Joshua E. Barnes
Galaxies come in a bewildering variety of shapes and sites, but a large majority - perhaps 80% - possess a disk of some kind. Galactic disks are thin, basically circular distributions of stars, gas, and dust; this material moves on nearly circular orbits about a common center. Many disks exhibit beautiful spiral patterns as a result of this rotation, and some have pronounced bars crossing their centers. Other disks, however, are nearly featureless, and can only be identified by a characteristic falloff of brightness with radius.
The evolution of disk galaxies is inextricably bound up with the highly controversial problem of galaxy formation. This entry focuses on the history of disks in galaxies such as the Milky Way, as a way of distinguishing the present subject matter from larger issues. Progress in this field has come by combining detailed studies of the motions, compositions, and ages of stars in the solar neighborhood with less-detailed knowledge of the overall structure of the Milky Way and global observations of other galaxies. This approach, while productive, is fraught with uncertainties. Presently one cannot offer a definitive account of the evolution of disk galaxies.
The mere existence of a galactic disk has two basic implications. First, the gas from which the disk formed must have settled into circular orbits before the disk stars were born; once a star is formed its future path is determined entirely by gravitational forces, and gravity cannot circularize a random distribution of stellar orbits. Second, since the disk formed, the gravitational field has not undergone any sudden, dramatic changes, which would disrupt the circular pattern of stellar orbits. We cannot, however, rule out the possibility that the mass distribution, and hence the gravitational field, has evolved slowly.
The more massive a star, the brighter it shines, the bluer its color, and the shorter its lifetime. Thus it is relatively easy to tell the age of a system in which all stars formed at the same time by observing the colors of the brightest stars still on the main sequence. Stellar associations and open clusters in the disk of the Milky Way yield ages between 3 x 106 and 6 x 109 yr, and the oldest disk stars are at least 1010 years old. These widely ranging ages imply that star formation in the Milky Way started when the Universe was less than half its present age and continues up to the present time.
Observations of other galaxies show that the Milky Way is hardly unique in this respect. The broadband colors of a galaxy depend largely on the rate of star formation averaged over the last ~ 108 yr; higher rates of star formation yield bluer colors. Late-type galaxies (type Sc and Irr) indeed have rather blue colors, suggesting a constant rate of star formation. In very early-type disk galaxies (type SO), star formation has largely ceased, although it is hard to tell how long ago this occurred. Finally, the intermediate colors of galaxies like the Milky Way (type Sb) are consistent with a present star formation rate ~ 3 times lower than the average rate over the lifetime of the galaxy.
How long can galaxies continue to form stars at these rates? In all but the most extreme cases, the gas is 15% of the total mass in stars. This suggests that late-type galaxies are literally about to run out of gas, ending their phase of star formation. It seems unlikely, however, that we find ourselves at such a unique moment in cosmic evolution. Alternatively, galaxies may accrete gas from their surroundings, their present gas content representing a rough balance between income and expenditure. This accretion hypothesis solves several problems in galactic evolution, but at present there is little direct evidence that galaxies such as the Milky Way are accreting significant amounts of gas.
Some of the gas invested in stars is returned to the galactic reservoir as the stars age, having been enriched in metals (elements heavier than H and He). Massive stars very rapidly become Type II supernovae, spewing a wide range of elements back into the clouds from which they form, whereas close, intermediate-mass binaries may evolve into Type I supernovae, favoring production of iron-group elements. Modeling galactic chemical evolution is in principle just a matter of bookkeeping, but in practice uncertainties in the physics of evolved stars make detailed predictions difficult. Some simple models, however, can at least be ruled out.
Simplest is the closed-box model, in which gas neither enters nor leaves the galaxy; as the metals must be built up over time, the closed-box model predicts large numbers of metal-poor, low-mass disk stars. In fact, only 2% of the low-mass stars in the solar neighborhood have less than a quarter of the solar fraction of metals, compared to the 44% predicted by the closed-box model. This is known as the G-dwarf problem; the scarcity of metal-poor disk stars indicates that a closed-box model is inappropriate for our galaxy.
One obvious solution to the G-dwarf problem is to supply metals from the outside. In addition to the disk, the Milky Way has a spheroidal component, which is older than the disk and massive enough to have contaminated the protodisk with a significant quantity of metals. This disk-spheroid model also explains why the metal fraction of a disk typically increases towards the center, because that is where the bulk of the mass in the spheroid component is found.
Paradoxically, another way to solve the G-dwarf problem is to slowly but steadily build the disk from metal-poor gas. In this case the metal fraction soon reaches roughly the present value, and then remains constant. By the present epoch most stars will have formed during the phase in which the metal fraction is constant, and metal-poor stars will be rare.
Which solution is preferred? Data for F stars show the fraction in metals increasing with time up until ~ 3 x 109 years ago, and then leveling off. If real, this leveling off indicates that our disk has only just reached the constant-metals phase, suggesting a compromise between the disk-spheroid and accretion solutions. These two hypotheses may be complementary sides of the same story; both challenge the assumption that galactic disks are closed systems.
