Phase-Mapper: An AI Platform to Accelerate High Throughput Materials Discovery

Xue, Yexiang; Bai, Junwen; Bras, Ronan Le; Rappazzo, Brendan; Bernstein, Richard J.; Björck, Johan; Longpre, Liane; Suram, Santosh K.; Dover, R. B. van; Gregoire, John M.; Gomes, Carla P.

doi:10.48550/arxiv.1610.00689

Cited by 2 publications

(4 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The unified COS, PCC, and JSD measure dissimilarity between vectors from cosine metric, linear correlation, and information entropy perspectives, respectively, and are not involved with physical meanings. In our future work, clustering techniques through injecting physical information into machine learning methods, like Gibb's phase rule imposed in AgileFD, 21 are more desirable due to their better interpretability.…”

Section: ∑ ∑ ∑mentioning

confidence: 99%

“…And successfully applied unsupervised learning methods are primarily based on non-negative matrix factorization (NMF) and hierarchy clustering analysis (HCA). CombiFD 20 and AgileFD 21 were both developed on the basis of NMF and performed well in grouping XRD patterns, 22 and the latter carried sufficient physical significance by incorporating Gibbs's phase rule. As for HCA, it produces dramatic outcomes with different dissimilarity measures.…”

Section: ■ Introductionmentioning

confidence: 99%

See 1 more Smart Citation

An Improved High-Throughput Data Processing Based on Combinatorial Materials Chip Approach for Rapid Construction of Fe–Cr–Ni Composition-Phase Map

et al. 2019

View full text Add to dashboard Cite

The combinatorial materials chip approach is vastly superior to the conventional one that characterizes one sample at a time in the efficiency of composition-phase map construction. However, the resolution of its high-throughput characterization and the correct rate of automated composition-phase mapping are often affected by inherent experimental limitations and imperfect automated analyses, respectively. Therefore, effective data preprocessing and refined automated analysis methods are required to automatically process huge amounts of experiment data to score a higher correct rate. In this work, the pixel-by-pixel structural and compositional characterization of the Fe–Cr–Ni combinatorial materials chip annealed at 750 °C was performed by microbeam X-ray at a synchrotron light source and by electron probe microanalysis, respectively. The severe baseline drift and system noise in the X-ray diffraction patterns were successfully eliminated by the three-step automated preprocessing (baseline drift removal, noise elimination, and baseline correction) proposed, which was beneficial to the subsequent quantitative analysis of the patterns. Through the injection of human experience, hierarchy clustering analyses, based on three dissimilarity measures (the cosine, Pearson correlation coefficient, and Jenson–Shannon divergence), were further accelerated by the simplified vectorization of the preprocessed X-ray diffraction patterns. As a result, a correct rate of 91.15% was reached for the whole map built automatically in comparison with the one constructed manually, which confirmed that the present data processing could assist humans to improve and expedite the processing of X-ray diffraction patterns and was feasible for composition-phase mapping. The constructed maps were generally consistent with the corresponding isothermal section of the Fe–Cr–Ni ternary alloy system in the ASM Alloy Phase Diagram Database except the inexistence of the σ phase under insufficient annealing.

show abstract

Section: ∑ ∑ ∑mentioning

confidence: 99%

Section: ■ Introductionmentioning

confidence: 99%

An Improved High-Throughput Data Processing Based on Combinatorial Materials Chip Approach for Rapid Construction of Fe–Cr–Ni Composition-Phase Map

et al. 2019

View full text Add to dashboard Cite

show abstract

“…For a comprehensive description of the mathematics of AgileFD, we refer the reader to ref. 20 Perhaps the most substantial advancement of AgileFD for the phase mapping problem is the efficient modeling of Peak Shifting, which we enable by applying a log-transformation to the XRD patterns. Each XRD pattern can be represented as the scattering intensity (I) as a function of q, the magnitude of the X-ray scattering vector.…”

Section: ■ Introductionmentioning

confidence: 99%

“…In ref. 20 we derive AgileFD's custom update rules for both H and W, which are derived from a loss function based on the generalized Kullback−Leibler divergence, an advancement that is critical for the phase mapping problem and is generally applicable for other source separation problems. With these lightweight update rules, this incorporation of Peak Shifting is vastly more efficient than the previously proposed time warping techniques.…”

Section: ■ Introductionmentioning

confidence: 99%

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

et al. 2016

Self Cite

View full text Add to dashboard Cite

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.

show abstract

Phase-Mapper: An AI Platform to Accelerate High Throughput Materials Discovery

Cited by 2 publications

References 8 publications

An Improved High-Throughput Data Processing Based on Combinatorial Materials Chip Approach for Rapid Construction of Fe–Cr–Ni Composition-Phase Map

An Improved High-Throughput Data Processing Based on Combinatorial Materials Chip Approach for Rapid Construction of Fe–Cr–Ni Composition-Phase Map

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

Contact Info

Product

Resources

About