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

Abstract

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 MnVO. The open-source family of AgileFD algorithms can be implemented into a broad range of high throughput workflows to accelerate materials discovery.

Machine Learning for Perovskites' Reap-Rest-Recovery Cycle

“…55 AI has been applied toward screening materials for light-absorbing applications, where high-throughput XRD measurements enabled the identification of the phase diagram for a family of Nb-V-Mn oxides from their composition and structural characterization data. 57 Overall, ML has the potential to hasten the energy-related materials development timeline by R10 times, if infrastructure and human-capital investments are adequately placed. 58 ML is starting to be implemented in perovskite research, with a modest number of very insightful publications.…”

Analyzing machine learning models to accelerate generation of fundamental materials insights

“…The field of combinatorial materials science comprises an experimental strategy for materials exploration and establishment of composition-structure-property relationships via systematic exploration of high-dimensional materials parameter spaces. 18,22,23 High-throughput experimentation can be used to accelerate such materials exploration [23][24][25][26][27][28] and enables generation of sufficiently large datasets to utilize modern machine learning algorithms. The dataset in the present work was generated using high-throughput synthesis, structural characterization, and photoelectrochemical performance mapping of BiVO 4 -based photoanodes 29,30 as a function of composition in Bi-V-A and Bi-V-A-B compositions spaces where A and B are chosen from a set of five alloying elements.…”

Tracking materials science data lineage to manage millions of materials experiments and analyses

“…The development of a high throughput synthesis 23 and screening [23][24][25][26] pipeline in the Joint Center for Artificial Photosynthesis (JCAP) enhanced the ability to explore new materials spaces and also introduced substantial data management challenges. Although the design of both experiments and data analysis in this effort were dictated by a specific target technology (solar fuel generators), the importance of re-analysis with evolving algorithms or for different target applications (phase mapping is an illustrative example [27][28][29][30][31] ) motivated the establishment of an experiment-centric data organization as opposed to a materialscentric organization. Materials-centric databases such as the ICSD 2 and computational materials databases enable retrieval of properties of a given composition and crystal structure.…”