Professor Christofidou delivered her talk titled ‘Making it Real: modern metallurgy in the age of ChatGPT’, where she discussed the barriers and immense opportunities of using machine learning and AI tools in alloy discovery as well as the disconnect to manufacturing. Reflecting on the event, Professor Christofidou shared, "This event aims to raise the profile of materials science in the UK academic and industrial communities. It’s always an honour to be able to come back to Cambridge, where I did my PhD, and share my research with my peers. This event is also a fantastic platform for early career researchers to present their work and build their networks in the field. Ben, Josh, and Xianyuan all gave really interesting talks - I was amazed how they managed to fit so much content in their allocated three-minute slots!"
The forum provided an excellent platform for ECRs to present their research in concise, three-minute slots. Sheffield had a strong presence, with PhD student Ben Jolly (supervised by Dr. Lewis Owen), Senior AI Research Engineer Dr. Xianyuan Liu (from the Centre for Machine Intelligence), and Postdoctoral Research Associate Dr. Josh Berry (working with Prof. Christofidou) all attending and presenting.
The opportunity to present to a diverse audience was highly valued by all three. Ben highlighted that "the best thing about attending the event was the opportunity to present some of my work to an audience who might not be as familiar with the topic. It forced me to reconsider my research in a wider context, and also generated some really lovely discussions in the networking sessions."
Xianyuan, with his AI background, found it rewarding to present interdisciplinary work at a materials-focused event, explaining that "the best part for me was presenting at a materials-focused forum and seeing how AI is connecting with materials science in meaningful ways." He added that "communicating this within just three minutes was a challenging but rewarding experience," and that the subsequent discussions "deepened my perspective on how AI and materials science can work together, not to replace scientists, but to empower them."
Similarly, Josh appreciated the opportunity to tailor his technical research for a wider audience, stating, "It was a great opportunity to practise tailoring my research for a broader crowd and to gain experience communicating technical work in a more accessible way." All three emphasised the benefits of networking and fostering new connections, with Josh noting, "It was also a great opportunity to network and meet new people."
Presentation abstracts
- Professor Katerina Christofidou: Making it Real: Modern Metallurgy in the Age of ChatGPT
Machine learning and artificial intelligence (AI) are no longer futuristic concepts but pervasive realities shaping our world. Ignoring their potential is no longer an option; it is imperative to strategically harness them. This talk will delve into the applications of Large Language Models for efficient data extraction from the vast landscape of metallurgical literature. Furthermore, we will explore the possibilities of employing machine learning tools to predict critical material properties, paving the way for innovative frameworks in forward material design. Acknowledging the inherent challenges in translating computational predictions to experimental outcomes, particularly in synthesis, this presentation will highlight the existing disconnect between prediction and empirical validation. Addressing this gap requires a focus on the development and implementation of high-throughput experimentation and quantifiable synthesisability and sustainability metrics to ensure the practical realisation of AI-driven metallurgical advancements.
- Ben Jolly: Atomic short-range ordering in multi-principal element alloys: A total-scattering investigation
Ben is a 4th -year PhD candidate in the MOSAIC Group, supervised by Dr Lewis Owen in the School of Chemical, Materials and Biological Engineering.
Multi-principal element alloys—MPEAs—are an emerging class of metallic materials that demonstrate unique, advantageous mechanical properties, frequently out-performing traditional alloying systems across a range of applications. Often, these exceptional properties are linked by researchers to the local ordering behaviour of constituent atoms, though many of these studies are contradictory or reliant on the inference of ordering by indirect characterisation. In this work, we demonstrate the use of total-scattering diffraction as a method for the direct observation of local ordering in chemically complex MPEAs, using a CrCoNi alloy as a case-study. A reverse Monte Carlo (RMC) fitting is employed in the generation of large-box models from diffraction data, allowing for the direct interrogation of atomic ordering. Examinations of the CrCoNi alloy are then performed using a novel Clapp configurational analysis, marking the first time that this method has been successfully applied to higher-order chemical compositions.
- Dr Joshua Berry: Intermetallic suppression in multi-principal element alloys via carbon reinforcement
Josh is a PDRA in the IDEAs group, supervised by Prof. Katerina Christofidou in the School of Chemical, Materials and Biological Engineering.
Our sustainability revolution relies on the development and implementation of novel alloy technologies. Multi-principal element alloys offer immense opportunities to deliver improved performance, sustainability, and cost effectiveness by exploiting a vast compositional design space. However, the prevailing emphasis on single-phase solid solutions contradicts conventional alloy design, where multiple phases are utilised to achieve the desired properties. Simultaneously the formation of brittle secondary intermetallic phases poses a significant challenge, compromising the design of ductile and damage-tolerant materials required in advanced engineering applications. Carbon reinforcement during the solidification of multi-principal element alloys enables powerful phase control by suppressing brittle intermetallic phase formation while promoting desirable carbide phases. The resulting microstructures enhance mechanical properties and align with the demands of additive manufacturing, meeting key requirements such as phase stability, resistance to hot cracking, and solidification control, thereby offering a robust route for scalability.
- Dr Xianyuan Liu: Mining the Literature: Building complex concentrated alloy property databases with large language models
Xianyuan is a Senior AI Research Engineer, with Prof. Haiping Lu. He is part of the Centre for Machine Intelligence and the School of Computer Science.
Complex concentrated alloys (CCAs) comprise multiple principal elements in substantial concentrations. They provide a versatile platform for designing alloys suited to demanding service conditions, crucial for extending infrastructure lifespan. Machine learning can accelerate CCA discovery by exploring vast compositional spaces in silico and reducing reliance on costly experiments. Progress is hindered by the scarcity of large, high-quality experimental datasets. To address this, we use large language models (LLMs) to automatically extract key compositional and property data from the scientific literature. Our approach processes entire PDF documents and generates structured outputs, such as JSON files. We validate our approach with existing CCA datasets and by testing various LLM techniques, including prompt engineering and fine-tuning. This work supports faster data curation and advances the development of high-performance alloys for next-generation Applications.
About the event:
The Armourers and Brasiers' Cambridge Forum is held annually at the Department of Materials Science & Metallurgy in Cambridge with the aim of raising the profile of materials science in the UK academic and industrial communities, while being international in scope. The Forum attracts high-level involvement from industry, research councils and other influential bodies. It incorporates the Kelly Lecture and the Gordon Seminars, inaugurated in 1999 to mark the opening of the Gordon Laboratory in the Department. It is generously supported by the Armourers and Brasiers' Livery Company and a number of other sponsors.