Design and Fabrication of a Tunable Fabry-Perot Interferometer/Photodiode Microspectral Image Sensor

The design and fabrication of a tunable Fabry-Perot interferometer (FPI) åÁV photodiode (PD) spectral image sensor using a complementary metal oxide semiconductor (CMOS) compatible process on silicon is reported in the present work. The system is designed to be integrated on a cellular neuralonlinear network (CNN) visual microprocessor chip which operates at 20,000 frames per second and serve as the image sensor in the visible and near IR wavelength range. A planar interdigitated PIN silicon photodiode was fabricated. The DC photoresponse, frequency response, and spectral response of the photodiode have been characterized. The PD demonstrated sufficient bandwidth (tens of MHz) for the application of this project and appreciable quantum efficiency (>50%) from 400 nm to 900 nm. An FPI with Al mirrors was fabricated using surface micromachining techniques on a silicon substrate. The tuning of the FPI at different bias voltages was demonstrated by measuring the reflectance spectrum of the FPI. The result showed that for an FPI with 2.94 µm original effective cavity length, 10 V of applied bias voltage can tune the FPI over a 0.4 µm change in effective cavity length. The quantum efficiency of an FPI/PD device was predicted to be 22% at the wavelength of 450 nm based on the PD and FPI performance. The concept of making an FPI/Diffraction grating/PD Array microspectrometer was evaluated. Such a microspectrometer is able to provide a better tuning range than the FPI alone and better resolution than the diffraction grating alone. A design example, simulation, and fabrication process flow were also discussed.