Event

Abstract: Field-effect transistor (FET) biosensors promise label-free, electronic, and CMOS-scalable detection of biochemical signals, but their translation to continuous, multiplexed, point-of-care monitoring has been limited by sensor drift, few-device architectures that lack statistical noise redundancy, and polymer packaging that absorbs and leaches interfering species. This thesis develops and validates a scalable CMOS-integrated ion-sensitive field-effect transistor (ISFET) platform for continuous, label-free, multiplexed profiling of inflammatory cytokines and apoptotic biomarkers, advancing the state of the art across physical modeling, surface chemistry, wafer-scale fabrication, and drift-compensated circuit design.

First, the thesis develops a coupled electrolyte–semiconductor model of the ISFET that combines site-binding surface chemistry, Gouy–Chapman–Stern double-layer electrostatics, and compact transistor behavior. Validated against commercial devices, the model analyzes temporal and thermal drift and guides the selection of gate-dielectric material and thickness, establishing silicon nitride as the sensing membrane used throughout the platform.

Second, the thesis establishes a reproducible amine-coupling surface-functionalization protocol on silicon-nitride surfaces and demonstrates it on Annexin V–based detection of phosphatidylserine, an early marker of apoptosis, achieving near-Nernstian sensitivities and nanomolar detection limits with independent microscopy validation.

Third, the thesis introduces a wafer-scale post-processing flow that makes foundry CMOS compatible with electrolyte operation, and integrates a 128×128 (16,384-pixel) CMOS ISFET array with a low-sorption thermoplastic-elastomer (FlexDym) microfluidic overlay. Whole-well spatial averaging and on-array reference channels enable simultaneous, label-free detection of interleukin-6 and interleukin-10 with a sub-picomolar detection limit.

Finally, the thesis presents a drift- and noise-compensated ISFET/REFET readout chip in a 180 nm CMOS process that combines analog current subtraction, in-pixel chopping, and a current-starved ring-oscillator counter with an I²C interface. Subtraction rejects common-mode drift and temperature while chopping suppresses 1/f noise, lowering the achievable detection limit beyond what either technique provides alone. Together, these contributions deliver a validated, calibrated platform for rapid, bedside biomarker detection, with applications in monitoring cytokine release syndrome and sepsis and in real-time profiling of apoptotic and inflammatory pathways.