Resonating with replicability: factors shaping assay yield and variability in microfluidics-integrated silicon photonic biosensors

The integration of biosensors and microfluidics has facilitated the development of compact analytical devices capable of performing automated and information-rich detection of myriad targets, both in the lab and at the point of need. However, optimization of microfluidics-integrated biosensor systems and replicability challenges present roadblocks in validation and commercialization. Understanding factors contributing to yield and replicability in biosensor performance is key to the development of biosensor optimization frameworks and technology translation beyond the research setting. Hence, for the first time, we present a detailed analysis of factors affecting performance, assay yield, and intra- and inter-assay replicability in microfluidics-integrated silicon photonic (SiP) evanescent-field microring resonator biosensors. Strategies for mitigating bubbles--a major operational hurdle and contributor to instability and variability in microfluidics-integrated biosensors--are analyzed to improve assay yield. Effective bubble mitigation is demonstrated by combining microfluidic device degassing, plasma treatment, and microchannel pre-wetting with surfactant solution. Both intrinsic and analyte-detection performance metrics and their replicability are quantified for the first time for sub-wavelength grating-based SiP biosensors, highlighting a path to further optimization. Lastly, the effects of sensor functionalization on analyte detection performance and replicability are evaluated. We compare polydopamine-vs. Protein A-mediated bioreceptor immobilization chemistries and spotting-vs. flow-mediated bioreceptor patterning approaches. We find that simple polydopamine-mediated, spotting-based functionalization improves spike protein (1 {micro}g/mL) detection signal by 8.2x and 5.8x compared to polydopamine/flow and Protein A/flow approaches, respectively, and yields an inter-assay coefficient of variability below the standard 20% threshold for immunoassay validation. Overall, this work proposes a practical framework through which evanescent-field SiP biosensors, and microfluidics-integrated biosensors more generally, can be characterized, compared, and optimized to facilitate efficient biosensor development.