The starburst galaxy, M82, has one of the earliest known and best studied examples of a galactic wind. Both mechanical luminosity in the form of stellar winds and supernovae and radiative luminosity from starbursts are feedback mechanisms that can potentially disrupt star formation and end the starburst phase. Yet there are galaxies, such as the Antennae, observed to have thousands of SSCs spread over regions of hundreds of pc extent, with cluster ages spanning many tens of Myr during tidal interactions lasting ~ 100 Myr. Evidently, episodes of intense star formation can take place over periods of time far longer than the lifetimes of individual massive stars in spite of feedback. The subject of galactic winds and feedback is a large and active one, and has been recently reviewed by Veilleux et al. (2005). The effects of feedback on denser molecular gas in ESF regions is not yet well characterized, and it is the molecular gas from which the future generations of stars will form.
Starburst feedback and star formation occur on different spatial scales. Giant molecular clouds consist of clumps that are governed by turbulence; only a small fraction of these clumps contain cores, which are the regions that collapse to form stars or star clusters (McKee and Ostriker 2007). Current computational models of turbulent clouds can explain the canonical star formation efficiencies of a few percent as that fraction of turbulent clumps that become dense enough for gravity to dominate (Padoan and Nordlund 2002, Krumholz et al. 2006, Padoan et al. 2007). Once a core begins to collapse, free fall is rapid and there is little time for feedback to operate. It is more likely that feedback operates on longer-lived and lower density molecular cloud envelopes dominated by turbulence, but how this large scale effect communicates down to the small and rapidly collapsing star forming cores is unclear (Elmegreen 2007, Krumholz and Tan 2007, Padoan et al. 2007). If rich star clusters form stars for several dynamical times, the energetic input feedback could become important (Tan et al. 2006).
One surprising characteristic of the interstellar medium in regions of ESF is the ubiquity of dense (ne ~ 104-5 cm-3), "compact" H II regions, a stage of star formation that should be fleeting and relatively rare. First detected spectroscopically in dwarf galaxies, such as NGC 5253 and He 2-10 (Turner et al. 1998, Kobulnicky and Johnson 1999, Beck et al. 2000), these nebulae are detected by their high free-free optical depths at cm wavelengths. These are the ESF analogs of dense Galactic "compact" H II regions, only much larger in size because of the high Lyman continuum rates from these large clusters. If the expansion of H II regions is governed by classical wind bubble theory, then the dynamical ages implied by the sizes of these H II regions are extremely short, tens of thousands of years. M82-A1 is a young SSC with both a visible H II region and a cluster, in which the dynamical age of the H II region is too small for the cluster age (Smith et al. 2006). The H II region around the Galactic cluster NGC 3603 is also too small for the cluster age (Drissen et al. 1995). These may be scaled-up versions of the classic Galactic ultracompact H II problem, in which there are "too many" compact H II regions given their inferred dynamical ages (Dreher et al. 1984, Wood and Churchwell 1989). A possible explanation for the long lifetimes of these H II regions is confinement by the high pressure environment of nearby dense (nH2 > 105 cm-3) and warm molecular clouds (e.g., de Pree et al. 1995, Dopita et al. 2005, 2006). Radiative cooling may provide an important energy sink for the output of young super star clusters (Silich et al. 2007). However, the standard wind-blown bubble theory that is generally applied to the development of Galactic H II regions (Castor et al. 1975, Chevalier and Clegg 1985) may not always apply to the H II regions surrounding SSCs, which are massive enough to exert a non-negligible gravitational pull on their H II regions (Kroupa and Boily 2002). The "supernebula" in NGC 5253 appears to be gravity-bound, if not in equilibrium (Turner et al. 2003). Gravity could either stall the expansion of an SSC H II region's expansion or create a cluster wind akin to a stellar wind, depending on conditions. It is clear that the high interstellar pressures in starburst regions are critical to their development, and to the evolution of the nearby molecular clouds.
Molecular gas has a tremendous ability to absorb energy and radiate it away. This could explain the ability of galaxies to sustain starburst events of extended duration such as the one that has produced the thousands of SSCs in the Antennae. Irradiation causes heating, ionization, dissociation, and pronounced chemical changes in molecular clouds. It also provides us with a rich spectrum of potential diagnostics of radiative feedback.
