Lin, Binbin (2023)
Data-driven Analysis of Microstructure-Property Relation in Functional Materials.
Technische Universität Darmstadt
doi: 10.26083/tuprints-00024475
Ph.D. Thesis, Primary publication, Publisher's Version
Text
Lin_Thesis_Final.pdf Copyright Information: CC BY-SA 4.0 International - Creative Commons, Attribution ShareAlike. Download (43MB) |
Item Type: | Ph.D. Thesis | ||||
---|---|---|---|---|---|
Type of entry: | Primary publication | ||||
Title: | Data-driven Analysis of Microstructure-Property Relation in Functional Materials | ||||
Language: | English | ||||
Referees: | Xu, Prof. Bai-Xiang ; Banerjee, Prof. Sarbajit | ||||
Date: | 25 September 2023 | ||||
Place of Publication: | Darmstadt | ||||
Collation: | VIII, 185 Seiten | ||||
Date of oral examination: | 17 July 2023 | ||||
DOI: | 10.26083/tuprints-00024475 | ||||
Abstract: | The interplay between structure and property is a fundamental research topic in materials science and engineering. Materials possess diverse microstructures, and effectively characterizing, representing, and correlating them with properties poses significant challenges. As a result, the understanding of the microstructure-property relation relies primarily on empirical approaches, which limits its application in materials optimization and design. However, the emergence of machine learning and data science methods in recent years has provided powerful tools with immense potential to advance materials research and design principles. These approaches offer promising opportunities to develop materials that meet future needs and have the capability to revolutionize traditional methods of materials research. This thesis focuses on the application of machine learning techniques to explore the relationships between microstructure and properties. Three prototype microstructural systems are studied: nanowire structure in lithium-ion cathode material, fibrous network structure in paper material, and grain/phase structure in dual-phase steel. The present work investigates different forms of microstructure representation across multiple microstructure levels. These include the use of deep neural networks to derive geometric descriptors to characterize nanowire morphology based on particle-level microscopy images, the derivation of descriptors from the complex fibrous network structure of paper materials at the network level, and the use of image-based latent features at the microstructure domain level for dual-phase steel. The material properties considered in this work are electrochemical properties obtained from experimental assessments, as in the case of battery cathode material, or from sound physical simulation data generated by sophisticated material models and simulations, as demonstrated for paper material and dual-phase steel. This thesis convincingly demonstrates that the use of machine learning-based techniques enables effective microstructural characterization, extraction of microstructural features, rapid prediction of material response, and ultimately the establishment of microstructure-property relations to facilitate improved material optimization and design. |
||||
Alternative Abstract: |
|
||||
Status: | Publisher's Version | ||||
URN: | urn:nbn:de:tuda-tuprints-244751 | ||||
Classification DDC: | 500 Science and mathematics > 500 Science 500 Science and mathematics > 510 Mathematics 600 Technology, medicine, applied sciences > 620 Engineering and machine engineering |
||||
Divisions: | 11 Department of Materials and Earth Sciences > Material Science 11 Department of Materials and Earth Sciences > Material Science > Mechanics of functional Materials |
||||
Date Deposited: | 25 Sep 2023 12:03 | ||||
Last Modified: | 26 Sep 2023 08:40 | ||||
URI: | https://tuprints.ulb.tu-darmstadt.de/id/eprint/24475 | ||||
PPN: | 511892233 | ||||
Export: |
View Item |