INTRODUCTION

The aim of this paper is to briefly explain what is meant by spectroscopy—stellar spectroscopy in particular—and to summarize the results of an actual research project involving the observation and photography of the spectra of two first-magnitude stars, namely, Sirius and Betelgeuse. The theoretical function of spectroscopes is also described, as well as types of this instrument which are currently utilized by astronomers.

Among the earliest experiments on spectroscopy were Isaac Newton's 1666 observations of the solar spectrum. He used a prism to disperse the Sun's rays, obtaining a luminous band with a continuum of colors, very much like a rainbow. Though he never detected features such as lines or bands in the solar spectrum, he correctly established that what we call white light is actually a mixture of six primary or fundamental colors.

The first detailed observations of the solar spectrum were made by the German optician Josef von Fraunhofer in 1814. Using a prism spectroscope he was able to see a large number of dark absorption lines in the Sun's spectrum (which had been discovered by the British chemist William H. Wollaston around 1802, but disregarded as gaps between colors). Though lacking the necessary theoretical knowledge to account for these lines, he catalogued about 600 of them [Freedman, Kaufmann, 2002], introducing in the process an alphabetical classification system that is still sometimes used.

Later, Fraunhofer made pioneering observations of some of the brighter stars, which became a puzzle to him because of the significant differences he noticed among the various stellar spectra. It would not be until half a century later that an appropriate explanation was given for the mysterious lines seen in the stars' spectra.

Years of experimentation—and classification of the lines of the elements known at the time—revealed the existence of three fundamental kinds of spectra. These are correspondingly described by what we now call Kirchoff's laws:

1st law— A hot solid, or an opaque gas under high pressure, produces a continuous spectrum (that is, a complete array of colors devoid of any dark or bright lines).

2nd law— A hot, transparent gas, produces an emission-line spectrum (that is, a spectrum characterized by bright, narrow lines against a dark background, which depend on the elements present within the gas).

3rd law— A cool, transparent gas, in front of a source of a continuous spectrum, produces an absorption-line spectrum (that is, a complete array of colors with dark, narrow lines, depending on the elements present within the gas).

With Kirchoff's laws it was finally understood that the emission spectra seen from vaporized substances in laboratories and the absorption spectra seen in stars were actually related to each other. It may be said that the modern science of spectroscopy was born as a result of these three laws.

Another half a century later, in 1913, the Danish physicist Niels Bohr would give a new model of the atom that would finally explain why specific elements actually absorb or emit light at discrete wavelengths. He stated that electrons within an atom do not move randomly around the nucleus, but in orbits at specific distances where their position is stable. A change or transition from an orbit into another would require a change in energy, which would then be manifested as a photon, namely, a particle of light. An electron jumping from an inner orbit would then emit a photon with a specific wavelength—corresponding to its energy—thus causing an emission line, while an electron moving to an outer orbit would, in turn, absorb a photon, producing an absorption line at the corresponding wavelength. We call these jumps electronic transitions.

Later, in the 1890's, the American astronomer Edward Pickering—together with Annie Jump Canon, Williamina Fleming, Antonia Maury, and others—began compiling the well-known Henry Draper Catalogue, which—once finished—would give spectral types for about 225,300 stars [Kaler, 1998, Stars]. Pickering's early system was similar to Secchi's scheme, but used instead capital Roman letters, running from A to Q. Annie Jump Canon later refined the system, dropping a few letters, to get the standard O-B-A-F-G-K-M spectral sequence that we have today, and also added sub-classes—labeled as numbers from 0 to 9—for better precision.

The modern spectral sequence is essentially a two-ended continuum defined by the relative strength of hydrogen absorption lines in a star's spectrum. Because the electronic transitions that occur in the atoms of stellar atmospheres greatly depend on temperature, the sequence is also one of temperature and color.

