Automating microfluidic part verification

Abstract

Microfluidic devices promise to significantly advance the field of medicine by providing diagnostic tools that may be administered at the point-of-care (POC). Currently, a major barrier to entry for POC microfluidic diagnostic technologies is part qualification where specific features of a microfluidic technology must be qualified during the manufacturing process to ensure the device performs properly, e.g., Tantra and van Heeren (Lab Chip 13:2199, 2013). Additionally, microfluidic device research is moving toward therapeutic applications where quality control requirements may be more stringent. In this paper, we use soft embossing—a replication technology—to manufacture high aspect ratio thermoplastic deterministic lateral displacement (DLD) parts, where DLD is a size-based microfluidic particle separation technology based on shift post arrays. Morphological data are collected using optical profilometry, and key metrics are automatically extracted using a novel image analysis algorithm. This algorithm allows us to rapidly quantify the average post height, post shape, array pitch and surface roughness in order to compare a lot of 12 devices to the original design and the deep reactive ion etched silicon master mold. During the run of 12 devices, the post height was 63.0 ± 5.1 μm (mean ± 6σ), the pitch was 35.6 ± 0.31 μm, the post eccentricity was 0.503 ± 0.197, the post circularity was 1.195 ± 0.084 and the root-mean-square surface roughness was 4.19 ± 3.39 μm. Importantly, this automated part qualification technique reduced data analysis time by 75- to 100-fold while allowing for additional qualification metrics and significantly reducing the need for manual measurement. Overall, this paper indicates that it is possible to improve the quality and efficiency of microfluidic part qualification using optical profilometry coupled with a novel image analysis algorithm.