Ph.D. Public Defense
High Performance THz Detection in Standard CMOS Technology
Supervised by Professor Zeljko Ignjatovic
Wednesday, April 17, 2019
Computer Studies Building, Room 426
FET-based THz detection has received a great amount of attention over the past two decades due to MOSFETs superior THz response and their easy integration with signal conditioning circuits on a single chip. The efforts originally started with explaining the THz detection mechanism in strongly-inverted High Electron Mobility Transistors (HEMTs) using plasma wave theory. Since then, researchers have tried to generalize the idea to characterize the detection mechanism in Silicon MOSFETs that are operated in sub-threshold regime, where a stronger response compared to the conventionally strongly inverted FET detectors is observed. However, the rectification nature in the sub-threshold region and its associated phenomena, such as noise, have not been fully explored yet.
In this dissertation, we explain the THz detection mechanism in sub-threshold Si MOSFETs by THz-Induced-Potential-Barrier-Lowering (TIBL) and diffusion current over the lowered barrier. By biasing the detector FET in sub-threshold regime using a current source (photo-conductive mode), we designed a THz pixel that improves responsivity and NEP by factors of 900 and 7, respectively, com- pared to the conventional cold-FET (photo-voltaic) detector, at the expense of more power consumption. Based on TIBL detection theory, we proposed source extension as a simple modification to device geometry that improves the THz response up to a factor of two. Adopting a bottom-up approach, we took the pro- posed THz pixel as the building block and designed and fabricated a 10x10 THz camera in 350 nm CMOS technology. We equipped the camera with adjustable gain on-chip Correlated Double Sampling (CDS) amplifiers for DC cancellation and amplification of pixels’ response. Pixel calibration, a major issue related to photo-conductive THz detector that results from intra-pixel process variation, is addressed. An efficient SAR-based algorithm is proposed/implemented to calibrate the pixel array for optimum THz response and increased dynamic range in the output of the amplifiers. For test and THz imaging experiments, a controller board is designed that interfaces with the camera chip through three microcontrollers. The controller board provides control for pixel array calibration, imaging signals and data acquisition through communication with a custom-designed computer interface. An imaging setup consisting of the THz camera, a 200 GHz source and a collimating lens is created. Using the setup, we have successfully taken images of metal objects concealed in a cardboard box.