CAREER: A Probabilistic Framework for the Nucleation of Recrystallization

Non-Technical Summary

When metals are heated and deformed into engineering components by processes such as forging, it is not just a macroscopic shape change—the properties of the metal itself can be tuned by controlling the local temperatures and deformation rates. These external parameters can be used to control the material’s microscale structure ("microstructure"); however, the relationships are complicated as there are numerous mechanisms simultaneously active.

In this research program, PI Miller uses statistical methods and machine learning to identify the most likely sites in the material for the microstructure changes to initiate. The ability to predict the initiation sites for microstructure changes can help to prevent material failure during service in extreme environments, such as in aircraft engines or powerplants.

It can also be used to optimize materials processing during manufacturing, potentially reducing the cost of metals processing. The core concept of this research—measurement and quantification of the micro-scale structure of materials—has been integrated into education and outreach modules for use from middle school to university. Through partnerships with existing programs at the University of Florida, the PI will distribute kits to middle school teachers.

These programs specifically target districts with a high fraction of historically under-represented groups or a low socioeconomic status, seeking to improve the diversity of students exposed to materials science early in their academic careers. Additionally, digital versions of the modules targeted to different age groups will be broadly disseminated through a relevant professional organization, the International Metallographic Society.

Technical Summary

This research program enables unprecedented control of solid-state microstructural evolution during thermomechanical treatment by introducing a probabilistic framework for the prediction of likely recrystallization sites. This is a key component of PI Miller’s career-long goal of developing a quantitative, mechanism-based understanding of the statistical accumulation of dislocations and deploy this knowledge to define processing paths that result in microstructures with unique properties.

A priori prediction of likely nucleation sites allows for greater fidelity in modeling of recrystallization-related failure mechanisms or the as-recrystallized microstructure. The PI hypothesizes that the likelihood of recrystallization at a given microstructural site can be captured by a "recrystallization indicator parameter (RXIP)", similar to the concept of a fatigue indicator parameter in fatigue and fracture.

This dimensionless parameter describes the probability of nucleation at a given microstructural site by quantifying and weighting the contributions of local features that promote or inhibit nucleation, e.g., curvature, slip system alignment, Schmid factor, misfit, etc. The core concept of this research—measurement and quantification of the microstructure of materials—has been integrated into education and outreach modules for use from middle school to university.

The PI will partner with existing programs at the University of Florida to pilot and iteratively refine physical kits through teacher training programs funded by the Department of Education and the Cornell Lending Library of Experiments. For broader dissemination, digital demonstrations and data-processing only versions of the modules will be distributed as an educational tool through the International Metallographic Society.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.