Preliminary characterization results of a large format 4k x 4k EMCCD

ABSTRACT

Scientific EMCCD cameras have demonstrated excellent imaging performance under extreme low light conditions. Photon counting capability combined with a very low dark current offered by the CCD technology have made EMCCDs the detector of choice for high-performance applications such as time resolved spectroscopy and low light imaging. However, future astronomical instrumentation requires high spatial resolution while commercially available EMCCD devices are limited by a relatively modest area format of (1k×1k). To address this requirement, the Universite de Montr ´ eal and Teledyne-e2v have jointly developed a 4k ´ ×4k EMCCD, the CCD282. This paper presents the results of cryogenic characterization of the CCD282 operated with Nuv¨ u Cam ¨ eras’ CCD Controller for Counting ¯ Photons version 3. The advantages of a novel large format EMCCD over existing technology for high resolution spectroscopy are discussed.

INTRODUCTION

Since the introduction of the first commercial Low Light Level CCD by Marconi in early 2000,1 and shortly after the Impactron2 commercialized by Texas Instruments, several devices exhibiting an Electron Multiplication register to generate sub-electron read-out noise were developed. These Electron Multiplying CCDs (EMCCDs), from the CCD65 which was built in two variants for the PAL and NTSC video standards, to the CCD201-20 (introduced by e2v Technologies) and TC285 (introduced by Texas Instruments) of 1k × 1k image area, were falling short of pixels for scientific applications in extreme faint flux conditions requiring large field of view and high pixel count. For instance, very high resolution spectroscopy (R > 30000) can not be efficiently accomplished on a 1k×1k device. On the other hand, although wide field surveys could be accomplished with several small devices, the form factor of the existing EMCCDs (non-buttable) renders the use of the focal plane inefficient and reduces the advantages of using an EMCCD.

For those reasons, the Universite de Montr ´ eal, through a funding by the ´ Canadian Foundation for Innovation (CFI), awarded a contract to e2v Technologies (now Teledyne-e2v) to develop a large format EMCCD. The end result, the CCD282, is a 4k × 4k device capable of generating ∼5 images per second with sub-electron read-out noise. This work presents the experimental setup developed at Nuv¨ u Cam ¨ eras to test this novel device as well as the preliminary characterization ¯ results.

2 CCD282 ARCHITECTURE

The CCD282 is a standard silicon split-frame transfer EMCCD with an active region of 4096×4096 square pixels of 12µm3 (49.15 mm side length). The split-frame transfer architecture requires twice the area of the active pixels, since half of the pixels are masked by the store shield. Hence, the CCD282 is a large device, with a die size of 104×51 mm, and with a package size of 125×70 mm, which compares to the size of today’s smartphones.

The image and store sections are designed to operate in 2-phase mode, each phase being clocked in pair (Figure 1). It provides 8 Electron Multiplying outputs that are specified to operate at up to 15MHz, which should yield 4–5 full frames per second at this frequency. Its large size limits the vertical line transfer rate to 166 kHz (6 µs per line). This yields a frame transfer time of ∼12 ms.

By comparison with well know EMCCDs such as the CCD201-20, the 1k × 1k device of Teledyne-e2v, the CCD282 has a few notable design differences. First, the CCD282 does not provide non-electron multiplication (conventional) output. Conventional read-out of the device is be possible by using a high voltage clock amplitude that is low enough to prevent Electron Multiplication. In this mode, no Excess Noise Factor will be generated. It must be noted that the device provides one dummy output for every output for optional differential use. The differential use does however increase the read-out noise by a factor of √ 2 as the noise of both outputs adds-up quadratically.

Next, the Dump Gate, which allow for dumping unwanted lines of signal, is not present. Although it is a useful feature to properly get rid of the dark current that has accumulated in the storage area during the integration period (which pile up in the horizontal register during the initial Frame Transfer), it does increase the capacitance of the horizontal register, which in turn renders the high speed operation of the device more challenging. On the other hand, an Overspill Drain has been included, which prevent an excessive level of charge from being generated in the Electron Multiplying register. This helps limit (or eliminate) the ageing of the multiplication register. Moreover, it does prevent, at high EM gain, the energetic cosmic rays from generating an excessive number of electrons, which eventually saturates the multiplication register and requires several clock cycles (and therefore, pixels) to be cleared (Figure 2).

Для продолжения чтения вы можете скачать полную версию материала по ссылке ниже