Spherical feedstock powders with controlled particle sizes offer significant advantages for many technologies, including printable inks, enabling higher solids loadings and enhanced resolution in additively manufactured (AM) items. One material-agnostic method to produce such powders is emulsion-assisted milling. The characteristics and quality of the produced powders are affected by the milling process parameters. Understanding these effects is challenging as they are governed by multiple powder descriptors and many process parameters. Here, this challenge is addressed by processing and analyzing large sets of electron microscope images of the powder particles. StarDist, a deep-learning model plugin in FIJI, served to extract diverse shape and size descriptors for individual particles. Three powders of interest to the field of energetic materials: Al·Fe 2 O 3 , Al·PVDF, and elemental boron were produced with variable process parameters. Integrated descriptors of the powder quality (figures of merit) were developed for each case based on the individual descriptors obtained from the image analysis. These figures of merit were correlated with the milling process parameters. The findings identified a generalized process parameter χ, depending on the milling time, ball-to-powder mass ratio, volume ratio of the droplet to continuous phase in the emulsion, rate of the milling vial rotation, and the vial volume, that captures the correlations between milling conditions and product quality. Such process-property correlations will guide the predictive design and manufacturing of feedstock powders for many powder technology applications. • Different powders were produced via emulsion-assisted milling with varied process parameters. • Particle size and shape descriptors were obtained from AI-based processing of SEM images. • Figures of merit developed to characterize the powder quality using the statistical analysis of multiple particle descriptors. • A generalized process parameter valid for different materials combining multiple individual parameters was introduced. • Figures of merit show maxima for the same range of the generalized process parameter for different materials.
Ozsoysal et al. (Wed,) studied this question.