Local Lyman break galaxy analogs:
The impact of massive star-forming clumps on the interstellar medium and
the global structure of young, forming galaxies
We report on the results of HST optical and UV imaging, Spitzer mid-IR photometry and optical spectroscopy of a sample of 30 low-redshift (0.1 to 0.3) galaxies chosen from the SDSS and GALEX surveys to be accurate local analogs of the high-redshift Lyman Break Galaxies (LBGs). The Lyman Break Analogs (LBAs) are similar in stellar mass, metallicity, dust extinction, SFR, physical size and gas velocity dispersion, thus enabling a detailed investigation of many processes that are important in star forming galaxies at high redshift. The main optical emission line properties of LBAs, including evidence for outflows, are also similar to those typically found at high redshift. This indicates that the conditions in their interstellar medium are comparable. In the UV, LBAs are characterized by complexes of massive clumps of star-formation, while in the optical they most often show evidence for (post-)mergers and interactions. In six cases, we find a single extremely massive (up to several ) compact (radius pc) dominant central object (DCO). The DCOs are preferentially found in LBAs with the highest mid-IR luminosities ( to L) and correspondingly high star formation rates (15 to 100 M per year). We show that the massive star-forming clumps (including the DCOs) have masses much larger than the nuclear super star clusters seen in normal late type galaxies. However, the DCOs do have masses, sizes, and densities similar to the excess-light/central-cusps seen in typical elliptical galaxies with masses similar to the LBA galaxies. We suggest that the DCOs form in the present-day examples of the dissipative mergers at high-redshift that are believed to have produced the central-cusps in local ellipticals (consistent with the disturbed optical morphologies of the LBAs). More generally, the properties of the LBAs are consistent with the idea that instabilities in a gas-rich disk lead to very massive star-forming clumps that eventually coalesce to form a spheroid. Finally we comment on the apparent lack of energetically significant AGN in the DCOs. We speculate that the DCOs are too young at present to be growing a supermassive black hole because they are still in a supernova-dominated outflow phase (age less than 50 Myr).
Subject headings:galaxies: starburst — galaxies: active — galaxies: bulges — galaxies: peculiar — galaxies: high redshift
One of the most remarkable features of high redshift () galaxies is their great potential for vigorous star formation. Moreover, observations with the Hubble Space Telescope (HST) indicate that that star formation largely occurs in an extremely “clumpy” and compact mode (e.g. Cowie et al., 1995; Franx et al., 1997; Elmegreen et al., 2005; Stark et al., 2008). These clumps are reminiscent of star-forming complexes in local H regions, but their total masses and scale sizes are several orders of magnitudes larger. As shown by Elmegreen et al. (2009), clumps with masses of and sizes of 1 kpc are a very typical constituent of extended, irregular galaxies at in the Ultra Deep Field (UDF, Beckwith et al., 2006). The relatively high gas velocity dispersions observed in some high redshift galaxies (e.g. Law et al., 2007, 2009; Förster Schreiber et al., 2009) coupled with their large inferred gas column densities and gas fractions can explain the formation of the clumps and their bulk properties from gravitational instabilities in gas-rich, Jeans-unstable disks. This process may be instigated, for example, by the continuous accretion of gas or by the delivery of gas-rich material in small merger events (Bournaud & Elmegreen, 2009; Elmegreen et al., 2009). In any case, the compactness of the stellar clumps furthermore implies that they can form relatively unhampered by velocity shear or supernova feedback. Simulations show that for some galaxies the coalescence of clump systems may lie at the root of the formation of galaxy bulges, and possibly even their supermassive nuclear black holes if it is assumed that each clump gave rise to its own intermediate mass black hole prior to their migration to the center (Noguchi, 1999; Immeli et al., 2004; Elmegreen et al., 2008a, b).
It would be extremely valuable to study similar objects in nearby galaxies where multi-waveband data with relatively high spatial resolution and sensitivity would allow the relevant physical processes to be probed in much greater detail. While local starburst galaxies used in previous studies have some similarities to star-forming galaxies at high redshift, there are some important differences: local starbursts having SFRs of 100 yr are very dusty systems (the so-called Ultra-Luminous Infrared Galaxies, ULIRGs). The intense star formation occurs in compact, sub-kpc scale regions in the centers of interacting galaxies, and are usually inconspicuous in the far-UV (Heckman et al., 1998; Goldader et al., 2002). Although very dusty starbursts clearly make up a non-negligible fraction of high redshift galaxies (e.g. Hughes et al., 1998; Huang et al., 2005; Papovich et al., 2006; Caputi et al., 2007; Burgarella et al., 2009; Daddi et al., 2009), most of the systems contributing to the cosmic SFR at high redshift are accounted for in deep UV-selected samples of Lyman Break Galaxies (LBGs) (e.g. Reddy et al., 2006; Reddy & Steidel, 2009; Bouwens et al., 2009), indicating that the amount of dust extinction in LBGs is typically much lower: the average ratio of their bolometric IR-to-UV luminosities, , is much lower () compared to local galaxies of the same (e.g. Buat et al., 2007; Reddy et al., 2006; Burgarella et al., 2007). Locally calibrated dust corrections that are based on the reddening of the UV continuum slope as derived by Meurer et al. (1999) seem to apply to typical LBGs (Reddy et al., 2006) with some exceptions (e.g. Baker et al., 2001; Reddy et al., 2006; Siana et al., 2008), and can be used to show that, despite their UV selection, LBGs at still put out most of their energy in the mid- and far-IR (Bouwens et al., 2009).
The star formation in LBGs typically occurs over the entire extent of the galaxy ranging from a few hundred pc to several kpc (e.g. Franx et al., 1997; Giavalisco, 2002; Bouwens et al., 2004; Ferguson et al., 2004; Stark et al., 2008). The relatively low extinctions and high UV surface brightnesses seen in LBGs are characteristic of local blue compact dwarf galaxies (BCDs). However, typical BCDs have star formation rates up to two orders of magnitude smaller than LBGs and occur in galaxies with much lower mass (e.g. Telles & Terlevich, 1997; Hopkins et al., 2002), although some may resemble fainter LBGs at high redshift in their basic properties (see, e.g., Östlin et al., 2009). The main differences between LBGs and typical local starbursts studied previously may be due to a systematically lower metallicity and higher gas-mass fraction in the LBGs and/or due to a different triggering mechanism (e.g. Erb et al., 2006a, b; Law et al., 2009).
The “Lyman break analogs” (LBA) project was designed in order to search for local starburst galaxies that share typical characteristics of high redshift LBGs (Heckman et al., 2005). The LBA sample and main properties are given in Heckman et al. (2005), Hoopes et al. (2007), Basu-Zych et al. (2007, 2009) and Overzier et al. (2008, Paper I), and we note that the LBAs are identical to the sample of “supercompact UV-luminous galaxies” (ScUVLGs) referred to in the papers listed above. In brief, the UV imaging survey performed by the Galaxy Evolution Explorer (GALEX) was used in order to select the most luminous ( ) and most compact ( kpc) star-forming galaxies at . Galaxies having these high are already extremely rare (local space density of Mpc), and the surface brightness requirement further reduces their absolute number density by a factor of . These objects tend to be much more luminous in the UV than typical local starburst galaxies studied previously (Heckman et al., 1998; Meurer et al., 1997, 1999) consistent with high SFRs and relatively little dust extinction. The median absolute UV magnitude of the sample is –20.3, corresponding to , where is the characteristic luminosity of LBGs at (Steidel et al., 1999, i.e., ). Analysis of their spectra from the Sloan Digital Sky Survey (SDSS) and their spectral energy distributions from SDSS and GALEX subsequently showed that the LBAs are similar to LBGs in their basic global properties, including: stellar mass, metallicity, dust extinction, SFR, physical size and gas velocity dispersion. In Overzier et al. (2008) we showed that most of the UV emission in a preliminary sample of 8 LBAs originates in highly compact burst regions in small, clumpy galaxies that are also morphologically similar to LBGs. We demonstrated that if LBGs at high redshift are also small merging galaxies similar to the LBAs, unfortunately this would be very hard to detect given the much poorer physical resolution and sensitivity. Furthermore, for the present paper it is important to point out that the LBAs also occupy an “offset” region in the main optical emission line diagnostics diagram (i.e. log([N]/H) versus log([O]/H), or the “BPT” (Baldwin et al., 1981) diagram) analogous to galaxies at high redshift. This suggests that similar physical conditions may apply to their interstellar medium (ISM) as well. In summary, the LBAs appear indeed good local analogs to the LBGs.
In this paper we will present new results from our HST and Spitzer imaging campaigns and spectroscopic follow-up in order to highlight some of the remarkable properties of the LBA sample: we will compare the UV–optical sizes and morphologies, showing evidence for massive star-forming “clumps” that dominate the UV (§3). We explore possible connections between the starburst regions and the general properties of the optical emission line gas (§3.5 and §4), as well the relationship between the formation and evolution of massive stellar clumps in the context of the main structural components of young galaxies (§5). We conclude the paper with a summary and final remarks (§6). Forthcoming work on the LBA sample currently in preparation includes a detailed morphological analysis, multi-waveband modeling, mid-IR spectroscopy, VLBI radio imaging, X-ray spectroscopy, and integral field and long-slit optical and near-IR spectroscopy.
