«RICHARD B. LARSON Yale Astronomy Department, Box 208101, New Haven, CT 06520-8101 1 Introduction: Basic Problems Galaxies are, in their observable ...»
STAR FORMATION AND GALACTIC EVOLUTION
RICHARD B. LARSON
Yale Astronomy Department, Box 208101, New Haven, CT 06520-8101
1 Introduction: Basic Problems
Galaxies are, in their observable constituents, basically large bound systems of stars and
gas whose components interact continually with each other by the exchange of matter and
energy. The interactions that occur between the stars and the gas, most fundamentally the continuing formation of new stars from the gas, cause the properties of galaxies to evolve with time, and thus they determine many of the properties that galaxies are presently observed to have. Star formation cannot be understood simply in terms of the transformation of the gas into stars in some predetermined way, however, since star formation produces many feedback eﬀects that control the properties of the interstellar medium, and that thereby regulate the star formation process itself. A full understanding of the evolution of galaxies therefore requires an understanding of these feedback eﬀects and ultimately of the dynamics of the entire galactic ecosystem, including the many cycles of transfer of matter and energy that occur among the various components of the system.
Present data show strikingly that the structure and dynamics of the Galactic interstellar medium (ISM) are extremely complex, and it is clear that no simple model can represent adequately all of its important properties. This complexity is well illustrated, for example, by the results of the recently completed Leiden/Dwingeloo survey of the atomic hydrogen which is the dominant component of the Galactic ISM (Hartmann 1994; Hartmann & Burton 1995);
six images representing diﬀerent velocity slices from this survey are shown in the special color section of this volume. Even a superﬁcial glance at these results shows immediately that the ISM is in a violently turbulent state, and on closer inspection one sees that features such as bubbles, loops, and ﬁlaments are almost ubiquitous, demonstrating the role of explosive energy input from massive stars in keeping the ISM in a vigorously bubbling and turbulent state.
Also evident in this and other surveys is the intricate wispy structure of interstellar clouds, which resembles that of many terrestrial clouds. Observation of the terrestrial clouds with wispy shapes shows that they are in a highly dynamic state and that individual features are very transient, constantly forming, changing in structure, and evaporating. Much of the cloudy structure of interstellar clouds must also be very transient, so it is important to understand both the dynamical processes responsible for forming and structuring interstellar clouds and the thermal processes by which they may exchange matter with the surrounding more diﬀuse medium.
The system of interacting stars and gas phases in a galaxy may resemble in some respects a biological ecosystem, in which a great variety of interactions and cycles of exchange of matter and energy can occur among the various components. This makes possible very complex behaviors for the system, but under normal circumstances such an ecosystem tends to evolve toward a quasi-steady state in which all processes are more or less in balance and everything is eﬀectively regulated by everything else, making it diﬃcult to separate cause and eﬀect. Galaxies, too, tend to settle into such quasi-equilibrium states in which they evolve only slowly and the entire complex system behaves in a regular and predictable way. This is demonstrated, for example, by the UBV colors of galaxies, which provide evidence about their star formation histories; although some peculiar galaxies may show a large scatter in color, the great majority of normal galaxies fall along a single well-deﬁned sequence in the two-color diagram and have colors that apparently depend on a single parameter which is closely correlated with Hubble type (de Vaucouleurs 1977; Larson & Tinsley 1978). This color sequence is well accounted for by simple models of galactic evolution in which the only parameter that varies is the decay time for a postulated exponential decline of the star formation rate with time; the observed color range is reproduced if this decay time varies from less than 2 Gyr for elliptical galaxies through about 10 Gyr for intermediate-type galaxies like our own to eﬀectively inﬁnity for the latest-type spiral and irregular galaxies (Kennicutt 1986, 1992, and this conference; Larson 1991a, 1992b.) This trend reﬂects an increase along the Hubble sequence in the timescale for the conversion of gas into stars, which in turn probably reﬂects an increase along this sequence of the various dynamical timescales relevant to star formation; the most important underlying physical parameter may in fact just be the average density of galaxies, which decreases systematically along the Hubble sequence (Larson 1977, 1988, 1992b). In any case, it is clear that the timescale for gas depletion is the property of star formation that is most important for the overall evolution of galaxies. The various physical eﬀects that determine this timescale will be discussed further in Section 2.
A second basic property of star formation which, together with the timescale, determines the colors and various other observed properties of galaxies is the distribution of masses or ‘initial mass function’ (IMF) with which stars are formed. A question which has long been debated is whether the IMF is universal or whether it varies importantly with location; although this issue has not yet been completely settled, most of the present evidence seems consistent with a nearly universal IMF that has two basic characteristics: (1) an approximate power-law form for masses above one solar mass, similar to the power law originally proposed by Salpeter (1955), and (2) a turnover at a mass somewhat below one solar mass where the IMF begins to fall well below an extension of this power law (Scalo 1986; Larson 1991b, 1992b; Zinnecker, this conference).
The lower end of the IMF remains uncertain and may even be variable, but nevertheless it now seems clear that there cannot be much mass in very low-mass stars or ‘brown dwarfs’, and that with a standard IMF, most of the mass goes into stars with masses not very far from one solar mass. In other words, star formation makes stars with a characteristic mass of the order of one solar mass. This fact is obviously of great importance for astronomy generally, so it should be explainable by any quantitative understanding of star formation. The origin of this characteristic stellar mass and its possible relation to the properties of the interstellar medium will be discussed further in Section 3.
