Photon Counting and Precision Photometry for the Roman Space Telescope Coronagraph
ABSTRACT
The Nancy Grace Roman Space Telescope will include, as one of its two instruments, the highest contrast coronagraph ever attempted, with sensitivity down to Jupiter class planets. With flux ratio below 1e-8, these planets will be exceedingly dim, so that signal rates are as low as 0.01 electrons per second at the imaging detector. These rates necessitate ultra-low noise detectors and methods. For its science imaging camera, the Roman Coronagraph will employ an electron multiplication CCD (EMCCD), to achieve near-zero read noise. EMCCD’s, however, deliver the low read noise at the cost of amplification of all other noise, because of the stochastic nature of the electron multiplication process. To circumvent this next-order challenge, a thresholding technique called photon counting can be used. The resulting image has no read noise and no excess noise factor (ENF). The remaining challenge, for precision photometry, is to account for the undercount and overcount effects inherent to photon counting. These arise primarily from the inefficiency of thresholding itself, and coincidence loss, where multiple-electron events are not distinguished from single-electron ones. Here we present a detailed description of the photon counting algorithm and the corrections necessary to achieve photometric accuracy below 0.5%.
1. INTRODUCTION
High contrast imaging in space will take a major leap forward with the Coronagraph Instrument (CGI) on the Nancy Grace Roman Space Telescope.1 Equipped with deformable mirrors, and focus and tip tilt active control, this technology demonstration instrument will achieve at least two orders of magnitude higher contrast than any existing instrument, and will potentially have sensitivity to image and obtain spectra, in the visible band, from Jupiter class planets. High contrast imaging coronagraphs offer versatility in targeting and longer mission life, since they do not require formation flying as do external occulters.2 Along with this comes a disadvantage of generally lower throughput, particularly when the telescope has significant obscurations as is the case for Roman. A Jupiter class planet, depending on its albedo and phase function, will have a flux ratio around 5 ·10−9 (5 ppb), or a delta-magnitude of ∼ 21. The typical target star brightness will be around 6 mag in the visible band, so that the planet will be at a relative magnitude of about 27. In direct imaging, this results in signal rates in the 10 milli-photon per second regime, necessitating ultra low noise sensors. In this paper we will describe the low noise sensor approach used by Roman CGI, the EMCCD, and derive the image processing steps need to provide precision (0.5%) photometry of the low noise images.
2. THE EMCCD
The best CCD’s currently achieve about 3 electrons (e −) of read noise, and even at this rate, read noise becomes by far the dominant noise source. Since the instrument is being operated in the space environment, with significant rates of cosmic rays, frame times must be kept relatively short. Integration with higher frame rate makes the per-read classes of noise, chief among them read noise, more important. Electron multiplication CCD’s (EMCCD’s) can dramatically reduce read noise by amplifying the signal through successive impact ionization events. Each pixel’s charge packet in these devices is clocked out as usual but in the last stage, it is passed through a serial gain register prior to read out. Roman CGI uses a mission-optimized version of the Teledyne e2v CCD-201 EMCCD, designated as the CCD-301, featuring mitigation for charge traps and cosmic ray tails. In the CGI’s two cameras, this EMCCD will be operated with gains as high as 7500. The amplified image, sometimes also called the ’analog’ or ’proportional’ image, is enhanced by a factor of the EM gain, GEM, so that the best estimate image is obtained by dividing every pixel by the EM gain (a single scalar number for the whole image).
Для продолжения чтения вы можете скачать полную версию материала по ссылке ниже
