Quantum efficiency

The composition of a CCD is essentially pure silicon. This element is thus ultimately responsible for the response of the detector to various wavelengths of light. The wavelength dependence of silicon can be understood in an instant by glancing at Figure 3.1. Shown here is the length of silicon needed for a photon of a specific wavelength to be absorbed. Absorption length is defined as the distance for which 63% (1/e) of the incoming photons will be absorbed. Figure 3.1 clearly shows that, for light outside the range of about 3500 to over 8000 A, the photons (1) pass right through the silicon, (2) get absorbed within the thin surface layers or gate structures, or (3) simply reflect off the CCD surface. At short wavelengths, 70% or more of the photons are reflected, and for very short wavelengths (as for long wavelengths) the CCD becomes completely transparent. Thus the quantum efficiency of a typical CCD device will approximately mirror the photon absorption curve for silicon. Shortward of ~ 2500 A (for thinned devices) or about 25 A (for thick devices) the detection probability for photons increases again. However, owing to their much higher energy, these photons lead to the production of

100 nm

10 nm

1 nm

100 nm

10 nm

1 nm

Fig. 3.1. The photon absorption length in silicon is shown as a function of wavelength in nanometers. From Reicke (1994).

Fig. 3.1. The photon absorption length in silicon is shown as a function of wavelength in nanometers. From Reicke (1994).

multiple electron-hole pairs within the silicon and may also produce damage to the CCD itself (see Chapter 7).

CCD quantum efficiencies are therefore very dependent on the thickness of the silicon that intercepts the incoming photons. This relation between absorption probability and CCD thickness is why front-side illuminated (thick) devices are more red sensitive (the photons have a higher chance of absorption) and why they have lower overall (blue) QEs (since the gate structures can be close to or even exceed the necessary absorption depths of as small as only a few atomic layers). A few front-side CCDs have been produced with special gate structures that are transparent to incoming blue and UV photons. In thinned devices, the longest wavelength photons are likely to pass right through the CCD without being absorbed at all.

Figure 3.2 shows the quantum efficiencies for various imaging devices. Note that the y scale is logarithmic and the much superior QE provided by CCDs over previous detectors. Figure 3.3 shows a selection of modern CCD QEs. The large difference in QE that used to exist between thinned and thick CCDs is now mostly eliminated due to manufacturing processes

CHARGE-COUPLED DEVICE (THINNED)

CHARGE-COUPLED DEVICE (THINNED)

Fig. 3.2. QE curves for various devices, indicating why CCDs are a quantum leap above all previous imaging devices. The failure of CCDs at optical wavelengths shorter than about 3500 A has been essentially eliminated via thinning or coating of the devices (see Figure 3.3).

350 400 450 500 550 600 650 700 750 800 850 900 950 1000

Wavelength (nm)

Fig. 3.3. QE curves for a variety of CCDs. WFPC2 is the second generation wide-field/planetary camera aboard HST, CAT-C is a new generation SITe CCD used in a mosaic imager at the University of Arizona's 90" telescope on Kitt Peak, MIT-LL is a CCD produced at the MIT Lincoln Laboratories and optimized for red observations, ACS is the Hubble Space Telescope Advanced Camera for Surveys SITe CCD, LBL is a Lawrence Berkeley Lab high resistivity, "deep depletion" CCD with high red QE, and MAT is a front-side, processed CCD showing high blue QE.

350 400 450 500 550 600 650 700 750 800 850 900 950 1000

Wavelength (nm)

Fig. 3.3. QE curves for a variety of CCDs. WFPC2 is the second generation wide-field/planetary camera aboard HST, CAT-C is a new generation SITe CCD used in a mosaic imager at the University of Arizona's 90" telescope on Kitt Peak, MIT-LL is a CCD produced at the MIT Lincoln Laboratories and optimized for red observations, ACS is the Hubble Space Telescope Advanced Camera for Surveys SITe CCD, LBL is a Lawrence Berkeley Lab high resistivity, "deep depletion" CCD with high red QE, and MAT is a front-side, processed CCD showing high blue QE.

and coatings although other differences (such as location of peak QE, cosmic ray detection, etc.) remain. Quantum efficiency or QE curves allow one quickly to evaluate the relative collecting power of the device as a function of wavelength. Measured QE curves, such as in Figure 3.3 and those shown in the literature, are generally assumed to be representative of each and every pixel on the device, that is, all pixels of a given device are assumed to work identically and have the same wavelength response. This is almost true, but it is the "almost" that makes flat fielding of a CCD necessary. In addition, the QE curves shown or delivered with a particular device may only be representative of a "typical" device of the same kind, but they may not be 100% correct for the exact device of interest.

The quantum efficiency of a CCD is temperature sensitive especially in the red wavelength region. It has long been known that measurement of the QE at room temperature is a poor approximation to that which it will have when operated cold. Thus QE curves should be measured at or near the operating temperature at which the CCD will be used. As an example of the temperature sensitivity of the efficiency of a CCD, Figure 3.4 shows

Wavelength (nm)

Fig. 3.4. Sensitivity of the quantum efficiency of a MIT/LL CCD for three operating temperatures. The blue sensitivity is little affected by a change in operating temperature but the red QE can change by a factor of two. The use of such devices requires a balance of higher operating temperature and keeping the dark current under control.

