Quantitative SEM morphometry for monitoring laser powder-bed-fusion metal-powder feedstock

In laser powder-bed fusion (L-PBF), feedstock reuse is important for production efficiency, but repeated handling, sieving and exposure can alter the particle population entering subsequent builds. Reliable SEM-based morphometry can support additive-manufacturing powder monitoring by quantifying particle size, shape and morphology across reuse cycles, although manual annotation remains slow and operator dependent. This work presents PowderAnnotator, an open-source workflow for SEM analysis of metal AM powders. The workflow combines a released pre-trained Segment Anything Model (SAM) checkpoint, conventional image-processing steps and user-trained classifiers to segment particles, extract morphometric descriptors and assign morphology labels. The contribution is the AM-specific integration, validation and release of these existing models and methods for reused-powder analysis, rather than a new foundation model or classifier architecture. The workflow is demonstrated on gas-atomised Inconel 718 L-PBF feedstock sampled at Reuse 0, 2, 5 and 6. On 23 annotated SEM images containing 1981 particles, enhanced SAM achieved instance F1 = 0.688 at IoU 0.50, while the Random-Forest classifier achieved F1 = 0.78 for spherical particles and 0.17–0.31 for minority morphologies.