Deep learning-based precipitate quantification in STEM images of complex steel microstructures
-
M. Müller
Martin Müller studied materials science and engineering at Saarland University. He then worked in the forging and ring rolling industry before returning to Saarland University to do his PhD on AI-based microstructure analysis in collaboration with MECS. He now heads the steel and AI research projects there.
,
J. Barrirero
Jenifer Barrirero has 10+ years of experience in research and knowledge transfer, with over 30 scientific publications. She holds a PhD in Materials Science and works at the Chair of Functional Materials in Saarbrücken and the Steinbeis Center MECS. Her expertise lies in high-resolution materials characterization, metallurgy and a recent focus on circularity in EV batteries.
Abstract
Understanding precipitate evolution in microalloyed high-strength low-alloyed (HSLA) steels is essential for optimizing their processing and mechanical properties. In this work, we present an automated workflow for precipitate quantification in scanning transmission electron microscopy (STEM) images of niobium and titanium HSLA steels, combining deep learning-based semantic segmentation with conventional image processing techniques. Among the different STEM image contrasts tested, the high-angle annular dark-field-based model delivered the best segmentation performance, achieving a mean intersection over union of 0.8111 and a deviation of only 5.37 % in the measured mean particle diameter. The workflow’s robustness and generalizability were demonstrated on unseen images of different HSLA steels, with particle counts and size distributions aligning well with expectations based on the respective alloying concepts.
Kurzfassung
Für eine Optimierung der Verarbeitung und der mechanischen Eigenschaften von mikrolegierten HSLA-Stählen (High-Strength Low-Alloyed (hochfest niedrig legiert) Stählen ist das Verständnis der Entwicklung von Ausscheidungen entscheidend. In dieser Arbeit stellen wir einen automatisierten Workflow zur Quantifizierung von Ausscheidungen in rastertransmissionselektronenmikroskopischen Aufnahmen (Scanning Transmission Electron Microscopy, STEM) von HSLA-Niob- und Titanstählen vor, in dem Deep Learning-basierte semantische Segmentierung und herkömmliche Bildverarbeitungsverfahren miteinander kombiniert werden. Mit einer mittleren Intersection over Union (IoU, auch Jaccard-Index) von 0,8111 und einer Abweichung von lediglich 5,37 % im gemessenen mittleren Partikeldurchmesser erzielte von den verschiedenen getesteten STEM-Bildkontrasten das HAADF-basierte Modell die beste Segmentierungsleistung. Robustheit und Generalisierbarkeit des Workflows wurden anhand bisher unbekannter Aufnahmen verschiedener HSLA-Stähle aufgezeigt, wobei Partikelanzahl und -größenverteilungen im Einklang mit den Erwartungen für die jeweiligen Legierungskonzepte stehen.
About the authors
Martin Müller studied materials science and engineering at Saarland University. He then worked in the forging and ring rolling industry before returning to Saarland University to do his PhD on AI-based microstructure analysis in collaboration with MECS. He now heads the steel and AI research projects there.
Jenifer Barrirero has 10+ years of experience in research and knowledge transfer, with over 30 scientific publications. She holds a PhD in Materials Science and works at the Chair of Functional Materials in Saarbrücken and the Steinbeis Center MECS. Her expertise lies in high-resolution materials characterization, metallurgy and a recent focus on circularity in EV batteries.
References / Literatur
[1] Loy, A. C. M.; Ng, W. L.; Bhattacharya, S.: Advanced characterization techniques for the development of Subatomic scale catalysts: One step closer to industrial scale fabrication. Materials Today Catalysis 4 (2024). DOI:10.1016/j.mtcata.2023.100033.10.1016/j.mtcata.2023.100033Search in Google Scholar
[2] Hossain, N. et al.: Advances and significances of nanoparticles in semiconductor applications – A review. Elsevier B. V. (2023). DOI: 10.1016/j.rineng.2023.101347.10.1016/j.rineng.2023.101347Search in Google Scholar
[3] Mohammed, H.; Mia, M. F.; Wiggins, J.; Desai, S.: Nanomaterials for Energy Storage Systems – A Review. Multidisciplinary Digital Publishing Institute (MDPI) (2025). DOI:10.3390/molecules30040883.10.3390/molecules30040883Search in Google Scholar PubMed PubMed Central
[4] Al Mahmud, M. Z.: A Concise Review of Nanoparticles Utilized Energy Storage and Conservation. Hindawi Limited (2023). DOI:10.1155/2023/5432099.10.1155/2023/5432099Search in Google Scholar
[5] Webel, J.; Mohrbacher, H.; Detemple, E.; Britz, D.; Mücklich, F.: Quantitative analysis of mixed niobium-titanium carbonitride solubility in HSLA steels based on atom probe tomography and electrical resistivity measurements. Journal of Materials Research and Technology 18 (2022), pp. 2048–2063. DOI:10.1016/j.jmrt.2022.03.098.10.1016/j.jmrt.2022.03.098Search in Google Scholar