Wavelength (nm)

Fig. 3.4. Sensitivity of the quantum efficiency of a MIT/LL CCD for three operating temperatures. The blue sensitivity is little affected by a change in operating temperature but the red QE can change by a factor of two. The use of such devices requires a balance of higher operating temperature and keeping the dark current under control.

three QE measurements of the same CCD for temperatures of +20° C (~room temperature), -40°C, and -100°C (^operating temperature). Note that the variations are small at 8000 A but increase to 20% at 9000-10 000 A. The curves in this plot would lead one to the conclusion that operating a CCD at room temperature is the best thing to do. However, Section 3.5 will show us why this is not the case and a compromise between operating temperature (i.e., dark current) and delivered QE must be used.

A recent advance in the manufacture of CCDs is to use high resistivity silicon. Typical CCDs you have used have a resistivity of 20-200 ohm-cm or maybe up to 300 ohm-cm and are made on 10-40 micron epi.1 The above resistivity and thickness values for Si wafers are fairly standard today and lend themselves to easy etching for thinning (10-20 micron final thickness). Starting with bulk silicon and a new process called the float-zone technique, resistivities up to 5000-10000 ohm-cm are possible. Adding in a bias voltage to the optically transparent back-side substrate and using a thick layer of 45 to 350 micron epi, each pixel in a high resistivity CCD can be fully depleted resulting in very high red QE and deep pixel wells. Figure 3.5 shows the relationship between Si resistivity, depletion depth, and the bias voltage used.

1 epi (pronounced "ep'-pea") is CCD lingo for epitaxial silicon, which is the Si wafer type used to make CCDs. An example was shown in Figure 2.8.

Depletion Depth vs Resistivity and Bias 30.0 I

0 100 200 300 400 500 600 700 800 Resistivity

Fig. 3.5. Laboratory measurements of the depletion depth in a pixel vs. the CCD silicon resistivity for three different bias voltages. We see that one can deplete deeper (larger) pixels with higher resistivity silicon assuming the use of a larger bias voltage.

These "deep depletion" CCDs collect up to about 300 000 photoelectrons deep in a pixel where they are more likely to stay given the high resistivity. Higher resistivity silicon wafers require very special care in their production and much higher purity tolerances of the Si wafers, thus are more costly to produce. The internal Si lattice structures must be highly uniform and the level of unwanted impurities in the Si must be very near zero. Until the past few years, production of such Si was not possible and today the Lawrence Berkeley Lab (LBL) and MIT/Lincoln Labs are the leaders in making such devices. Figure 3.3 shows the superior red QE of a LBL high resistivity CCD. As we noted in our discussion of Figure 3.1, the thickness of a CCD is important in the QE it attains, and thus deep depletion devices have large well depths to aid in the improvement of their red QE. The resistivity of the Si and the deep pixels both come into play when one considers charge diffusion within a CCD. We will discuss charge diffusion in some detail below.

Placing an antireflection (AR) coating on a CCD (both for visible and near-UV light) increases the QE and extends the range of good QE well into the 3000 A region. The need for AR coatings comes from the fact that silicon, like most metallic substances, is a good reflector of visible light. If you ever have a chance to hold a CCD, you will easily see just how well the surface does indeed reflect visible light. All the QE curves in Figure 3.3 have the overall shape expected based on the absorption properties of silicon as shown

Depletion Depth vs Resistivity and Bias 30.0 I

0 100 200 300 400 500 600 700 800 Resistivity in Figure 3.1. Graphical illustrations of QE curves almost always include photon losses due to the gate structures, electron recombination within the bulk silicon itself, surface reflection, and, for very long or short wavelengths, losses due to the almost complete lack of absorption by the CCD. Given that all these losses are folded together into the QE value for each wavelength, it should be obvious that changes within the CCD structure itself (such as radiation damage or operating temperature changes) can cause noticeable changes in its quantum efficiency.

Measurement of the quantum efficiency of a CCD is usually performed with the aid of complicated laboratory equipment including well-calibrated photodiodes. Light at each wavelength is used to illuminate both the CCD and the photodiode, and the relative difference in the two readings is recorded. The final result of such an experiment is an absolute QE curve for the CCD (with respect to the calibrated diode) over the range of all measured wavelengths.

To measure a CCD QE curve yourself, a few possibilities exist. You may have access to a setup such as that described above. Measurements can also be made at the telescope itself. One good method of QE measurement for a CCD consists of employing a set of narrow-band filters and a few spectrophotometry standard stars. Performing such a task will provide a good relative QE curve and, if one knows the filter and telescope throughput well, a good absolute QE curve. A detailed reference as to what is involved in the measurement of a spectrophotometry standard star is provided by Tug et al. (1977). When producing a QE curve by using the above idea, the narrowband filters provide wavelength selection while the standard stars provide a calibrated light source. A less ambitious QE curve can be produced using typical broad-band (i.e., Johnson) filters, but the final result is not as good because of the large bandpasses and throughput overlap of some of the filters. In between a detailed laboratory setup and the somewhat sparse technique of using filters at the telescope, another method exists. Using an optics bench, a calibrated light source covering the wavelength range of interest, and some good laboratory skills, one can produce a very good QE curve for a CCD and can even turn the exercise into a challenging classroom project.

Was this article helpful?

0 0
Telescopes Mastery

Telescopes Mastery

Through this ebook, you are going to learn what you will need to know all about the telescopes that can provide a fun and rewarding hobby for you and your family!

Get My Free Ebook


Post a comment