Engineering topics

Szymanski, Nathan J

Publications and source records attributed to Szymanski, Nathan J.

2 records

DOE OSTI2025

Alternative Solid‐State Synthesis Route for Highly Fluorinated Disordered Rock‐Salt Cathode Materials for High‐Energy Lithium‐Ion Batteries

Abstract Fluorination has been identified as a key element for enabling the stable cycling of earth‐abundant manganese‐based disordered rock salt (DRX) cathodes. However, fluorination in the DRX bulk remains a challenge for scalable solid‐state synthesis. In this study, a tailored reaction pathway is proposed to synthesize a highly fluorinated DRX. It is demonstrated for the first time that the unconventional precursors, Li 6 MnO 4 , MnF 2 , and TiO 2 , can avoid the formation of Mn‐based intermediates (such as Li 2 (Mn,Ti)O 3, LiMnO 2 , and Mn 3 O 4 ), which, once formed, persist until the synthesis temperature reaches close to or above that required for fluorine volatility. Therefore, this method can form a highly fluorinated DRX with a composition of Li 1.23 Mn 0.40 Ti 0.37 O 2−y F y ( y = 0.29–0.34) at a low temperature (800 °C) relative to that required for conventional DRX solid‐state reactions (≥900 °C). Li 1.23 Mn 0.40 Ti 0.37 O 2−y F y ( y = 0.29–0.34) delivers a specific capacity above 300 mAh g −1 and a specific energy of 980 Wh kg −1 at 30 °C. Detailed characterization reveals that this DRX phase reversibly utilizes Mn 2+/3+ redox in the low‐voltage region and Mn 3+/4+ redox in the middle‐voltage range, whereas reversible oxygen redox is observed at high potentials.

Avvaru, Venkata Sai↗

DOE OSTI2025

Leveraging unlabeled SEM datasets with self-supervised learning for enhanced particle segmentation

Scanning Electron Microscopes (SEMs) are widely used in experimental science laboratories, often requiring cumbersome and repetitive user analysis. Automating SEM image analysis processes is highly desirable to address this challenge. In particle sample analysis, Machine Learning (ML) has emerged as the most effective approach for particle segmentation. However, the time-intensive process of manually annotating thousands of SEM images limits the applicability of supervised learning approaches. Self-Supervised Learning (SSL) offers a promising alternative by enabling knowledge extraction from raw, unlabeled data. This study presents a framework for evaluating SSL techniques in SEM image analysis, focusing on novel methods leveraging the ConvNeXtV2 architecture for particle detection. A dataset comprising 25,000 SEM images is curated to benchmark these proposed SSL methods. The results demonstrate that ConvNeXtV2 models, with varying parameter counts, consistently outperform other techniques in particle detection across different length scales, achieving up to a 34% reduction in relative error compared to established SSL methods. Furthermore, an ablation study explores the relationship between dataset size and SSL performance, providing actionable insights for practitioners regarding model selection and resource efficiency. This research advances the integration of SSL into autonomous analysis pipelines and supports its application in accelerating materials science discovery.

Rettenberger, Luca↗