Thesis Subject: Bragg coherent diffractive imaging driven by machine learning: from data collection through reduction
The ESRF ID01 beamline is a strain microscope, which exploits the intrinsic coherent properties of the X-ray source to image the strain in materials at high spatial resolution (<10nm) using the Bragg Coherent Diffraction Imaging (BCDI) method. This characterization tool provides a route to materials design through strain engineering. With the realisation of the ESRF-EBS and improvements in data acquisition, a data deluge is imminent for BCDI experiments at the ID01 beamline. It is timely that a solution is sought concerning all aspects of the data pipeline from sample selection and data acquisition, through data pre-processing and the subsequent image generation and analysis. Machine learning (ML) has the potential to unlock the sought physical parameters and rapidly put them at scientists’ fingertips, dramatically improving both the throughput, design and output of such experiments.
At the core of the proposal is the BCDI method. BCDI allows the scientist to image the strain distribution in 3D, operando or in-situ. Strain signatures are common in the presence of defects and typically extend beyond the spatial resolution of the method, thus the defects themselves can be located and identified, alongside strain associated to facets, substrate interactions, phase changes and so forth. The goal of this PhD is to exploit machine-learning solutions to streamline all stages of the experimental process: data collection, data processing, phase retrieval and eventual data interpretation.
Several attempts have already been made to use ML methods to tackle the phase retrieval step, the image creation, but there is significant work to be done before less tailored samples can be routinely solved. Specifically, the goals of the project are to augment:
The ML developments will be applied to experimental datasets collected on metal (Pt, Ni, Au, Cu) nano-crystals, and will also be tested within the scope of the ERC project Carine.
For further information contact:
Tobias Schulli (tel.: +33 (0)4 76 88 22 80, email: email@example.com)
Vincent Favre-Nicolin (tel.: +33 (0)4 76 88 28 11, email: firstname.lastname@example.org)
Steven Leake (tel.: +33 (0)4 76 88 1939, email: email@example.com)
The successful candidate will be enrolled at the Université Grenoble-Alpes. The candidate will be hired by the ESRF (Grenoble, France). The contract is of two years renewable (subject to satisfactory progress) for one year.
The ESRF is an equal opportunity employer and encourages diversity.
If you are interested in this position, please apply via the ENGAGE website (not on the ESRF website) by 15 March 2022
This PhD is co-funded by the Marie Skłodowska-Curie COFUND project ENGAGE (grant agreement101034267), and applicants must follow the associated rules notably regarding mobility (see https://ec.europa.eu/research/participants/data/ref/h2020/other/guides_for_applicants/gfa_h2020-msca-cofund-2020_en.pdf). In particular, applicants must not have resided or carried out their main activity (work, studies, etc.) in France for more than 12 months in the last 3 years.
The European Synchrotron, the ESRF, is an international research centre based in Grenoble, France.
Through its innovative engineering, pioneering scientific vision and a strong commitment from its 700 staff members, the ESRF is recognised as one of the top research facilities worldwide. Its particle accelerator produces intense X-ray beams that are used by thousands of scientists each year for experiments in diverse fields such as biology, medicine, environmental sciences, cultural heritage, materials science, and physics.
Supported by 21 countries, the ESRF is an equal opportunity employer and encourages diversity.