In Figure 6 is shown a schematic of the ionization structure of a photodissociation region (PDR), adapted from Tielens and Hollenbach (1985). (For a full description of PDRs, see Tielens 2005 or Hollenbach and Tielens 1999 and also Tielens and Hollenbach 1985, Bertoldi and Draine 1996, Draine and Bertoldi 1996, Kaufman et al. 1999, Kaufman et al. 2006.) Molecular clouds tend to form H2 at AV 1. The translucent edge to the molecular cloud can be quite warm, a few hundred K, warm enough for excited H2 to be visible. Between AV = 1 and AV ~ 5, while the clouds are molecular, they also have significant abundances of heavy ions such as C+ and S+. The presence of ions can drive a rich gas-phase chemistry through ion-molecule reactions. The high temperatures can also liberate molecules that have formed on the surfaces of grains in the form of ices in the coldest clouds. Warming the grains in either shocks or radiatively in PDRs brings these molecules into the gas phase. PAH emission is also bright from these regions, and dominates Spitzer images in the 8 µm IRAC band where it shows a close association with star-forming regions (Peeters et al. 2004, Galliano et al. 2008). These chemical diagnostics have been used to great effect in modeling the effects of protostars on their surrounding protostellar disks, and in determining the shapes and orientations of disks (van Dishoeck and Blake 1998). Clouds near sources of high X-ray radiation are subject to a similar phenomenon, but with chemistry that is driven by a hard radiation field; these regions are called "XDRs" (Maloney et al. 1996, Lepp and Dalgarno 1996, Meijerink and Spaans 2005, Meijerink et al. 2006).
Figure 6. Schematic of the ionization structure of the Orion photodissociation region (PDR), with relative elemental abundance plotted versus visual extinction. Adapted from Tielens & Hollenbach (1985) and Tielens (2005).
It might seem that processes that occur on scales of Av ~ 1 - 5 would be difficult to detect in other galaxies, on the spatial scales of GMCs, but this is not the case. Molecular clouds are porous, and there are many surfaces within clouds of relatively low Av individually; an estimated 90% of molecular gas is in PDRs (Hollenbach and Tielens 1999). Some of the first indications of the importance of PDR chemistry were the detections of warm CO and the tracers of warmed, potentially shocked, gas such as the C II 158 µm line in starburst galaxies (Mauersberger and Henkel 1991, Stacey et al. 1991). Temperatures as high as 400 - 900K have been inferred from lines of interstellar ammonia (Mauersberger et al. 2003).
Lines of numerous heavy molecules have been detected from other galaxies, and the brightest sources are the star-forming galaxies. Molecules such as formaldehyde (H2CO), methanol (CH3OH), and cyanoacetylene (HC3N) have been detected in nearby starburst galaxies (Martín et al. 2006, Mangum et al. 2008, Meier and Turner 2008). Many of these models can provide discriminants between PDR and XDR-heated gas (Meijerink et al. 2006, Aalto et al. 2007). A spectral line survey of NGC 253 in the 2mm atmospheric window reveals 111 identifiable spectral features from 25 different molecular species; the spectrum suggests that the molecular abundances in NGC 253 are similar to those of the Galactic Center, with a chemistry dominated by low-velocity shocks (Martín et al. 2006, 2008).
Imaging adds another dimension to the molecular line spectra. In Figure 7 are shown interferometric images in several molecules of the nuclear "minispiral" of the nearby Scd galaxy, IC 342. The lines shown are all at = 3mm, and have comparable excitation energies and similar critical densities. The images show a clear variation in cloud chemistry and molecular abundances with galactic location. A principal component analysis (Meier and Turner 2005) shows that the molecules N<>2H+, HNC, and HCN have similar spatial distributions to CO and its isotopologues, and are good tracers of the overall molecular cloud distribution. The molecule CH3OH (methanol) is well known from Galactic studies to be a "grain-chemistry" molecule, which forms on grain surfaces and is liberated by shocks or warm cloud conditions; here, methanol and HNCO follow the arms of the minispiral. Methanol and HNCO appear to be tracing the gentle shocks of the gas passing through the spiral arms. The final group of molecules, C34S and C2H are found in the immediate vicinity (50 pc) of the nuclear cluster; these molecules presumably reflect the intense radiation fields of the nuclear star cluster.
Figure 7. Spatially-resolved (5" = 75 pc) chemistry of the central 300 pc of the Scd spiral galaxy IC 342. Meier & Turner (2005). Panels at top show molecules that are overall molecular gas tracers. Bottom, left: C2H and C34S trace clouds in high radiation fields (PDRs); right: CH3OH (methanol) and HNCO trace grain-chemistry along the arms of the minispiral.
The next decade will see a blossoming of molecular line studies of ESF. The current state of molecular line work in galaxies has been recently reviewed by Omont (2007), and also is represented by contributions in the volume by Bachiller & Cernicharo (2008). Spectroscopy of ESF gas in galaxies with the APEX, ASTE, CARMA, IRAM 30 m, Plateau de Bure, CARMA, SMA, Spitzer, and the VLA telescopes will continue to probe the star-forming ISM in nearby galaxies through the next decade. Herschel and SOFIA will soon provide spectroscopy of the important 158 µm line of C II from PDRs in ESF galaxies. ALMA will add tremendous sensitivity, milliarcsecond resolution, km/s velocity resolution, access to the southern hemisphere, and submillimeter capability, allowing us to study extreme star formation and extreme star-forming gas and its effects on galaxies in exquisite detail.
I am grateful to Nate McCrady, David S. Meier, and Andrea Stolte for their helpful discussions and comments, and to Xander Tielens for his good humor and patience.