How a Spectrograph Works

First, let us get some basic definitions, often encountered in spectroscopy, out of the way. A spectrograph is a device that can produce a graph of the intensity of light as a function of color or wavelength (i.e., a spectroscope that produces a graph). A spectrometer is a device that measures only one selectable color, whereas a monochromator is a device that transmits only one color. The basic components of a single prism spectrograph are shown in Figure 8.3. Essentially, the aim is to gather as much light as possible from the star being observed (and not from anything else), split the star's light up into a spectrum, and focus the spectrum. If a single, narrow beam of light from an intensely bright point-like source like the sun was being examined, all you would need is a chink of light and a prism. However, for astronomical spectroscopy with a telescope, where the star is much fainter (and there may be other stars nearby), you need to channel parallel light from the star through a prism and then use a lens to focus the red end of the spectrum at one end of the CCD detector chip and the blue end at the other. This is the simplest, most efficient and practical way to capture the spectrum. Moving from left to right in the figure, we first come to the slit. In a normal telescope, this is where the eyepiece would focus or the CCD would be placed, that is, the focal plane, where the image of the star field exists. The purpose of the slit is to reduce background noise from the rest of the sky and to reduce any overlap from adjacent wavelengths. The narrower the slit, the better the spectrum is resolved, but, if the slit is narrower than the focal plane star diameter, light will be lost. The collimator is simply a lens designed to ensure that parallel light enters the prism. Once the parallel light has been split into a spectrum by the prism, the spectrograph's own mini telescope

3 CP

Amateur Spectrograph
Figure 8.3. The basic components of a spectrograph. Diagram courtesy of Prof. Chris Kitchin.


Figure 8.4. The production of multiple spectra of different orders by a diffraction grating. Diagram courtesy of Prof. Chris Kitchin.

lens, or imaging lens, easily focuses the red light to one end of the CCD and the blue to the other; so the spectrum is nicely spread out along the chip.

So, those are the basics dealt with. In practice, though, there are many variants. For example, the prism can be replaced with a diffraction grating, which disperses the light in a slightly different way. With a diffraction grating, dispersions are conveniently greater than with a single prism (older spectroscopes often used several prisms in sequence) but they produce two sets of spectra, each with several "orders" of spectra (see Figure 8.4). The majority of the light goes into the white light "zero order" spectrum. Thus, the spectra are not as bright. However, if the grating is of the "blazed" type (more often found in reflection gratings), the indi-

Figure 8.5. SBIG's autoguiding spectrograph attached to an SBIG ST7 CCD camera. Light travels from A (telescope interface) via a slit to B (mirror) to C (collimating mirror) to D (the grating carousel). The spectrum produced is then directed to the second half of the collimating mirror (E), which then focuses it into the CCD camera (G) via another mirror (F). Photo: Courtesy Maurice Gavin.

Figure 8.5. SBIG's autoguiding spectrograph attached to an SBIG ST7 CCD camera. Light travels from A (telescope interface) via a slit to B (mirror) to C (collimating mirror) to D (the grating carousel). The spectrum produced is then directed to the second half of the collimating mirror (E), which then focuses it into the CCD camera (G) via another mirror (F). Photo: Courtesy Maurice Gavin.

vidual grating line surfaces are angled to direct the majority of the light into the y spectrum. Obviously, to take advantage of this, the grating has to be angled accu- aop rately to direct the bright spectrum at the detector. ovsc

The next issues to be addressed are how well can the spectrum be resolved, what rnotr focal length should the spectrograph telescope lens be, and how much of the spec- a. <u trum will fit onto the length of the CCD? With the typical prisms or gratings avail- «a «¡^

