Rebirth

The remarkably sensitive CCD camera has also become a tool for amateur astronomers, and it has reawakened their interest in spectroscopy. Now a modest backyard telescope can study the planetary atmospheres of Venus, Titan, Uranus, and Neptune, or record the redshifts of quasars at the edge of the universe. The power of the Internet makes it possible to post an observation worldwide almost instantly, invoking a vital and stimulating exchange of news and ideas.

Early amateur spectroscopists did not have to contend with the modern scourge of light pollution. Fortunately, most of the faint targets of interest to amateurs are starlike point sources. Placing a slit at the entrance of the spectroscope will help mask degrading skylight. It is also possible to take a series of short CCD exposures and add them together for a long effective exposure. Both techniques are used successfully by amateur spectroscopists observing from suburban locations.

There is always a tradeoff between spectral resolution (the detail visible within a spectrum) and the brightness of the target. Spreading light into a long spectrum improves the resolution but greatly reduces the intensity. The trick is to match the spectrograph to the task at hand, since no single instrument will cope with every application.

In the 1980s I built a solar spectrograph that produced a spectrum 40 inches long and could resolve detail as fine as 0.04 angstrom on film. Segments of this colorful spectrum were captured in snapshots of less than 1/10 second. Recently I recorded the spectra of distant quasars several billion times fainter than the Sun using a simple CCD spectrograph and exposures less than an hour. The complete spectrum was just 2 millimeters long with a dispersion of 40 angstroms per pixel on the camera's CCD chip. Nevertheless, this low resolution was still adequate for detecting the quasar's strong hydrogen emission lines shifted toward the red end of the spectrum.

Getting precious photons onto the CCD is the prime objective of stellar spectroscopy. Light can be dispersed into a spectrum with a prism or diffraction grating, and each method has pros and cons. The grating can be either a reflection or transmission type. While a prism disperses light into a single spectrum, a grating creates multiple spectra of varying lengths and intensities (normally only the brightest one is used, so the remaining light is lost). A grating, however, conveniently produces a spectrum with an essentially linear dispersion, while a prism compresses the spectrum at red wavelengths.

When one is working with small dispersions and very low resolutions, a target's real image (the so-called zero-order spectrum) produced by a grating can be recorded together with its spectrum on the CCD chip. The separation between this image and the spectrum is fixed and serves as a convenient reference when one is measuring wavelengths. This is a novel application, since it eliminates the need to have a reference spectrum (usually formed by feeding light from an emission lamp into the spectrograph, and this is not possible when the spectrograph has no slit).

A CCD with high quantum efficiency and broad spectral response is desirable but often quite expensive. Nevertheless, the chips in today's popular CCD cameras are more than adequate for spectroscopy, and they offer a huge improvement over conventional photographic film. CCDs also have extended red sensitivity compared with film. For example, my Starlight Xpress camera with its Sony camcorder chip has fairly even response between 3970 angstroms in the violet and 9300 angstroms in the near-infrared.

There is a small downside to the CCD's extended sensitivity. Some spectrograph designs use a conventional lens from a 35-millimeter camera to focus the spectrum onto the CCD. The focus of these lenses will usually need to be adjusted when you are recording the infrared region of the spectrum, as few lenses yield sharp images at visible and infrared wavelengths using the same setting.

A low-resolution spectrograph for point-source targets can be made by placing a prism or grating just ahead of a CCD at a telescope's focal plane. Because of its thickness, however, a prism must be placed in a collimated beam of light and the resulting spectrum focused onto the CCD with an additional lens. The results are very similar to those obtained by using a costly objective prism in front of the telescope's aperture. A single exposure records the spectra of all stars in the field of view. However, just as when using the telescope for astrophotography, exposure durations are limited by skyglow.

Such a system works well with telescopes of 2-meter focal length and less. Because the resulting spectrum is only a few pixels high (regardless of its length), the spectrum is kept relatively bright and requires the minimum exposure. The narrow spectrum can be electronically stretched with image-processing software to show as a traditional rectangle, which easily reveals any bright or dark lines.

High spectral resolutions with dispersions of, say, 4 angstroms per pixel, typically require advanced equipment designs that incorporate a slit at the spectrograph's entrance. The tracking accuracy needed to keep a target on the narrow slit is a daunting task for a typical amateur telescope.

There are, however, three ways to solve this problem. The slit can be aligned east-west in the telescope's focal plane since most periodic drive errors are in this direction. As such, errors simply broaden the spectrum while the detail remains sharp. Another technique is to have some form of autoguider to keep the target on the slit. The last method is to use a fiber-optic cable to pipe light from a circular area in the focal plane to a "slit" formed by positioning the cable's individual fibers in a line at the spectrograph's entrance.

The autoguider method is the one chosen for the new commercial spectrograph developed by Santa Barbara Instrument Group (SBIG). The fiber-optic approach is used in the Nu-VIEW spectrograph sold by Sivo Scientific and based on a prototype described in the February 1998 issue of Sky & Telescope (page 134). You can find out more about these spectrographs at www.sbig.com and www.sivo.com/sivosci.

The introduction of two spectrographs aimed at amateur astronomers is a promising sign. Nevertheless, because the demand is relatively small the market is not overwhelmed with choice. As a result, many amateurs build their own equipment. In addition to keeping costs down, this approach is both rewarding and instructive.

Just about anyone with a CCD camera can make an almost instant introduction to astronomical spectroscopy with the purchase a transmission grating, such as the one sold by Rainbow Optics. Placed an inch or so in front of the CCD chip and attached to a telescope (preferably f/6 to f/10), it yields satisfactory low-resolution spectra that will have any would-be spectroscopist up and running after an evening or two.

How faint a star will a spectrograph record? This depends on many factors, including the telescope aperture, spectrograph design and dispersion, and the CCD. There is, however, a simple way to estimate what exposure is required to capture a star of given magnitude. Start by finding the proper exposure necessary to record the spectrum of a bright star. Then for each whole-magnitude drop in brightness, you multiply the exposure duration by 2½ times.

With my 12-inch Meade Schmidt-Cassegrain telescope and a Rainbow Optics transmission grating arranged to yield a low-resolution spectrum having a dispersion of 40 angstroms per pixel, I can record the spectrum of zero-magnitude Vega with a 1/100-second exposure. A 6th-magnitude star requires 1 or 2 seconds, and a 15th-magnitude source needs about an hour-long exposure. This holds quite well to my "rule." For higher resolutions, say 4 angstroms per pixel (which is enough to detect the radial velocities of stars in the solar neighborhood), amateur equipment is usually confined to naked-eye stars.