10.1888/0333750888/1583
Starburst eaa.iop.org
DOI: Galaxies
T Heckman
From
Encyclopedia of Astronomy & Astrophysics
P. Murdin
IOP Publishing Ltd 2006 ISBN: 0333750888
Institute of Physics Publishing Bristol and Philadelphia Downloaded on Thu Mar 02 23:48:03 GMT 2006 [131.215.103.76]
Terms and Conditions
Starburst Galaxies
E N C Y C LO P E D IA O F A S T R O N O M Y AN D A S T R O P H Y S I C S
Starburst Galaxies
A starburst galaxy is one undergoing a brief episode of intense STAR FORMATION, usually in its central region. The massive stars in the burst generate most of the total luminosity of the entire galaxy. Starburst galaxies are fascinating objects in their own right and are the sites where roughly 25% of all the massive stars in the local universe are being formed. They offer unique laboratories for the study of the formation and evolution of massive stars, the effects of massive stars on the interstellar medium and the physical processes that were important in building galaxies and chemically enriching the INTERGALACTIC MEDIUM.
What is a starburst?
Overview There is no rigorous definition of a starburst galaxy, but several different criteria are often used. These are clear conceptually, but each is subject to uncertainties in its application. Starbursts are usually located in the centers of galaxies, and have typical radii of 1001000 pc (110% of the size of their `host' galaxy). Despite their small size, they are converting gas into massive stars at a rate that exceeds that found throughout the rest of their host galaxy. The starburst is `fueled' by a copious supply of interstellar gas (primarily in the form of molecular hydrogen) that has been accumulated in the center of the galaxy. The available gas supply is sufficient to sustain the current rate of star formation for only of order 108 yr (1% of the age of the universe). The dust grains associated with the molecular gas usually absorb most of the radiation produced by the burst's stars. This can make it difficult to determine many of the basic properties of starbursts. Star-formation rates We do not directly measure the star-formation rate in a starburst but must infer it from measurements of the starburst's luminous output. Two techniques are routinely used. The first is to use the bolometric luminosity of the starburst (Lbol ), which is the luminosity integrated over the entire electromagnetic spectrum. Lbol for a starburst is dominated by the emission from hot, massive, short-lived stars. These have masses greater than about 8 times the mass of the Sun (M > 8M ) with lifetimes of less than about 40 million years. The bulk of the radiation from these stars is emitted in the ultraviolet part of the spectrum. Dust grains in interstellar medium of the starburst absorb this radiation, are heated, and cool by emitting far-infrared radiation. Thus, a good approximation of Lbol is given by the sum of the starburst's ultraviolet and infrared luminosities. The second technique uses measurements of the amount of ionizing radiation produced by the starburst (often denoted by Q, the total number of photons produced per second that are capable of ionizing atomic
hydrogen). The production of these ionizing photons is dominated by the most massive, shortest-lived stars (M > 25M and lifetimes less than 7 million years). Measurements of Q are based on measurements of the rate at which H ions and electrons recombine, which in turn is measured by the luminosity of the so-called recombination emission lines of hydrogen (after the luminosity is corrected for absorption due to dust inside the starburst). Given a measurement of Lbol or Q, we can estimate the mass of the population of massive stars required to produce this quantity. If we divide this mass by the lifetime of the appropriately massive stars we obtain an estimate of the mean formation rate of the massive stars. This technique tells us almost nothing about the rate at which low-mass stars (like our Sun) are being formed, since these stars contribute only a negligible amount to Lbol or Q. Indeed, it is entirely possible that starbursts form only massive stars (unlike the mode of star formation in normal galaxies like our own). If we assume that starbursts form a normal complement of low-mass stars, the implied starformation rates usually range from 1M to 100M yr-1 . The burst intensity A useful way to define a starburst is to consider the burst intensity--the rate of star formation per unit area ( SFR -- typically given in units of M per year per kiloparsec2 ). In normal star-forming galaxies like our own Milky Way, the star-formation rate is a few M per year throughout a galactic disk with a radius of about 10 kpc (e.g. SFR has a typical value of 10-2 M yr-1 kpc-2 ). In a typical starburst galaxy, the star-formation rate would be 10M yr-1 in a region with a radius of 0.5 kpc. The implied SFR is then 10M yr-1 kpc-2 , or 103 times greater than in a normal galaxy. It is important to emphasize that there is no particular `magic' value for SFR that separates normal and starburst galaxies. A continuum of values is observed, spanning at least 6 orders of magnitude from the most quiescent starforming normal galaxies to the most intense starbursts. Although calculating a global value for SFR is a useful way to quantify a starburst, star formation is not uniformly distributed throughout the bursting region. Instead, the star formation occurs both in compact (few parsec scale) star clusters and in a more smoothly distributed mode. The most massive CLUSTERS (the `super star clusters') have estimated masses of (105 106 ) M and may be close analogs to young GLOBULAR CLUSTERS. The burst duration A starburst is by definition a transient event. The duration of a starburst ( t) must be much smaller than the age of the galaxy in which it occurs ( t 1010 yr). Another way of stating this requirement is that the present rate of star formation must greatly exceed the past rate averaged over the age of the galaxy. Unfortunately, it is difficult to accurately determine how long starbursts last.
