.2 eV, less than half the value of the oxide's band gap. These findings motivate further exploration of the visible light response of this material that is predicted to be stable over a wide range of electrochemical potential.

Automated Phase Mapping with AgileFD and its Application to Light Absorber Discovery in the V-Mn-Nb Oxide System

Rapid construction of phase diagrams is a central tenet of combinatorial materials science with accelerated materials discovery efforts often hampered by challenges in interpreting combinatorial X-ray diffraction data sets, which we address by developing AgileFD, an artificial intelligence algorithm that enables rapid phase mapping from a combinatorial library of X-ray diffraction patterns. AgileFD models alloying-based peak shifting through a novel expansion of convolutional nonnegative matrix factorization, which not only improves the identification of constituent phases but also maps their concentration and lattice parameter as a function of composition. By incorporating Gibbs' phase rule into the algorithm, physically meaningful phase maps are obtained with unsupervised operation, and more refined solutions are attained by injecting expert knowledge of the system. The algorithm is demonstrated through investigation of the V-Mn-Nb oxide system where decomposition of eight oxide phases, including two with substantial alloying, provides the first phase map for this pseudoternary system. This phase map enables interpretation of high-throughput band gap data, leading to the discovery of new solar light absorbers and the alloying-based tuning of the direct-allowed band gap energy of MnV₂O₆. The open-source family of AgileFD algorithms can be implemented into a broad range of high throughput workflows to accelerate materials discovery.

High Throughput Light Absorber Discovery, Part 1: An Algorithm for Automated Tauc Analysis

High-throughput experimentation provides efficient mapping of composition-property relationships, and its implementation for the discovery of optical materials enables advancements in solar energy and other technologies. In a high throughput pipeline, automated data processing algorithms are often required to match experimental throughput, and we present an automated Tauc analysis algorithm for estimating band gap energies from optical spectroscopy data. The algorithm mimics the judgment of an expert scientist, which is demonstrated through its application to a variety of high throughput spectroscopy data, including the identification of indirect or direct band gaps in Fe₂O₃, Cu₂V₂O₇, and BiVO₄. The applicability of the algorithm to estimate a range of band gap energies for various materials is demonstrated by a comparison of direct-allowed band gaps estimated by expert scientists and by automated algorithm for 60 optical spectra.

High Throughput Light Absorber Discovery, Part 2: Establishing Structure-Band Gap Energy Relationships

Combinatorial materials science strategies have accelerated materials development in a variety of fields, and we extend these strategies to enable structure-property mapping for light absorber materials, particularly in high order composition spaces. High throughput optical spectroscopy and synchrotron X-ray diffraction are combined to identify the optical properties of Bi-V-Fe oxides, leading to the identification of Bi₄V_1.5Fe_0.5O_10.5 as a light absorber with direct band gap near 2.7 eV. The strategic combination of experimental and data analysis techniques includes automated Tauc analysis to estimate band gap energies from the high throughput spectroscopy data, providing an automated platform for identifying new optical materials.

Stability and self-passivation of copper vanadate photoanodes under chemical, electrochemical, and photoelectrochemical operation

Self-passivation under operational conditions is observed for several copper vanadate photoanodes, demonstrating their viability for durable solar fuels devices.

Generating information-rich high-throughput experimental materials genomes using functional clustering via multitree genetic programming and information theory

High-throughput experimental methodologies are capable of synthesizing, screening and characterizing vast arrays of combinatorial material libraries at a very rapid rate. These methodologies strategically employ tiered screening wherein the number of compositions screened decreases as the complexity, and very often the scientific information obtained from a screening experiment, increases. The algorithm used for down-selection of samples from higher throughput screening experiment to a lower throughput screening experiment is vital in achieving information-rich experimental materials genomes. The fundamental science of material discovery lies in the establishment of composition-structure-property relationships, motivating the development of advanced down-selection algorithms which consider the information value of the selected compositions, as opposed to simply selecting the best performing compositions from a high throughput experiment. Identification of property fields (composition regions with distinct composition-property relationships) in high throughput data enables down-selection algorithms to employ advanced selection strategies, such as the selection of representative compositions from each field or selection of compositions that span the composition space of the highest performing field. Such strategies would greatly enhance the generation of data-driven discoveries. We introduce an informatics-based clustering of composition-property functional relationships using a combination of information theory and multitree genetic programming concepts for identification of property fields in a composition library. We demonstrate our approach using a complex synthetic composition-property map for a 5 at. % step ternary library consisting of four distinct property fields and finally explore the application of this methodology for capturing relationships between composition and catalytic activity for the oxygen evolution reaction for 5429 catalyst compositions in a (Ni-Fe-Co-Ce)Ox library.

