INTRODUCTION In its most general form, confocal microscopy encompasses all optical techniques whose illumination and detection scheme exam- ines each point in an object in the absence of interfering informa- tion from neighboring points. Recently the technique has expanded to encompass not only morphology but also disciplines as far afield as physiology, spectroscopy, fluorescence lifetime analysis, and even DNA sequencing. As a result, the requirements and design constraints on appropriate detection systems are as varied as the fields to which the technique is applied. In the familiar case of fluorescent imaging, for example, con- focal illumination and detection is used to reduce the back- ground fluorescence from out-of-focus planes to obtain better image quality than that achieved in conventional fluorescence microscopy using Köhler illumination. For many confocal micro- scopes employing either disk (Petrán ˘ et al., 1968) or line scan tech- niques, the primary detector is the human eye. It is attractive in terms of its quantum efficiency: roughly 16% of the photons inci- dent on the cornea are perceived. The number and size of detector elements is high, on the order of 10 6 , and this massive parallelism, joined with higher-order processing, results in the relatively rapid perception of an image. Any confocal technique using visible light that projects points in an object coherently onto conjugate points in an image plane can employ this detector. If the scanning is rapid enough, a stable, full-field image will be seen. When such microscopy is combined with focusing through the object, it is possible to readily identify and characterize its three-dimensional (3D) structure. Under relatively high illumination, such full-field confocal systems are in many ways the most general and suitable for a wide variety of tasks. The detection problem, if you will, has been solved for us over the last 10 9 years of evolution. Under low-light conditions and for other more specialized tasks such as spectroscopy or fluorescent lifetime analysis, the eye is far from the perfect detector. Because it is highly adaptive and can work over nine log units of intensity, it is also less than ideal at quantitative comparisons of absolute intensity. Nor is the eye able to implement averaging and filtering techniques to enhance the signal-to-noise ratio (S/N) in the image. Finally, and often most significantly, some confocal systems have slow scanning rates or are designed to project points in the object plane back onto a single point, and it is impossible in such cases for the eye to perceive a coherent image. For these reasons and others, it is necessary to develop tech- niques that approach, and in many respects surpass, the signal pro- cessing capabilities of the eye and that are amenable to further digital enhancement. The task is important because, at present, the design goal of achieving the highest S/N in a confocal image for a given dose to the specimen is often degraded by elements in the detection system. For example, when light levels are low (<20 photons/pixel), significant improvements in S/N can often be made by employing the greatest possible gain in the detector or by exploiting photon-counting, rather than analog measurement schemes. It should also be remembered that the particular detec- tion scheme chosen is constrained not only by the detectors but also by the particular confocal technology employed, the levels of illumination that can be used, and limitations imposed by the nature specimen. At the outset it should be noted that simple point detectors, such as photomultiplier tubes (PMT), are often used in conjunc- tion with bandpass filters to quantify the number of photons at dif- ferent wavelengths in a typical two- or three-channel laser scanning microscope. However, the recent interest in characteriz- ing the full spectrum of dyes in situ, as well as a desire to sepa- rate labels whose wavelength of peak emission may differ by a few nanometers, has led to the development of sophisticated spectro- scopic array detectors that divide the incident light into bands a few nanometers wide. In addition, continued improvements in multi-point or line-scanning confocal microscopy have resulted in the use of sophisticated array detectors that simultaneously accu- mulate data across the image plane. In all of these contexts, the overarching goal is to choose a detection scheme that yields the highest quantum efficiency, the lowest background levels, and the highest S/N. In this chapter we will consider the quantal nature of light and the interaction of photons with materials. We will compare a number of possible detectors in terms of their quantum efficiency, responsivity, spectral response, inherent noise, response time, and linearity. We will then consider the design constraints in terms of the front-end circuitry that digitizes the data. The figures of merit for detection are usually that the estimate of the signal is limited either by the noise within the signal or by the background radia- tion. No physical detector can improve on these limits. Finally, we will suggest future directions toward more perfect detectors with signal processing capabilities limited only by the stochastic nature of the signal. We will begin by considering the kinetic energy of photons. THE QUANTAL NATURE OF LIGHT At very low light levels, two aspects of the quantal nature of light can be demonstrated. First, each particle, or photon, has an asso- ciated kinetic energy. An incident photon stream transfers kinetic energy to a material and gives rise to the variety of effects used in light detection. Second, at low light levels it is apparent that even 12 Photon Detectors for Confocal Microscopy Jonathan Art 251 Handbook of Biological Confocal Microscopy, Third Edition, edited by James B. Pawley, Springer Science+Business Media, LLC, New York, 2006. Jonathan Art • University of Illinois College of Medicine, Chicago, Illinois 60612