Digital Protocols for Statistical Quantification of Microstructures From Microscopy Images of Polycrystalline Nickel-Based Superalloys

This paper presents new digital image analysis protocols and workflows for the quantification of size distributions of key microstructural features (e.g., grain size, γ ’ size) and γ ’ volume fraction in polycrystalline nickel-based superalloys. These digital protocols leverage recently established image analyses algorithms that have been shown to be computationally efficient and scalable. They allow consideration of very large sample sizes (both larger scan sizes and larger number of images), and the extraction of reliable and reproducible microstructure statistics. For grain size analysis, electron backscatter diffraction was used to characterize the polycrystalline microstructure, and angularly resolved chord length distribution (AR-CLD) analysis was performed to reliably quantify the grain size and morphology distribution. For γ ’ size and volume fraction analysis, back-scattered electron imaging in the scanning electron microscopy (BSE-SEM) was used to characterize the microstructure at the sub-grain level, and a robust image segmentation workflow was developed and used to identify the γ ’ pixels in the gray-scale image. AR-CLD analysis was also performed on these segmented images to determine size distributions of the γ ’ precipitates. The digital protocols developed in this work are demonstrated on three different nickel superalloy specimens subjected to very different cooling rates (labeled as slow-cool, moderate-cool, and fast-cool), which resulted in vastly different γ ’ size distributions. It is demonstrated that the effect of the cooling rates on the γ ’ size distributions can be reliably quantified using the digital protocols presented in this study.