Copyright Nature Publishing Group 2001 Brunel Road, Houndmills, Basingstoke, Hampshire, RG21 6XS, UK Registered No. 785998 and Institute of Physics Publishing 2001 Dirac House, Temple Back, Bristol, BS1 6BE, UK
1
Starburst Galaxies
E N C Y C LO P E D IA O F A S T R O N O M Y AN D A S T R O P H Y S I C S
Perhaps the most commonly used technique to estimate t is to calculate the gas-depletion time: the mass of interstellar gas in the starburst divided by the present rate of star formation. This is then a rough estimate of how much longer the starburst can be sustained before running out of gas. Gas-depletion times in starbursts are usually of order 108 yr, but they are highly uncertain for many reasons. For one thing, the mass of molecular gas is difficult to determine to better than a factor of a few since the determination relies on indirect arguments and observations of a trace molecule (usually CO). The estimated molecular gas masses range from 108 M to 1010 M , increasing as a function of the luminosity of the starburst. In addition, the star-formation rate itself is quite uncertain, primarily because we do not know the rate at which low-mass stars are being formed (see above). More sophisticated estimates of the duration of starbursts utilize measurements of Q, Lbol and the starburst mass plus detailed information on the stellar population (for example the relative numbers of massive main sequence and post-main-sequence stars such as red supergiants and WolfRayet stars). These measurements are then compared with models for the evolution of a population of massive stars (see STELLAR EVOLUTION). Resulting estimates for burst duration range from a few million years for the smallest and least powerful starbursts (which occur in dwarf galaxies), up to 107 or 108 yr for powerful starbursts. More generally, the minimum possible duration of a starburst is set by considerations of causality: the duration of star formation cannot be significantly less than the time for gas on one side of the starburst to `communicate' with gas on the other side. Such `signals' will travel at speed of sound in the gas or the velocity induced by the gravitational field in the starburst. These velocities will be of order 102 km s-1 , so t > 107 yr for a starburst with a diameter of 1 kpc. The most powerful and intense starbursts seem to turning gas into stars as fast as allowed by causality.
stellar photons with energies sufficient to photoionize neutral hydrogen (E > 13.6 eV, or < 912 ). The subsequent recombination of hydrogen or helium ions and resulting radiative cascade produce H and He recombination emission lines. The free thermal electrons in the gas can also collisionally excite ions, whose radiative decay produces emission lines. Many of the strongest lines of both types are in the visible part of the spectrum, and together constitute a few per cent of the bolometric luminosity of the starburst. Thus, another way to find starbursts is to search for galaxies with unusually bright emission lines. The dust grains in the starburst are effective at absorbing ultraviolet photons of all wavelengths. The grains are heated by this radiation and cool by emitting radiation. The equilibrium temperatures that result from balancing heating and cooling rates are usually in the range 10100 K, and the emitted radiation therefore lies in the mid- and far-infared spectral region ( 30300 m). The survey by the INFRARED ASTRONOMY SATELLITE (IRAS) in the 1980s (which surveyed nearly the entire sky in the mid and far infrared) has produced the most extensive and best-studied sample of starbursts. These three types of surveys select samples of starbursts that overlap one another but nevertheless have important systematic differences. The most direct difference is that dusty starbursts are preferentially detected in the far-infared surveys, while the less dusty starbursts are preferentially found by the ultraviolet and emission-line surveys. Dust grains are made of elements heavier than H or He (`metals'). Thus, the dust-content of starbursts correlates well with chemical composition: typically only a few per cent of ultraviolet radiation escapes `metal-rich' starbursts (having chemical composition like the Sun) while the majority of the ultraviolet light escapes the most `metal-poor' starbursts. Metal-poor starbursts tend to be less powerful to and occur in smaller and less massive galaxies (as a result in part of the well-known massmetallicity relation for galaxies).
How do we find starburst galaxies?
The massive stars that power starbursts are mostly very hot (T 20 00050 000 K) and emit most of their radiation in the ultraviolet between 912 and roughly 2000 . Thus, one technique for finding starbursts is to search for galaxies that are unusually bright in the ultraviolet. The spectral region below about 3200 is inaccessible from the surface of the Earth, so most ultraviolet surveys for starbursts in the local universe have been conducted in the near-ultraviolet region just longward of the atmospheric cut-off. This will change in the near future as spacebased ultraviolet imaging surveys are conducted at shorter wavelengths. Starbursts are rich in interstellar gas and dust (the raw material for star formation). The primary radiative output from the massive stars (ultraviolet radiation) is absorbed by this interstellar material which then re-radiates the energy in other forms. The gas absorbs nearly all the
What causes a starburst?
