Materials informatics study on magnesium based alloy design optimisation: a grain-resolved statistical study on recrystallisation and grain growth

Abstract

Magnesium (Mg) and its alloys are the lightest structural metal in common engineering applications, offering substantial weight reductions compared with aluminium and steel. Therefore, they are attractive for lightweight design in sectors such as automotive and aerospace. However, wider adoption remains limited by relatively unsatisfying mechanical properties, which makes the improvement of Mg alloys important for reducing vehicle mass and associated carbon emission. Most of the strengthening strategies are related to thermal mechanical processing, where plastic deformation and subsequent annealing will determine the final microstructure. During annealing, two microstructural evolution progresses—recrystallisation and grain growth, together decide the final crystallographic texture and grain size, which are two critical microstructural features for Mg alloys. Therefore, a mechanistic understanding of recrystallisation and grain growth is essential for optimising heat treatment parameters and achieving improved properties.
Nevertheless, despite decades of efforts, the mechanisms controlling recrystallisation and grain growth in Mg alloys remain disputed. For example, the origin of rare‑earth texture has been variously attributed to solute drag and particle pinning, preferential nucleation and oriented growth, etc. In practice, these factors coexist and interact during annealing, making it difficult to separate an individual factor for investigation. Despite recent development in advance characterisation, the ability of efficiently quantifying microstructural evolution to perform factor resolved study is still in absence. Accordingly, the key gap is grain‑resolved informatics in time‑resolved datasets, i.e. tracking of individual grains throughout large datasets and synthesise statistical outcome.

Therefore, the aim of this thesis is to develop deliciated grains tracking method and combine with recent advancements in two-dimensional (2D) and three-dimensional (3D) characterisations to revisit long-standing questions, within the annealing related microstructural evolution of Mg. The outcome contains well-structured, time-resolved dataset with tracked information of each individual grain, which not only clarify debated questions in physical metallurgy, but also accelerated materials informatics approaches for the optimisation of Mg alloys design.
For recrystallisation, quasi-in-situ electron backscatter diffraction (EBSD) was used to capture grain-scale evolution during annealing. Three datasets were collected from two alloys: Mg–3Al–1Zn (AZ31, one dataset) and Mg–2.4Zn–0.2Ce (wt.%) (ZE20, two datasets). To enable statistically robust analysis, an automated grain tracking toolbox, Track-Rex, was developed to track individual grains across large quasi-in-situ datasets. Tracking was performed for more than 40,000 grains within approximately ten minutes and outputs structured datasets are suitable for straight forward analysis. Application of Track-Rex clarifies several debated aspects of recrystallisation, including growth and shrinkage behaviour of recrystallised grains, texture evolution in rare earth containing alloy systems, and the contribution of preferential nucleation sites.

For grain growth, quasi-in-situ laboratory-based diffraction contrast tomography (LabDCT) was employed to investigate microstructural evolution in four-dimensional space (3D, 3D + time). Six interrupted annealing steps under argon flow was performed on a WE43 alloy, an effective scanning protocol and reconstruction procedure was established to yield a large, trackable 4D grain growth evolution dataset, across a 4 mm-hight, 0.8 mm-diameter rod. Track-4DGG was built for with capability of tracking grains individually, as clusters, or for user-defined subsets. The resulting statistical datasets enabled analysis of 4D growth kinetics and the behaviours of shrinking, normal growth, and abnormal growth, including a correlation with spatial position.

Overall, this work establishes practical interrupted annealing and characterisation workflows for generating large microstructural evolution datasets and delivers automated tracking toolboxes that convert such datasets into structured, analysis-ready formats. The combined experimental and informatics framework advances understanding of recrystallisation and grain growth in Mg alloys, while providing a scalable route towards predictive modelling and data-driven processing/design optimisation.

More information

Identifiers

ORCID for Dikai Guan: orcid.org/0000-0002-3953-2878

ORCID for Haiming Liu: orcid.org/0000-0002-0390-3657

Catalogue record

Export record

Altmetrics