The standard spectral classification system—as finally defined by Canon—runs as follows:

Spectral class O—
Blue stars with temperatures over 31,000 K, sometimes called helium stars. Spectrum is characterized by ionized and neutral helium absorption lines, with little or no hydrogen.

Spectral class B—
Blue-white stars with temperatures ranging from 9,750 to 31,000 K. Spectrum is characterized by neutral helium absorption lines—no ionized helium—with hydrogen becoming more noticeable.

Spectral class A—
White stars with temperatures ranging from 7,100 to 9,750 K. Spectrum is characterized by very strong hydrogen absorption lines—the Balmer series, strongest between A1 and A2—with no helium lines. Ionized metals begin to appear in the more latter types.

Spectral class F—
Yellow-white stars with temperatures ranging from 5,950 to 7,100 K. Spectrum is characterized by hydrogen absorption lines—not as strong as in class A—and ionized metals, such as the Calcium II lines (Fraunhofer H and K).

Spectral class G—
Yellow stars with temperatures ranging from 5,250 to 5,950 K. Spectrum is characterized by weakening hydrogen absorption lines and strengthening metallic lines, in both neutral and ionized states.

Spectral class K—
Orange stars with temperatures ranging from 3,950 to 5,250 K. Spectrum is characterized by strong metallic lines—Calcium II is strongest in K1—both in neutral and ionized states. Molecular lines (mainly Titanium Oxide) begin to appear in latter types.

Spectral class M—
Red stars with temperatures ranging from 2,000 to 3,950 K. Spectrum is characterized by neutral metal and molecular lines. Titanium Oxide absorption bands are very prominent. Hydrogen lines are gone.

[Temperature values in the table above have been taken from Kaler, 1998, Spectra.]

Other spectral types exist, such as WC and WN for blue, hot Wolf-Rayet stars, featuring emission lines in their spectra, and classes C—formerly N and S—and R, for red giant variables. Two new classes, namely L and T, have been recently introduced for cool dwarfs shining feebly in the red and infrared parts of the spectrum. These feature absorption lines and bands due mostly to molecules such as metallic hydrides and methane.

Now that we have seen the fundamentals of stellar spectroscopy, I will be describing the kind of instrument that actually makes this possible, and present a few results of what actually can be done even with relatively modest equipment.

Two types of spectroscopes are now currently in use among astronomers: prism-based spectroscopes and grating-based spectroscopes. While early spectroscopes were mostly of the prism type, most units now employ exclusively diffraction gratings (or in special cases, a combination of both). A brief description of each type follows:

Prism-based spectroscopes—
This was the kind of device originally used by Newton and by major observatories of the late 19th and early 20th centuries. A prism is a triangular-shaped piece of glass using the principle of refraction—as stated by Snell's law—to disperse a light beam into its components. This happens because a ray of light is refracted (or bent) as it passes from air to glass, and this refraction is slightly different, depending upon the specific wavelength of light.

One big disadvantage of prism-based spectroscopes, however, is that the incoming light is not linearly dispersed, meaning that the distances between the various constituent wavelengths are not equal. The blue end of the spectrum tends to be stretched, while the red end is compressed.

Grating-based spectroscopes—
Their invention is credited to Fraunhofer, but it was not until well into the 20th century that these devices came into frequent use. A diffraction grating is a solid surface with a large number of evenly spaced, parallel lines etched on one side (usually from 200 to 1,000 per mm). It uses the principle of diffraction—as stated by Bragg's Law—to disperse the constituent colors of light. This happens because a ray of light is subjected to bending by diffraction (or interference) as it reaches the grooves on the grating, to an amount that is dependent, again, upon the specific wavelength of light.

Gratings are further classified as either reflection or transmission gratings. Reflection gratings are made of solid, metallic surfaces—where light rays are reflected—but they are very fragile. Transmission gratings, on the other side, let the light rays pass through them. They are more resistant, but also very expensive, which is why they are not frequently seen in astronomical applications.