Colorimetric screening for high-throughput discovery of light absorbers

High-throughput screening is a powerful approach for identifying new functional materials in unexplored material spaces. With library synthesis capable of producing 10(5) to 10(6) samples per day, methods for material screening at rates greater than 1 Hz must be developed. For the discovery of new solar light absorbers, this throughput cannot be attained using standard instrumentation. Screening certain properties, such as the bandgap, are of interest only for phase pure materials, which comprise a small fraction of the samples in a typical solid-state material library. We demonstrate the utility of colorimetric screening based on processing photoscanned images of combinatorial libraries to quickly identify distinct phase regions, isolate samples with desired bandgap, and qualitatively identify samples that are suitable for complementary measurements. Using multiple quaternary oxide libraries containing thousands of materials, we compare colorimetric screening and UV-vis spectroscopy results, demonstrating successful identification of compounds with bandgap suitable for solar applications.

Combinatorial thin film composition mapping using three dimensional deposition profiles

Many next-generation technologies are limited by material performance, leading to increased interest in the discovery of advanced materials using combinatorial synthesis, characterization, and screening. Several combinatorial synthesis techniques, such as solution based methods, advanced manufacturing, and physical vapor deposition, are currently being employed for various applications. In particular, combinatorial magnetron sputtering is a versatile technique that provides synthesis of high-quality thin film composition libraries. Spatially addressing the composition of these thin films generally requires elemental quantification measurements using techniques such as energy-dispersive X-ray spectroscopy or X-ray fluorescence spectroscopy. Since these measurements are performed ex-situ and post-deposition, they are unable to provide real-time design of experiments, a capability that is required for rapid synthesis of a specific composition library. By using three quartz crystal monitors attached to a stage with translational and rotational degrees of freedom, we measure three-dimensional deposition profiles of deposition sources whose tilt with respect to the substrate is robotically controlled. We exhibit the utility of deposition profiles and tilt control to optimize the deposition geometry for specific combinatorial synthesis experiments.

Statistical analysis and interpolation of compositional data in materials science

Compositional data are ubiquitous in chemistry and materials science: analysis of elements in multicomponent systems, combinatorial problems, etc., lead to data that are non-negative and sum to a constant (for example, atomic concentrations). The constant sum constraint restricts the sampling space to a simplex instead of the usual Euclidean space. Since statistical measures such as mean and standard deviation are defined for the Euclidean space, traditional correlation studies, multivariate analysis, and hypothesis testing may lead to erroneous dependencies and incorrect inferences when applied to compositional data. Furthermore, composition measurements that are used for data analytics may not include all of the elements contained in the material; that is, the measurements may be subcompositions of a higher-dimensional parent composition. Physically meaningful statistical analysis must yield results that are invariant under the number of composition elements, requiring the application of specialized statistical tools. We present specifics and subtleties of compositional data processing through discussion of illustrative examples. We introduce basic concepts, terminology, and methods required for the analysis of compositional data and utilize them for the spatial interpolation of composition in a sputtered thin film. The results demonstrate the importance of this mathematical framework for compositional data analysis (CDA) in the fields of materials science and chemistry.

High-throughput on-the-fly scanning ultraviolet-visible dual-sphere spectrometer

We have developed an on-the-fly scanning spectrometer operating in the UV-visible and near-infrared that can simultaneously perform transmission and total reflectance measurements at the rate better than 1 sample per second. High throughput optical characterization is important for screening functional materials for a variety of new applications. We demonstrate the utility of the instrument for screening new light absorber materials by measuring the spectral absorbance, which is subsequently used for deriving band gap information through Tauc plot analysis.

