Deconstructing the Milky Way Galaxy

Henry Freudenreich. American Scientist. Volume 87, Issue 5. Sep/Oct 1999.

Our sun resides in an unremarkable part of a vast stellar system, the Milky Way Galaxy. Shaped like a great circular disk, our galaxy has a breadth nearly 10 times greater than its thickness. Within the plane of the disk, long “arms”dense concentrations of bright, young stars, dust and giant gaseous clouds-spiral outward from its center. Astronomers think there are four spiral arms in the Milky Way, and the sun appears to be on a spur connecting two of them. Beyond this simple description, our species knows surprisingly little about the galaxy we live in.

The appearances of other galaxies in the universe provide few clues to the structure of the Milky Way. Remarkably, all spiral galaxies share only one trait in common-they consist of a flattened disk whose brightness falls rapidly with distance from its center-but otherwise they differ in almost every respect. Some are very dusty, some are almost clean. Some have two majestic out-spiraling arms, some a messy multitude of limb fragments. Some have a spheroidal nucleus in their center, some a lens-shaped bulge, some a cigar-shaped bar and some a combination of these or nothing of note. Such diversity makes for an interesting galactic zoo, but it provides few constraints on the structure of our own galaxy.

The central reason astronomers have been slow to understand the Milky Way is simply because we are deep in the thick of things: The other stars, the gas and especially all the dust in the disk prevent us from seeing the full extent of the galaxy’s structure. We are in the middle of a forest and we cannot make out its full extent for all the trees.

Astronomers have an advantage over the forest-bound cartographer, however, because we can exploit the fact that longer wavelengths of the electromagnetic spectrum can slip past the dust. Dust also gives its presence away by making stars appear redder than they should be and because it “glows” in the far-infrared. Knowing where the dust is, we can effectively subtract it from what we see.

I was able to exploit some of these properties as a team member on the Diffuse Infrared Background Experiment (DIRBE), a near-infrared telescope that was part of the Cosmic Background Explorer satellite (COBE) that began its mission in 1989. Maps of the entire sky made by DIRBE provided the clearest and most complete images yet of our galaxy’s central bulge. I used these maps to create a model of the Milky Way, which I describe here.

Mapping a Dusty Galaxy

The tenuous matter between the stars in our galaxy is about 90 percent hydrogen and 9 percent helium. The remaining 1 percent consists of heavier elements collected into fine particles astronomers call dust. Most of these particles are less than a micrometer long and are rich in carbon and silicates. Interstellar dust might be described poetically as the smoke and ashes of burned-out stars or prosaically as a mixture of ultra-fine sand and soot that contains some of the same carcinogens found in cigarette smoke. Although there is a diffuse distribution of dust throughout the galaxy, most of the dust is collected into ragged clouds of various sizes and densities. The interiors of the more massive clouds, which are shielded from starlight, are relatively cool and dense. Here hydrogen atoms may combine to form hydrogen molecules (H2), and for this reason they are known as molecular clouds. Molecular clouds are the birthplaces of stars.

Unlike stars, which only interact gravitationally, dust particles have complex dynamic interactions. They collide with each other and also with neutral and ionized atoms of gas, so they are subjected to hydrodynamic and magnetic forces in addition to gravity. Dust is also pushed around by stellar winds (mostly protons and electrons) and the photon pressure of starlight. Because of these forces, galactic dust exhibits a variety of large-scale features-huge bubbles, tendrils and wavy sheets-that are unique among the galaxy’s components. Some of these wispy dust formations are reminiscent of cirrus clouds on earth, and hence they are named cirrus. Graceful sweeps of galactic cirrus are evident in infrared images of our galaxy.

Lovely as these structures may be, dust itself poses a problem for the would-be galactic cartographer. Since dust can both absorb and scatter light—what astronomers call extinction-it hides much that is of interest. Extinction is greatest when the wavelength of light is the same size or smaller than the dust grains. The longer the wavelength, the fewer the grains large enough to affect it, and the lesser the extinction. In visible light, where the wavelengths are smaller than the great bulk of the galactic dust particles, we can barely see beyond the sun’s backyard.