[6] Vervynckt, S.; Verbeken, K.; Thibaux, P.; Houbaert, Y.: Recrystallization-precipitation interaction during austenite hot deformation of a Nb microalloyed steel. Materials Science and Engineering A 528 (2011) 16–17, pp. 5519–5528. DOI:10.1016/j.msea.2011.03.087.10.1016/j.msea.2011.03.087Search in Google Scholar
[7] Saaim, K. M.; Afridi, S. K.; Nisar, M.; Islam, S.: In search of best automated model: Explaining nanoparticle TEM image segmentation. Ultramicroscopy 233 (2022). DOI:10.1016/j.ultramic.2021.113437.10.1016/j.ultramic.2021.113437Search in Google Scholar PubMed
[8] Zhou, L.; Wen, H.; Kuschnerus, I. C.; Chang, S. L. Y.: Efficientand Robust Automated Segmentation of Nanoparticles and Aggregates from Transmission Electron Microscopy Images with Highly Complex Backgrounds. Nanomaterials 14 (2024) 14. DOI:10.3390/nano14141169.10.3390/nano14141169Search in Google Scholar PubMed PubMed Central
[9] Zhang, Y.; Zhang, H.; Liang, F.; Liu, G.; Zhu, J.: The segmentation of nanoparticles with a novel approach of HRU2-Net†. Sci Rep 15 (2025) 1, p. 2177. DOI:10.1038/s41598-025-86085-w.10.1038/s41598-025-86085-wSearch in Google Scholar PubMed PubMed Central
[10] Rühle, B.; Krumrey, J. F.; Hodoroaba, V. D.: Work-flow towards automated segmentation of agglomerated, non-spherical particles from electron microscopy images using artificial neural networks. Sci Rep 11 (2021) 1. DOI:10.1038/s41598-021-84287-6.10.1038/s41598-021-84287-6Search in Google Scholar PubMed PubMed Central
[11] Monteiro, G. A. A.; Monteiro, B. A. A.; Dos Santos, J. A.; Wittemann, A.: Pre-trained artificial intelligence-aided analysis of nanoparticles using the segment anything model. Sci Rep 15 (2025) 1, p. 2341. DOI:10.1038/s41598-025-86327-x.10.1038/s41598-025-86327-xSearch in Google Scholar PubMed PubMed Central
© 2025 Walter de Gruyter GmbH, Berlin/Boston, Germany
Articles in the same Issue
- Contents
- Editorial
- Editorial
- Evaluation of damage mechanisms in full-forward rod extruded AISI 5115 steel by AI-based image segmentation
- A roll-headed pin from Getzersdorf, Austria, opens up interesting questions about iron metallurgy in the Hallstatt period
- Magneto-optical Kerr effect analysis of strain-induced martensite formation during flow forming of metastable austenitic steel AISI 304L
- Materials scientific challenges of a circular economy of high-alloy tool steels
- Cavitation-induced damage on a ship propeller’s Kort nozzle made of AISI 316L and investigation of the near-surface microstructural changes
- Preparation of aluminum thin films and laminates
- Preparation and microstructural examination of additively manufactured alloys 316L and 17-4PH
- Results of a round robin test on grain boundary determination by image analysis in the DGM working group ‘Quantitative Structural Analysis’
- Deep learning-based precipitate quantification in STEM images of complex steel microstructures
- A bronze fibula foot from the Roman Carnuntum, Lower Austria
- Optimization of preparation methods for EBSD analysis of highly metastable austenitic steels
- Picture of the Month
- Picture of the Month
- People
- Interview with Gaby Ketzer-Raichle on the occasion of the new edition of “Schumann Metallography”, published by Heinrich Oettel and Gaby Ketzer-Raichle
- News
- News
- Meeting Diary
- Meeting Diary
Articles in the same Issue
- Contents
- Editorial
- Editorial
- Evaluation of damage mechanisms in full-forward rod extruded AISI 5115 steel by AI-based image segmentation
- A roll-headed pin from Getzersdorf, Austria, opens up interesting questions about iron metallurgy in the Hallstatt period
- Magneto-optical Kerr effect analysis of strain-induced martensite formation during flow forming of metastable austenitic steel AISI 304L
- Materials scientific challenges of a circular economy of high-alloy tool steels
- Cavitation-induced damage on a ship propeller’s Kort nozzle made of AISI 316L and investigation of the near-surface microstructural changes
- Preparation of aluminum thin films and laminates
- Preparation and microstructural examination of additively manufactured alloys 316L and 17-4PH
- Results of a round robin test on grain boundary determination by image analysis in the DGM working group ‘Quantitative Structural Analysis’
- Deep learning-based precipitate quantification in STEM images of complex steel microstructures
- A bronze fibula foot from the Roman Carnuntum, Lower Austria
- Optimization of preparation methods for EBSD analysis of highly metastable austenitic steels
- Picture of the Month
- Picture of the Month
- People
- Interview with Gaby Ketzer-Raichle on the occasion of the new edition of “Schumann Metallography”, published by Heinrich Oettel and Gaby Ketzer-Raichle
- News
- News
- Meeting Diary
- Meeting Diary
