Optimization of Signal-to-Noise Ratio in Semiconductor Sensors via On-Chip Signal Amplification and Interface-Induced Noise Suppression.

Abstract

Radiation detectors are now used in a large variety of fields in science and technology, and the number of applications is growing continually. This thesis presents the development of a wide band-gap solid state photomultiplier (SSPM) and the performance improvement of Si radiation detector with respect to noise suppression and resolution enhancement. Recently developed advanced scintillators, which have the ability to distinguish gamma-ray interaction events from those that accompany neutron impact, require improved quantum efficiency in the blue or near UV region of the spectrum. We utilize AlGaAs photodiode elements as components in a wide band-gap SSPM as a lower-cost, lower logistical burden and higher quantum efficiency replacement for the photomultiplier tube (PMT). We demonstrate that the diodes are responsive to blue and near UV in both linear and breakdown modes with robust electrical characteristics, which includes the leakage current and the onset of breakdown against geometric alteration in the diode design. For semiconductor direct-conversion radiation detectors, we investigated the performance enhancement of the detector via the suppression of noise induced from the semiconductor interface and the resolution improvement with on-chip amplification. The properties of the phonon-based noise are studied and methods to quench the charge mobility fluctuation via surface control, evaluating acoustic reflectance at the semiconductor metal interface by calculating reflectance coefficient via the roaming phonon microgradient (RPMG) model. Si radiation detectors are fabricated and the hypotheses evaluated with different geometries and metal types. In addition to the noise suppression, we also sought to increase the device signal by integrating an amplifying junction as part of the detector topology so that the SNR could be maximized. From this research, we demonstrated the feasibility of improving the energy resolution relative to those low-noise designs that don’t possess on-chip amplification by modeling, fabricating, and characterizing proof-of-concept planar and partitioned detectors. From the fabricated detectors, a semi-empirical result shows that the energy resolution for 81 keV gamma-rays can be reduced from 2.12 % to 0.96 % (for a k = 0.2) with a gain of ~8, which shows the best SNR optimization from our modeling.