At much longer wavelengths, in the radio part of the electromagnetic spectrum, extinction is negligible. Unfortunately, most stars are effectively invisible at radio wavelengths. Instead, the radio sky shines with the emissions of atoms and molecules in interstellar gas and with the radiation produced by energetic electrons. Astronomers have used the emissions of atomic hydrogen atoms and carbon monoxide molecules to map the distribution of gas over most of the galaxy. Such maps do not necessarily correspond to the distribution of stars in the galaxy, so radio surveys only provide a partial picture. Apart from the large-scale cirrus, gas and dust tend to settle into a thinner, broader layer in the galactic disk compared to stars.

The distribution of gas is also not so useful for mapping the inner part of the galaxy (within about 15,000 light-years of the center). Although radioastronomy has provided a wealth of data about the inner galaxy, making sense of it has been like putting together a jigsaw puzzle with missing pieces and no picture for reference. We do know that the density of gas and dust increases as we move inward, peaking at a distance of about 12,000 light-years from the center of the galaxy, where it forms a collection of clouds known as the “molecular ring.” The density is again lower inside the ring.

Radio observations of the emitting gas can also tell us its velocity along our line of sight-how fast it is approaching or receding. If the gas moves along a known trajectory, say a circular orbit, knowledge of its velocity in a given direction is enough to specify its distance. Such velocity maps can be used to plot the locations of the spiral arms and the molecular ring. Inside the molecular ring, the motions of the gas are peculiar and poorly understood.

Explorations in the Infrared

The infrared region of the spectrum would seem an obvious choice for galactic exploration, since many stars and heated dust shine brightly at these wavelengths. There are practical difficulties, however. Infrared detectors are noisy unless kept very cold, and the atmosphere both absorbs and emits infrared radiation, so it is best to lift the telescope above as much of the atmosphere as possible. In the 1970s, teams of Japanese astronomers used balloon-borne telescopes to map the inner galaxy at a wavelength of 2.4 micrometers. The Japanese experiments produced the first pictures of the galactic bulge (the massive central portion of the galaxy), revealing a bright ovoid, about 20 degrees wide and 10 degrees high, which was somewhat flattened at the top and bottom.

In 1983 the Infrared Astronomical Satellite (IRAS) carried a telescope into orbit. IRAS observed in the mid-to-far infrared wavelengths (about 8 to 120 micrometers) at which stars are dim and dust is bright. However, some stars in a late stage of their evolution form thick shells of warm dust that could be detected by IRAS from halfway across the galaxy. Plotting their positions in galactic latitude and longitude created a “pointillist” picture of the Milky Way, revealing a narrow, slightly warped disk (not perfectly flat) with a boxy bulge in its center.

The existence of the bulge was firmly established by the late 1980s, but its three-dimensional shape and its nature were not. Most astronomers considered the bulge to be an oblate spheroid, similar to the galaxy’s halo-the vast, faint region that surrounds the entire galactic disk, and which consists of ancient globular clusters and widely scattered stars. Because the bulge was also thought to be ancient, it was believed that the halo and the bulge were part of the same structure. But there was strong evidence that the inner galaxy is not so simple. Odd motions of interstellar gas in the bulge hinted that it was not a spheroid, but rather an elongated bar, similar to that seen in some other spiral galaxies. Indeed, the molecular ring 12,000 light-years from the galactic center is reminiscent of rings found encompassing the bars in other spiral galaxies. Near-infrared surveys that counted stars in the bulge also suggested that the stars were distributed in a “cigar-shaped” bar.

Further support for the notion that the Milky Way was actually a barredspiral galaxy came from the DIRBE. The DIRBE was a small telescope that observed simultaneously in 10 infrared bands, ranging from 1 micrometer to 300 micrometers. Although DIRBE had a modest 19-centimeter primary mirror, its correspondingly large field of view is exactly what is needed when the goal is to map the entire sky several times in less than a year. The DIRBE’s near-infrared sky maps provided stunning low-extinction images of the Milky Way and the most complete images yet of its central bulge. The maps were made in four bands: J (1.25 micrometers), K (2.2 micrometers), L (3.5 micrometers) and M (4.9 micrometers). These bands are dominated by red-giant stars, which represent a late stage in the evolution of a wide variety of stars (the bulk of those that lie along the so-called main sequence). Since they are bright and relatively common, red giants are excellent tracers of galactic structure.

