An advanced microstructured semiconductor radiation detector for neutron imaging and oil well logging

Abstract

The work presented in this dissertation will serve as an essential tool for modern nuclear engineers to perform radiation transport with import of complex computer aided design (CAD) models efficiently. Performing radiation transport with complex CAD geometries and obtaining accurate detector response functions has been a major challenge. For micro-structured semiconductor detectors, results obtained from radiation transport alone are not enough to provide a clear picture of expected performance. Hence, the work presented in this dissertation is divided into two major research areas (RAs). RA 1 describes methods for sensor response modelling with hybrid Geant4 geometries consisting of both conventional C++ based models and complex CAD based models in a single simulation. This RA will address the methods developed for radiation transport models and benchmarking experimental detector responses. Additionally, this RA will further discuss how these models were later utilized for finalizing design parameters of a semiconductor radiation detection system (Timepix3 coupled with Dual-Sided Microstructured Semiconductor Neutron Detector). Multiple bugs were identified in Geant4 source code as a result of the work presented in RA 1. RA 2 of this dissertation will provide detailed analysis of the existing boundary conditions of the DS-MSND and MSND detectors. Finite Element Analysis based semiconductor physics modeling was performed to investigate various parameters of the Si-SiO₂ interface. The radiation transport models were developed for simulating the Kansas State University Materials Interrogation (KSUMI) test facility that was set up to enable bulk-material irradiation experiments that replicate similar oil-well logging scenarios. These experiments were conducted with an aim to address the problem of replacement of conventional radioisotope sources commonly used in oil-well logging industries. An exploration tool similar to an oil-well logging tool was used to conduct experiments with water and sand as testing materials. The facility includes a 2500-gallon concrete test chamber to be filled with testing material with an aluminum pipe going horizontally through it, permitting placement of a neutron source and radiation sensors. A machine-based 14.1 MeV deuterium-tritium neutron source was used to irradiate the 2500-gallon testing material. Experiments were performed with tap water and sand as bulk testing materials. Irradiation was performed for one hour and results were obtained from a BF₃/³He neutron sensor, a BF₃ neutron sensor, and two NaI gamma sensors placed at different locations within the exploration tool. Geant4, a Monte-Carlo based toolkit, was deployed on a high-performance computing system to simulate the entire experiment in order to benchmark the experimental responses obtained from the photon and neutron sensors. The facility was modeled in detail with accurate dimensions and material compositions. Materials such as tap water, high-density polyethylene, and aluminum metal were modeled with thermal neutron scattering cross-sections. The reference physics list QGSP_BIC_HP along with G4NDL and S(α,β) cross-sections were found to be appropriate for simulation of neutron interrogation experiments with neutron energies lower than 20 MeV. The experimental results obtained were successful in characterizing the bulk testing materials, and results obtained from Geant4 were found to be in good agreement with the experimental results in most cases. The source code developed for the above method was then utilized to design the physical parameters for an X-DS-MSND system. These detectors have the benefit of doubling neutron detection efficiency as compared to single-sided devices (X-MSND) by staggering ⁶LiF-filled trenches between the top and bottom surface of a silicon diode. This produces a more complex electric field distribution and depletion characteristics in the diode and creates an indirect path for signal carrier transport between device electrodes. Geant4 was extensively utilized for radiation transport and interaction modeling, and Allpix Squared was used for mobile charge carrier transport and total charge collection. The results of this simulation work provided an estimate of charge cluster shape and intensity for a pixel array configuration corresponding to Timepix3 read-out system. In addition, imaging performance for transmission radiography was demonstrated with a simple two-dimensional shape in a Gaussian beam of thermal neutrons. The simulated neutron detection efficiency was 57.6%. It was discovered that the Allpix Squared is not optimal for simulation of complex boundary conditions and charge carrier transport inside the DS-MSND. To understand the behavior of charge carriers in the presence of a complex electric field and boundary conditions, a simpler non-conformally doped MSND geometry was fabricated and analyzed. This sensor geometry produced a complex electric field distribution and depletion characteristics within the diode. The fixed oxide charge present at the trench walls and its effects on the depletion characteristics and electric field were analyzed. The capacitance-voltage and current-voltage curve characterizations were performed for these prototypes and compared with the COMSOL Multiphysics simulations. An ²⁴¹Am alpha particle source was further utilized to perform spectral analysis. Geant4 was utilized for radiation transport, interaction modeling, and benchmarking the spectral data. The results of this simulation work provided a reasonable confidence in capability to obtain and benchmark electric fields and spectral data for complex micro-structured semiconductor radiation detectors. Further, COMSOL Multiphysics was used to account for time-dependent charge carrier motion and detector response for these boundary conditions. As a result of work presented in RA 1 and RA 2, important features in the MSND/DS-MSND geometry were discovered that could significantly impact the device's performance. These features were incorporated in the updated Monte-Carlo model of the DS-MSND sensor. Simulated response of the updated DS-MSND sensor as a replacement of BF₃/³He was then analyzed for the oil-well logging irradiation environment using Geant4 simulations. The results obtained from these simulations establish DS-MSND as a strong contender for replacing BF₃/³He sensors for oil well logging experiments.