2.1. HST and Spitzer Imaging
We observed 30 LBAs using HST (Programs 10920/11107, PI: Heckman). In order to probe morphologies in the UV, twenty-four objects were observed using the ACS Solar Blind Channel (SBC) in the filter F150LP (Å), six objects were observed using the High Resolution Camera (HRC) in the filter F330W (Å). The exposure times per object were typically 2500 s. These UV images offer a tremendous gain in resolution over that of the GALEX images ( compared to 5″ for GALEX). Matching observations in the rest-frame optical were carried out either using the Wide Field and Planetary Camera 2 for 3600 s in F606W (Å; twenty-four objects) or using the ACS Wide Field Channel for 2200 s in F850LP (Å; seven objects). The ACS and WFPC2 observations were divided into, respectively, three and six dithered exposures and were combined using multidrizzle (Koekemoer et al., 2002).
We make use of photometric data obtained with the Multiband Imaging Photometer (MIPS) on the Spitzer Space Telescope at 24 m (Program ID 20390, PI: C. Hoopes). Photometry was obtained by performing point source extractions on the post-Basic Calibration Data.
2.2. SDSS and VLT/FLAMES Spectroscopy
We further use integrated optical emission line fluxes and flux ratios measured from the SDSS spectra. In order to account for stellar absorption lines, the emission-line parameters were obtained from Gaussian fits after subtracting the continuum based on the best-fitting spectrum using stellar population models (see Tremonti et al., 2004).
The fraction of the total (line and continuum) light that is included within the 3″ diameter SDSS fiber apertures is found to be , estimated from the difference in fluxes measured in a 3″ aperture compared to the total SDSS photometric aperture111The fraction of fiber-to-total-light in the SDSS -band is estimated by . We estimate that this estimate is in fact a lower limit to the fraction of UV and line emission included within the SDSS fiber aperture, based on the compactness of the star-forming regions in the HST images and follow-up spectra.
In Sect. 4.1 we present a preview of data222ESO Program 082.B-0512(A) taken with the Fibre Large Array Multi Element Spectrograph (FLAMES) on the European Southern Observatory (ESO) Very Large Telescope (VLT) during the night of Oct 30 2008. We used the GIRAFFE/ARGUS integral field spectrograph (IFS) to obtain seeing-limited spectra resolving the H and [N]6548Å,6584Å line complex at a spectral resolution of and a spatial plate scale of or pixel (depending on the seeing). Each source was observed in 4 exposures of 400 s each. The data was pipeline reduced and further cleaned of cosmic rays and the background using our own routines.
More details on the LBA sample selection, observations and all measurements used in this paper are given elsewhere (Paper I; Hoopes et al., 2007; Basu-Zych et al., 2007; Overzier et al., 2009). The object coordinates, redshifts and basic quantities used in this paper are listed in Table 1. Emission line measurements are listed in Table 2.
2.3. Measuring Stellar Masses
We will use stellar mass estimates that were obtained from the SDSS/DR7 stellar mass catalogue333Available at http://www.mpa-garching.mpg.de/SDSS/DR7/. These masses are derived from fits to the SDSS photometry using a large grid of models constructed using the Bruzual & Charlot (2003, BC03) stellar synthesis library. Although the BC03 models may not include the most up-to-date treatment of the thermally pulsating (TP)-AGB phase, it has been shown that this phase mostly affects the stellar mass-to-light ratio in the rest-frame near-infared and bands (e.g., see Eminian et al., 2008). Because near-infared data is not being used in the fits, we believe that our stellar mass estimates are relatively accurate. However, in order to further reduce the uncertainty in these galaxy stellar mass estimates, we will derive a “best” mass by taking the average of the SDSS photometric mass and a dynamical mass that we calculate from the H emission line velocity dispersion and the optical half-light radius. The two mass estimates typically agree to within 0.3 dex with no systematic differences (see Table 1).
2.4. Measuring Star Formation Rates
We will make use of three different estimates of the integrated star formation rate. We use the most recent calibrations from literature as tabulated in Calzetti et al. (2009), and valid for a Kroupa (2008) initial mass function (IMF). The resulting SFRs are lower by a factor of 1.5 compared to a Salpeter (1955) IMF.
SFR is calculated from the H line luminosity and applying a correction for dust based on the Balmer decrement () following the recipe in (Calzetti, 2001):
where , and
with and . We apply a small correction factor of typically due to the flux expected outside the SDSS fiber.
2.4.2 H24 m
An improvement to the above method is given by SFR, which is based on a combination of the uncorrected H luminosity (), and the 24m luminosity () related to the emission of dust heated by young stars. It has been shown that SFR is a robust indicator for the total SFR calibrated against Pa measurements in a large sample of nearby galaxies (Calzetti et al., 2007):
As the exact, individual infrared spectral energy distributions are currently uncertain, we choose not to apply any -corrections to the observed 24m fluxes. Based on the average corrections found for a collection of spectral templates presented by Rieke et al. (2009, see their Figure 9) we estimate that the actual SFRs could be between 0.0 and 0.3 dex higher depending on the redshift, spectral shape, and total IR luminosity.
We will make use of SFR, which is based on the dust-corrected far-UV luminosity using the empirical correlation between the attenuation of the UV and the IR emission from dust heated by obscured stars (the “– relation”, Meurer et al., 1995). The correlation has been shown to work well for relatively unobscured star-forming and starburst galaxies at both low and high redshift (e.g. Meurer et al., 1995, 1999; Seibert et al., 2005; Reddy et al., 2006; Salim et al., 2007), but it must be noted that the situation is more complex for galaxies that have more complicated star formation histories (e.g. Kong et al., 2004; Johnson et al., 2007), very young ages (e.g. Reddy et al., 2006; Siana et al., 2008) or that are heavily obscured (e.g. Meurer et al., 1999; Goldader et al., 2002; Reddy et al., 2006). We use the calibration between UV colour and from Treyer et al. (2007), mag, finding:
Using the spectral slope with we further applied a small -correction to obtain the monochromatic luminosity at rest-frame 1500Å, .
3.1. UV–optical morphologies
In Fig. 1 we show 6″6″ UV-optical false-color postage stamps of 30 LBAs, including the 8 objects studied in Paper I. In all cases, both the UV and the optical light is dominated by the intense light from young massive stars in very compact starburst regions. The optical images furthermore allow us to see the faint emission from older stars displaying a wide range of (often complex) morphologies, which include clumpy or chain-like galaxies, interacting pairs, a ring galaxy, and perturbed (spiral) disks. In Fig. 2 we plot the half-light radii versus the stellar masses. The global median half-light radii of the galaxies are 1 kpc and 1.5 kpc in the UV and optical respectively. The galaxy mass estimates range from a few times to (Table 1). The sizes and the stellar masses of these galaxies correspond well to those of typical Lyman Break Galaxies (e.g. Shapley et al., 2001; Papovich et al., 2001; Giavalisco, 2002; Ferguson et al., 2004; Stark et al., 2009).
As noted already in Paper I, a significant fraction of both the UV and optical emission is seen to come from luminous, very compact clumps. The much larger size of the sample presented here allows us to investigate this in more detail. In Figs. 3 and 4 we show some close-ups of the rest-frame optical HST images at high contrast. LBAs typically consist of a collection of centrally-located compact clumps having a range in magnitudes (e.g., Fig. 3). However, in a minority of cases the LBA contains a single, dominant luminous point-like source at or near the center of the galaxy (i.e., 005439, 021348, 080232, 080844, 092159, and 210358). In these sources, the central source typically contains 50% (20 and 77% being the lowest and the highest, respectively) of all the UV light. In the optical, the contribution from the host galaxies is larger, somewhat reducing the light fraction of the central source to %. Here we have used an aperture of 02 radius for the central point-like source and an aperture of 10″ to estimate the total light of the galaxy. Because of their visual prominence and in anticipation of some of their remarkable properties that will be investigated in the following sections, we will henceforth refer to this subset of the clumps as “dominant central objects” (DCOs). A collage of images of the six objects hosting DCOs is shown in Fig. 4.
3.2. Color-magnitude diagrams: estimated clump masses
We have measured the magnitudes and colors of each of the clumps and DCOs identified by eye inside circular diameter apertures. The backgrounds were determined inside a local annulus and subtracted. We did not make a correction for the (small) amount of flux expected to fall outside the aperture due to the HST PSF. In Fig. 5 we show the color-magnitude diagrams for all objects for which two filters were available, either and (shown on the left) or and (shown on the right). Open and filled squares indicate the measured values, while the dotted lines indicate the dust-corrected values based on the Balmer decrement (/) using a constant relation between reddening of the gas and the stellar continuum444 (Calzetti, 2001). Unfortunately we need to assume that the Balmer decrement as measured from the integrated SDSS spectrum applies to the much smaller clumps as well, although this is not necessarily correct. The results are tabulated in Table 3. The clumps are very luminous with a median absolute magnitude in (approximately rest-frame ) of mag, and very blue (). The DCOs are the most luminous having –20 to –22 mag. The small gap in the magnitude histograms in the top panels of Fig. 5 may suggest that the DCOs are not just a bright tail of the general clumps distribution, but perhaps a distinct class of object.
