Akiko Kumada

Machine learning methods for predicting the electron ionization mass spectra from molecular structures have shown promise for environmental chemical identification, but their performance under domain-specific data scarcity remains poorly understood. We systematically compare a conventional multilayer perceptron model (NEIMS) with a Transformer-based chemical foundation model (MolFormer-XL) for the electron ionization mass spectrometry spectrum prediction under controlled few-shot conditions. Using fluorine-containing molecules as a broader proxy domain, including a PFAS-like subset, motivated by the practical challenge of detecting novel fluorinated contaminants with limited reference data, we vary the number of domain-specific training examples from 5 to 175 while maintaining fixed validation and test sets. Across all few-shot conditions and three of four evaluation metrics (weighted cosine similarity, intensity-weighted precision, and top-10 precision), MolFormer-XL consistently outperforms NEIMS, while intensity-weighted recall remains comparable between the two models. The largest performance gaps are observed in extreme data-scarcity regimes. These results demonstrate that MolFormer-XL, which combines pre-trained molecular representations with a Transformer-based architecture and learned SMILES embeddings, provides a promising approach for transfer under severe domain-specific data scarcity in environmental mass spectrometry.

Polymers are widely used as electrical insulating materials in the power and energy sector. As data-driven approaches to materials development are attracting increasing attention, two major challenges remain: the limited availability of polymer data and the complex hierarchical structures of polymers. In this study, we conduct high-throughput quantum chemistry calculations and molecular dynamics (MD) simulations using a single force field for approximately 5,500 homopolymers to construct a dataset of computed polymer properties. We develop machine learning models to predict MD-calculated polymer properties using quantum chemistry-derived physical quantities and MD force field parameters as descriptors. We also compare their performance with models using structural descriptors. While interpolation performance is comparable between molecular structure–based models, including a fingerprint model and a graph convolutional neural network, and models using quantum chemistry and MD force-field descriptors, the latter shows superior extrapolation performance in our evaluated setting for MD-computed density. In addition, we confirm that increasing the amount of data improves the accuracy of both interpolation and extrapolation predictions. The polymer dataset and predictive approach proposed in this study may help inform computational screening toward the development of high-performance polymer materials; experimental validation remains necessary.

In Direct Current Gas Insulated Switchgear (DC-GIS), surface charge accumulation on insulating spacers is a significant challenge. Accumulated charges due to high DC electric fields distort the electric fields and decrease the flashover voltage. With surface charge measurements and numerical simulations, it is revealed that multiple factors, such as electric field strength and temperature, affect the charging profile. The objective of this paper is to investigate the charge accumulation characteristics of insulating spacers with different shapes through experiments and calculations, and to elucidate the charging mechanism. Surface charge distributions on downsized cone-type and disk-type model spacers under a simulated DC-GIS environment are measured and compared. Additionally, simulations considering the spacer bulk and insulating gas are conducted. Based on the measurements and simulations, the dominant charge supply sources and charge transport paths responsible for the charging phenomena are discussed. The results show that the surface charge distribution profiles of the two types of spacers differ in terms of charge quantity, charging area, and time variations. These results indicate that the charges injected from the electrode are predominantly transported through the spacer bulk in cone-type spacers and through the spacer surface in disk-type spacers. The dominant charge transport path and the intensive charge accumulation are greatly influenced by the initial electric field distribution, determined by the spacer geometry.

While directly connecting gas-insulated switchgear (GIS) to a DC de-icing circuit improves operational efficiency, the subsequent power restoration process imposes coupled stresses from elevated temperature and DC-AC switching electric fields on the insulators, creating a significant risk of surface flashover. In this study, the influence of such coupled conditions on the surface flashover characteristics of epoxy (EP) insulators was investigated through a combination of experimental testing and multiphysics simulations. A coupled test platform incorporating DC-AC switching electric fields and thermal fields was established, and a transient simulation model of gas–solid composite insulation with temperature-dependent parameters was developed. The results show that high temperature and prolonged DC operation intensify the accumulation of homopolar surface charges. Upon switching to AC voltage, the residual charges interact with the applied field, leading to severe local electric field distortion, particularly when the AC field opposes the charges polarity. This effect, exacerbated by temperature-induced increases in material conductivity and carrier mobility, leads to a sharp rise in flashover probability. At 80 °C with 5 min of DC pre-charging, the flashover probability reaches up to 60%. This study reveals the underlying physical mechanisms of flashover under coupled thermal and DC-AC switching electric field conditions, providing critical theoretical guidance for insulation design and safety assessment in DC de-icing applications of GIS.

The electron density and electric field govern physical and chemical reactions in a secondary streamer discharge under atmospheric-pressure air. We present direct combinational measurements of these essential quantities and demonstrate their consistency by solving the continuity equation for electrons. The measurements enabled validity examination of previous computational studies and clarified contradictions; the prominence was a near-cathode phenomenon measured during primary-to-secondary transition, which caused unpredicted ionization inside the positive secondary discharge. The thorough comparison revealed that a fundamental governing mechanism to be established for the secondary discharge is current continuity.

Accurately predicting the properties of polymers is essential for data-driven materials design. However, such predictions are often challenged by the limited availability of polymer-related data and the fact that high-performance polymers of interest typically lie outside the distribution of existing datasets. In this study, we develop a machine learning model that enhances extrapolative prediction accuracy beyond the training data by leveraging hierarchical, physics-informed descriptors. Specifically, we utilize quantum mechanical (QM) descriptors derived from density functional theory calculations, molecular dynamics (MD) descriptors representing structural and dynamical properties, and force field (FF) descriptors characterizing the interaction parameters used in MD simulations. We investigated two types of extrapolation tasks: extrapolation beyond the range of physical properties and extrapolation to structurally dissimilar molecules. By systematically evaluating all non-zero combinations of QM, MD, and FF descriptors, we find that selected subsets often outperform models using the full descriptor set. This highlights the critical role of dimensionality reduction and descriptor relevance, especially under data-scarce conditions. Comparisons with structure-based models employing molecular fingerprints or molecular graphs further demonstrated the superiority of the proposed model based on selected physics-based descriptors.

The electric-field-induced second-harmonic generation (E-FISH) method has gained attention as a non-contact, non-invasive, and highly sensitive technique for measuring electric fields. In this method, the electric field profile along the laser path influences the second harmonic (SH) signal and must be carefully characterized. Several approaches have been proposed to infer electric-field profiles from SH distributions obtained by changing the focus position along the laser optical axis. These include optimization-based methods and techniques that assume a predefined electric field shape. However, optimization-based methods are prone to local optima, while assumption-based methods cannot be applied to unknown electric fields. To address these challenges, we propose a novel phase-sensitive approach to infer an unknown electric-field profile without requiring optimization. We generated an additional second-harmonic reference beam using a local oscillator (LO) with a β–BBO crystal and observed interference fringes by superimposing the LO with the E-FISH beam. By analyzing changes in the interference fringes, we determined the phase distribution of the E-FISH beam along the laser optical axis. While larger errors were observed in regions of low electric field strength, using the obtained phase distribution, we successfully inferred the electric-field profile with a peak error of 0.3%. This work marks the first successful demonstration of a technique enabling unique measurements of arbitrary electric-field profiles.