“Solidification of metallic alloys: 4D X-ray tomography and machine learning”
When a metal is cooled from the liquid during castings or additive manufacturing, the solid phase evolves via the growth and coarsening of dendrites. These tree-like microstructures exhibit diverse morphologies and have a strong impact on the physical, chemical, and mechanical properties of the resulting macroscale material. In-situ observation of dendrite growth is possible using high temporal and spatial resolution 4D synchrotron x-ray computed tomography (XCT). Solidification events were imaged at Argonne National Laboratory, then reconstructed using the novel TIMBIR algorithm, and segmented using an optimized convolutional neural network machine learning approach. This study showcases the first in-situ observation of free-growing “hyperbranched” dendrites in aluminum-zinc alloys. By combining interface energy anisotropy calculations with the XCT experiments, the full evolution of the crystallographic orientations of the entire dendritic structure is obtained. The dendrites have four-fold symmetry with primary arms growing in directions that are 10 degrees offset from <100>. Surprisingly, hyperbranched dendrites exhibit self-similarity during growth. These findings provide insights into the interface energy anisotropy and further the fundamental understanding of solidification in metallic systems.