A comparison with Starburst99 5.1 (Vázquez & Leitherer, 2005) stellar synthesis models indicates that the stellar masses of the clumps shown in Fig. 5 are typically a few times to a few times , while the DCOs have typical masses near or in excess of (see Table 3). It is important to note that these mass estimates are relatively insensitive to dust extinction, since the dust extinction vector runs nearly parallel to the age vector in these plots. As shown in Fig. 6, the masses are relatively insensitive to the star formation history (SFH) that we adopt. The figure also illustrates that it is much less straightforward to determine the typical stellar age of the clumps without accurate information about the extinction and SFH. For our adopted extinction corrections, the instantaneous burst models imply ages in the range from 5 to 100 Myr (red points in Fig. 6). Alternatively, if we adopt a continuous SFH the ages would be substantially higher, in the range 100–1000 Myr (blue points), while the SFRs would have to be on the order of 10 yr in order to explain the high luminosity of the brightest clumps (left blue track in Fig. 5). The masses would be higher on average compared to those derived for instantaneous bursts (Table 3). However, there are several arguments to make in favor of an instantaneous SFH (or a SFH having a very short -folding time). We note that at least three of the sources show prominent Wolf-Rayet features indicating the presence of very massive stars at a special phase only 3–4 Myr after the onset of a massive burst (e.g. Brinchmann et al., 2008b). This would favor the instantaneous burst model. Also, we note that the very compact sizes of these objects (and thus the short dynamical times) are consistent with a short formation time ().
3.3. Clump sizes
It is clear from simple visual inspection of the HST images that the DCOs are marginally-resolved at best (note the strong diffraction spikes associated with the DCOs in Fig. 3). Many of the other clumps appear either as faint point-like sources or larger spatially resolved regions (see examples in Fig. 3).
To obtain the best constraints on the sizes of the DCOs, we have first examined the radial flux profile of the three brightest examples, using the ACS UV images as they have smaller pixels and a sharper PSF compared to the optical images. The result is shown in Fig. 7. Each panel shows the measured count rate (plusses) compared to the measured profile of a model PSF (dashed line) modeled using the TinyTim 3.0 software (Krist, 1995). We also simulated a range of 2D Gaussian models having a FWHM of – and convolved them with the model PSF. The results are shown using coloured lines. In all three cases, the data is most consistent with an object no larger than (FWHM), corresponding to effective radii no larger than 70–160 pc at ).
In order to be consistent across our entire clump sample and with our earlier definition of the clumps, we have measured the physical sizes of the brightest clumps in each LBA by measuring the optical half-light radii of each clump within the same fixed angular apertures used in Fig. 5. The radii were deconvolved assuming a PSF of (FWHM). In Fig. 8 we plot the clump radii versus the extinction-corrected clump absolute magnitudes. The typical clumps have radii of 200 to 400 pc, while the DCOs have radii of 100-200 pc. Comparing the sizes obtained here with those obtained for the three DCOs using our more careful model simulations shown in Fig. 7 suggests that the more crude analysis may somewhat overestimate the true sizes of at least the DCOs. In any case, Fig. 8 shows an inverse correlation between clump luminosity and size, largely defined by the extreme properties of the DCOs. Their high luminosities (corresponding to very high stellar masses) and compact sizes imply very high densities. We will discuss the implications of this remarkable result later in the paper.
3.4. The relation between clumps and their host galaxies
We have demonstrated above that massive star-forming clumps are an essential ingredient in our sample of local Lyman Break Galaxy analogs, and have found that a subset contain a single extremely massive (up to several ) compact (radius pc) dominant central object (DCO). How do the properties of these compact structures relate to the overall properties of the host galaxies?
The results are shown in the six panels of Fig. 9. In all panels we plot the extinction-corrected absolute -band magnitudes of the clumps measured within a fixed radius of 500 pc, in order to account for the effect that the angular sizes change by a factor of 2.5 over the redshift range . The results are as follows.
Galaxy Concentration.– In panel (a) we plot clump versus the optical concentration parameter555To calculate the concentration we follow the procedures described in full detail in Lotz et al. (2006) and Paper I. First, we use SExtractor to make an object segmentation map and mask out neighboring objects. The image is background subtracted, and we calculate an initial Petrosian radius ( with ) using the object center and (elliptical) shape information from SExtractor. We then smooth the image by and create a new segmentation map by selecting those pixels that have a surface brightness higher than the mean surface brightness at the Petrosian radius. We recalculate the object center by minimizing the second order moment of the flux, and then recalculate the Petrosian radius in the original image using this center. The total flux is defined as the flux within a radius of and is then calculated in circular apertures containing 20 and 80% of the light. of the host galaxy, log. In general the more luminous clumps occur in the LBAs with larger concentrations. The six dominant central objects (filled squares) are the most extreme. Inspection of Fig. 1 shows that these objects are indeed systematically found inside larger, optically faint disks compared to the sample on average.
Galaxy Half-light radius.– In panel (b) we plot versus the half-light radii for each galaxy measured in the UV images. This plot shows that the total UV light in galaxies hosting a DCO is dominated by the DCO itself, albeit with one notable exception (DCO object 080232 having kpc).
Stellar mass.– In panel (c) we plot versus the stellar mass of the hosts. The brighter clumps are predominantly found in the more massive objects in our sample. The DCOs are found in galaxies with masses .
Mid-IR luminosity.– In panel (d) we show the relation between optical clump luminosity and the host galaxy luminosity at 24m. We find a correlation with indicating that the galaxies hosting the brighter clumps seen in the UV–optical are also the stronger IR sources. Even though selected on the basis of their high luminosity in the far-UV, the galaxies with DCOs have mid-IR luminosities similar to those of luminous infrared galaxies (LIRGs).
SFR.– These high mid-IR luminosities imply high star-formation rates (SFR), as shown in panel (e). The star formation rates are systematically larger in the LBAs with the more luminous clumps, and the six galaxies with DCOs have star formation rates ranging from 15 to nearly 100 per year.
Comparing SFR indicators.– In panel (f) we plot versus the ratios of the two commonly used SFR indicators SFR and SFR. For the LBAs with the lower luminosity clumps, the majority show good agreement between the two SFR indicators. The LBAs with the brighter clumps are sytematically displaced towards low values of SFR/SFR, with the galaxies having a DCO being among the most extreme. To further investigate the discrepancy between SFR and SFR for the brightest sources, we compare SFR/SFR with the ratio SFR/SFR. The result is shown in Fig. 10 comparing the three SFR indicators. For the majority of sources all three indicators agree to within a factor of 2 to 3, but there are outliers for which the SFRs derived from the extinction-corrected H luminosity are significantly lower than either both or one of the other two SFRs. In other words, the ratio of the dust-corrected H luminosity relative to the luminosities of the mid-IR and (intrinsic) far-UV continua is systematically small in the galaxies with DCOs compared to normal star forming galaxies and the less extreme LBAs. We will refer to these offsets in Sections 4.1 and 4.2 below.
4. Discussion of the Emission Line Properties of LBAs
The most significant results provided by our HST data are:
1) Highly luminous, compact star-forming “clumps” appear to be the
defining feature of the UV morphology of the LBAs.
2) In the most extreme cases, the UV images are dominated by a single
extremely bright and compact object surrounded by a disturbed envelope
or disk. We have called these dominant central objects (DCOs).
3) The star-formation rate derived from the extinction-corrected
H emission-line luminosity tends to be systematically smaller
than that derived using the mid-IR and/or the extinction-corrected
far-UV continuum luminosities (see panel f of
Fig. 9 and Fig. 10). More properly, the
ratio of the H luminosity to either the mid-IR or far-UV
continuum is smaller on-average than in typical star-forming
For our discussion below it is important to address first whether any of these properties could be explained by the contribution from a Type 1 (unobscured) or Type 2 (obscured) AGN.
4.1. Are they AGN?
4.1.1 Type 1 AGN
Could the DCOs be Type 1 AGN in which we have a direct view of the accretion disk and broad emission-line region? This would account for the exceptionally compact morphology as well as for the significant 24m “excess”. We note that the initial LBA selection process excluded any obvious Type I AGN in the first place based on the absence of obvious broad (several thousand km s) Balmer emission-lines from the AGN Broad Line Region in the SDSS spectra (Hoopes et al., 2007). We can quantify this and exclude the possibility that some of our galaxies are of type Seyfert 1.8 or 1.9, with weak broad emission lines. In the event that possible broad emission was somehow being missed in the integrated SDSS spectrum, we have obtained integral field spectra using FLAMES on the VLT. In Fig. 11 we show the preliminary results of the H[N] line profile as measured in the central 1″ region centered on the clumps in four LBAs hosting some of the most luminous clumps (three of which are DCOs). The line profiles were fitted with a series of narrow and broad Gaussian components indicated by red and blue dashed lines, respectively. The fits show that each source has significant line asymmetries on the blue side of the narrow lines that are consistent with a relatively broad, blueshifted component (Overzier et al., in prep.). Such blue asymmetries of width of a few hundred km s in both the permitted and forbidden are not related to the much broader line profiles from Type 1 AGN, which are seen only for the permitted lines. Instead, the broad blueshifted wings seen on the emission-line profiles in our objects (Fig. 11) are signposts of galactic winds and are commonly observed in intense local and high redshift starbursts (e.g. Heckman et al., 1990; Lehnert & Heckman, 1996; Lehnert et al., 2009; Shapiro et al., 2009).