While grating-based spectroscopes—unlike prisms—actually generate a proportionally dispersed spectra, they do suffer from the fact that the resulting dispersion is spread among a number of different spectra. Successive, fainter spectra are referred to as a first-order spectrum, second-order spectrum, and so on. This effect, of course, may give rise to significant light losses.

An important part of many spectroscopes is the slit, a narrow rectangular aperture, whose primary purpose is to act as the entrance gate of the spectroscope. A slit is necessary to obtain the spectra of extended objects such as nebulae and galaxies, but at the same time that it cuts through an object's excessive size, it also reduces its brightness, allowing the detection of between 1 to 10% of the incident light [Kitchin, 1998]. Stellar spectroscopy may sometimes be done without a slit, allowing in this case the detection of as much as 75% of the incident light.

All kinds of spectroscopes in use today operate either on the refraction or dispersion principle. Many special adaptations of the prism and grating types—have been constructed, however, to suit special needs. Echelle spectroscopes, for example, use diffraction gratings in which the surface grooves are further apart than with ordinary gratings. This allows spectra of high dispersion to be obtained, though limited to a small wavelength range. A more complete spectrum may be obtained, however, by overlapping the successive wavelength ranges produced by adjacent orders of the spectrum. The spectrohelioscope (an instrument invented in 1924 by the American astronomer George Elery Hale) uses a diffraction grating along with several lenses to achieve a very high dispersion image of the Sun, which is tunable over the whole range of visible light.

Prisms and gratings have been combined to produce hybrid spectroscope designs. A grism, for example, is a right-angle prism with a transmission grating at its base. Grisms have been successfully used with telescopes where in-line presentation of the spectrum is required—the Hubble Space Telescope NICMOS camera, for example—but like prism spectroscopes, they suffer from non-linear dispersion.

It has also been noted that an ordinary CD or DVD compact disk—which has approximately 600 grooves per mm—may act as a very modest diffraction grating, revealing some of the more prominent features of the solar spectrum [Tonkin, 2002]. Pictures have even been taken this way showing a good number of absorption lines, but precautions should be taken not to view the Sun directly using this technique.

Instrumentation—
I used a Celestron 200 mm f/10 Schmidt-Cassegrain telescope, together with a 26 mm Plössl eyepiece and a standard Celestron (mirror-type) star-diagonal. A Nikon FM-2 manual SLR camera was utilized for recording the spectra, using Fujicolor Superia 400 color negative film, and Kodak Tri-X 400 black and white negative film. The spectroscope itself consists of a diffraction grating for stellar spectroscopy made by Rainbow Optics, which screws into the bottom of a standard 1.25-inch eyepiece. This is a transmission grating with about 200 grooves per mm [Gavin, 2003], and it has been blazed to concentrate about 75% of the incoming light into just one of the first-order spectra. The spectroscope also includes a cylindrical (or astigmatic) lens which elongates the resulting spectra to allow for direct visual observation.

Observing site and circumstances—
The observing site is located within the urban area of San Juan, Puerto Rico, at a longitude 66° 04' west of Greenwich, and latitude 18° 22' north of the equator. The observations were carried out during the nights of November 18, 2003 and November 20, at approximately 02:30 and 01:30 local time, respectively. Thin high clouds—the remains of a strong low pressure system that affected the eastern Caribbean region during the first two weeks of November—prevailed during each of the observing sessions, but did not appreciably affect the visual or photographic appearance of the spectra.

Technique—
I targeted two first-magnitude stars of different spectral type: Sirius, a bluish-white, main-sequence star, and Betelgeuse, a well-known red supergiant. Each of the two stars was observed within one hour of its local transit, so that they would be visible at a sufficiently high elevation above the horizon (55° and 79°, respectively), in order to minimize any possible atmospheric degradation. I attached a 26 mm eyepiece to the telescope, which gave an exit pupil of 2.6 mm—ideal for spectroscopic work—and then screwed the diffraction grating at the far end of the star diagonal, in order to produce a longer, better resolved spectral image. Focusing was made by using the star itself, although the final precise focus was achieved by moving the focuser very slightly to the inside—which is recommended by the grating's manufacturer—in order to allow the stellar absorption lines to become sharply focused.