In the DIRBE sky maps the bulge resembles an oblong box. In the J and K bands there seems to be a small notch in the top of the bulge, which led some astronomers to speculate that the bulge was actually shaped like a peanut. However, Janet Weiland, Eli Dwek and other DIRBE team members at NASA’s Goddard Space Flight Center proved that the notch was attributable to extinction by a molecular cloud in the foreground. The boxy shape of the bulge could be explained by a long bar pointing 13 degrees away from our line of sight to the galactic center.

A Model of the Disk and Bar

I was a member of the DIRBE team at this time, but my attention was turned toward the outer galaxy Here the Milky Way seemed “warped.” Warps in a galaxy’s layer of gas and dust are fairly common, and they are sometimes also detected in the faint outer regions of a galaxy’s stellar disk. Dark matter-an unobserved something whose presence is inferred from the fact that galaxies have more mass than the mass of their known constituents-is considered to be responsible. The Milky Way is embedded in a much larger and more massive body of dark matter. Computer simulations show that if a galaxy’s disk is tilted with respect to the symmetry plane of the dark matter, or rotates about a different axis, it will warp.

What I had noticed is that the stellar disk of the galaxy seems to bend upward on one side and downward on the other side. A similar effect had been noted in star counts and in the IRAS pointillist map, and it was also known that the gas layer is warped, though not necessarily to the same extent. Are the gaseous and stellar warps the same? Or could it be that the solar neighborhood is just coincidentally tilted with respect to the rest of the disk? We don’t know.

The DIRBE sky maps themselves tell us nothing about an object’s distance. The DIRBE summed the light from all sources regardless of distance. Consequently, interpreting a DIRBE map requires a model that describes how the sources are distributed in space. This is basically the amount of starlight produced per cubic light year as a function of position. I made such a model of the galactic disk and eventually expanded it to include the inner galaxy.

It is more difficult to model the disk than to model the bar alone because the disk is much fainter than the bar, and because we are immersed in the disk. One of the confounding factors is the zodiacal light, the glow of dust within our own solar system. Throughout much of the sky the zodiacal light is comparable in brightness to the galactic disk. In a map based on galactic coordinates, the zodiacal light forms a sigmoid-shaped feature that traces the path of the sun across the sky. There is no such feature in the images shown here because the DIRBE team modeled the zodiacal light and subtracted it from the sky maps. I took particular care to remove any residue.

Our location within the disk is problematic because it lends undue weight to nearby sources, which may not be representative of the galaxy as a whole. The DIRBE could see individual stars only within a distance of about 1,500 light-years, a relatively small volume in which statistical fluctuations in star numbers and types could be significant. Therefore I filtered out all stars that appeared as points, and in the modeling program accounted for the loss by omitting locally produced starlight. Nearby molecular clouds are also troublesome. They are too irregular and complicated to model and they only describe the local structure of the dust layer. Most lie within “Gould’s Belt,” an association of bright young stars, which is tilted about 18 degrees with respect to the galactic plane. Fortunately, molecular clouds give themselves away by their color-as determined by the ratio of one wavelength band to another. A high ratio of the L band intensity compared to that of the K band is characteristic of molecular clouds. So I could remove molecular clouds from the model simply by avoiding regions in the sky where the L/K ratio is high.

With so much data available, a model of the disk and the bar could be made that was more sophisticated and realistic than any before. The modeled disk could permit a warping of its outer region and a non-circular hollowing of its center, effectively a “hole,” as is seen in many barred spirals. The model allowed a variety of bar forms, including an abrupt cutoff on the bar’s long axis and a shape that ranged from rectangular through elliptical to a diamond. And if the model included more than one wavelength band, the differences in extinction between the bands would allow it to describe the distribution of the dust layer.

