ICCD compared to EMCCD
Within the last years, electron multiplying CCD cameras (EMCCD) have become very popular. In a variety of applications which do not require fast gating the EMCCD has started to compete with the ICCD. In some specific astronomical long time observations extremely cooled custom-built EMCCD cameras yet are superior to ICCD camera systems.
Quantum Efficiencies of ICCD and EMCCD
Since the back illuminated EMCCD sensors were introduced to the market it can be heard frequently, that the sensitivity of EMCCD cameras would outperform ICCD systems and therefore EMCCD cameras would be the better choice in low light level applications. This point of view is mainly made up from considering the quantum efficiencies of both camera types.
As shown in the above diagram back illuminated EMCCD chips reach a quantum efficiency of up to 92% peak value. In contrast, third generation image intensifier photocathodes reach 52% and second generation image intensifier photocathodes mostly lie below 20%. Indeed, it can be said that, as a spectral average, the quantum efficiency of EMCCD cameras is by a factor 4 better than Gen II equipped ICCD cameras and a factor 2 better than Gen III systems.
The main functional difference between ICCD and EMCCD cameras
But this is only one aspect, it is the aspect which only concerns the first process step of both camera types, the arriving-photon detection. The second step, which comprises the amplification of the detected light, is achieved in very different ways in both systems. The final camera performance will of course result from the coaction of both operation steps. For this reason, the two workig principals shall be explained qualitatively to some extend in the following.
Let us consider an arriving signal consisting of 4 photons per pixel within the exposure time. As shown in the pictures below, maybe the EMCCD collects all 2 photons due to its high quantum efficiency. However, a Gen III equipped ICCD may only detect 1 out of these 2 photons. Whereas the EMCCD pixel now contains 2 electrons, the appropriate ICCD pixel will be filled up with let's say 1 x 500 x 40% = 200 electrons as a result of the image intensification.
When the exposure time is expired the CCD sensors will be read out. The readout process itself contributes with several noise sources to the number of electrons actually stored in the pixel. When we consider an uncooled ICCD camera operated at maximum frame rate there maybe 50 noise electrons contributed to the already existing 200, resulting in a ratio of 200/50 = 4. An EMCCD cooled down to 200 Kelvin and operated at standard frame rates will only generate about 1 electron from charge transfer, i.e. clock induced charge noise, and particularly from readout darkcurrent. The ratio will in this case amount to 2/1 = 2.
Please keep in mind that the comparison of the signal to noise ratios of the different camera types is a complex issue. This simple example shall only give you a more concrete idea of the principal situation and the scale of the values. Despite the enormous number of noise electrons generated in the ICCD camera compared to the EMCCD, it nevertheless achieves the better ratio between the signal and the technical noise contributions. It is a very important difference that the ICCD amplifies the measurement signal prior to the readout process and the large number of noise electrons are added to the even higher number of signal electrons. This makes the ICCD system insensitive against the sensor temperature and high readout frame rates. However, the strongly cooled EMCCD system only adds an extremely small amount of noise electrons to the measurement signal yet prior to the amplification process. So, the noise contributions are amplified to the same extent as the signal itself.
A detailed examination of this issue can be found in following publications:
Title: Noise performance comparison of ICCD with CCD and EMCCD cameras
Author: David Dussault, Paul Hoess
Institute: Paul Hoess KG, Stanford Computer Optics, Inc.
Title: ICCDs edge out electron-multiplying CCDs in low light
Author: David Dussault, Paul Hoess
Institute: Paul Hoess KG, Stanford Computer Optics, Inc.