Record #238104 - iMPULSE

DFG project G:(GEPRIS)413993866

Kerr microscopy with machine learning domain detection for in-situ magnetic materials analysis (MaKerr)

Coordinator	Professor Dr. Gerhard Schneider
Grant period	2019 - 2025
Funding body	Deutsche Forschungsgemeinschaft
	DFG
Identifier	G:(GEPRIS)413993866

Note: This project proposal concerns a high resolution light-optical Kerr microscope specifically designed for insitu characterization of the magnetization of hard- and soft magnetic materials under real life conditions (temperature, magnetic fields, and mechanical loads) as found in automotive applications, e.g. traction motors for electric cars. Key element is the intelligent combination of in-situ microscopy with digital microscopy, automated quantification tools and machine learning approaches to obtain a deeper knowledge of these materials in their application environment based on efficient analysis of their microstructures. This requires a holistic approach to consider the system and then the visualization process: sample preparation – microscope – camera – software.An important element of sample preparation is contrast enhancement and surface preservation by dielectric anti-reflection coatings. Core element of the microscope concept is the implementation of multiple testing modules operated in a high vacuum chamber, including a hot stage, mechanical micromanipulator (e.g. tensile stress), electromagnets for generation of homogenous and tunable magnetic fields as well as fast and high resolution digital camera technology. The flexible design allows the concurrent operation of the testing modules, e.g. investigation of soft magnetic electrical steel under application of tensile stress at elevated temperatures in the presence of magnetic fields. Usage of lenses with high numerical aperture guarantees high optical resolution of the microscope also when operated in combination with the vacuum chamber (up to 230 nm without, 430 nm with vacuum chamber). State of the art high speed digital cameras allow analysis of physical processes with time resolutions of up to 10 ms. Specifically designed machine learning software routines enable highly accurate detection of magnetic domain patterns even in materials with less favorable signal to noise ratio, e.g. materials with weak intrinsic magnetic contrast or images acquired in fast time lapses. Thus, automated detection and quantification of domain patterns acquired over a large sample area is possible as a function of time and/or temperature.The applied for microscope system is exceptionally well suited to efficiently investigate important scientific questions arising in the development of sustainable powerful electric energy converters, such as in depth analysis of magnetic reversal and demagnetization processes in permanent magnets or magnetic losses in soft magnetic components. In addition, the magnetic potential of novel hard magnetic phases found by experimental high throughput screening can be evaluated, analyzing the domain structure. Here, the contrast and size of the domain pattern and its movement are of special significance.

Recent Publications

There are no publications

Record created 2023-01-20, last modified 2025-08-19

Similar records

Rate this document:

(Not yet reviewed)

Add to personal basket
Export as Author List with IDs BibTeX (UTF-8), EndNote XML, EndNote Text, RIS, MARC, Print MARC, MARCXML, DC,
Request correction
Submit fulltext

guest :: login iMPULSE
		Search		Submit		Personalize Your alerts Your baskets Your searches		Help