Progress towards a microchannel plate detector with AlGaN photocathode and cross-strip anode for ultraviolet astronomy

TL;DR

The paper tackles the need for high-quantum-efficiency UV detectors for astronomy by pursuing Al$_{x}$Ga$_{1-x}$N photocathodes coated directly on MCPs, enabling NEA surfaces for efficient photoemission in the FUV/EUV. It combines materials development—growth of $Al_{x}Ga_{1-x}N$ on MgO with controlled Al content and identification of a cubic gamma phase—with an FPGA-implemented, non-iterative centroiding algorithm for a coplanar cross-strip anode, aiming for high throughput and low power. Key results include evidence of cubic gamma stabilization at high Al content, bandgap measurements, and a non-iterative centroiding method that nearly matches Gaussian fitting in software and shows promising FPGA performance with planned optimizations. The work advances compact, low-power UV detector architectures suitable for missions like LyUV, LyRIC, SIRIUS, and Habitable Worlds Observatory by delivering both improved photocathode materials and efficient, FPGA-based readout.

Abstract

Microchannel plates (MCPs) were the driving detector technology for ultraviolet (UV) astronomy over many years, and still today MCP-based detectors are the baseline for several planned UV instruments. The development of advanced MCP detectors is ongoing and pursues the major goals of maximizing sensitivity, resolution, and lifetime, while at the same time decreasing weight, volume, and power consumption. Development efforts for an MCP-based detector system for the UV are running at IAAT at the University of Tübingen. In this publication, we present our latest results towards coating aluminum gallium nitride (AlGaN) photocathodes directly on MCPs, to improve quantum detection efficiency in the far- and extreme-UV. Furthermore, we report on the implementation of a non-iterative centroiding algorithm for our coplanar cross-strip anode directly in an FPGA.

Figures (8)

• Figure 1: Sketch of the MCP detector principle for the UV: The window (1) is optional and only applicable for wavelengths larger than about 118 nm. Incoming UV photons hit a large-bandgap photocathode and release photoelectrons (2). These photoelectrons are accelerated in the channels of an MCP stack (3) and generate a charge cloud. The position of the charge cloud exiting the MCP stack is determined with a position-sensitive anode (4).
• Figure 2: Plot of the lattice constants of MgO substrate and Al$_{x}$Ga$_{1-x}$N films for different Al fractions $x$. The lowest lattice mismatch was found at $x=0.25$. For small values of $x$ a strong dependence of the measured bandgap on the Al fraction is visible while for $x>0.5$ the bandgap is almost stable.
• Figure 3: Photographs of the IAAT coplanar cross-strip anode produced in an LTCC process. Left: Complete anode with gold electrodes on a blue LTCC substrate. The diameter of the substrate is about 8.8 cm. Right: Close-up of the electrode structures. While the strips that code the y-position are complete lines, the x-strips are interrupted and form small rectangles that are connected within the substrate.
• Figure 4: Image of an etched metal grid placed in front of the detector to compare centroiding methods: Gaussian fitting (left panels) and the non-iterative approach presented in Sect. \ref{['ssec:algo']} (right panels). Both were implemented in software and run on pre-processed events taken with a detector prototype. While the top panels show the full detector images, the lower panels zoom on the illuminated region. The etched grid was tilted by 45° to detect possible non-linearities that could arise from the interpolation of the centroiding process between two anode strips. These slight nonlinearities can be seen as a small distortion of the grid and a faint dark and light pattern in the horizontal and vertical directions. (Coincidentally, the diagonal crossing points of the grid have nearly the same distance as the anode strips.) In both images the hexagonal substructure of the MCPs is clearly visible. However, the contrast is still slightly higher for Gaussian fitting, leaving room for optimization of the non-iterative approach.
• Figure 5: Distribution of the position differences between the Gaussian fit positions and the non-iterative algorithm position, in software (left) and in the FPGA (right). Two curves are shown: one for the x positions (blue) and one for the y positions (orange). Units of the horizontal axis are the distance between two anode strips (=1.0). As we interpolate to 1/32 between two strips, one pixel width corresponds to about 0.03 on the horizontal axis. The inset lists the widths of the distributions as full width at half maximum (FWHM).