Advancing Nanoplasmonics-enabled Regenerative Spatiotemporal Pathogen Monitoring at Bio-interfaces

Non-invasive and continuous spatiotemporal pathogen monitoring at biological interfaces (e.g., human tissue) holds promise for transformative applications in personalized healthcare (e.g., wound infection monitoring) and environmental surveillance (e.g., airborne virus surveillance). Despite notable progress, current receptor-based biosensors encounter inherent limitations, including inadequate long-term performance, restricted spatial resolutions and length scales, and challenges in obtaining multianalyte information. Surface-enhanced Raman spectroscopy (SERS) has emerged as a robust analytical method, merging the molecular specificity of Raman spectroscopy's vibrational fingerprinting with the enhanced detection sensitivity from strong light-matter interaction in plasmonic nanostructures. As a receptor-free and noninvasive detection tool capable of capturing multianalyte chemical information, SERS holds the potential to actualize bio-interfaced spatiotemporal pathogen monitoring. Nonetheless, several challenges must be addressed before practical adoption, including the development of plasmonic bio-interfaces, sensitive capture of multianalyte information from pathogens, regeneration of nanogap hotspots for long-term sensing, and extraction of meaningful information from spatiotemporal SERS datasets. This dissertation tackles these fundamental challenges. Plasmonic bio-interfaces were created using innovative nanoimprint lithography-based scalable nanofabrication methods for reliable bio-interfaced spatiotemporal measurements. These plasmonic bio-interfaces feature sensitive, dense, and uniformly distributed plasmonic transducers (e.g., plasmonic nano dome arrays, optically-coupled plasmonic nanodome and nanohole arrays, self-assembled nanoparticle micro patches) on ultra-flexible and porous platforms (e.g., biomimetic polymeric meshes, textiles). Using these plasmonic bio-interfaces, advancements were made in SERS signal transduction, machine-learning-enabled data analysis, and sensor regeneration. Large-area multianalyte spatiotemporal monitoring of bacterial biofilm components and pH was demonstrated in in-vitro biofilm models, crucial for wound biofilm diagnostics. Additionally, novel approaches for sensitive virus detection were introduced, including monitoring spectral changes during viral infection in living biofilms and direct detection of decomposed viral components. Spatiotemporal SERS datasets were analyzed using unsupervised machine-learning methods to extract biologically relevant spatiotemporal information and supervised machine-learning tools to classify and predict biological outcomes. Finally, a sensor regeneration method based on plasmon-induced nanocavitation was developed to enable long-term continuous detection in protein-rich backgrounds. Through continuous implementation of spatiotemporal SERS signal transduction, machine-learning-enabled data analysis, and sensor regeneration in a closed loop, our solution has the potential to enable spatiotemporal pathogen monitoring at the bio-interface.