arXiv subjects

Ryoji Asahi

Publications and source records attributed to Ryoji Asahi.

9 recordsLinked to original sources

arXiv2026

Unveiling and quantifying the topology-dependent pre-melting of nanoparticles

The melting of metallic nanoparticles is governed by surface premelting, a phenomenon traditionally modeled as the isotropic growth of a uniform liquid shell. Challenging this classical view, we report facet-dependent premelting in hexagonal close-packed Co nanoparticles, arising from the structural heterogeneity of their surface. In molecular dynamics simulations (587 to 13047 atoms), the onset of surface mobility is observed as low as 20% of the bulk melting point, driven by the early disordering of stepped $\{01\bar{1}1\}$ facets. These facets consistently melt nearly 150 K below flat $\{0001\}$ facets, regardless of particle size. We show that both surface and facet melting temperatures scale with nanoparticle size through the Gibbs-Thomson effect, and determine a size-dependent critical liquid layer thickness that triggers complete melting of the nanoparticle, which saturates near three atomic layers. Our results confirm recent experimental observations of surface premelting and extend the framework to anisotropic particles with facet-orientation-dependent behavior.

cond-mat.mtrl-sci↗

arXiv2025

Hierarchical Stacking Optimization Using Dirichlet's Process (SoDip): Towards Accelerated Design for Graft Polymerization

Radiation-induced grafting (RIG) enables precise functionalization of polymer films for ion-exchange membranes, CO2-separation membranes, and battery electrolytes by generating radicals on robust substrates to graft desired monomers. However, reproducibility remains limited due to unreported variability in base-film morphology (crystallinity, grain orientation, free volume), which governs monomer diffusion, radical distribution, and the Trommsdorff effect, leading to spatial graft gradients and performance inconsistencies. We present a hierarchical stacking optimization framework with a Dirichlet's Process (SoDip), a hierarchical data-driven framework integrating: (1) a decoder-only Transformer (DeepSeek-R1) to encode textual process descriptors (irradiation source, grafting type, substrate manufacturer); (2) TabNet and XGBoost for modelling multimodal feature interactions; (3) Gaussian Process Regression (GPR) with Dirichlet Process Mixture Models (DPMM) for uncertainty quantification and heteroscedasticity; and (4) Bayesian Optimization for efficient exploration of high-dimensional synthesis space. A diverse dataset was curated using ChemDataExtractor 2.0 and WebPlotDigitizer, incorporating numerical and textual variables across hundreds of RIG studies. In cross-validation, SoDip achieved ~33% improvement over GPR while providing calibrated confidence intervals that identify low-reproducibility regimes. Its stacked architecture integrates sparse textual and numerical inputs of varying quality, outperforming prior models and establishing a foundation for reproducible, morphology-aware design in graft polymerization research.

cs.LG↗

arXiv2025

Extracting ORR Catalyst Information for Fuel Cell from Scientific Literature

The oxygen reduction reaction (ORR) catalyst plays a critical role in enhancing fuel cell efficiency, making it a key focus in material science research. However, extracting structured information about ORR catalysts from vast scientific literature remains a significant challenge due to the complexity and diversity of textual data. In this study, we propose a named entity recognition (NER) and relation extraction (RE) approach using DyGIE++ with multiple pre-trained BERT variants, including MatSciBERT and PubMedBERT, to extract ORR catalyst-related information from the scientific literature, which is compiled into a fuel cell corpus for materials informatics (FC-CoMIcs). A comprehensive dataset was constructed manually by identifying 12 critical entities and two relationship types between pairs of the entities. Our methodology involves data annotation, integration, and fine-tuning of transformer-based models to enhance information extraction accuracy. We assess the impact of different BERT variants on extraction performance and investigate the effects of annotation consistency. Experimental evaluations demonstrate that the fine-tuned PubMedBERT model achieves the highest NER F1-score of 82.19% and the MatSciBERT model attains the best RE F1-score of 66.10%. Furthermore, the comparison with human annotators highlights the reliability of fine-tuned models for ORR catalyst extraction, demonstrating their potential for scalable and automated literature analysis. The results indicate that domain-specific BERT models outperform general scientific models like BlueBERT for ORR catalyst extraction.

cs.CL↗

arXiv2024

Dielectric tensor of perovskite oxides at finite temperature using equivariant graph neural network potentials

Atomistic simulations of properties of materials at finite temperatures are computationally demanding and require models that are more efficient than the ab initio approaches. Machine learning (ML) and artificial intelligence (AI) address this issue by enabling accurate models with close to ab initio accuracy. Here, we demonstrate the utility of ML models in capturing properties of realistic materials by performing finite temperature molecular dynamics simulations of perovskite oxides using a force field based on equivariant graph neural networks. The models demonstrate efficient learning from a small training dataset of energies, forces, stresses, and tensors of Born effective charges. We qualitatively capture the temperature dependence of the dielectric tensor and structural phase transitions in calcium titanate.

cond-mat.mtrl-sci↗

arXiv2024

Representing Born effective charges with equivariant graph convolutional neural networks

