The postdoc will work on the ANATOMIX and PSICHÉ beamlines under the joint supervision of Andrew King (PSICHE beamline scientist, imaging coordinator) and Timm Weitkamp (ANATOMIX beamline manager).
She/He will actively participate to the user-support program (X-ray computed tomography and materials science) and will be involved in the scientific, technical, and methodological activities of the beamlines. That means preparing the beamlines for the experiments, helping the users with the management of experimental setups, data acquisition and data treatment if necessary.
She/He will develop the AIQuAM3D research program (see next section). She/He will be granted in-house research beam time while also submitting proposals to the peer review committees. She/He will publish her/his results and present them at national and international conferences. She/He may also pursue their own research interests and collaborations in addition to this program.
In Laser-based Powder Bed Fusion (L-PBF), the additive manufacturing (AM) process with the highest industrial maturity and spread, there are more than 50 adjustable process parameters that influence parts quality. Despite increasing scientific effort, the L-PBF process chain still cannot be fully simulated with state-of-the-art tools, making quality inspection even more important, especially for safety-critical application areas such as the aerospace, automotive and medical sectors. For this, the support of μCT is needed because there is no other method to provide local part density information in a non-destructive, fast and economic manner. In this project an AI-model will be developed to correlate multiple features generated from L-PBF in-situ monitoring data with defect labels obtained by μCT. The ultimate goal is to implement an embedded model for real-time defect prediction in a L-PBF system and validate it for reference samples and a complex geometry demonstrator. The post-doc will contribute to these studies via μCT studies of AM parts.
μCT is widely used as a tool to investigate three-dimensional morphology of materials in industrial research and development (R&D) and non-destructive testing. Synchrotron-based μCT is capable of very fast data acquisition at high spatial resolution, allowing in-situ imaging and high sample throughput. However, so far synchrotron μCT has remained mainly an R&D tool despite its advantages. One obstacle to its wider adoption is the delay between the design of an experiment and the final result. This is partly due to the quantity of data, which is often the bottleneck in the pipeline between data collection and result: the number of 3D image volumes can exceed 100 samples per 8-hour shift, and the corresponding data volume can be over 1 TB per hour of beamtime. Additionally, access to synchrotron facilities is more limited than for conventional μCT scanners: measurement campaigns typically must be planned weeks or months ahead, and beamtimes are short and require stringent preparation and execution. As a result, scan failures due to technical problems or mistakes during the acquisition can put at risk the success of the campaign altogether, unless the error is immediately discovered during the experiment. To ensure the efficient use of synchrotron beamtime for high-throughput analyses, real-time analysis or at least near-real-time automatic quality control of the acquired scan volumes would therefore be a tremendous advantage. In this project software tools (both conventional and using artificial intelligence) will be developed to automate and streamline both data analysis and data quality control/instrument feedback.
The candidate should hold a PhD in mechanics of materials/physics applied to material science or equivalent and have experience in tomography imaging. His/her skills must also include image and volume data processing with Python or some other language, as well as knowledge in AI modelling, training and inference.
The postdoc will join an enthusiastic and growing pluri-disciplinary team, benefiting from multiple national and international partnerships including the AIQuAM3D project.
We’re looking for someone curious, dynamic, autonomous, and who enjoys working in a team.