In-Situ Bacterial Detection in Liquids Using Novel Bio-Free and Bio-Hybrid Biosensors

Abstract

The detection of bacteria in liquid is crucial to a wide range of applications, from clinical diagnostics and water safety to food processing and pharmaceutical quality control. While each application involves a different liquid environment, the core sensing mechanisms can remain platform-consistent when carefully designed. This study therefore focuses on developing two adaptable biosensing platforms that can operate effectively in complex liquid samples. To validate performance and showcase real-world utility, urinary tract infection (UTI) diagnosis is chosen as a test scenario, representing a medically relevant and chemically challenging background. The study develops and evaluates two biosensing strategies: a bio-hybrid magnetostrictive particle (MSP) sensor and a novel, bio-free dielectric spectroscopy-based approach, each designed to independently address the challenge of in-situ bacterial detection in liquid environments. In the first project, an optimized MSP biosensor was developed and tested for detecting Escherichia coli (E. coli) in urine. Extensive fabrication refinements were introduced to improve sensor uniformity and resonance stability. These enhancements significantly reduced baseline drift and mechanical noise. The resulting MSP sensors, functionalized with specific antibodies, were integrated into a closed-loop fluidic system. When tested with urine samples spiked with E. coli, the system reliably detected bacterial concentrations as low as 10⁴ CFU/mL within 30 minutes and revealed that slope provided sensitive quantification during early-stage binding. By systematically varying bacterial concentrations, a semi-quantitative measurement protocol was validated and showed clear dose-dependent shifts. These findings validated the MSP platform’s capability for rapid, real-time diagnostics. The second project explored the development of a bio-free, dielectric spectroscopy platform using rod, plate, and interdigitated electrode configurations. Dielectric/electric parameters were measured across multiple media, bacterial concentrations. Electrode material and geometry were shown to influence bacterial response by altering the relative contributions of bulk liquid versus electrochemical double layer (EDL) and modifying the structure and thickness of the EDL itself. These shifts provide strong evidence that geometry and surface chemistry act as tunable filters, selectively amplifying bacterial-induced changes in dielectric spectra and make them suitable as spectral fingerprints for the presence of bacteria. Interdigitated electrodes at microscale added additional complexity, capturing subtle changes in spectral shape through derivative analysis. Further, Direct Current (DC) bias was introduced as a tunable stimulus to modulate the EDL and track dynamic spectral shifts. Bacterial suspensions displayed distinct stabilization behaviors under DC stimulation, particularly in capacitance, offering a new temporal axis for bacterial detection. Across dielectric sensor platforms, machine learning models were trained to recognize spectral patterns linked to bacterial presence. These findings confirm that dielectric responses encode both immediate and time-resolved information that can be exploited for robust, pattern-based classification. Together, the MSP and dielectric systems represent two independent sensing strategies: one leveraging biochemical binding for selective detection, and the other employing bio-free mechanisms based on intrinsic properties. Each system offers unique strengths and constraints, and their development highlights how distinct physical principles can be harnessed to address common diagnostic challenges in complex liquid environments, paving the way for a new generation of bacterial sensors combining hardware engineering, signal modeling, and data-driven intelligence.