Since the alternative suggestion has been made that the typical stellar mass is determined by the internal physics of stars and not by the star formation process or the properties of the ISM (Shu, Adams, & Lizano 1987; Silk, this conference), it is worth noting that there is now some empirical evidence bearing on this question. An analysis of the spatial distribution of the T Tauri stars in the Taurus-Auriga clouds shows two distinct regimes, a regime of self-similar clustering at large separations and a regime of typically binary systems at smaller separations, with a clear break at a separation of about 0.04 pc (Larson 1995). The existence of this break implies that the clustering hierarchy is built up of basic units with a characteristic radius of about 0.04 pc, and this dimension thus appears as an intrinsic length scale in the star formation process. The associated star-forming cloud substructures can be identiﬁed with the dense ‘ammonia cores’ of the Taurus-Auriga clouds, which have typical diameters of about
0.1 pc and masses of about 1 M (Myers 1985, 1987). Since star-forming units of this size form on the average just two stars, i.e. binary systems (Larson 1995), the expected mass of an individual star is about half of this core mass or about 0.5 M, in good agreement with the typical masses of the observed T Tauri stars. Thus there is evidence, at least in the TaurusAuriga region, that star formation occurs in units of a characteristic size that is closely related to the typical stellar mass.
For the standard IMF mentioned above, only about 10 percent of the mass goes into those stars with masses greater than 10 M that are responsible for most of the important feedback eﬀects. This fraction depends on the form of the upper IMF, which is therefore another feature of star formation that is important for galactic evolution. Unfortunately, the properties of the upper IMF cannot be predicted at present because a quantitative understanding of the formation of massive stars does not yet exist, and is probably well beyond the current state of the art. The only approach to understanding this subject that is possible at present is therefore one based on phenomenology, and in this regard it is noteworthy that massive stars appear to form only in large clusters or associations along with large numbers of less massive stars. The formation of massive stars is thus evidently closely associated with the formation of clusters and associations. Some phenomenological aspects this subject will be discussed brieﬂy in Section 4.
2 Star Formation Rates and Interstellar Recycling
The star formation rate in a galaxy depends on two types of processes: (1) eﬀects that drive star formation by creating massive, dense star-forming clouds, and (2) negative feedback eﬀects that limit the eﬃciency of star formation by destroying these clouds before most of their matter has been turned into stars. Although star formation may in some circumstances also produce positive feedback eﬀects that stimulate further star formation (see Section 4), the net feedback eﬀect of star formation must be negative rather than positive, otherwise all of the gas in galaxies would have been consumed long ago in a runaway process lasting only a small fraction of the age of the universe.
2.1 Eﬀects Driving Star Formation
As reviewed by Larson (1988, 1992b), it seems likely that star formation is normally driven primarily by the large-scale self-gravity of the gas layer in a galaxy, perhaps assisted or organized by density waves. A gas disk is gravitationally unstable if the stability parameter Q = cκ/πGµ is smaller than about unity, where c is the velocity dispersion of random gas motions, κ is the epicyclic frequency, and µ is the surface density of the gas layer. Swing ampliﬁcation, a kind of truncated instability that ampliﬁes shearing density perturbations by a ﬁnite amount (Goldreich & Lynden-Bell 1965; Toomre 1981, 1990), can occur for somewhat larger values of Q up to about 2, and is probably the mechanism usually responsible for collecting the interstellar gas in galaxies into the large complexes or spiral arm segments in which most star formation occurs. The requirement that Q must be less than a critical value of the order of unity for such large-scale gravitational instability eﬀects to occur implies that, for given values of c and κ, large gas complexes should form and extensive star formation should ensue only if the gas surface density µ in a galaxy exceeds a minimum or threshold value. There is indeed evidence that signiﬁcant star formation in galaxies occurs only where the gas surface density exceeds a threshold value that agrees quantitatively with the value predicted if star formation is driven by swing ampliﬁcation eﬀects (Kennicutt 1989; Larson 1991a, 1992b).
The time required to collect the interstellar gas into large cloud complexes is expected to be comparable to the growth time of the above instability, τ ∼ c/πGµ = Q/κ, and this is about 50 Myr locally. The timescale for converting the gas into stars will, however, be much longer than this because star formation is observed to be a very ineﬃcient process that converts only about 2 percent or less of the gas in a star-forming complex into stars before the rest is dispersed by the processes to be discussed below (Myers et al. 1986; Leisawitz et al. 1989;
Evans & Lada 1991). In addition, with a standard IMF, only about half of the mass turned into stars remains permanently locked in low-mass stars, and the rest is eventually returned to the ISM by stellar mass loss; therefore only about 1 percent of the gas in a star-forming complex is permanently removed from the ISM. The expected timescale for gas depletion is then two orders of magnitude longer that the timescale for cloud formation, or about 5 Gyr locally. This predicted timescale for gas depletion is almost the same as the empirical timescale of about 7 Gyr inferred from the observed properties of galaxies like our own (Larson 1991a, 1992b), so it appears that this basic timescale for galactic evolution can be understood, at least in order of magnitude, if star formation is driven by the large-scale self-gravity of the interstellar medium.
2.2 Feedback Eﬀects and the Eﬃciency of Star Formation