able to amateurs, the middle of the visual spectrum can be resolved as finely as 1 angstrom. But these same prisms or gratings typically disperse the spectrum such that 1 angstrom of the spectrum subtends an angle of, say, only 2 arc-seconds. Thus, the spectrograph's imaging lens will need a focal length of a meter to capture 1 Angstrom of resolution per 10 micron CCD pixel. At this scale, however, a 500-pixel-long CCD array will only capture 500 angstroms of visual spectrum, compared with the whole visual spectrum of 4,000 to 7,000 angstroms, that is, 3,000 angstroms. SBIG's Self-Guiding Spectrograph (Figure 8.5) features a choice of two dispersing diffraction gratings, offering 1 angstrom per pixel and 4.3 angstroms per pixel. But it also features an ingenious double concave mirror arrangement, which acts both as collimator and imaging lens and keeps the unit's size compact. However, for the homemade spectrograph builder who wants to keep the imaging lens focal length short, settling for a resolution of a few angstroms per pixel and using a CCD with small pixels will help keep the system compact. It is important not to get confused here between spectral resolution and dispersion. Let's look again at SBIG's Self-Guiding Spectrograph to clarify matters. The SBIG unit, like all spectroscopes, has a spectral resolution set by the diffraction grating's performance, but this can be compromised if the slit is widened (to reduce exposure times) and by instrumental deficiencies. However, to actually capture the resolution on the CCD, the dispersion and the focal length of the imaging lens/mirror must deliver a small enough "angstroms per pixel" scale. The SBIG Self-Guiding unit features a choice of two diffraction gratings of 150 and 600 lines per millimeter with corresponding resolutions of 10 and 2.4 angstroms with the narrow, 18-micron slit (18 microns = 2 arc-seconds at 2 m focal length). With the wide, 72-micron slit, the 150 and 600 line gratings deliver resolutions of 38 and 10

angstroms. The dispersions of these gratings, combined with the focal length of the imaging lens/mirror, give image scales of 4.3 angstroms per pixel with the 150 grating and 1 angstrom per pixel with the 600 grating. The image scale is always of finer resolution than the spectral resolution to ensure that all the resolution available from the instrument is captured at the CCD. Obviously, if the spectrum is analysed at a higher resolution on the CCD surface, the spectrum will be dimmer and longer exposures will be needed to capture a supernova's "Type" classification. In spectroscopy, everything is a trade-off between spectral resolution and brightness, but if your system reliably auto-guides on a star (like SBIG's SGS unit), long exposures can be combined with good resolution, and the crucial details can be resolved, even for supernovae that are below the visual threshold in the same instrument. Many amateur supernova discoveries are as faint as magnitude 17. With short exposures of a minute or two in length, it is virtually impossible for amateur spectrographs to get this faint on 0.3- or 0.4-m apertures. In practice, telescopes in the 1- to 2-m range are needed to routinely obtain such spectra with relatively short exposure times.

For the DIY spectroscope builder, optimum grating/prism assemblies are rarely available; likewise for the collimating and imaging lenses. It's usually a case of buying cheap components and bolting them together to see what happens. Amateur spectrographs are rarely designed precisely. Fortunately, diffraction gratings of 600 lines/mm can be purchased for as little as $25 and adjustable slits can be made from two razor blades. In addition, secondhand camera lenses can be called into service for the collimating and imaging lenses, leaving the CCD as the most expensive component. But there are other technical considerations, too. For example, how do you actually keep the telescope guided so that the star being analyzed is kept in the slit? One way of doing this is to focus a guiding eyepiece or telescope on the outer surface of the slit; this surface, if highly polished, will easily show the outer overspill of the star's disc. It is actually advantageous to let the star's right ascension drift trail back and forth along the slit length as this produces the height of the spectrum. With perfect tracking, the spectrum would be an almost infinitesimally thin line and very hard to analyze. The slit in the SBIG SGS unit is formed from two halves of a plane mirror, which reflects the image at the focal plane to the separate guiding CCD. Thus, while the main CCD collects the spectra, the guiding CCD shows the field, with a dark line (or white if back-illuminated) showing the position of the slit; perhaps the ultimate spectroscope luxury.

Few amateurs will want to spend $5,000 on the auto-guiding SBIG spectrograph, but, fortunately, the basic components of a spectrograph are easily available and just need loads of experimentation and patience to fine tune.

+1 0

Post a comment