Simulations and analysis of fluorescence effects in semiconductor x-ray and gamma-ray detectors

Photon-counting semiconductor detectors are a key technology for reducing dose in clinical x-ray imaging procedures, such as CT, and improving performance in gamma-ray imaging procedures such as SPECT. These detectors offer excellent energy resolution and high spatial resolution. To stop high-energy photons, high-Z semiconductors must be used, such as CdTe, TlBr or other emerging candidates. These crystals often suffer from poor hole transport due to hole trapping, which can greatly affect signal, even when data is primarily collected from anodes. There are many interesting challenges in the production of these detectors as well as in developing complete quantitative models of the photon-matter interaction, charge transport, and signal induction. Prior work in our group has focused on optimal ways to estimate photon interaction parameters (x,y,z) and energy (E). This work is based on statistical models and calibration data. In recent work we are exploring a method to account for k x-ray fluorescence and to model signals induced on a double-sided strip detector. Our approach is Monte-Carlo sampling of interaction details, followed by charge transport and signal induction modeling via weighting potentials. First our simulation creates first and second order statistics for three charge induction cases: simple transport, charge sharing, and x-ray fluorescence. Using mean signals and covariance matrices from these cases we build a likelihood that can be used with maximum likelihood methods to estimate the primary interaction location and classify whether the event's energy deposition involved fluorescence. In planned work we will test the model against experimental semiconductor detector data.