Preparation strategy for statistically significant micrometer-sized particle systems suitable for correlative 3D imaging workflows on the example of X-ray microtomography

• Published in: Powder technology 2022-01, Vol.395, p.235-242
• Main Authors: Ditscherlein, Ralf, Leißner, Thomas, Peuker, Urs A.
• Format: Article
• Language: English
• Subjects: Computed tomography; Ion beams; Learning algorithms; Machine learning; Machine learning assisted segmentation; Nanoparticles; Particle characterization; Sample preparation; Statistical analysis; Statistical methods; Statistical significance; X ray microtomography; XRM
• ISSN: 0032-5910; 1873-328X
• DOI: 10.1016/j.powtec.2021.09.038

[Display omitted] •Preparation of homogeneous particle samples using a low X-ray attenuating matrix.•Forced distancing by addition of low X-ray attenuating graphite nanoparticles.•Generation of particle discrete datasets by machine learning assisted image data segmentation.•Validation of reproducible particle size distribution measurement via X-ray microtomography. The characterization of multidimensional particle property distributions through computed tomography requires an adapted sample preparation strategy. This strategy should both generate as many spatially separated particles as possible in the smallest achievable volumes and also enable mechanically and vacuum-stable samples that are suitable for correlative measurement, for example with high-energy ion beam methods. In the present study an epoxy-based method is presented that minimizes the negative influence of particle sedimentation by adding very low X-ray absorbing graphite nanoparticles as spacer. A machine learning-based method is presented to discretize the particle system. Results are compared with data from 2D SEM validation measurements and data of a previous study.