Graph convolutional neural networks have been instrumental in machine learning of material properties. When representing tensorial properties, weights and descriptors of a physics-informed network must obey certain transformation rules to ensure the independence of the property on the choice of the reference frame. Here we explicitly encode such properties using an equivariant graph convolutional neural network. The network respects rotational symmetries of the crystal throughout by using equivariant weights and descriptors and provides a tensorial output of the target value. Applications to tensors of atomic Born effective charges in diverse materials including perovskite oxides, Li3PO4, and ZrO2, are demonstrated, and good performance and generalization ability is obtained.

cond-mat.mtrl-sci↗

arXiv2024

Polyvalent Machine-Learned Potential for Cobalt: from Bulk to Nanoparticles

We present the development and applications of a quadratic Spectral Neighbor Analysis Potential (q-SNAP) for ferromagnetic cobalt. Trained on Density Functional Theory calculations using the Perdew-Burke-Ernzerhof (DFT-PBE) functional, this machine-learned potential enables simulations of large systems over extended time scales across a wide range of temperatures and pressures at near DFT accuracy. It is validated by closely reproducing the phonon dispersions of hexagonal close-packed (hcp) and face-centered cubic (fcc) Co, surface energies, and the relative stability of nanoparticles of various shapes. An important feature of this novel potential is its numerical stability in long molecular dynamics simulations. This robustness is exploited to compute the heat capacity of nanoparticles containing up to 9201 atoms, showing convergence to less than 2 J.K-1.mol-1 after 100 ns. Computations of the melting temperature of nanoparticles as a function of their size revealed a convergence to the bulk limit in excellent agreement with the experimental value. Thus, the new, highly accurate machine-learned potential for Co opens exciting opportunities for further applications such as the dynamics of nanoparticles in catalytic reactions.

cond-mat.mtrl-sci↗

arXiv2022

Doping effect and Li-ion conduction mechanism of ALi6XO6 (A = K or Rb, and X = pentavalent): A first-principles study

Recent theoretical and experimental evaluations demonstrated that KLi6TaO6 is a good Li-ion conductor. In this study, the energetics and detailed mechanism of Li-ion migration, relevant to the point defects of KLi6TaO6, were analyzed by first-principles calculations. Defect formation energy analysis suggested that it has limited chemical potential conditions for attaining Li-excess conditions through doping (substituting tetravalent elements for Ta). The formation of other native defects, such as Li vacancies, hinders the stabilization of the dopant and compensates for the interstitial Li. When the doping is successful, the interactions between the coexisting dopant and interstitial Li can increase the migration energy barrier of the interstitial Li. This phenomenon limits the factors responsible for achieving high Li-ion conductivity in this material. Based on the results of the investigations on KLi6TaO6, isostructural materials of the form ALi6XO6, with various combinations of constituent elements A and X, were each screened on the basis of high stability and low Li-ion migration energy. Twelve structures of the form (A = K or Rb)Li6XO6 were suggested, of which X was pentavalent. They also exhibited limited chemical potential conditions for achieving Li-excess conditions through doping. Combinations of the suggested isostructural oxides and dopants were identified to reduce the interactions between interstitial Li and dopant. Some isostructural oxides were doped using Sn and were evaluated using first-principles molecular dynamics; their Li-ion conductivities at room temperature were found to be comparable with those of garnet-type Li-ion conductors.

cond-mat.mtrl-sci↗

arXiv2020

Transfer learning for materials informatics using crystal graph convolutional neural network

For successful applications of machine learning in materials informatics, it is necessary to overcome the inaccuracy of predictions ascribed to insufficient amount of data. In this study, we propose a transfer learning using a crystal graph convolutional neural network (TL-CGCNN). Herein, TL-CGCNN is pretrained with big data such as formation energies for crystal structures, and then used for predicting target properties with relatively small data. We confirm that TL-CGCNN can improve predictions of various properties such as bulk moduli, dielectric constants, and quasiparticle band gaps, which are computationally demanding, to construct big data for materials. Moreover, we quantitatively observe that the prediction of properties in target models via TL-CGCNN becomes more accurate with an increase in size of training dataset in pretrained models. Finally, we confirm that TL-CGCNN is superior to other regression methods in the predictions of target properties, which suffer from small amount of data. Therefore, we conclude that TL-CGCNN is promising along with compiling big data for materials that are easy to accumulate and relevant to the target properties.

cond-mat.mtrl-sci↗

arXiv2018

Water Facilitated Electrochemical Reduction of CO2 on Cobalt-Porphyrin Catalysts

Cobalt-porphyrin catalyzed reductive decomposition of CO2 to CO is investigated based on Koper's water facilitated CO2 reduction mechanism using a simple but accurate protocol based on thermodynamics. In our protocol, accurate predictions of standard redox potentials and free energy differences are achieved by combining strengths of both density functional theory and experimental observations. With the proposed protocol, we found that the proton transfer from H2O takes place at -0.80 V vs. RHE at pH=3 through a concerted pathway and, as a result, the key intermediate for the CO generation, i.e., [CoP-COOH]- is formed. Since the redox potential of the proton transfer agrees well with experimentally observed CO2 reduction potential, we successfully clarified that H2O plays an important role in the reductive decomposition of CO2 to CO. This result is valuable not only for understanding the cobalt-porphyrin catalyzed reductive decomposition of CO2 but also as a guide for the development of new catalysts.

cond-mat.mes-hall↗