CFR502-PhD Student

As a PhD student based at the ESRF in Grenoble, you will develop neural-network-driven reconstruction tools for Scanning 3D X-ray Diffraction (S3DXRD), a technique capable of mapping grain-resolved orientation and strain fields in crystalline materials non-destructively at sub-micron resolution. Working with beamline scientists on ID11, ID03, and the ESRF Algorithms & scientific Data Analysis group, you will generate synthetic diffraction training data from simulated microstructures and use these to train neural networks that replace the current expert-intensive reconstruction process with something fast, robust, and accessible to industrial users. The project spans methods development, experimental validation, and open-source software release, sitting at the intersection of synchrotron science, materials characterisation, and scientific machine learning. The PhD is hosted by Université Grenoble Alpes (UGA) within the Physics doctoral school.

Responsibilities include:

·         Develop a phantom microstructure generation pipeline and validate synthetic diffraction output against real ID11 datasets

·         Train and validate neural network indexing models, progressing from box-beam to full scanning geometry

·         Benchmark the automated pipeline against conventional reconstruction across a range of materials including deformed and additively manufactured samples

·         Collaborate with crystal plasticity simulation groups to ensure training microstructures capture realistic orientation gradient fields

·         Package, document and release an open-source reconstruction toolkit for the broader diffraction microstructure imaging community

Further information may be obtained from James Ball (tel.: +33 (0)4 76 88 22 73, email: james.ball@esrf.fr).