High-end commercial CCDs can provide excellent sensitivity and detection limits for spectroscopy applications.
Dr. Antoinette O’Grady, Princeton Instruments Inc./Acton Research Corp.
Recent developments in CCD detectors for spectroscopy have produced a bewildering range of options, from the standard, front-illuminated CCD with its moderate quantum efficiency and limited wavelength coverage, through the broadband open-electrode device with relatively good quantum efficiency and wavelength coverage, to the ultimate sensitivity of the back-illuminated CCD. Add to these the intensified and electron-multiplying CCDs, and the choices may seem daunting.
Figure 1. Recent developments in CCD detectors for spectroscopy have produced a bewildering range of options. Making the right choice requires an understanding of basic device parameters and of how they affect a real experimental setup.
Choosing the correct detector for an application requires an understanding of the basic device parameters involved and of how they affect a real experimental setup. The first step is to clarify the difference between detection limit and sensitivity.
A CCD’s detection limit is the smallest signal that the device can detect. Manufacturers usually define this in terms of the noise specifications of the device because a signal would have to be greater than the noise floor of the system to be detectable.
A CCD’s quantum efficiency — that is, its ability to convert a photon of light into an electrical signal — determines its sensitivity. It establishes how much of the signal is detected within the silicon. Sensitivity usually is defined in practical situations in terms of signal-to-noise values.
Traditionally, high-end spectroscopy CCD detectors have been devices with a high dynamic range — typically, 16 bits — and a low noise level. A high dynamic range enables the user to detect weak signals adjacent to intense ones.
Consider the choices
In recent years, standard devices have achieved readout noise levels as low as 2 e2 rms. The resolution of such detectors usually is between 20 and 26 μm to match the slit width in most spectrometers. The time resolution is such that they can easily achieve 500 spectra per second in standard mode. A detector with true 16-bit dynamic range will have slightly higher readout noise and larger pixels.
Figure 2. Standard front-illuminated CCDs achieve readout noise levels as low as 2 e2 rms. The resolution usually is between 20 and 26 μm to match the slit width in most spectrometers. The time resolution of such detectors, such as the Princeton Instruments Pixis:100, enables achievement of 500 spectra per second in standard mode.
Intensified CCDs consist of an intensifier, usually fiber optically coupled to a CCD. The high electron gain of the intensifier’s microchannel plate enables it to amplify a signal far above the noise floor, essentially bestowing these devices with single-photon detection capability. However, the main benefit of intensified CCDs in spectroscopy is their gating ability, which allows the intensifier to act as a very fast shutter, providing time resolution on the scale of 1 ns. This is useful in applications that feature a short event within a broadband continuum with a longer lifetime. These events usually are repetitive, and the intensifier can capture them and accumulate their signal while ignoring the broadband continuum.
Before the advent of electron-multiplying CCDs, intensified CCDs were perceived to offer the ultimate in sensitivity. However, this was the result of confusing their sensitivity and their detection limit. Although their detection limit is excellent (a very small signal can be detected because it is amplified significantly above the noise floor), their sensitivity, or signal-to-noise ratio, is not as good as that for conventional CCDs because intensified CCDs have a significantly lower quantum efficiency and suffer from effects such as additional noise, crosstalk, “chicken wire,” scintillation and halos. Moreover, intensified CCDs are expensive and complex, have finite lifetimes and can be damaged when overexposed.
Electron-multiplying CCDs amplify the signal above the readout noise. This allows them to run at much faster readout speeds while retaining an excellent detection limit. Because their quantum efficiency is the same as that of conventional CCDs, their sensitivity is comparable at very low light levels. At light levels above the readout noise, however, the additional noise sources of electron-multiplying CCDs degrade their signal-to-noise ratio (Figure 3).
Figure 3. A comparison of the detection limit of electron-multiplying and traditional CCDs reveals that the former has an excellent detection limit at high speeds but that its multiplier noise factor degrades the signal-to-noise ratio at higher light levels. The signal-to-noise ratio for theback-illuminated CCD is the same as that for the ideal detector until the signal level drops below its readout noise.
Electron-multiplying CCDs are manufactured using standard fabrication techniques. They differ from conventional CCDs in that they have an additional, or gain, register inserted between the end of the usual shift register and the amplifier. Higher voltage amplitudes than normal are used in the gain register to generate electrons (the gain) by impact ionization. Because this is done before the amplifier, a signal that is lower than the readout noise can be amplified by up to 1000 times and can be identified easily over the same, provided it is greater than the electron-multiplying CCD’s generated noise factor.
Electron-multiplying CCDs are said to unite the detection capabilities of an intensified CCD with the inherent advantages of a CCD for imaging applications. To an extent this is true, as they have the resolution and quantum efficiency of a CCD without a lot of the disadvantages of an intensified CCD.
The electron-multiplying CCD’s ability to maintain a high detection capability is extremely beneficial in imaging applications, particularly as biological applications often require qualitative rather than quantitative data. However, for spectroscopy applications, this is less useful because traditional CCDs already offer high spectral rates coupled with high sensitivity.
A key disadvantage of electron-multiplying and intensified CCDs is the noise factor associated with the amplification process, which determines their detection limit and sensitivity. Furthermore, the nature of the amplification process means that these devices are not quantitative, which limits their versatility in spectroscopy applications. Thus, for different photon levels, it is not possible to accurately determine the incident number of photons, even in a ratio.