4.1.2 Type 2 AGN
Our spectra show clearly that the optical continuum must be dominated by the light of young stars (with no detectable contribution from an unobscured AGN). However, is it possible that the DCOs are associated with the presence of an obscured (Type 2) AGN? In Fig. 12 we show the log([N]/H) versus the log([O]/H) diagnostic “BPT” diagram that is most commonly used for identifying the main source of ionization among samples of star forming galaxies and AGN (Baldwin et al., 1981; Kauffmann et al., 2003; Kewley et al., 2006). Because it is based on ratios of bright emission lines that are observable out to , the diagram is widely used to characterize galaxies at both low and high redshift. It has been known for some time that intensely star-forming galaxies at high redshift such as Lyman breaks, “s”, and distant red galaxies are often peculiarly offset towards higher line ratios (e.g. Teplitz et al., 2000; Erb et al., 2006a; Kriek et al., 2007; Lehnert et al., 2009; Hayashi et al., 2009) with respect to the mean SDSS star-forming population at low redshift (Fig. 12, dotted line). Similar offets, albeit perhaps slightly less extreme, have been observed in local “warm” IR-luminous galaxies (Kewley et al., 2001) and Wolf-Rayet galaxies (Brinchmann et al., 2008b), and DEEP2 galaxies at (Shapley et al., 2005b; Liu et al., 2008). At present it is not known what physical mechanisms lie at the root of these offsets (see the detailed discussion in Brinchmann et al., 2008a), but it is important to note that our sample (filled circles) shows similar offsets in the BPT diagram (see also Hoopes et al. (2007) and Paper I). The BPT displacement is in the sense of enhancements in one or both of the two line ratios. If we focus on the locations of the host galaxies of the DCOs in Fig. 12 (filled squares), we see that they are more significantly displaced (in log[[N]/H]) with respect to the locus of normal star-forming galaxies and the other LBAs. Typical galaxies in the SDSS with similar emission-line ratios are almost certainly composite objects in which the SDSS fiber encompasses both gas near the nucleus that is excited by an AGN and normal regions of star formation in the surrounding galaxy (e.g. Kauffmann et al., 2003; Kewley et al., 2006; Kauffmann & Heckman, 2009). Could the DCOs harbor a Type 2 AGN surrounded by a starburst?
To answer this question we further investigate the BPT offsets in Fig. 13. In panel (a) we plot for each clump the dimensionless parameter , defined as the relative offset its host galaxy has in the BPT diagram. This offset distance is measured perpendicular to the SDSS star-forming “ridge” (dotted line in Fig. 12). This clearly demonstrates a relation between the offsets and the luminosity of the clumps. Panel (b) shows the offsets along log([N]/H) only, but we note that the correlation here is driven mainly by the well-known relation between mass and metallicity. In panel (c) we plot versus the electron density () of the gas as derived from the density sensitive [S]6713Å,6731Å line doublet assuming a gas temperature of K (Osterbrock & Ferland, 2006). The densities derived from the integrated SDSS spectrum correlate with the of the clumps over almost two orders of magnitude (the lowest and the highest being 30 and 2000 cm) albeit with quite a large scatter. The implied gas pressures are to K cm, several orders of magnitude higher than in normal star forming galaxies in the local universe.
The Type 2 AGN interpretation would be consistent with the high electron densities seen in the most extreme cases ( cm), since such densities are typical in the narrow emission-line region in AGN (Osterbrock & Ferland, 2006). It would also be consistent with the blue-asymmetric emission-line profiles seen in Fig. 11, which are characteristic of outflows in emission-line AGN (e.g. Heckman et al., 1981; Whittle et al., 1988). The AGN hypothesis could also explain why the ratio of the 24 micron and extinction-corrected H luminosity is higher than in typical star forming galaxies (Fig. 10), since AGN are characterized by strong emission by warm dust peaking in the mid-IR (e.g. Spinoglio et al., 2002). It would not however explain the result in Fig. 10 that the ratio of the extinction-corrected far-UV and H luminosity is also higher on-average in the galaxies with DCOs than in typical star-forming galaxies, since in a Type 2 AGN the continuum from the AGN itself will be very heavily obscured in the far-UV.
To further test this starburst/AGN composite hypothesis, we can use other emission-line diagnostic diagrams that are complementary to the BPT diagram. Following Veilleux & Osterbrock (1987) and Kewley et al. (2006) we show in Figure 14 plots of [S]/H vs. [O]/H and [O]/H vs. [O]/H. Bona fide starburst-AGN composite systems will lie in the region intermediate between the locus of pure star-forming galaxies and pure AGN in these plots. Instead, we see that the galaxies with the DCOs tend to lie in a kind of “no man’s land” displaced from the locus of star forming galaxies in a direction opposite to that of the AGN (they have weaker [S] and [O] emission than normal star forming galaxies). Their anomalous behavior is even more dramatically illustrated in the right-most panel in Fig. 14 where we plot [N]/[S] vs. [O]/H. As indicated in Figs. 12 and 14 (small grey circles), similar behavior is seen in local IR-warm starbursts (Kewley et al., 2001).
Thus, the hypothesis that the compact clumps are found in starburst/AGN composite systems does not explain the properties of these objects. In future papers we will describe observations with XMM-Newton, the Spitzer Infrared Spectrograph, and the European VLBI network that are designed to search for more direct signatures of the presence of an AGN.
4.2. Further Clues from the Emission Line Properties
We have estimated the ages of the clumps and DCOs to be 5–100 Myr assuming a model of an instantaneous burst of star formation (see Table 3). We have also shown that there exist striking correlations between the properties of the clumps (e.g., luminosity), the properties of their host galaxies (e.g., mass, radius, and concentration) and the physical properties of the surrounding gas (e.g. ionization and density). The LBAs with DCOs are particularly extreme in these plots. Below we will discuss the most likely physical mechanisms that can explain these correlations.
The results presented in Section 3 showed that there are significant
correlations between the properties of the massive compact clumps and
the physical properties of the surrounding ionized gas as traced by
the nebular emission-lines. In particular, the DCOs tend to be found
in galaxies in which:
1. The star-formation rate derived from the extinction-corrected
H emission-line luminosity tends to be systematically smaller
than that derived using the mid-IR and/or the extinction-corrected
far-UV continuum luminosities (see panel (f) in
Fig. 9 and Fig. 10). More properly, the
ratio of the H luminosity to either the mid-IR or far-UV
continuum is smaller on-average than in typical star-forming
2. The location of the galaxy in the diagnostic plot of
[O]/H vs. [N]/H (BPT diagram, Fig. 12)
is more significantly displaced with respect to the locus of normal
star-forming galaxies (see panels (a) and (b) in
Fig. 13). This displacement is in the sense of
enhancements in one or both of the two above ratios.
3. The electron density in the ionized gas is significantly higher
than in the other galaxies (panel (c) in
Occam’s razor drives us to ask if there is a single explanation for all of these results. In Section 4.1 we have already argued that the contribution from an obscured, low-luminosity AGN could explain some but not all of the properties observed. Here we will consider two alternative possibilities: 1) The DCOs are short-duration starbursts caught at a time after the peak in the O-star population, and 2) The DCOs include “leaky” systems in which a large fraction of the Lyman continuum is escaping.
4.2.1 The Aged Starburst Hypothesis
The DCOs are extremely massive ( M), compact ( pc) objects. These high densities ( M pc) imply very short dynamical times ( 1 Myr). On simple grounds of causality, it is therefore plausible that these objects could have formed all their stars in a Myr, a timescale significantly less than the characteristic lifetimes of the O stars that produce nearly all the ionizing radiation in a starburst. In such a case, the ratio of the ionizing luminosity to the non-ionizing far-UV luminosity changes rapidly with time, and falls dramatically for ages greater than about 5–6 Myr (e.g. Leitherer & Heckman, 1995). The de-reddened UV/optical colors of the massive compact clumps shown in Fig. 5 (while uncertain) are consistent with those of such “aged” instantaneous starbursts with ages of one-to-a-few tens-of-Myr (see Table 3). This is consistent with Basu-Zych et al. (2007) who compared the SFRs derived from the UV, IR and radio. We should note however that at least in one case (210358) we find evidence for WR stars in the SDSS spectrum indicating that its SFH must include a much more recent ( Myr) burst.
