A Butterfly's Eye Camera for Intensity Interferometry with Cherenkov Telescopes

TL;DR

The paper addresses the sensitivity and imaging limitations of optical intensity interferometry with IACTs by introducing the Butterfly's Eye I3T concept, which partitions the primary mirror into submirrors directed to separate camera pixels. This enables narrower spectral filtering and photon-counting detectors, delivering a $4$–$6\times$ sensitivity gain and offering angular imaging from $2$ to $40$ mas, with practical implications for stellar surface imaging and nova ejecta studies. Through practical MAGIC/LST considerations and Python-based photon-arrival simulations, the authors quantify timing performance for three Butterfly configurations, revealing sub-nanosecond to tens-of-picoseconds time spreads depending on submirror count and staggering. The work demonstrates how aperture segmentation can transform IACTs into diffraction-limited optical interferometers with imaging capabilities, potentially impacting key science cases like red-giant surfaces, Be star envelopes, and fast-rotating star disks while informing design choices for CTA-era arrays. The approach hinges on photon-counting operation, narrow-band filtering, and precise submirror alignment to manage NSB and timing, enabling high-resolution optical interferometry on existing Cherenkov facilities.

Abstract

In recent years, imaging atmospheric Cherenkov telescopes (IACTs) have emerged as promising platforms for optical interferometry through the use of intensity interferometry. IACTs combine large segmented mirrors, photodetectors with nanosecond-scale time response capable of detecting signals from just a few photo-electrons, and array configurations with baselines of hundreds of meters. As a result, all major IACT facilities have now been upgraded to function also as optical intensity interferometers, achieving sensitivities an order of magnitude better than their predecessor, the Narrabri Stellar Intensity Interferometer. However, further improvements in sensitivity are currently limited by key IACT design constraints, namely the combination of poor optical quality and small focal ratios. Here we present three practical implementations of the "I3T concept", in which segments of the IACT primary mirror are focused onto different pixels of its camera. This approach yields several unexpected but significant advantages. Optics with larger focal ratios allow to integrate narrow-band optical filters, while lower photon fluxes enable to deploy next-generation photodetectors operating in photon-counting mode. We demonstrate that this so-called "Butterfly's Eye" configuration enhances the sensitivity of IACT-based intensity interferometers by a factor between 4 and 6. Moreover, as originally envisioned, the I3T design introduces imaging capabilities on angular scales from 2 to 40 milliarcseconds, unlocking new scientific opportunities such as direct surface imaging of nearby red giants. Besides, realistic simulations show that it can have a transformative impact on at least two key science cases: imaging the earliest stages of nova ejecta, and measuring the oblateness and circumstellar disks of fast-rotating stars.

Figures (5)

• Figure 1: Photoelectron rates expected in a pixel as a function of star's B magnitude. The uppermost solid line corresponds to the current MAGIC telescope, that is, full reflector focused into the pixel, 25 nm filter passband and the PDE of the current photomultipliers. The other three solid lines correspond to the three Butterfly configurations that are described in the text, all with a narrower filter passband of 2 nm and enhanced PDE but with three different submirror areas. The NSB rate under dark sky conditions for the four configurations is also drawn as a reference. We have shadowed the area where the rate is too high to operate in photon-counting mode (i.e. rate above 100 MHz).
• Figure 2: An illustration of the I3T concept applied to a MAGIC IACT. Each of the tiles of the tesselated 17 m diameter primary mirror is focused to a pixel. In this figure, the pixels are spread over a very large area to better illustrate the concept. As we shall see practical implementations do not require such a large camera.
• Figure 3: A very simplified scheme of the optics: light from a 1 m$^2$ or 4 m$^2$ section of the MAGIC reflector ('submirror', corresponding to 1 or 4 mirror tiles) is focused into a pixel which has been aligned in the direction of the submirror. A 30 mm diameter lens is employed to condense the light into a photodetector. Note that light coming from the sub-mirror covers essentially the whole lens due to the large PSF.
• Figure 4: Illustration of the random stagger of the mirror tiles caused by the rough precision of the mechanical installation that is common in Cherenkov telescopes. tiles are staggered respect to the ideal paraboloid of focal length 17 m. For the second MAGIC telescope the stagger roughly follows a gaussian distribution of 3.5 mm standard deviation, as illustrated in the inset figure. Note that the magnitude of the stagger has been exaggerated in the main figure.
• Figure 5: Simulated arrival time distribution, in ps, of photons arriving at a pixel. The origin of the X axis is arbitrary because we are only interested on the shape of the time distribution. Distributions for simulations with random stagger are colored orange, those without stagger are colored blue. The upper left panel corresponds to a 1 m$^2$ submirror (Butterfly 1) at the center of the reflector and a pixel at the camera center. The upper right one to a 36 m$^2$ submirror (Butterfly 3) for the same pixel and a submirror again at the center of the reflector. The lower two plots corresponds to 4 m$^2$ submirrors (Butterfly 2). On the left, the submirror is at the center of the reflector and the pixel at the camera center. On the right, the submirror is at the edge and the pixel is 24 cm away from the camera center.