8 ADVANCED MATERIALS & PROCESSES | NOVEMBER/DECEMBER 2025 EXPLORING HYDROGEN EMBRITTLEMENT Scientists at Empa’s Joining Technology and Corrosion Laboratory, Switzerland, are investigating an aspect of hydrogen embrittlement that involves the interaction of hydrogen with the passivation layer on steel. This thin layer forms naturally on the surface of most metals and alloys, giving stainless steels their corrosion resistance. The type and composition of this layer differs from steel to steel with some oxides significantly more resistant to hydrogen than others. Researcher Chiara Menegus developed an electrochemical cell in which the steel sample is mounted. Water is placed on one side of the sample and argon gas on the other. By applying an electrical voltage, atomic hydrogen is generated from the water. The hydrogen diffuses through the thin sample until it reaches the oxide layer on the opposite TESTING | CHARACTERIZATION AI SUPPORTS MICROSTRUCTURE ANALYSIS Imagic Bildverarbeitung AG, Zurich, and the Materials Engineering Center Saarland (MECS), Germany, began a strategic partnership in September. The collaboration combines MECS’ expertise in materials science with Imagic’s software expertise in micro- scopy, image analysis, and image data management. This combination creates a holistic ecosystem for materials analysis, from image capture with microscopes and cameras from various manufacturers to AI-based interpretation and automated report generation. The result is accelerated and scalable analysis processes. A key benefit is the complete integration of AI analysis into the on-premises solution of the Imagic IMS software. Notably, data sovereignty always remains with the end user including full control over sensitive image and research data. The first joint application, AI-supported grain size evaluation, is setting new standards, say researchers. “Our approach makes it possible to precisely define the database required for an AI model, including the fundamental truth and a deep understanding of the data generation process. This ensures that the models achieve the highest accuracy, robustness, and generalization in series production,” says Frank Mücklich, FASM, institute director of MECS. The system achieves extreme precision and reproducibility in the detection and quantification of grain boundaries, even in complex cases. This means that complex analyses can be carried out in a fraction of the time previously required, along with increased objectivity, reliability, and significantly greater analytical depth. The application portfolio for AI-supported microstructure analysis will be continuously expanded over the next few years. www.mec-s.de. Testbed 80. Courtesy of Rolls-Royce. Chiara Menegus (back) and Claudia Cancellieri investigate how hydrogen interacts with the thin oxide layers on high-strength steels. Courtesy of Empa. From left: Martin Mueller, Tobias Fox, Björn Bachmann, Dominik Britz, (all with MECS), Patrik Wermelinger (Imagic), and Frank Mücklich (MECS). Oak Ridge National Laboratory built a non-nuclear test bed to simulate the conditions of a space nuclear reactor and avoid the cost and regulations required for testing in a reactor environment. The facility combines hardware with computer models to reproduce space conditions and enable NASA and others to develop autonomous controls using cost-effective components and open-source software. ornl.gov. MIT is the 16th member to join an international consortium building the $2.6 billion Giant Magellan Telescope in Chile. The telescope’s 25.4-meter aperture will have five times the light-collecting area and up to 200 times the power of existing observatories. It is already 40% under construction, with major components being manufactured across 36 U.S. states. Completion is expected by the 2030s. mit.edu. BRIEFS
RkJQdWJsaXNoZXIy MTYyMzk3NQ==