Machine Learning Integrated CMOS Terahertz Focal Plane Arrays

Prixarc proposes to develop and commercialize novel terahertz (THz) focal-plane array (FPA) that utilizes 3D microstructures, smart readout integrated circuits, and allows efficient integration with processors that incorporate machine learning to increase the data collection efficiency. We plan to collaborate with University of Miami (UM) and Kansas State University (KSU) for this project. We propose to use novel multidirectional ultraviolet (UV) lithography to fabricate 3D periodic microstructures with spatial resolutions of several tens of microns and vertical structures of a few hundred microns across a comparatively large surface area that will enable the development and commercialization of very sensitive THz FPA devices for imaging. Our proposed THz FPA can operate either as a frequency-tunable continuous-wave detector or a broadband-pulsed detector by using appropriately designed 3D microstructures integrated within a microlens on top of the CMOS FPA. The 3D microstructures integrated on top of the CMOS FPA can be designed to operate over a 0.1 – 3 THz frequency range. The 3D microstructures can be tuned to optimally harvest THz radiation energy to be converted to radio frequency (RF) signals that can be converted to DC signal output by the CMOS FPA, and this configuration would offer more than 30 decibel (dB) dynamic range per pixel. Our current prototype design includes 32 x 32 array (1024 pixels) FPA with a frame rate of 10 Hz. Our approach would allow the use of pixel-level, chip-level and/or system-on-chip level advanced signal processing concepts to include machine learning and adaptive signal processing to enable faster-frame rate imaging, event-recognition, edge-detection, and object-recognition capabilities. This SBIR project would thus help to accelerate development of military and DoD-centric systems among other benefits including reducing technical risks and product costs.