About the project
This PhD project aims to accelerate the discovery of heat-resistant austenitic alloys by integrating machine learning with high-throughput combinatorial experiments. Through iterative design, synthesis, and validation, the project will develop advanced materials for high-temperature reactors, significantly reducing alloy development time and enhancing structural integrity under creep-fatigue conditions.
The long-term structural integrity of steam generators in high-temperature reactors critically depends on the performance of advanced austenitic stainless steels, particularly under creep and creep-fatigue conditions. However, the conventional development of such alloys has relied heavily on trial-and-error exploration of a vast compositional space defined by elements such as iron (Fe), Chromium (Cr), Nickel (Ni), Manganese (Mn), and Molybdenum (Mo). Although this approach has led to the development of alloys such as Type 316, Alloy 617, 800H, and 709, the process remains slow, expensive, and inefficient.
This project aims to revolutionise alloy development by establishing an accelerated discovery protocol that integrates machine learning (ML) with combinatorial high-throughput experimentation in a closed-loop framework. The goal is to streamline the identification and validation of new austenitic stainless steels with superior high-temperature performance.
You will initiate the process by developing an ML model trained on a combined dataset of historical alloy performance data and CALPHAD-based high-throughput thermodynamic simulations. This first-generation ML model will be used to predict new alloy compositions, which will then be experimentally validated through a suite of high-throughput experiments. These experimental results will serve as feedback to iteratively retrain the ML model, enhancing its predictive accuracy.
Specifically, three material synthesis routes will be employed to construct compositional libraries:
- compositionally graded thin films, deposited onto a metallic substrate using a unique high-throughput physical vapour deposition (HT-PVD) system available at Southampton
- compositionally graded bulk samples
- bulk samples containing discrete alloy compositions, both fabricated using laser-based directed energy deposition (DED) additive manufacturing
You will benefit from access to advanced research infrastructure at the University of Southampton, including the Testing and Structures Research Laboratory (TSRL), the Material Innovation Laboratory, and the Royce Advanced Metals Processing Facility (e.g., BeAM instrument for DED) located at Sheffield. To this end, trainings will be provided by senior experimental experts.
