Abstract
In applications such as atmospheric monitoring of greenhouse gases and pollutants, the detection and identification of trace concentrations of harmful gases is commonly achieved using non-dispersive infrared (NDIR) sensors. These devices employ a broadband infrared emitter, thermopile detector, and a spectrally selective bandpass filter tuned to the vibrational resonance of the target analyte. However, the fabrication of these filters is costly and limited to a single frequency. This limitation introduces a fundamental tradeoff, as broadening the optical passband width enhances sensitivity but compromises selectivity, whereas narrowing improves selectivity at the expense of sensitivity. In this work, we validate a filterless NDIR approach using a multi-peak thermal emitter developed through inverse design. This emitter enhances detection sensitivity by targeting multiple absorption bands, demonstrated through the creation of a sensor designed for the C-H vibrational modes of propane. Additionally, a set of single-peak emitters were developed to showcase the capability of designing highly selective sensors operating within close spectral proximity. These emitters, targeting the stretching modes of carbon monoxide and carbon dioxide, exhibit Q-factors above 50 and minimal crosstalk, enabling accurate detection of the target gas without interference from gases with spectrally adjacent absorption bands. This is enabled by the implementation of an aperiodic distributed Bragg reflectors, which allows for higher Q-factors with fewer layers than a periodic Bragg reflector using the same materials and number of layers, thereby reducing fabrication complexity and cost. Experimental results validate that this approach breaks the tradeoff between sensitivity and selectivity. This work highlights the potential of optimized thermal emitters for more efficient and compact gas sensing applications.