Automated high-throughput data collection and analysis of composition-process-structure relations by utilizing casting segregation in a Ni-base superalloy

We attempt to address the problem of generating large, reliable experimental datasets that interrelate composition, processing and microstructure in a short timeframe by combining a heterogeneous sample with a gradient heat treatment. Composition-Process-Structure datasets are essential for both physical and purely statistical models aiming to predict material behavior. Using the segregated microstructure of a cast single-crystal sample of Ni-base superalloy MGA1400 in combination with a temperature gradient furnace, we cover a whole range of heat treatment temperatures and compositions according to elemental partitioning. Data collection uses automatic, correlative FE-SEM and EPMA analysis for microstructure images with corresponding local concentration. Image analysis combined with precipitate shape fitting using superellipses allows evaluation of phase fractions, precipitate size/shape/aspect distributions as well as inter-precipitate distances and relative angles for investigating clustering and alignment phenomena. As a result, we were able to generate a pseudobinary phase diagram indicating the γ′ solvus. The activation energy for coarsening of γ′ was found to be mostly independent of as-segregated composition. Further, results showed lattice misfit between the γ and γ′ phase to cause precipitate shape to vary with treatment temperature, and precipitate aspect ratio to vary between the dendrite core and interdendrite regions, indicating dendritic stresses from solidification.