The Sun and nearby disk stars share a common orbital motion about the galactic center, but in addition each has a small random velocity, reflecting the fact that their orbits are not perfect circles. On the average, older stars have larger random velocities; for stars less than 109 Years old, rms velocities toward or away from the galactic center are R 10 km s-1, while for the oldest disk start, we find R 40-60 km s-1.
The most likely explanation for the trend of random velocity with age is that stars are born on nearly circular orbits, and are subsequently deflected onto more random orbits by fluctuations in the galactic gravitational field. This led Lyman Spitzer, Jr. and Martin Schwarzschild to postulate the existence of giant molecular clouds, long before such clouds were detected. Present calculations indicate that clouds of mass 106 M can produce random velocities of up to ~ 30 km s-1, but not the higher velocities seen in the oldest disk stars. In addition, the Spitzer-Schwarzschild mechanism predicts that random velocities grow rather slowly, roughly as t0.25, and the observations are better fit by t0.5. Another source of fluctuations is needed; transient spiral structure, to be discussed next, may fill the bill.
Many different kinds of spiral structure are seen in disk galaxies. Most photogenic are the grand-design two-armed spiral galaxies such as M51, but far more common are ragged or flocculent spirals made up of many short arms. The diversity of spiral galaxies is paralleled by the diversity of theories of spiral structure. Grand-design spirals are often discussed in terms of the Lin-Shu theory (after Chia-Chiao Lin and Frank H. Shu), which views the spirals as slowly turning wave patterns maintaining their form for many rotation periods. However, classic grand-design spirals like M51 often have close companions, and it is possible that such spirals are actually excited by tidal interactions. Flocculent spirals, on the other hand, are generally thought to evolve over time, with individual spiral arms constantly forming and dissolving.
Computer models of rotating disks can produce spiral patterns similar to those seen in real galaxies. In these models, thousands of particles represent the disk; each particle moves in the net gravitational field produced by all the others. If the particles start out in nearly circular orbits with small random velocities, striking multiarmed spiral patterns soon develop. These spirals result from the gravitational amplification of small fluctuations in a disk that rotates differentially (i.e., not like a solid body). As a result of these ever-changing spiral patterns, particles acquire increasingly large random velocities. After a few rotation periods the random velocities become large enough to shut off the gravitational amplifier, and the spiral-making activity dies away.
Transient spiral structure can in principle provide the fluctuating gravitational field needed to generate the random motions of old disk stars, and a theoretical analysis even predicts t0.5. However, the spiral activity seen in the simplest computer models lasts only a few rotation periods, whereas in real galaxies it persists more than 10 times as long. Computer experiments show that spiral structure can be maintained by adding stars to the disk on nearly circular orbits, consistent with the above discussion of random velocities. Moreover, the kind of spiral pattern produced depends on the rate at which stars are injected; high rates produce open, well-defined patterns typical of late-type spirals, and slower injection results in weaker, tightly wound spirals like those seen in early-type disk galaxies. These results support the view that accretion provides a disk galaxy with the shot in the arm needed to promote vigorous development of spiral structure, the type of the resulting spiral depending largely on the rate of accretion.
Complementing the mechanisms which build up galactic disks are those which destroy them. According to the accretion hypothesis, spiral galaxies are susceptible to starvation: If the inflow of raw material for new stars is cut off, the spiral soon fades, leaving a smooth disk resembling an SO galaxy. Indeed, SO galaxies are generally found in high-density regions where starvation is likely. A disk galaxy that has the misfortune to fall into a rich cluster may be swept clean of interstellar material by the ram pressure of the hot, low-density gas pervading such clusters. Alternatively, the overpressure of the cluster gas may compress molecular clouds within the galaxy, provoking a burst of rapid star formation. This process may account for some unusually blue galaxies observed in high-redshift clusters.
Finally, instead of accreting gas, a disk galaxy may ingest a companion. Computer simulations show that interactions between galaxies often result in mergers, the outcome depending on the mass ratio of the colliding systems. Large disk galaxies can swallow small companions, of less than ~ 10% their mass, with only minor damage: Random motions of disk stars increase, and the disks become thicker. A number of galaxies, including the Milky Way, are reported to have thick disks which may have been produced in this way. Mergers between disk galaxies of comparable mass have a very different outcome. So violent is the interaction that neither disk survives; such mergers may in fact produce elliptical galaxies.
Disk galaxies are thus rather fragile and delicate objects. The evidence, while fragmentary and largely circumstantial, suggests that these galaxies grow best in quiet, undisturbed locations where their disks can develop slowly without outside perturbations. When such galaxies become involved with others, they run the risk of violent transformation. But from the wreckage of such cosmic accidents a new disk galaxy may arise, given only time and a sufficient supply of raw materials.