Abstract Deep learning (DL) models currently employed in materials research exhibit certain limitations delivering meaningful information for interpreting predictions and comprehending the relationships between structure material properties. To address these limitations, we propose an interpretable DL architecture that incorporates attention mechanism to predict properties gain insights into their structure–property relationships. The proposed is evaluated using two well-known datasets (the...

Abstract Oxygen storage and release with oxygen diffusion in the bulk of cerium–zirconium solid solution oxide Ce 2 Zr O x ( = 7–8), which possesses an atomically ordered arrangement cerium zirconium atoms, is key to three-way exhaust catalysis. proceeds via heterogeneous into vacant sites 7 particles, but track erased after bulk. Here we show three-dimensional hard X-ray spectro-ptychography clearly visualize valence map unsupervised learning reveals concealed oxygen-diffusion-driven...

Catalyst degradation at the cathode of a membrane electrode assembly (MEA) remains critical issue for practical polymer electrolyte fuel cell (PEFC) operation, but such wet systems impede detailed visualization events in during its operation. In this work, first time, operando spectroimaging (X-ray absorption near-edge structure–computed tomography) was used to produce clear three-dimensional (3D) images morphology, Pt and Co distributions, Co/Pt atomic ratio, valence state Pt–Co/C catalyst...

Discovering novel high-entropy alloys (HEAs) with desirable properties is challenging due to the vast compositional space and complex phase formation mechanisms. Efficient exploration of this requires a strategic approach that integrates heterogeneous knowledge sources. Here, we propose framework systematically combines extracted from computational material datasets domain distilled scientific literature using large language models (LLMs). A central feature explicit consideration element...

We have developed a descriptor named Orbital Field Matrix (OFM) for representing material structures in datasets of multi-element materials. The is based on the information regarding atomic valence shell electrons and their coordination. In this work, we develop an extension OFM called OFM1. shown that these descriptors are highly applicable predicting physical properties materials providing insights space by mapping into low embedded dimensional space. Our experiments with transition...

Existing data-driven approaches for exploring high-entropy alloys (HEAs) face three challenges: numerous element-combination candidates, designing appropriate descriptors, and limited biased existing data. To overcome these issues, here we show the development of an evidence-based material recommender system (ERS) that adopts Dempster-Shafer theory, a general framework reasoning with uncertainty. Herein, without using model, collect combine pieces evidence from data about HEA phase existence...

A heterogeneous phase/structure distribution in the bulk of spinel lithium nickel manganese oxides (LNMOs) is key to maximizing performance and stability cathode materials lithium-ion batteries. Herein, we report use two-dimensional ptychographic X-ray absorption fine structure (XAFS) visualize density valence maps as-prepared LNMO particles unsupervised learning classify three-phase group terms different elemental compositions chemical states. The described approach may increase supply...

Abstract In practical applications of polymer electrolyte fuel cells (PEFC), the wide variations local structures and environment electrocatalysts bring about complex reaction behaviours inside a membrane electrode assembly (MEA). The coordination structure redox response Pt/C cathode catalyst in an MEA before after typical accelerated degradation test (ADT) were three‐dimensionally visualized by operando Pt L III ‐edge computed‐tomography X‐ray absorption fine (CT‐XAFS) imaging under PEFC...

Abstract In this article, we propose a query-and-learn active learning approach combined with first-principles calculations to rapidly search for potentially stable crystal structure via elemental substitution, clarify their stabilization mechanism, and integrate SmFe $$_{12}$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:msub> <mml:mrow/> <mml:mn>12</mml:mn> </mml:msub> </mml:math> -based compounds ThMn structure, which exhibits prominent magnetic properties. The proposed...

Abstract The Curie temperature ( T C ) of RT binary compounds consisting 3 d transition-metal and 4 f rare-earth elements R is analyzed systematically by a developed machine learning technique called kernel regression-based model evaluation. Twenty-one descriptive variables were designed assuming completely obtained information the . Multiple regression analyses with different types: cosine, linear, Gaussian, polynomial, Laplacian kernels implemented examined. All possible variable...

