Dynamic Characterization of a Fast-Responding Nanophotonic Gas Sensor Using Optimization-Based System Identification

This article develops and parameterizes a dynamic model of a novel nanophotonic gas sensor. The sensor employs a functionalized microring resonator to measure CO2 concentrations in biomedical applications. The literature presents both computational and experimental approaches for characterizing nanophotonic sensor response times. However, it can be challenging to distinguish between the dynamics of a fast-responding sensor versus the dynamics of the setup used for characterizing it. We address this challenge using optimization-based system identification. Specifically, we construct a test rig that supplies mixed N2/CO2 flow to the sensor and measures the sensor’s voltage amplitude response at a given laser excitation wavelength. Step response experiments are conducted using this test rig, at different gas flow rates and concentrations. A state-space model is then constructed, capturing the sensor’s first-order dynamics as well as the gas transport, manifold filling, and first-order valve actuator dynamics of the test setup. A particle swarm optimizer is used for least-squares model parameterization. The resulting residuals are reasonable in magnitude, two observations being that their magnitude changes with CO2 concentration and that they exhibit some coloring, potentially due to the setup’s signal processing filters. The estimated time constant has reasonable Cramér-Rao bounds and is close in magnitude to finite-element predictions.