Monte Carlo Simulations of Microchannel Plate–Based, Time-Gated X-ray Imagers

Abstract

In the last two decades, high-speed, time-gated Microchannel plate (MCP) x-ray detectors have proven to be powerful diagnostic tools for two-dimensional, time-resolved imaging and time-resolved x-ray spectroscopy in the field of laser-driven inertial confinement fusion and fast Z-pinch experiments (McCarville et al., 2005; Oertel et al., 2006; Bailey et al., 2004). These detectors’ quantitative measurements are critical for a comprehensive understanding of the experimental results. To assist their characterizations and to aid design improvements, a more comprehensive Monte Carlo simulation model for the MCP detector is needed. The MCP detectors are widely used as electron, ion, and x-ray detectors, as well as imaging tools in many areas of scientific research. Their principles of operation have been documented in the literature (Wiza, 1979; Fraser et al., 1982; Fraser et al., 1984; Kilkenny, 1991) as have extensive research efforts to characterize detection sensitivity (Ze et al., 1999; Landen et al., 1993; Landen et al., 1994), angular and energy dependences (Hirata et al., 1992; Landen et al., 2001; Rochau et al., 2006), and temporal and spatial resolution (Robey et al., 1997). In many previous studies, a discrete dynode gain model was used to describe the MCP gain dependence on the applied voltage (Eberhardt, 1979). This dependence is extremely nonlinear. The discrete dynode model assumes that the MCP can be treated as a conventional, discrete-stage electron multiplier with a fixed number of stages. This gain model uses a few free parameters, chosen to best fit a certain MCP’s data. The discrete dynode model seems to work well to describe the behavior of MCPs under some circumstances, but several factors are not included when inferring the secondary electron yield from gain measurements. For example, the discrete dynode model assumes that the number of dynode stages is independent of the applied voltage on the MCP (the number of stages is chosen to best fit the gain vs. voltage data), which is unlikely to be valid. In addition to the discrete dynode model, previous researchers have also performed Monte Carlo–based computer simulations of the MCP response to a steady-state voltage for straight and tilted microchannels (Guest, 1971; Ito et al., 1984; Choi & Kim, 2000; Price & Fraser, 2001). These simulations apparently did not include the constraint of energy conservation for the secondary electrons. This constraint prevents the aggregate energy of the emitted electrons from exceeding the primary electron energy. Furthermore, these previous efforts omit the effects of low-energy electrons’ elastic scattering from the channel walls, potentially an important effect for the low-impact energies prevalent in an MCP electron cascade. A further difficulty encountered by all such previous efforts (and, indeed,