WHY A CCD?

To be succinct, a Charge-Coupled Device (CCD) camera is simply a photon-counting device that is capable of making a picture. A CCD chip is divided up into several little pixels (our chip is a rectangle having a size of 1024 x 768 pixels). Each pixel keeps track of the number of photons that hit it (actually, it is not efficient enough to count every single photon, but rather some percentage of the total number of photons hitting it). The better (and more expensive!) the CCD, the higher this percentage is. Of course, having a 100% efficiency would be wonderful (as well as impossible!). When the contents of all the pixels are dumped out (the technical word is "read out") after exposing the CCD to a light source (like a galaxy), a digitized image of that object is produced. CCD images have many advantages over photographic film. Perhaps the greatest advantage is that the image is in digital form, and can be mathematically manipulated via a computer. Digitizing photographic images by scanning them can cause loss of resolution, and also introduce more uncertainty in the scanned flux values. Not to mention that this is one more time-consuming step that a CCD completely eliminates. Another big advantage that CCDs have over photographic film is that CCDs have a "linear response". What this means is that the number of counts collected in the pixels (after accounting for bias and dark current) is proportional to the exposure time of the image. Loosely speaking, for a CCD, a 10-minute exposure will have twice the number of counts that a 5-minute exposure would have, after taking into account dark and bias effects.

HOW A CCD WORKS

The CCD chip is a solid state device. Although the purpose of this lab is to characterize the performance of our CCD camera (and not to investigate the details of solid state physics), it is useful to have at least a rudimentary understanding of how a CCD knows when light has hit it!

A typical CCD camera consists of a silicon substrate, overlaid with an insulating layer of silicon dioxide, upon which an array of closely-spaced electrodes (which are set at a certain voltage level, or "gain") is placed. These electrodes are transparent to light, and are made of "polysilicon".

When light falls on the CCD, a fraction of the incident photons will pass through the polysilicon and silicon dioxide layers into the silicon substrate. Now, this silicon substrate is a type of material that hangs onto some of its outer (valance) electrons quite "loosely". Just a slight amount of energy (>1.14 eV) acquired by one of these outer electrons can displace the electron into the so-called "conduction band", which you can think of as being a "sea" of free electrons inside this semiconductor material. Because of these loosely-attached electrons, the photons that penetrate the substrate knock some of the outer electrons from the atoms. The number of dislodged electrons will be proportional to the number of incident photons. In a good CCD, as many as 60% of the incoming photons will knock an electron loose. The freed electrons (now in the conduction band) then migrate rapidly to the nearest electrode, where they collect in the electric potential well generated by the electrode. After a period of time (the exposure time), the pixels are read-out by the computer and the electrodes are cleared to allow a fresh batch of electrons to accumulate.

The material that a CCD is made of is wavelength-dependent. For example, photons of long wavelength (1.2 microns) will not have the minimum required energy to eject a silicon electron into the conduction band, and so the detector will not be able to record light beyond these wavelengths. For such light (which is in the infrared), CCDs are typically made of materials with smaller "band gap energies", such as germanium (having a band gap energy of 0.55 eV, which is the minimum required energy to eject an electron from the valance band). For photons having very small wavelengths (with energies > 10keV), their wavelengths are so small that the chance of interacting with a valence electron is greatly decreased, and so again the detector becomes "blind" to such photons.

WORKING AROUND THE "IMPERFECTIONS" OF A CCD

As you are already aware, there are several sources of "artificial" signal that a CCD generates by itself, that must be subtracted out in order to reveal the true amount of light that is coming from a particular astronomical object. Below, I will go through each source in detail:

1.     BIAS

As stated above, there is a certain amount of electric potential difference that must be imposed on the chip so that each pixel acts as a little "bucket" for gathering electrons. Well, the very act of putting a potential difference on the chip causes there to be some electrons sitting in the pixels, even if no light is hitting the CCD detector. Now, these electrons do not build up more and more over exposure time. In other words, the number of bias electrons remains constant no matter how long you expose the CCD chip, as long as the electronics connected to the chip do not change.

So how does one experimentally determine the so called "bias level"? Since you only want any electrons from any other sources to join the "bias" electrons, then the only way to get a "head-count" of the bias electrons is to read the chip without making an exposure. In other words, a bias level is obtained from an image of zero exposure time. What this will give you is a bias count for each pixel in the chip (we assume that the bias count will be different for each pixel, since the electonics will be slightly different).

2.     DARK CURRENT

Dark current is different from the bias level, in that the dark current can build up with exposure time. A "dark count" is the number of electrons that are randomly sucked down into the potential wells of CCD pixels, without the assistance of a photon. What actually happens is that the thermal motion of the silicon atoms themselves tends to knock off some electrons, in addition to the electrons knocked off by photons. The more time that goes by, the more likely that such an occasional event will take place. So, a 2-second exposure will have less dark counts than a 5-minute exposure. Fortunately, most modern CCD's have a very small dark count. Dark count is measured in terms of number of counts per second.

So how does one experimentally determine the "dark count"? The best way is to create a long exposure, with the chip in complete darkness. Of course, there will be bias electrons along with the dark electrons. But after accounting for the bias electrons (since you took a bias image), you can then figure out the number of dark electrons per second, for each pixel in the CCD. As for the bias count, we assume that each pixel can have its own unique value for the dark current.