EPSRC Centre for Doctoral Training in Fusion Power - EPSRC Centre for Doctoral Training in the Science and Technology of Fusion Energy

Fusion CDT
Menu

Advanced Fusion Metallurgy through Machine Learning – Enhanced Electron Hyperspectral Imaging: Bridging the gap from Electronic Density of States to Functional Integrity in Fusion components (Materials Strand Project)

Supervisors: Professor Cornelia Rodenburg, Professor Russell Goodall & Professor Chris Race (University of Sheffield).

Electron beams (e-beams) are critical for 3D printing of fusion reactor components or for electron beam welding of very dissimilar materials such as tungsten and copper. A further advantage of e-beams is the potential for in-situ inspection which has recently been attempted for e-beam 3D printing of tungsten. It has been found that e-beam imaging can detect more abnormal critical defects than other techniques, thus potentially eliminating the need for additional and time-consuming destructive post-process analysis.  However, the need for better detector setups, scanning strategies and parameters, and theoretical models for the image generation has been identified, to achieve a better resolution.

Many of the electrons emitted during e-beam processes are inelastically scattered secondary electrons (SEs) with very low energies, while higher energy electrons backscattered elastically (BSEs) have lower resolution. Therefore, SE emission might be used for detecting subtle chemical alterations or stress inhomogeneities, which are important for the welding of materials such as tungsten, copper alloys, and reduced-activation steels needed in fusion reactors. As well as joints between components made of these materials, heterogenous electron beam welded joints are also likely to be required for fusion reactor components. Could SE images, which also contain SE spectral information hold the key for in-situ evaluation as a universal solution for rapidly optimising the most challenging e-beam processing of advanced materials for fusion applications?

Secondary Electron Hyperspectral Imaging (SEHI) delivers SE spectra (rather than just the flux of SEs) for each incident point. SEHI was pioneered by the Rodenburg group. For lithium compounds, some precious metals and carbon materials (including graphite), SE peak positions have been shown to be related to the Density of States (DOS) as obtained from Density Functional Theory (DFT) models. Peak widths and position can be related to local disorder, surface chemical changes on the micron to the nanometre scale, stress, or changes in SE spectra due to beam damage.  Thus, SEHI is ideally placed to connect fundamental science that could improve metallurgical workflows in fusion reactor applications. However, SEHI has not yet been explored in this context.  This project aims to understand the SE emssion spectra of metal alloys used in fusion reactors and their relationships to structural changes (atomic, nano, micro through to weld joint scale) and functional property changes.

The initial focus is on simple model systems representative of e-beam joining (W/Cu, W/V) and deliberately generated defects, through targeted ion beam damage, proceeding to alloys of increasing complexity starting with V-4Cr-4Ti. SEHI analysis will build on Machine Learning (ML) for unsupervised feature extraction and  be combined with traditional metallurgical characterisation techniques, complemented by nanoscale testing methods (nanoindentation/AFM) to explore relationships to engineering properties.

Objectives and Expected Key Outcomes:

  • DFT calculations to determine the electronic density of states for the alloys of interest as a function of alloy chemistry, disorder and strain
  • ML framework for SEHI features extraction that enables mapping of alloy chemistry variation, disorder and strain
  • Establish relationships between electron emission spectra and alloy ab-initio models and performance (engineering properties and/or degradation behaviour), through local density of state (DOS) uniformity/variations derived from experimental SEHI data
  • Testing and validation of relationships, through controlled ion beam damage, metallurgical characterisation and localised mechanical property testing
  • A trained graduate with transferable skills and knowledge of electron microscopy, metallurgy, modelling and or machine learning, and Hyperspectral Imaging techniques for process monitoring in the context of fusion reactor components.

This project is ambitious but feasible through the CDTs training for the Materials Stream, the Sorby Centre for Electron Microscopy, the Sheffield Henry Royce Discovery Centre and a planned collaboration with the Ion Beam Center at the Helmholtz-Zentrum Dresden-Rosendorf (HZDR). Outputs will benefit other fusion-related projects in both the Materials and Plasma streams.

During the first six months of the PhD, materials strand students will typically travel to attend taught modules at all six of the Fusion CDT partner universities.

The project will be mainly based in Sheffield, but there might be opportunities for travel to conferences and collaborations with other groups.

This project may be compatible with part time study, please contact the project supervisors if you are interested in exploring this.

This project is offered by University of Sheffield. For further details please contact: Professor Cornelia Rodenburg (c.rodenburg@shef.ac.uk).

For details on how to apply, please visit: Apply