If, however, the aged starburst hypothesis is correct for other objects, this would naturally explain the relative weakness of the H emission-line in these galaxies. Such a model could also explain the peculiar location of these objects in the BPT diagram (Fig. 12). For an instantaneous burst in the age range between about ten and fifty Myr, the primary heating/ionizing source for the associated gas would be the mechanical energy supplied by supernovae rather than O star ionizing radiation (Leitherer & Heckman, 1995). Models of such shock-heated gas (e.g. Dopita & Sutherland, 1995; Allen et al., 2008) show an enhancement in the [O]/H and [N]/H ratios, as we observe (Fig. 12).
The expected supernova rate would be ( M) per decade. Such a high supernovae rate occurring in a very small region also naturally accounts for the high electron densities and thus the high gas pressures ( K cm) in these objects. Following Chevalier & Clegg (1985), in the absence of significant radiative losses, the central pressure inside a starburst region of radius (in units of 100 pc) with a supernova rate of per decade is given by K cm. Chevalier & Clegg (1985) also demonstrate that this high-pressure gas will subsequently expand outward to form a galactic wind with a ram pressure that drops like . This wind would be sufficient to produce the observed gas pressures over kpc-scales in the region surrounding the clump. The galactic wind will also accelerate emission-line clouds, leading to the broad blue-shifted wings seen on the emission-line profiles in our objects (Fig. 11). In fact, similar high gas pressures, broad blueshifted wings, and other signposts of galactic winds are common in intense local and high redshift starbursts (e.g. Lehnert & Heckman, 1996; Lehnert et al., 2009; Shapiro et al., 2009).
This model fails however to explain the relative weakness of the [S] and [O] lines seen in Fig. 14. Shock heating should lead to an enhancement in the relative strengths of these lines (see Fig. 2 in Dopita & Sutherland, 1995). We are left with an otherwise attractive model that does not naturally explain all the properties of these objects.
4.2.2 The Leaky Starburst Hypothesis
An obvious way to lower the amount of observed H emission would be a “leaky” interstellar medium in which a significant fraction of the ionizing radiation produced by O stars escapes the galaxy. This hypothesis has a plausible physical basis: the feedback effect of concentrating the release of the kinetic energy from of-order supernovae ( ergs) all within the exceptionally small volume of a DCO may create large holes in the surrounding interstellar medium through which ionizing radiation could then escape. Taken at face value, Fig. 10 would then imply that the majority of the ionizing radiation is escaping in our two most extreme objects (021348 and 080844). It is interesting to note that Shapley et al. (2006) find 2 out of 14 LBGs at with a significant escape fraction (%) as measured from the ratio of the fluxes below and above the Lyman limit in UV spectra. Similarly, Iwata et al. (2009) report the detection of large in 7 of 73 LBGs at . These fractions are similar to what is implied for our sample of LBAs under the leaky starburst hypothesis.
The leaky starburst hypothesis can readily account for the high electron densities (gas pressures) and the broad blue wings on the emission-line profiles: these are still intense starbursts and the same arguments as given above will apply. A leaky interstellar medium could also explain at least some aspects of the peculiar emission-line ratios we observe. Models of photoionized gas under “matter-bounded” conditions (e.g. Binette et al., 1996; Castellanos et al., 2002; Giammanco et al., 2005; Brinchmann et al., 2008a) show that the strengths of the [O] and [S] lines relative to H are decreased. This is because these lines are relatively strong in the warm partially neutral region lying beyond the edge of the Stromgren sphere, and this region is missing in matter-bounded conditions. These models show that the [O]/H ratio will also be enhanced in matter-bounded clouds, as we observe (Fig. 12). However, this model fails to explain the enhanced [N]/H ratios seen for the DCOs in the BPT diagram. Models (B. Groves, private communication) show that matter-bounded clouds in starbursts should have [N]/H ratios that are similar to or lower than ionization-bounded clouds.
A more fundamental objection to the leaky starburst hypothesis is that is difficult to envisage a situation in which a sizable fraction of the ionizing radiation escapes the galaxy, but the majority of the non-ionizing UV radiation is absorbed by dust inside the galaxy (as is required by the high mid-IR luminosities we see). For this reason, we regard the “aged starburst” hypothesis as the more likely one.
In the future, we will use longslit and integral field spectroscopy in order to spatially resolve and disentangle the line emission coming from the clumpy versus the diffuse regions seen in the HST images.
5. Discussion of the Structural Properties of LBAs
The peculiar emission line properties discussed above show the presence of significant feedback processes between the massive star-forming regions in LBAs and their surrounding ISM. Here, we will focus on the remarkable stellar properties of the massive clumps and the DCOs in particular, by asking the question to what kind of structural components of galaxies do they correspond?
The DCOs are only marginally resolved in the HST UV images, with implied radii of no greater than 100 pc (see Fig. 7). The fainter clumps are typically several times larger than this. The masses of the DCOs estimated in Section 3 range from a few to a few M, while the more typical clumps have masses in the range of a few to a few M. This implies very high stellar mass surface densities. For the typical clumps values for the effective surface mass density () range from to kpc, while for the DCOs to kpc. What then will the clumps and DCOs evolve into? Can we identify such evolved counterparts in local galaxies?
5.1. Super Star Clusters and Central Massive Objects?
Massive clusters are observed in and around the nuclei of many nearby late- and early-type galaxies, mergers and merger remnants, and AGN (e.g. Carollo et al., 1997; Geha et al., 2002; Böker et al., 2004; Maraston et al., 2004; Walcher et al., 2005; Seth et al., 2008). The largest stellar clusters typically have sizes of only a few pc (the largest being 10 pc) and masses rarely exceeding . Their structural properties are thus similar to those of massive globular clusters in the Milky Way and local galaxies, and also to the young “super star clusters” (SSCs) found in starburst galaxies (Meurer et al., 1995).
The most luminous star clusters in normal and starburst galaxies lie along a relation between the absolute optical magnitude of the cluster and the global SFR of the host galaxy (Larsen, 2002), a relation that is mainly driven by the relation between the luminosity of the brightest cluster and the total number of clusters formed in a galaxy (Whitmore et al., 2007). The correlation implies that for galaxies having SFRs as high as LBAs even the brightest clusters should only have to –14 (, see Weidner et al., 2004) compared to found in Fig. 5. This discrepancy could be partly reconciled given that the LBA clusters are systematically younger than the brightest clusters observed in normal galaxies, since a star cluster fades by mag in the first few 100 Myr. However, this scenario still largely falls short for any of the luminous clumps/DCOs having .
The most massive cluster-like object known to date, object W3 in the merger remnant NGC 7252, has a mass of and a radius of pc (Maraston et al., 2004). Interestingly, its high mass and small size are very much at odds with any kind of stellar cluster found to date. Curiously, the properties of object W3 are perhaps more similar to that of ultra-compact dwarf galaxies (e.g. Mieske et al., 2004). If we extrapolate from its current luminosity at an age of 400 Myr to that expected at 10 Myr, we find (Bastian, 2008), still over 2 magnitudes fainter than some of the DCOs.
Although the stellar mass surface densities that we derive ( kpc) are comparable to that of typical massive clusters mentioned above, the sizes are about an order of magnitude larger (100 pc) and the masses are up to two orders of magnitude larger ( ). Thus, the massive clumps and DCOs identified in the LBA sample appear as significantly scaled-up versions of known types of massive star clusters, and are not simply an extension of the cluster luminosity function as derived from nearby star-forming galaxies.
Furthermore, nuclear super star clusters and supermassive black holes are jointly referred to as central massive objects (CMOs). CMOs follow a tight correlation with the total mass of the galaxy over many orders of magnitude in stellar mass. The central star clusters in low luminosity early-types and spiral bulges, as well as the masses of the central supermassive blackholes in early-types all lie on the relation with small scatter (e.g. Magorrian et al., 1998; Marconi & Hunt, 2003; Côté et al., 2006; Ferrarese et al., 2006; Wehner & Harris, 2006; Rossa et al., 2006). Simulations of both isolated and merging galaxies reproduce the observed correlation well (e.g. Li et al., 2007). As we discuss below in more detail, the clumps and DCOs typically represent several percent of the total stellar mass of the surrounding galaxy. These mass fractions are therefore over an order of magnitude higher than in the CMOs.
Could the massive clumps or DCOs consist of multiple SSCs? If so, how long would it take this ensemble to merge into a single object? To address this, we assume that a clump of mass actually consists of 10–100 SSCs each having a mass of distributed over a region with a 100 pc radius within which the velocity dispersion km s. Then the timescale over which a single SSC migrates inward as a result of dynamical friction is on the order of only 10–20 Myr (Binney & Tremaine, 2008), comparable to the typical ages we have derived for the clumps. This suggests that a given massive clump (or DCO) is likely to have evolved into a single very massive object, even if it started out as an ensemble of SSCs.
We conclude that the clumps and DCOs are very different structurally from typical nuclear (super) star clusters.