New Nd–Fe–B crystal structures can be formed via the elemental substitution of LA – T X host structures, including lanthanides ( ), transition metals ) and light elements, = B, C, N O. The 5967 samples ternary materials that are collected then used as structures. For each structure, a substituted structure is created by substituting all lanthanide sites with Nd, metal Fe light-element B. High-throughput first-principles calculations applied to evaluate phase stability newly 20 them found...

Abstract We propose a data-driven method to extract dissimilarity between materials, with respect given target physical property. The technique is based on an ensemble Kernel ridge regression as the predicting model; multiple random subset sampling of materials done generate prediction models and corresponding contributions reference training in detail. distribution predicted values for each material can be approximated by Gaussian mixture models. contributed model that accurately predicts...

A method has been developed to measure the similarity between materials, focusing on specific physical properties. The information obtained can be utilized understand underlying mechanisms and support prediction of properties materials. consists three steps: variable evaluation based nonlinear regression, regression-based clustering, measurement with a committee machine constructed from clustering results. Three data sets well characterized crystalline materials represented by critical...

In this study, we investigate the structure–stability relationship of hypothetical Nd–Fe–B crystal structures using descriptor-relevance analysis and t-SNE dimensionality reduction method. 149 are generated from 5967 LA–T–X host in Open Quantum Materials Database by elemental substitution method, with LA denoting lanthanides, T transition metals, X light elements such as B, C, N, O. By borrowing skeletal structure each materials, a is created substituting all lanthanide sites Nd, metal Fe,...

We investigate the correlation between geometrical information, stability, and magnetization of SmFe12-based structures using machine learning-aided genetic algorithm structure generation first-principle calculation. In parallel with inherited USPEX program, a pool is created for every population sub-symmetry perturbation method. A framework embedded orbital field matrix representation as fingerprint Gaussian process predictor has been applied to ranking most potential stability structures....

Measuring the similarity between materials is essential for estimating their properties and revealing associated physical mechanisms. However, current methods measuring rely on theoretically derived descriptors parameters fitted from experimental or computational data, which are often insufficient biased. Furthermore, outliers data generated by multiple mechanisms usually included in dataset, making data-driven approach challenging mathematically complicated. To overcome such issues, we...

In this paper, the influence of rare earth (RE) on microstructure and mechanical properties austenitic high manganese steel (HMnS) Mn15Cr2V were investigated. The results showed that microstructure, hardness impact strength RE modification sample is finer better than non-modified sample. Under effect load, depth work-hardening layer modified was higher steel, thereby, value microhardness in surface 420 HV while it only 395 up to 132 J/cm2 compared 115 J/cm2. Moreover, after austenite...

Abstract Deep learning (DL) models currently used for materials research have limitations in providing meaningful information interpreting predictions and understanding the relationships between structure material properties. To address this, we propose a DL architecture that incorporates attention mechanism to predict properties gain insights into their structure-property relationships. The proposed is evaluated using four datasets: QM9 molecule dataset three in-house-developed...

We developed a method for measuring the similarity between materials, focusing on specific physical properties. The obtained information can be utilized to understand underlying mechanisms and support prediction of properties materials. consists three steps: variable evaluation based non-linear regression, regression-based clustering, measurement with committee machine constructed from clustering results. Three datasets well-characterized crystalline materials represented by critical atomic...

We propose a data-driven method to extract dissimilarity between materials, with respect given target physical property. The technique is based on an ensemble Kernel ridge regression as the predicting model; multiple random subset sampling of materials done generate prediction models and corresponding contributions reference training in detail. distribution predicted values for each material can be approximated by Gaussian mixture model. contributed model that accurately predicts property...

New Nd-Fe-B crystal structures can be formed via the elemental substitution of LATX host structures, including lanthanides LA, transition metals T, and light elements X as B, C, N, O. The 5967 samples ternary materials that are collected then used structures. For each structure, a substituted structure is created by substituting all lanthanide sites with Nd, metal Fe, element B. High throughput first-principles calculations applied to evaluate phase stability newly 20 them found potentially...