From a purely empirical point of view, the causes of starbursts become increasingly clear for starbursts of greater and greater luminosity. The most luminous starbursts in the local universe are the so-called `ultra-luminous infrared galaxies', which have bolometric luminosities of roughly 1012 L (nearly all of which is emitted in the midand far-infrared). The power source is deeply buried inside a dense, dusty region of molecular gas only a few hundred parsecs in size and may consist of a combination of a starburst and a dust-enshrouded QUASAR. The mass of molecular gas (1010 M ) is comparable with the entire mass of the interstellar medium in a big SPIRAL GALAXY. These ultraluminous galaxies almost invariably have highly disturbed morphologies that are strongly suggestive of the ongoing or recently completed merger of two large DISK GALAXIES. Specific morphological structures indicative of mergers include long narrow `tidal tails'
Copyright Nature Publishing Group 2001 Brunel Road, Houndmills, Basingstoke, Hampshire, RG21 6XS, UK Registered No. 785998 and Institute of Physics Publishing 2001 Dirac House, Temple Back, Bristol, BS1 6BE, UK
2
Starburst Galaxies
E N C Y C LO P E D IA O F A S T R O N O M Y AN D A S T R O P H Y S I C S
of stars and gas (the remnants of the outer disks of the merging galaxies) and double nuclei. For less powerful starbursts the evidence suggests that more mild gravitational interactions between galaxies are an important triggering mechanism (e.g. the close passage of two galaxies, without their subsequent merger). During the close passage of two galaxies, tidal stresses act to strongly perturb the orbits of the stars and gas in the galaxy disk. The dissipation of kinetic energy as gas collides with gas allows the gas to become sufficiently displaced from the stars that gravitational torques act between the stars and gas to transfer significant amounts of angular momentum from the gas to the stars. The gas can thereby flow into the center of the galaxy, where it can fuel a starburst. If the passage of the two galaxies is slow and interpenetrating enough, dynamical friction can transfer enough kinetic energy from the stars to the galaxy darkmatter halos to allow the two galaxies to merge into a single galaxy. Such mergers or strong interactions should take a few times the galaxy rotation period (about 109 yr), with the intense starburst phase being significantly shorter (see GALAXIES: INTERACTIONS AND MERGERS). These timescales are loosely consistent with independent estimates of starburst lifetimes discussed above. However, many starbursts are not found in obviously interacting systems. In such cases, a stellar bar is often present in the inner disk of the starburst galaxy. A bar can act to rob gas of its angular momentum, and transfer gas into the center of the galaxy where it can fuel a starburst. The mechanism for triggering starbursts is most uncertain in the lowest-power starbursts, which occur in DWARF GALAXIES. In some cases it appears that the dwarf may have recently collided with an extragalactic gas cloud, but in most cases there is no clear evidence for any type of interaction or stellar bar.
Thus, starbursts are indeed highly significant components of the universe, past and present. Starbursts as analogs to high-redshift galaxies As described above, starbursts in the local universe are often selected by either their ultraviolet or their farinfrared continuum emission or by visible-band emission lines. At high redshifts (z > 2) the rest-frame ultraviolet, visible, and far-infrared emission from a star-forming galaxy will be observed in the visible near-infrared, and submillimeter spectral regions respectively (see GALAXIES AT HIGH REDSHIFT). With the Hubble Space Telescope and modern 10 m class ground-based telescopes operating in the visible and near infrared, and rapid advances in submillimeter astronomy, it is now possible to detect and study such high-redshift galaxies. The ultraviolet-selected galaxies at high redshift strongly resemble similarly selected local starbursts: they have similar values for SFR , similar ultraviolet colors (suggesting a similar amount of reddening due to dust absorption), and their ultraviolet and visible spectra show that they have similar stellar populations and gas dynamics. One important difference is that the regions of star formation in the high-redshift galaxies are typically larger and more luminous than in ultraviolet-selected local starbursts. Depending on the uncertain corrections for dust extinction, the ultraviolet luminosities of the most powerful high-redshift galaxies imply star-formation rates that can reach several hundred M per year over a region a few kpc in size. To date, less is known about the nature of the submillimeter-selected galaxies at high redshift. The available information suggests that these objects resemble the local ultraluminous galaxies described above. It appears that the physical processes that we can study in considerable detail in local starbursts are directly applicable to high-redshift galaxies. Galactic 'superwinds' One of the most important processes that has been o...