5.2. Extra light and central “cusps” in elliptical progenitors?
For the discussion below it is important to point out that the DCOs (and, more generally, the most luminous clumps) tend to appear in the more massive galaxies in our LBA sample (see panel c in Fig. 9). The DCOs also each sit in the middle of extended, perturbed disk-like structures (hence the correlation between clump luminosity and the optical concentration index, panel a of Fig. 9). Another characteristic (by definition) is that the DCOs are single objects rather than the multiple (fainter) clumps as seen in many of the other objects.
In Fig. 15 we plot the clump mass () and effective radius () versus the effective mass density () for the brightest clump in each LBA, and compare with a collection of results from the literature. The comparison data shown was extracted from a compilation of literature results presented in Fig. 45 in Hopkins et al. (2009), showing globular clusters (crosses), nuclear star clusters (triangles), and the extra-light components in early-type galaxies from the samples of Kormendy et al. (2008, stars) and Lauer et al. (2007, circles). The high masses of derived for the DCOs combined with the strong constraints on their small effective radii ( pc as evidenced by the HST diffraction spikes in Fig. 4) yield the highest mass surface densities found in our sample, kpc (see Table 4).
Hopkins et al. (2009) show that this region of parameter space in the and versus planes is almost exclusively occupied by the so-called central excess or “extra light” component seen in local elliptical galaxies (stars and circles in Fig. 15; Crane et al., 1993; Lauer et al., 1995; Kormendy et al., 2008, and references therein). This extra light is characterized as an inner (1 kpc) component that rises more steeply than expected from the extrapolated Sersic power-law that fits the outer profile of ellipticals well. The so-called “cusps” have effective radii of pc and show a trend of decreasing density with increasing mass and in that sense they lie on the same sequence formed by elliptical galaxies. The cuspy cores are furthermore typical for relatively low-mass () ellipticals, while they are relatively uncommon for the most massive ellipticals. Kormendy et al. (2008) suggest that the core–cusp dichotomy in local ellipticals may correspond to their formation history: the extra light or cusp components are very likely the remnant signature of a massive central starburst in a dissipative (“wet”) merger, while the extended cores in core-dominated ellipticals are consistent with being the relaxed descendants of dissipationless (“dry”) mergers.
While none of the clumps bear any structural similarities with both globular and nuclear star clusters (crosses and triangles), the DCOs in the , , plane appear structurally similar to the extra-light light components in early-type galaxies (stars and circles). Hopkins et al. (2009) investigate in detail how the central mass associated with the extra light component is related to the remnant of a massive central starburst fueled by dissipative merging, and use extensive simulations to show that such a mechanism successfully reproduces the observed profiles of elliptical galaxies (Hopkins et al., 2008, 2009). In particular, they find a strong correlation between the galaxy stellar mass and the fraction of the total stellar mass that is in the extra light component (), as well as a one-to-one correlation between the extra light mass fraction and the fraction of the total stellar mass that formed in the dissipative starburst (), with roughly a factor of intrinsic scatter in and a factor of scatter in at each . For stellar masses in the range of to the ranges from a few to 50%, consistent with the extra-light fractions measured in real ellipticals by Hopkins et al. (2009).
We can compare this with the ratio of the central clump mass to the total stellar mass we found for the DCOs. Due to the uncertainties in the mass estimates involved, we estimate the DCO mass by taking the average of the values derived using single burst and continuous SFHs. The result is shown in Fig. 16. We find mass fractions typically of a few percent (%), object 021348 being a notable exception (%, see also Table 4). Except for object 021348, which agrees well with the cuspy cores, the mass fraction of the DCOs appears to be relatively low compared to the early-type cusps plotted in Fig. 16 (stars and circles). However, the mass fractions we find are still significant and not inconsistent with the data.
Alternatively, we should note that the special region of parameter space in , and occupied by both the DCOs and the cuspy cores appears to represent a particular transition region between massive, cluster-sized objects on one hand and entire galaxies on the other. While low-mass star clusters follow the scaling relations of globular clusters for masses , a wide variety of higher mass objects ranging from the most massive nuclear star clusters, dwarf-galaxy transition objects, ultra-compact dwarf galaxies, dwarf elliptical nuclei, and the early-type cusps themselves all appear to lie on the same scaling relations for spheroidal galaxies (e.g. Kissler-Patig et al., 2006; Hopkins et al., 2009). Although this overlap may be partially coincidental, perhaps it can be explained if these objects are all the product of some generic dissipational process.
In contrast with the DCOs, the much less extreme clumps indicated in Fig. 15 are typically found in LBAs containing multiple clumps distributed within their central regions (see Figs. 1 and 3). For a clump of mass moving within a disk of mass and a disk radius of 1.5 kpc (the average optical half-light radius) we find km s and a dynamical friction time of 100–200 Myr. This is several times longer than the ages that we have derived for the individual clumps. Thus it is plausible that at this relatively early dynamical stage we usually see several such clumps within the effective radius of the galaxy. The diffuse light seen in the HST images may build up as the clumps lose stars due to tidal stripping while they interact with each other and with the disk and slowly migrate inward. This process may eventually lead to the formation of an inner bulge as well as the Sersic profile typical of disk galaxies (Noguchi, 1999; Immeli et al., 2004; Carollo et al., 1997; Bournaud et al., 2007, 2008; Elmegreen et al., 2008a, b, 2009).
5.3. Relation to Mergers and IR-Luminous Galaxies
Although most LBAs were unresolved in the original SDSS images, some showed evidence for close companions or faint extended emission. One of the major results from the HST images is that most of them show very complex morphologies at scales well below the SDSS seeing or sensitivity. We interpret the large array of morphologies in Fig. 1 as evidence that most LBAs are undergoing or have undergone an interaction or merger666Here we use the term “merger” in the widest sense of the word, i.e. it is not limited to “major mergers” but includes minor mergers and other mechanisms that may deliver a supply of gas-rich material.. We also see a small number of objects that do not show unambiguous evidence of merging such as tidal tails or companions, but that are nonetheless often highly peculiar as evidenced by ring or chain-like morphologies (e.g. object 102613 and objects 082001, 082550, 083803). These observations suggest that the starbursts in LBAs and the formation of the massive stellar clumps are somehow linked to the mergers (e.g. Larson & Tinsley, 1978; Lambas et al., 2003; Nikolic et al., 2004; Woods et al., 2006; Owers et al., 2007; Li et al., 2008). The LBAs must also be sufficiently gas-rich in order for any starburst trigger to be effective. Simulations show that the inflow of gas ensures that the pressure in the interstellar medium becomes larger than the internal pressure of giant molecular clouds, and that the clouds can collapse and form massive clusters before they are disrupted by supernova explosions (Bekki, 2004).
It is important to note that the spectral energy distributions of LBAs are very different from typical starburst galaxies in the local Universe. These are predominantly associated with the highly dust-obscured LIRGs ( ) and ULIRGs ( ). At least ULIRGs are known to be strongly associated with the different phases of galaxy–galaxy mergers, with about equal fractions showing dual nuclei (pre-merger) and showing indication of post-merging (after the separate nuclei have coalesced and a period of violent relaxation starts forming an extended spheroidal component) (e.g. Genzel et al., 2001; Tacconi et al., 2002; Dasyra et al., 2006a, b; Arribas et al., 2008). These mergers are found to be limited to galaxy pairs having near-equal mass ratios (1.5:1 on average, with a maximum of 3:1). The less luminous LIRGs are typically associated with much less violent events that are less effective at triggering a major starburst phase as observed in ULIRGs, perhaps as a result of smaller gas fractions, orbital parameters, and larger pair mass ratios (or a combination thereof). As shown in Arribas et al. (2008), about one third of LIRGs appear in single isolated (spiral) galaxies. While none of the LBAs qualify as ULIRGs, there is some overlap with the LIRGs since 30–50% of LBAs have . As such, LBAs likely represent a population of relatively unobscured mergers or merger remnants that share some characteristics with LIRGs. The number density of LBAs is much lower than that of LIRGs. If the two populations are somehow connected, the LBAs could represent a very short evolutionary phase during which the starburst can be seen relatively unobscured. Alternatively, or in addition, LBAs might be seen at a special viewing angle, perhaps along the axis where the starburst has blown a hole in the ISM and is leaking Lyman continuum photons.
5.4. Implications for Black Hole Formation
We have shown that the LBAs are consistent with galaxies that are in the very early stages of forming their central bulge and/or core excess. Local galaxies have tight correlations between the mass of their central black holes and that of their bulge and total mass. It is thus interesting to ask whether there is any evidence that a large black hole mass is accumulating in the LBAs as well. The DCOs are particularly intriguing in this regard: such very massive and compact objects appear to be the ideal breeding gounds for a supermassive black hole. As we have no means of directly detecting the gravitational signature of a black hole inside the massive clumps, the only way to infer the presence of a massive black hole would be through the spectral signature of an actively accreting AGN. As discussed in Sect. 4.1, we did not find any direct evidence for energetically significant AGNs in the DCOs.
