MEMS çubuk tabanlı algılayıcılar için optik okuma yöntemleri

Optical readout methods are attractive for micro-electro-mechanical system (MEMS) based sensors as they allow remote readout without electrical connections to the sensor chip and scalable architectures for sensor arrays. Optical readout noise is often the limiting factor in sensor applications. This thesis presents different optical readout methods that can improve the detection sensitivity and decrease the optical readout noise for MEMS sensors specifically for infrared (IR) detection and MEMS oscillator applications. MEMS based thermo-mechanical IR detection technology with optical readout can play a significant role for wide adaption of thermal detectors due to its low cost. In order to achieve sensitivities near the thermal noise limit, optical readout noise need to be reduced dramatically. For this purpose, firstly we demonstrated sensitivity improvements using a number of single sensor optical readout methods. Secondly, we developed both integrated and free-space based optical readout architectures for sensor arrays. Interferometric and optical beam deflection methods have been explored for a variety of single MEMS IR detectors. We developed AC-coupled detection methods to reduce the DC noise and increase the sensitivity to detect the thermal or spatial changes in the scene. Detailed noise characterizations of thermo-mechanical MEMS detectors with 35 µm pixels were performed. In addition, a single MEMS pixel with Fabry-Perot cavity type optical readout method was designed and fabricated. Lastly a novel prism-based optical-readout is proposed and demonstrated for a single lever MEMS pixel. The noise equivalent temperature differences (NETD) for different sensor designs were measured below 200 mK, with a best NETD performance of 150 mK. The measured noise levels were comparable to the state-of-the-art thermo-mechanical IR sensors for very small pixels. In order to improve the performance of the thermo-mechanical MEMS detector arrays, we developed two new approaches: i-) Compact optical Fourier filtering system. A convergent illumination system was developed, which reduced the size of the system and improved the SNR compared to the conventional 4f optical system. ii-) Two-wavelength based optical readout system. This method provided spatial auto-registration of two different color images on a single RGB camera. The sensitivity and dynamic operation range can be enhanced significantly by using two wavelengths for a non-uniform array. Integration of MEMS devices with CMOS electronics enables large array operation in a small low-cost package. MEMS-based sensor array with a large number of elements (64x64) is bonded at chip level with CMOS readout IC for the first time. A diffraction grating interferometer-based optical readout is realized by pixel-level illumination of the MEMS chip through the through-silicon via holes and by capturing the reflected light using a photodetector array on the CMOS chip. Lastly, two non-linear optical readout methods were developed for biological and chemical sensor arrays using dynamic MEMS cantilevers. A single controller is desired to monitor the entire array for tightly packed high density sensor arrays. A separate saturation mechanism and nonlinearity is required for each oscillating cantilever sensor. We used optical non-linearities to drive and sense multiple oscillators with a single controller. In the first method, MEMS cantilevers with embedded diffraction gratings were used in order to parallelize the array for multiple oscillations. In the second method, a prism-based optical readout method was proposed for simple cantilevers. Both methods were successfully implemented to actuate and monitor two nearly identical cantilevers with one actuation coil and photodetector.