ABSTRACT

Single electron detection is desirable in systems incorporating EBCCDs. The sources of noise  in single electron counting are discussed and an expression is given for the total noise.

INTRODUCTION

Electron bombarded charge coupled devices (EBCCDs) have been recognized as having performance advantages over competing technologies in terms of their high sensitivity, low noise and high spatial resolution. Over the past 15 years at least three distinct types of applications have emerged: as the target in two dimensional photon counters^1,2, as the anode in real-time image intensifiers^3,4 and as detectors in high performance CRT based instruments^5,6,7. Obviously single photo-electron counting ability is crucial for photon counters, but it is also desirable in intensifiers and CRTs. However, image intensifiers and CRTs also need good dynamic range and a good signal to noise ratio near full well. Unlike the CCDs in many high performance optical imaging systems, the EBCCD may be required to run at room temperatures and fast (video) readout speeds. Each of these applications requires the performance a backside-thinned, gain enhanced, n-buried channel CCD can bring.

The system designer has a few variables he or she can manipulate to optimize performance. While the temptation is to just turn up the electron beam energy until there is enough gain to "unambiguously" detect single electron events, this is done at the expense of the number of primary electrons it takes to reach full well. To properly "tune" the system the designer must know the effect of other variables on performance. At room temperature and long integration and readout times, CCD dark current is a noise contributor. There is charge spreading of the secondary gain electrons in a CCD so the charge generated by a single electron impact is spread over several pixels, reducing the signal in the pixel under impact. A light shield may be necessary on the CCD to remove unwanted optical signal. The read noise of a CCD running at video rates is much higher than for the same CCD running cooled and slow.

In this paper we will discuss the sources of noise present during single electron counting and present an expression for the total noise. First, however, some background is necessary.