It has long been suggested that the mass loss from stars and supernovae in a very massive compact cluster at the center of a galaxy may play a major role in the fueling of a supermassive black holes (Norman & Scoville, 1988). While the stellar winds from O and WR stars and the ejecta from core-collapse supernovae may in principle provide a significant amount of material available for accretion during the early stage of a massive central starburst, these are not expected to be viable fueling mechanisms because the speed of the outflowing stellar ejecta is far in excess of the escape velocity from the star cluster (Norman & Scoville, 1988). Norman & Scoville (1988) argue that the high mass-loss rates and the low mass-loss speeds of the winds from evolved AGB stars may offer a steady flow of material that will be retained within the potential well of the star cluster. It can then cool and (some fraction) can be accreted by the supermassive black hole. Thus, the growth phase of the supermassive black hole (a powerful AGN) is expected to occur only after the time that the mechanical feeedback energy input from core collapse supernovae has subsided and the progenitors of AGB stars have reached the end of their main sequence lifetimes. This will occur at 50 Myr after the onset of the starburst (Leitherer & Heckman, 1995), and this age constraint is consistent with the observed ages of the nuclear starbursts observed in a sample of local AGN (Davies et al., 2007). Under the assumption of an instantaneous burst, the ages of the DCOs shown in Fig. 4 and listed in Table 4 are generally much lower (5–29 Myr) than 50 Myr. Thus, it is perhaps not unexpected that under such a scenario the DCOs are currently not capable of efficiently fuelling a supermassive black hole.
We also note that in the “migrating–clump model” advocated by Bournaud et al. (2007, 2008) and Elmegreen et al. (2008b), supermassive nuclear black holes may form under the assumption that each of the clumps gave rise to an intermediate mass black hole prior to its migration toward the center where they can coalesce. Such a mechanism may be able to explain the presence of seed black holes of substantial mass in the centers of gas-rich galaxies dominated by massive, turbulence-induced star-forming clumps. Both mechanisms described above may therefore predict significant delay times between the onset of a starburst and that of AGN.
6. Summary and concluding remarks
We have presented the results of the analysis of a sample of 30 low-redshift () galaxies selected from the union of the GALEX and SDSS surveys to have the high far-UV luminosities and small sizes characteristic of the high-redshift Lyman Break Galaxies (Heckman et al., 2005; Hoopes et al., 2007; Basu-Zych et al., 2007; Overzier et al., 2008). These “Lyman Break Analogs” (LBAs) offer the opportunity to investigate the physical processes occurring in Lyman Break Galaxies (LBGs) in considerably more detail than is possible at high-redshift777Heckman et al. (2005) coined the term “living fossils” in relation to the Lyman break analogs sample then discovered. Indeed, an apt analogy in the field of evolutionary biology is presented by the Coelacanth. A modern-day analog of this species of fish, known from the fossil record and believed to be extinct since the end of the Cretaceous period, was discovered in 1938. The find illustrates the use of analogs in the study of cosmology. Even though the local analogs will differ from their distant look-a-likes to some level of detail, and even though their local environment has probably changed drastically, they offer an entirely new way of investigating the fossil record alongside with other, more traditional lines of research.. In this paper we reported on the internal structures of the LBA’s using HST optical and UV images. We complemented these data with mid/far-IR photometry and optical spectroscopy from the SDSS and the VLT.
We have shown that the morphologies of the LBAs are dominated by massive compact clumps of young stars. Our most unexpected and potentially important result is that we have found six galaxies in which the galaxy UV morphology is dominated by a single very compact object which is unresolved or only marginally resolved in the HST images. These dominant compact objects (“DCOs”) have estimated masses of few to few M and radii of-order 100 pc or less. In the other LBAs we instead find multiple clumps per galaxy, the brightest of which have estimated masses of typically a few and radii of several hundred pc.
We show that the galaxies containing the DCOs differ from the other
LBAs in many respects:
1. Their host galaxies tend to be more massive
than the other LBAs (median masses of
2. They have systematically higher central concentrations
of light than the other LBAs (the DCO is typically located near the
center of a larger disk or envelope).
3. The DCOs are found in
galaxies with high mid-infrared luminosities: L to , roughly an
order-of-magnitude more luminous than the other LBAs and comparable to
luminous infrared galaxies.
4. The galaxies with a DCO are more
strongly displaced from the locus of normal star forming galaxies in
the standard BPT diagnostic emission-line ratio diagram. This
displacement is in the sense of having larger [O]/H and/or
larger [N]/H ratios.
5. The DCOs tend to occur in galaxies
with higher gas densities and pressures as traced by the ionized gas
( cm K).
6. The DCOs are found in galaxies
in which the star-formation rate inferred from the dust-corrected
H luminosity tends to be smaller than that derived from the
mid-IR luminosity and/or the extinction-corrected far-UV
luminosity. In the most extreme cases, this discrepancy reaches
roughly an order-of-magnitude.
We have shown that the DCOs are not Type 1 (unobscured) AGN. We find no evidence for the presence of a very broad (several thousand km/s) H emission-line, with the most stringent constraints provided by our VLT GIRAFFE/ARGUS IFU spectra. Instead, these data show blue-asymmetric wings on the H and [N] line profiles indicative of a dusty outflow of ionized gas. While we can not entirely exclude the possibility that an energetically significant Type 2 (obscured) AGN is present in some or all of the DCOs, we regard this possibility as unlikely. In particular, the DCOs are associated with galaxies having unusually weak [S] and [O] emission-lines compared to normal star forming galaxies. This is the opposite of what is seen in starburst-AGN composite systems (e.g. Kewley et al., 2006), but similar to what is seen in low redshift IR-warm starbursts (Kewley et al., 2001).
The continuum of the DCOs is dominated by the light of young stars: these are clearly strong highly compact starbursts. This is consistent with both the high gas pressures and emission-line profiles: both are commonly observed in starbursts and are the signposts of supernova-driven galactic winds (e.g. Lehnert & Heckman, 1996). We have considered two specific scenarios under which the starbursts associated with the DCOs might be unusual. They may be evolved short-duration starbursts in which the rate of photoionization is relatively low because few O stars are present. They may also be “leaky” starbursts in which a significant fraction of the ionizing radiation escapes (matter bounded conditions). The latter scenario seems less likely, since it is difficult to envision circumstances in which most of the ionizing radiation escapes the galaxy, but most of the non-ionizing UV radiation is absorbed by dust and reprocessed in the mid/far IR.
We have compared the properties of the DCOs and other massive clumps to those of the nuclear star clusters typically seen in different kinds of nearby galaxies. The DCOs are highly massive and dense objects that do not appear to be related to single, young star clusters which are significantly less massive ( M) and smaller (10 pc) than the structures we see in the LBAs. However, the DCO structures are quite similar to predictions for the merger-driven starbursts that are believed to be the progenitors of the extra light or central cusps seen generically in lower-mass () elliptical galaxies (Kormendy et al., 2008; Hopkins et al., 2009). This would be consistent with our optical images of the LBAs, most of which strongly suggest on-going mergers. However, at present it is unclear whether the DCO hosts could passively evolve to form a full-grown elliptical without additional merging.
More generally, the properties of the LBAs are quite consistent with the idea that large Jeans instabilities in a gas-rich disk will lead to the formation of highly massive star forming clumps. At high redshift, it is not yet clear what the main source of this gas is: it could be replenished continuously, or come in the form of discrete accretion events of small galaxies or clouds (or, most likely, a combination thereof). These processes have been proposed to explain the properties of high- star-forming galaxies (e.g., see Overzier et al., 2008; Basu-Zych et al., 2009; Elmegreen et al., 2009; Förster Schreiber et al., 2009; Law et al., 2007, 2009, and references therein). In any event, it is expected that the massive stellar clumps will aggregate at the center of a galaxy via dynamical friction to ultimately form a bulge and supermassive black hole (Elmegreen et al., 2009, and references therein). The LBAs offer us the opportunity to study such processes in more detail, and compare samples across different redshifts.
We speculated about why we do not see clear evidence for the formation or rapid growth of a supermassive black hole in the DCOs (which appear to be the ideal black hole nurseries). One intriguing possibility is that these objects are simply too young for a black hole to be forming yet. Following Norman & Scoville (1988) and Davies et al. (2007), the growth of a supermassive black hole inside a very massive compact star cluster may not commence until the cluster is old enough for the core-collapse supernovae phase to end (cluster age 50 Myr). Subsequent to this, gas shed by evolved stars will be retained in the cluster’s potential well rather than being blown out. This gas can then cool, flow inward, and be used to grow the black hole. If correct, we should be able to identify a population of the descendants of the young DCOs in a phase as powerful AGN. This scenario might also account for the rarity of strong AGN in the LBG population (e.g. Steidel et al., 2002; Ouchi et al., 2008).
Follow-up UV spectroscopy using the HST Cosmic Origins Spectrograph has been allocated for selected targets in Cycle 17. The UV spectra will allow us to study many of the features discussed in this paper that are relevant to LBAs and to starbursts in general. These include the role of galactic outflows and low-luminosity AGN, the escape of ionizing radiation, the strength of Ly emission, stellar abundances and the initial mass function.
- Allen et al. (2008) Allen, M. G., Groves, B. A., Dopita, M. A., Sutherland, R. S., & Kewley, L. J. 2008, ApJS, 178, 20
- Armus et al. (1988) Armus, L., Heckman, T. M., & Miley, G. K. 1988, ApJ, 326, L45
- Armus et al. (2007) Armus, L., et al. 2007, ApJ, 656, 148
- Arribas et al. (2008) Arribas, S., Colina, L., Monreal-Ibero, A., Alfonso, J., García-Marín, M., & Alonso-Herrero, A. 2008, A&A, 479, 687
- Baker et al. (2001) Baker, A. J., Lutz, D., Genzel, R., Tacconi, L. J., & Lehnert, M. D. 2001, A&A, 372, L37
- Baldwin et al. (1981) Baldwin, J. A., Phillips, M. M., & Terlevich, R. 1981, PASP, 93, 5
- Bastian (2008) Bastian, N. 2008, MNRAS, 390, 759
- Basu-Zych et al. (2007) Basu-Zych, A. R., et al. 2007, ApJS, 173, 457
- Basu-Zych et al. (2009) Basu-Zych, A., et al. 2009, ApJ, 699, L118
- Beckwith et al. (2006) Beckwith, S. V. W., et al. 2006, AJ, 132, 1729
- Bekki (2004) Bekki, K. 2004, Publications of the Astronomical Society of Australia, 21, 167
- Binette et al. (1996) Binette, L., Wilson, A. S., & Storchi-Bergmann, T. 1996, A&A, 312, 365
- Binney & Tremaine (2008) Binney, J., & Tremaine, S. 2008, Galactic Dynamics: Second Edition, by James Binney and Scott Tremaine. ISBN978-0-691-13026-2 (HB). Published by Princeton University Press, Princeton, NJ USA, 2008.,
- Böker et al. (2004) Böker, T., Sarzi, M., McLaughlin, D. E., van der Marel, R. P., Rix, H.-W., Ho, L. C., & Shields, J. C. 2004, AJ, 127, 105
- Bournaud et al. (2007) Bournaud, F., Elmegreen, B. G., & Elmegreen, D. M. 2007, ApJ, 670, 237
- Bournaud et al. (2008) Bournaud, F., et al. 2008, A&A, 486, 741
- Bournaud & Elmegreen (2009) Bournaud, F., & Elmegreen, B. G. 2009, ApJ, in press (arXiv:0902.2806)
- Bouwens et al. (2004) Bouwens, R. J., Illingworth, G. D., Blakeslee, J. P., Broadhurst, T. J., & Franx, M. 2004, ApJ, 611, L1
- Bouwens et al. (2009) Bouwens, R. J., et al. 2009, ApJ, Submitted
- Brinchmann et al. (2008a) Brinchmann, J., Pettini, M., & Charlot, S. 2008a, MNRAS, 385, 769
- Brinchmann et al. (2008b) Brinchmann, J., Kunth, D., & Durret, F. 2008b, A&A, 485, 657
- Bruzual & Charlot (2003) Bruzual, G., & Charlot, S. 2003, MNRAS, 344, 1000
- Buat et al. (2007) Buat, V., Marcillac, D., Burgarella, D., Le Floc’h, E., Takeuchi, T. T., Iglesias-Paràmo, J., & Xu, C. K. 2007, A&A, 469, 19
- Burgarella et al. (2007) Burgarella, D., Le Floc’h, E., Takeuchi, T. T., Huang, J. S., Buat, V., Rieke, G. H., & Tyler, K. D. 2007, MNRAS, 380, 986
- Burgarella et al. (2009) Burgarella, D., Buat, V., Takeuchi, T. T., Wada, T., & Pearson, C. 2009, PASJ, 61, 177
- Calzetti (2001) Calzetti, D. 2001, PASP, 113, 1449
- Calzetti et al. (2007) Calzetti, D., et al. 2007, ApJ, 666, 870
- Calzetti et al. (2009) Calzetti, D., Sheth, K., Churchwell, E., & Jackson, J. 2009, The Evolving ISM in the Milky Way and Nearby Galaxies,
- Caputi et al. (2007) Caputi, K. I., et al. 2007, ApJ, 660, 97
- Carollo et al. (1997) Carollo, C. M., Stiavelli, M., de Zeeuw, P. T., & Mack, J. 1997, AJ, 114, 2366
- Castellanos et al. (2002) Castellanos, M., Díaz, Á. I., & Tenorio-Tagle, G. 2002, ApJ, 565, L79
- Charlot & Longhetti (2001) Charlot, S., & Longhetti, M. 2001, MNRAS, 323, 887
- Chevalier & Clegg (1985) Chevalier, R. A., & Clegg, A. W. 1985, Nature, 317, 44
- Conroy et al. (2008) Conroy, C., Gunn, J. E., & White, M. 2008, ApJ, submitted (arXiv:0809.4261)
- Conselice et al. (2008) Conselice, C. J., Rajgor, S., & Myers, R. 2008, MNRAS, 386, 909
- Côté et al. (2006) Côté, P., et al. 2006, ApJS, 165, 57
- Cowie et al. (1995) Cowie, L. L., Hu, E. M., & Songaila, A. 1995, AJ, 110, 1576
- Crane et al. (1993) Crane, P., et al. 1993, AJ, 106, 1371
- Daddi et al. (2009) Daddi, E., et al. 2009, ApJ, 694, 1517
- Dale & Helou (2002) Dale, D. A., & Helou, G. 2002, ApJ, 576, 159
- Dale et al. (2005) Dale, D. A., et al. 2005, ApJ, 633, 857
- Dasyra et al. (2006a) Dasyra, K. M., et al. 2006a, ApJ, 638, 745
- Dasyra et al. (2006b) Dasyra, K. M., et al. 2006b, ApJ, 651, 835
- Davies et al. (2007) Davies, R. I., Sánchez, F. M., Genzel, R., Tacconi, L. J., Hicks, E. K. S., Friedrich, S., & Sternberg, A. 2007, ApJ, 671, 1388
- di Matteo et al. (2007) di Matteo, P., Combes, F., Melchior, A.-L., & Semelin, B. 2007, A&A, 468, 61
- Dopita & Sutherland (1995) Dopita, M. A., & Sutherland, R. S. 1995, ApJ, 455, 468
- Elmegreen et al. (2005) Elmegreen, D. M., Elmegreen, B. G., Rubin, D. S., & Schaffer, M. A. 2005, ApJ, 631, 85
- Elmegreen et al. (2008a) Elmegreen, B. G., Bournaud, F., & Elmegreen, D. M. 2008a, ApJ, 684, 829
- Elmegreen et al. (2008b) Elmegreen, B. G., Bournaud, F., & Elmegreen, D. M. 2008b, ApJ, 688, 67
- Elmegreen et al. (2009) Elmegreen, B. G., Elmegreen, D. M., Fernandez, M. X., & Lemonias, J. J. 2009, ApJ, 692, 12
- Eminian et al. (2008) Eminian, C., Kauffmann, G., Charlot, S., Wild, V., Bruzual, G., Rettura, A., & Loveday, J. 2008, MNRAS, 384, 930
- Erb et al. (2006a) Erb, D. K., Shapley, A. E., Pettini, M., Steidel, C. C., Reddy, N. A., & Adelberger, K. L. 2006a, ApJ, 644, 813
- Erb et al. (2006b) Erb, D. K., Steidel, C. C., Shapley, A. E., Pettini, M., Reddy, N. A., & Adelberger, K. L. 2006b, ApJ, 647, 128
- Ferguson et al. (2004) Ferguson, H. C., et al. 2004, ApJ, 600, L107
- Ferrarese et al. (2006) Ferrarese, L., et al. 2006, ApJ, 644, L21
- Förster Schreiber et al. (2009) Förster Schreiber, N. M., et al. 2009, ApJ, submitted (arXiv:0903.1872)
- Franx et al. (1997) Franx, M., Illingworth, G. D., Kelson, D. D., van Dokkum, P. G., & Tran, K.-V. 1997, ApJ, 486, L75
- Geha et al. (2002) Geha, M., Guhathakurta, P., & van der Marel, R. P. 2002, AJ, 124, 3073
- Genzel et al. (2001) Genzel, R., Tacconi, L. J., Rigopoulou, D., Lutz, D., & Tecza, M. 2001, ApJ, 563, 527
- Giammanco et al. (2005) Giammanco, C., Beckman, J. E., & Cedrés, B. 2005, A&A, 438, 599
- Giavalisco et al. (1996a) Giavalisco, M., Steidel, C. C., & Macchetto, F. D. 1996a, ApJ, 470, 189
- Giavalisco (2002) Giavalisco, M. 2002, ARA&A, 40, 579
- Goldader et al. (2002) Goldader, J. D., Meurer, G., Heckman, T. M., Seibert, M., Sanders, D. B., Calzetti, D., & Steidel, C. C. 2002, ApJ, 568, 651
- Groves et al. (2006) Groves, B. A., Heckman, T. M., & Kauffmann, G. 2006, MNRAS, 371, 1559
- