High-Fidelity Measurements of Spatter Sizes, Shapes, and Landing Sites in a Laser Powder Bed Fusion Machine

Spatter particles from nickel superalloy 625 were captured and measured during the printing of a 25 mm x 10 mm rectangular pad in commercial laser powder bed fusion (PBF-LB) machine. Particles were captured with a glue-coated sticky plate that covered the entire powder bed, holding any landed particles at their original landing sites for ex-situ measurement in an optical microscope. Large-area images of the sticky plates were processed with a convolutional neural network (CNN) to automatically identify spatter particles. The landing positions, ferret diameters, and circularities of identified particles were then measured using a blob detection algorithm. This dataset includes the original images from the microscope, the labeled dataset for training and testing the CNN, the CNN-segmented images (identifying spatter particles), the final measurements of detected spatter particles, and several relevant scripts and figures to describe the experiment.