An image processing pipeline for in-situ dynamic X-ray imaging of directional solidification of metal alloys in thin cells

TL;DR

This paper introduces an open-source image processing pipeline for analyzing in-situ X-ray images of metal alloy solidification, enabling detailed morphological, grain, and convection analysis even with low-quality images.

Contribution

The authors developed a robust, open-source image processing algorithm that automates the analysis of solidification structures and fluid dynamics in X-ray imaging of metal alloys.

Findings

Successfully applied to gallium-indium alloy experiments

Extracted grain orientation and convection data

Demonstrated reliability with low-quality images

Abstract

We present an image processing algorithm developed for quantitative analysis of directional solidification of metal alloys in thin cells using X-ray imaging. Our methodology allows to identify the fluid volume, fluid channels and cavities, and to separate them from the solidified structures. It also allows morphological analysis within the solid fraction, including automatic decomposition into dominant grains by orientation and connectivity. In addition, the interplay between solidification and convection can be studied by characterizing convection plumes in the fluid, and solute concentrations above the developing solidification front. The image filters used enable the developed code (open-source) to work reliably even for single images with low signal-to-noise ratio, low contrast-to-noise ratio, and low image resolution. This is demonstrated by applying the code to several dynamic in situ X-ray imaging experiments with a solidifying gallium-indium alloy in a thin cell. Grain (and global) dendrite orientation statistics, convective plume parameterization, etc. can be obtained from the code output. The limitations of the presented approach are also explained.