Imaging by Detection of Infrared Photons Using Arrays of Uncooled Micromechanical Detectors

The objective of this dissertation was to investigate the possibility of uncooled infrared imaging using arrays of optically-probed micromechanical detectors. This approach offered simplified design, improved reliability and lower cost, while attaining the performance approaching that contemporary uncooled imagers. Micromechanical infrared detectors undergo deformation due to the bimetallic effect when they absorb infrared photons. The performance improvements were sought through changes in structural design such as modification and simplification of detector geometry as well as changes in the choice of materials. Detector arrays were designed, fabricated and subsequently integrated into the imaging system and relevant parameters, describing the sensitivity and signal-to-noise ratio, were characterized. The values of these parameters were compared to values published for other uncooled micromechanical detectors and commercial uncooled detectors. Several designs have been investigated. The first design was made of standard materials for this type of detectors - silicon nitride and gold. The design utilized changes in detector geometry such as reduction in size and featured an optical resonant cavity between the detector and the substrate on which arrays were built. This design provided decrease in levels of noise equivalent temperature difference (NETD) to as low as 500 mK. The NETD parameter limits the lowest temperature gradient on the imaged object that can be resolved by the imaging device. The second design used silicon dioxide and aluminum, materials not yet fully investigated. It featured a removed substrate beneath each detector in the array, to allow unobstructed transmission of incoming IR radiation and improve the thermal isolation of the detector. Second design also featured an amorphous silicon layer between silicon dioxide and aluminum layers, to serve as an optical resonant cavity. The NETD levels as low as 120 mK have been achieved. The only difference between the third and the second design was the modification of the geometry to minimize the noise. Successfully obtained thermal images and improved NETD values, approaching those of modern uncooled imagers (20 mK for commercial bolometer-based detectors), confirm the viability of this approach. With further improvements, this approach has a potential of becoming a lowcost alternative for uncooled infrared imaging.