The Formation of Star Clusters
Clouds in the summer sky provide clues about the organization of star populations
Hierarchical Groupings of Stars
Many star groups are visible with the naked eye or a modest telescope to anyone living under dark skies. The famous Pleiades cluster in the constellation Taurus and the double cluster (h and c Persei) in Perseus are obvious to even the most casual stargazer. Each of these open clusters consists of several hundred to a thousand stars that are gravitationally bound to each other in a volume of space no more than a few tens of light-years across. Their stars are typically born at about the same time (give or take a few million years) and may stay bound to each other for more than 100 million years. Located within the disk of the Galaxy, most of them are eventually disrupted by passing clouds of dense gas.
Other groups of stars in the Galaxy's disk, including many of the brightest stars in the constellation Orion, are more loosely distributed than the Pleiades cluster, with thousands of stars stretching across several hundred light-years. These giant stellar groups are not gravitationally bound, but can be recognized by the presence of bright, blue stars, known as O-type and B-type stars (according to their spectral properties). The stars in these so-called OB associations are recognized as belonging to a single group because they are concentrated in space and moving in the same general direction through the Galaxy. Since OB stars are massive, they burn their fuel very quickly, often within 10 million years of their birth. (At 5 billion years, our lower-mass sun is only halfway through its life span.) Such brief lives for OB associations make them convenient markers for the location of recent star formation.
Attempts to understand the distribution of open clusters and OB associations in the Galaxy led astronomers of the 1950s and '60s to turn their telescopes toward other galaxies in the hope that some would provide a better perspective. But identifying extragalactic OB associations proved to be challenging because individual stars in other galaxies were too faint to be characterized by their spectra at that time. An OB association in another galaxy could only be recognized by its relatively high stellar density.
The subjective methods of the astronomers produced a wide range of sizes for OB associations. In 1961 Paul Hodge and Peter Lucke of the University of Washington found that the average OB association in the Large Magellanic Cloud (a satellite galaxy of the Milky Way) was about 250 light-years across. But a decade earlier the Harvard astronomer Harlow Shapley found star groupings in the Milky Way's satellite that were 1,300 light-years across. Sidney van den Bergh, then at the David Dunlap Observatory in Toronto, found what he called "OB associations" that were 1,500 light-years across in the great galaxy in Andromeda (M31). At the time it was not clear whether these results were in conflict with each other or whether OB associations could actually have a wide range of sizes.
The reason for the diverging results became apparent in studies of OB associations in the 1970s and '80s. One of us (Efremov) and the Bulgarian astronomers Georgi Ivanov and Nikola Nikolov found small, compact groups inside nearly every large group of stars in the Andromeda galaxy. These small groups were also bluer and brighter, suggesting that they were younger than the large groups. The average size of the smaller groups was about 250 light-years, the same size as the OB associations in the Large Magellanic Cloud described by Hodge and Lucke. This is also about the size of the stellar associations in the Milky Way, including the Orion OB association.
How then could the large OB associations be explained? Since the small OB associations appeared to be parts of larger star groups in the Andromeda galaxy, the small Milky Way associations could be parts of larger groups here too. Evidence for this idea came from our work on Cepheid variable stars, massive stars that happen to be in a relatively late phase of stellar evolution as we observe them today. He found that Cepheid variables clump together in the Milky Way, much like OB stars, but on a larger scale, spanning 1,800 light-years or more. These large clumps, or star complexes, may be 50 million years old, about five times older than a typical OB association.
It turns out that the nearest star complex in the Milky Way corresponds to a great swath of stars surrounding us that was studied by the 19th-century American astronomer Benjamin Gould. This vast system, now called Gould's Belt, does indeed house several star-forming regions, including the Orion-, Perseus- and Scorpius-Centaurus OB associations. Furthermore, the Dutch astronomer Adriaan Blaauw recognized that there is also an older, dispersed association of stars, called Cas-Tau (after its location in the constellations Cassiopeia and Taurus). Star formation in the Cas-Tau association ended some 20 million years ago, so this grouping is now considered to be the first generation of star birth in Gould's Belt, whereas Orion and the other local OB associations belong to the second generation.
Efremov and his collaborators also determined that the large concentrations of OB stars found in the Andromeda galaxy by van den Bergh are clumps of Cepheids too, and are therefore much older than the smaller OB associations. These observations have been extended to other galaxies as well. The large, rich complexes of Cepheids in our Galaxy are the same size as giant "star clouds" detected along the spiral arms of several other galaxies. (Remarkably, the astronomer Frederic Seares suggested back in 1928 that these knots are similar to Gould's Belt.) It appears that about 90 percent of the OB associations and young clusters in the Milky Way, the Large Magellanic Cloud, the Andromeda galaxy and another nearby galaxy, M33, are all united into star complexes.
All of these observations imply that OB associations are merely the young cores of older and larger groupings of stars. They also suggest that there were probably other OB associations inside each Cepheid complex in the past, and that these older stars are now dispersed.
We now believe that star complexes are the fundamental scale for star formation in spiral galaxies and irregular galaxies, and that the construction of these complexes may extend up to the lengths of short spiral arms. Moreover, the hierarchical relation between OB associations and star complexes appears to continue in the opposite direction, toward smaller scales. Most OB associations contain clumps of smaller clusters (called subgroups by Blaauw), and these often contain even smaller collections of stars. (For example, the double cluster, h and c Persei, is part of an OB association discovered by William Bidelman of Case University in 1943.) These small collections are themselves made of multiple-star systems. All of these observations suggest that star formation is generally clumped into a hierarchy of structures, from small multiple systems to giant star complexes and beyond. OB associations are just one level in this hierarchy.
The hierarchical structure of star clusters helps to explain why astronomers had such difficulty defining star groups, since any such division is arbitrary. But it left an unsettling question: Why are stars arranged hierarchically when they form?