Signal over noise
In practical, experimental terms, the detection limit and the sensitivity of a system are determined by a combination of the exposure time, the total noise of the system and the quantum efficiency. This is measured experimentally by the signal-to-noise ratio of the sample of interest.
The exposure time must be as long as required to integrate the incident photons over the noise (readout noise or noise factor) of the system. It is related to the quantum efficiency. The higher the quantum efficiency, the shorter the exposure time for the same incident signal. In most spectroscopy applications, exposure times are seconds to minutes, so the readout time is not important.
The signal-to-noise ratio determines the detection limit of a CCD in spectroscopy. To optimize the signal-to-noise ratio, it is necessary to understand the contributions to the total noise of the system. The total noise (Ntotal) is made up of three components — the shot noise of the signal (Nsignal), the noise of the dark signal (Ndarksignal) and the readout noise (Nreadout) — and may be described as
or, in terms of signal (S), dark current (D) and readout noise, as
The shot noise typically is overcome by accumulating a number of spectra so that the noise relative to the signal decreases. The dark signal or dark current is minimized by cooling the detector. The readout noise, an effect of the readout electronics of the sensor and analog-to-digital converter, is always present and increases with readout speed.
Standard CCDs can achieve 2 to 4 e2 readout noise at slow speeds. At speeds of 2 MHz, readout noise of 12 e2 is typical. Electron-multiplying CCD technology does not lessen readout noise; it is still present. A sensor running at 5 MHz will have readout noise of approximately 50 e2 rms. Camera manufacturers usually state that the effective readout noise is subelectron because the electron-multiplying CCD amplifies the signal above the readout noise.
For a standard back- or front-illuminated CCD with pixels of 20 μm at 280 °C, the dark current is approximately 0.002 e2 per pixel per second. Thus, for a slit height of 2 mm, approximately 100 pixels would be binned, resulting in dark current of 0.2 e2 per second. With readout noise of 2 e2 rms, an exposure time of 10 s is possible before the noise of the dark current affects the detection limit.
Given this readout noise and negligible dark current, shot noise can be the dominant noise source, even for light levels as low as 10 photons. The signal, rather than the CCD or dark current noise, thus determines the detection limit:
The signal-to-noise ratio is:
The exposure time and quantum efficiency for the electron-multiplying CCD is the same as for a CCD. The quantum efficiency of an intensified CCD is significantly lower, but its gain is much higher.
The total noise of both intensified and electron-multiplying CCDs is affected by the respective gain (G) and noise associated with the electron-multiplication process. For an intensified CCD, this additional noise factor (F) is typically 2 to 3.5.
For an electron-multiplying CCD, it is typically 1.1 to 1.4. Thus, the total noise for these CCDs is given by
The calculations for Figure 3 assume single-pixel performance, sufficiently short exposure times such that the dark current level is below the limit of detection and an incident signal in electrons after quantum efficiency conversion. The effect of the total noise on the theoretical signal-to-noise values for a single pixel is presented in the graph for an electron-multiplying CCD at optimum gain, back-illuminated CCDs at different readout speeds and an ideal detector. The graph shows that the electron-multiplying CCD has an excellent detection limit at high speeds, but that the noise factor degrades the signal-to-noise ratio at higher light levels. The signal-to-noise ratio for the back-illuminated CCD imitates the ideal detector until the signal level is below its readout noise.
High-end commercial CCDs for spectroscopy provide excellent sensitivity and detection limits. For most applications, a front-illuminated or open-electrode CCD is more than sufficient, depending on the wavelength range of interest; back-illuminated CCDs, however, provide the sensitivity and detection limits needed for more demanding applications.
For those requiring nanosecond resolution, an intensified CCD is a good choice.
The ideal spectroscopy detector is a true quantitative device that combines high readout speeds with real sub-electron readout speeds and high sensitivity, resulting in ultimate detection limits without compromise. As new technology is being developed more and more rapidly, this could become reality in the not-too-distant future.
Further reading
J. Hynecek and T. Nishiwaki (January 2003). Excess noise and other important characteristics of low light level imaging using charge multiplying CCDs. IEEE T ELECTRON DEV., pp. 239-245.
J.R. Janesick (2001). Scientific Charge-Coupled Devices. SPIE Press.
P. Jerram et al (May 2001). The LLLCCD: Low light imaging without the need for an intensifier. Proc. SPIE, Vol. 4306, pp. 178-186.
M.S. Robbins and B.J. Hadwen (May 2003). The noise performance of electron multiplying charge-coupled devices. IEEE T ELECTRON DEV., pp. 1227-1232.
The use of multiplication gain in L3Vision electron multiplying CCD sensors (July 2003). Low-Light Technical Note 2, e2v technologies ltd. Available at www.e2v.com/applications/l3vision_tn.htm.
On-chip multiplication gain (2003). Technical note 14, Roper Scientific Inc. Available at www.piacton.com/products/photonmax/whitepapers.aspx.
Meet the author
Antoinette O’Grady is OEM business manager for Europe at Princeton Instruments Inc. in Tralee, Ireland; e-mail:[email protected]om.