Inverse methods to gradient etch three-dimensional features with prescribed topographies using abrasive jet micro-machining: Part II—Verification with micro-machining experiments

•This paper experimentally validates the methodologies of accompanying paper.•Inverse methods were used to machine microchannels with prescribed centerline depth.•Pockets with prescribed textures in two directions were presented.•Micro-channels with prescribed W-shaped cross sections were machined.•The effect of process parameters on accuracy of machined features was discussed.

Cross sectional shape and centerline waviness along the length of a micro-channel can affect different characteristics of microfluidic flow, including heat transfer, pressure distribution and dissipation, and separation of flow. Current existing technologies that allow such micro-machining have many limitations. The accompanying paper presented inverse techniques that can be used to predict the required non-uniform velocity to gradient etch, using abrasive jet micro-machining (AJM), micro-channels and pockets with a wide variety of prescribed textures and cross-sectional shapes. Methods to predict the final three-dimensional (3D) profiles of such features were also presented. In this paper, the velocity functions predicted using the inverse methods were used to machine micro-channels with prescribed centerline depths that varied linearly, parabolically and sinusoidally, pockets with prescribed textures in two directions, and micro-channels with prescribed W-shaped cross sections. Two different erosive efficacy sources were used, one resulting from an adjustable shadow mask and one using a maskless technique. The inverse and 3D shape prediction techniques were verified by comparing the measured feature topographies with those that were initially prescribed. The effect of process parameters such as source shape and the machined feature size on the accuracy of machined features and predictions of the models were also discussed. Overall, the inverse techniques were found to be very effective for predicting the process parameters required to machine a wide variety of desired micro-channel and pocket topographies.