There was no guarantee that one model would fit all four sky maps in the J, K, L and M bands. For example, one confounding possibility was that stars of differing intrinsic brightness might be segregated at different heights above the galactic plane. In one widely used model of the Milky Way, the brighter stars (such as red giants) are held to be more concentrated on the midplane of the disk, whereas faint stars (such as dwarfs) range more widely, and stars of middling brightness are dispersed to an intermediate degree. Even though red giants dominate all near-infrared bands, stars of different types and intrinsic brightness contribute different fractions of the total brightness in each band. If this model were true, the basic structure of the galaxy might appear different in each of the bands mapped by the DIRBE. After examining the data that went into this model, however, I concluded that the segregation of stars according to their brightness was the result of studying different types of stars at different distances from the midplane. Dwarfs near the midplane were compared to giants far above the plane. Comparing the DIRBE sky maps at different wavelengths reinforced this conclusion. In directions where there is little extinction, the maps are identical.

My model consists of a disk, a bar and a smooth layer of dust. According to the model, the disk is indeed warped, much like the gas layer, right to its edge—about 15,000 light-years outward from the sun. At that distance the number of stars drops sharply. The disks of other spiral galaxies are known to have a sharp outer edge, but counts of individual stars in our own galaxy had suggested this limit to be a few thousand light-years farther off. (The dust layer, like the gas, should extend well beyond the edge of the disk, but by that point the dust’s average density is too low for it to affect infrared observations.) The bar points 14 degrees from the line connecting the sun and the galactic center, and is in other respects similar to that found by Dwek and the DIRBE team. The sun is 54 light-years above the midplane, in good agreement with estimates made by others based on the fact that the sky appears fainter when looking up away from the plane than when looking down through it.

Some other current models of the Milky Way postulate a smaller bar and a disk without a hole. Sure enough, if the hole is removed from my model, there are fewer stars left for the bar, and it shrinks. (The quality of the fit to the data shrinks as well, however.) I find the number of stars in the bar to be roughly equal to the number the disk loses by having a hole. In other words, the stars that would have been in the hole go to form the bar. This, plus the fact that disk and bar have the same near-infrared color, support the theory (now widely accepted) that the bar formed from the disk.

My model fits the data very well. The mathematical description of the disk has a great deal of observational and some theoretical support; however, the characterization of the inner galaxy is purely empirical. There was actually more than one model. I tried three classes of function to describe the bar, and three to describe the hole in the disk. The best combination is presented here, but all led to very similar results, which argues that although I cannot claim to have the best possible model—the “truth”—I have probably arrived in its general vicinity.

Milky Way Leftovers

There is more to the Milky Way than a disk, a bar and a smooth layer of dust, as we see when the modeled version is subtracted from the real thing. Nearby stars, spiral arms, ragged dust clouds and the cumulative light of distant, young bright stars remain. The bright patch in the constellation Cygnus (longitude 85 degrees, latitude 0 degrees) appears because we are looking lengthwise through the collection of dust clouds that form the local spiral arm segment, or a spur reaching out from it. The most interesting feature is the bright “ridge” across the inner galaxy in the galactic plane. The ridge is brighter in the positive longitudes. Some of this excess brightness may be due to the Sagittarius Arm (the next arm inward from the sun), which lies in our line of sight of the galactic center. For that matter, an arm closer to the center of the galaxy or a segment of the molecular ring (which itself may be composed of arms) may contribute to the observed brightness. At the center of the galaxy lies a small bright nucleus. Could this be a second bar? There is no way of telling from these data if the nucleus is a distinct component of the galaxy or the product of an inadequately modeled bar.

We are progressing nicely, now that the DIRBE has provided a picture for the jigsaw puzzle. There are still some pieces missing, however, and others that have probably been jammed into the wrong spots. Other attempts to paint the Milky Way with a broad brush include automated surveys of red giants in the near infrared bands and dynamical simulations that attempt to both match the DIRBE sky maps and explain the data derived from radioastronomy This last approach is essential if we are ever to understand the Milky Way, as opposed to just describing it.

We need more sophisticated simulations and, above all, more facts to test them. It will take long years of painstaking work identifying objects-stars of different types, molecular clouds and planetary nebulae-determining their distances and cataloging them, before we can say with certainty, “This is what the Milky Way looks like, and this is our place in it.”