Geometry Driven Machine Learning Guided Fabrication of Uniform Sub-5 µm PLGA Microparticles for Balloon Catheter Drug Delivery

Our manuscript introduces a geometry-driven microfluidic platform that reliably generates stable sub-5 µm PLGA microparticles by integrating a shear-modulating obstruction within a cross-junction device. This obstruction enhances localized shear at the breakup interface, enabling stable droplet pinch-off under controlled conditions when paired with an optimized hydrophilic surface modification of the PDMS channels. Together, these features allow access to an ultrafine particle size regime that conventional microfluidic configurations have struggled to achieve.