The SLR camera was mounted on a separate tripod and pointed to the telescope by means of the afocal coupling technique. The resulting focal length of the system was 3,850 mm and the f-ratio, f/19.25. Four exposures were made of each spectrum—two each, in both color and monochromatic film—and the time was set at 8 seconds with no tracking (to allow the spectral image to gain width by the natural drifting due to the rotation of the Earth). The film was taken to a local photo laboratory for standard developing, and the resulting spectrograms were then scanned directly from the negative strip using a Nikon CoolScan IV film scanner. These were later computer-processed with the scanner's own NikonScan v3.1 software for increased contrast, and were also partially rotated to achieve a full horizontal orientation, using Corel PhotoHouse v2 software.

The spectrograms which resulted from my research appear below.

Figure 1. Color spectrogram of Sirius. Copyright © 2003 Armando Caussade.

Figure 2. Monochromatic spectrogram of Sirius. Copyright © 2003 Armando Caussade.

Figure 3. Color spectrogram of Betelgeuse. Copyright © 2003 Armando Caussade.

Figure 4. Monochromatic spectrogram of Betelgeuse. Copyright © 2003 Armando Caussade.

Although, at first sight, color spectrograms appear prettier (see figures 1 and 3), it becomes clear that the absorption lines reveal themselves much better in black and white. The color images also show a number of artifacts (dark bands at the transition areas of the main colors) that would actually hinder an analysis of the spectra [Tonkin, 2002]. So, I chose instead to use the monochromatic images for the analysis.

These spectrograms show a clear set of absorption lines and bands from which a spectral type may be ascertained. [Comparisons were made with sample spectra from Tonkin, 2002]

Sirius' spectral image, for example (see figure 2), has a very prominent set of lines that are—without a doubt—those of the Balmer series, which is a spectral signature of ionized hydrogen. Six such lines are clearly seen, converging rapidly towards the blue end of the spectrum, with a faint seventh and eighth line also visible at close inspection of the negative. No helium or metallic lines, however, appear to be visible. So, according to the spectral type descriptions given previously, it may be safe to establish Sirius' spectrum as of type A.

Betelgeuse's spectrogram (see figure 4) looks very different. Instead of lines, we mostly see dark bands, which is a signature of cool, red stars. The three prominent bands in the green—near the middle of the spectrum—are, very probably due to Magnesium Hydride and Titanium Oxide. The dark, wide feature on the right is harder to identify, but may be the Titanium oxide band located at 547.0 nanometers. No hydrogen lines appear to be present. Spectra such as this begin to appear in late class K—progressing into class M—but because of the strength of the molecular bands, it seems logical to establish Betelgeuse's spectral type as of type M.

A number of additional stars were also inspected visually, by attaching the cylindrical lens which came with the diffraction grating in front of the eyepiece. Canopus and Procyon, for example, also exhibited a set of Hydrogen (Balmer) lines, but these appeared weaker than those of Sirius, which is consistent with their spectral type of F0 and F5, respectively. Regulus—at spectral type B7—also showed a weak, although distinct set of Balmer lines.

The most striking of all spectra which I observed, however, was that of the 2nd magnitude Wolf-Rayet star γ Velorum. This object barely rises above the horizon for observers in the northern hemisphere (its elevation at the time of the observation was only about 15°) but nonetheless, its spectrum showed a very intense emission line in the blue, and two other bright, twin lines in the yellow. The British astronomer Ralph Copeland said—around 1867—of this star's spectrum, that it is "...incomparably the most brilliant and striking in the whole heavens."