High-throughput synchrotron X-ray diffraction for combinatorial phase mapping

Discovery of new materials drives the deployment of new technologies. Complex technological requirements demand precisely tailored material functionalities, and materials scientists are driven to search for these new materials in compositionally complex and often non-equilibrium spaces containing three, four or more elements. The phase behavior of these high-order composition spaces is mostly unknown and unexplored. High-throughput methods can offer strategies for efficiently searching complex and multi-dimensional material genomes for these much needed new materials and can also suggest a processing pathway for synthesizing them. However, high-throughput structural characterization is still relatively under-developed for rapid material discovery. Here, a synchrotron X-ray diffraction and fluorescence experiment for rapid measurement of both X-ray powder patterns and compositions for an array of samples in a material library is presented. The experiment is capable of measuring more than 5000 samples per day, as demonstrated by the acquisition of high-quality powder patterns in a bismuth-vanadium-iron oxide composition library. A detailed discussion of the scattering geometry and its ability to be tailored for different material systems is provided, with specific attention given to the characterization of fiber textured thin films. The described prototype facility is capable of meeting the structural characterization needs for the first generation of high-throughput material genomic searches.

High-throughput bubble screening method for combinatorial discovery of electrocatalysts for water splitting

Combinatorial synthesis and screening for discovery of electrocatalysts has received increasing attention, particularly for energy-related technologies. High-throughput discovery strategies typically employ a fast, reliable initial screening technique that is able to identify active catalyst composition regions. Traditional electrochemical characterization via current-voltage measurements is inherently throughput-limited, as such measurements are most readily performed by serial screening. Parallel screening methods can yield much higher throughput and generally require the use of an indirect measurement of catalytic activity. In a water-splitting reaction, the change of local pH or the presence of oxygen and hydrogen in the solution can be utilized for parallel screening of active electrocatalysts. Previously reported techniques for measuring these signals typically function in a narrow pH range and are not suitable for both strong acidic and basic environments. A simple approach to screen the electrocatalytic activities by imaging the oxygen and hydrogen bubbles produced by the oxygen evolution reaction (OER) and hydrogen evolution reaction (HER) is reported here. A custom built electrochemical cell was employed to record the bubble evolution during the screening, where the testing materials were subject to desired electrochemical potentials. The transient of the bubble intensity obtained from the screening was quantitatively analyzed to yield a bubble figure of merit (FOM) that represents the reaction rate. Active catalysts in a pseudoternary material library, (Ni-Fe-Co)Ox, which contains 231 unique compositions, were identified in less than one minute using the bubble screening method. An independent, serial screening method on the same material library exhibited excellent agreement with the parallel bubble screening. This general approach is highly parallel and is independent of solution pH.

Data mining for isotope discrimination in atom probe tomography

Ions with similar time-of-flights (TOF) can be discriminated by mapping their kinetic energy. While current generation position-sensitive detectors have been considered insufficient for capturing the isotope kinetic energy, we demonstrate in this paper that statistical learning methodologies can be used to capture the kinetic energy from all of the parameters currently measured by mathematically transforming the signal. This approach works because the kinetic energy is sufficiently described by the descriptors on the potential, the material, and the evaporation process within atom probe tomography (APT). We discriminate the isotopes for Mg and Al by capturing the kinetic energy, and then decompose the TOF spectrum into its isotope components and identify the isotope for each individual atom measured. This work demonstrates the value of advanced data mining methods to help enhance the information resolution of the atom probe.

Refining spatial distribution maps for atom probe tomography via data dimensionality reduction methods

A mathematical framework based on singular value decomposition is used to analyze the covariance among interatomic frequency distributions in spatial distribution maps (SDMs). Using this approach, singular vectors that capture the covariance within the SDM data are obtained. The structurally relevant singular vectors (SRSVs) are identified. Using the SRSVs, we extract information from z-SDMs that not only captures the offset between the atomic planes but also captures the covariance in the atomic structure among the neighborhood atomic planes. These refined z-SDMs classify the Δ(Δz) slices in the SDMs into structurally relevant information, noise, and aberrations. The SRSVs are used to construct refined xy-SDMs that provide enhanced structural information for three-dimensional atom probe tomography.

Informatics for Quantitative Analysis of Atom Probe Tomography Images

An informatics based approach to extract further refinements on the crystallographic information embedded in the Spatial Distribution Maps (SDMs) has been developed. The data mining based methods to generate and interpret spectra that de-convolute the SDMs are discussed. This work has resulted in a method to generate SDMs that can map three-dimensional crystallographic information as opposed to existing methods that map structural information on only one atomic plane at a time. The broader implications of this work on enhancing the interpretation and resolution of structural information in atom probe tomography studies is also discussed.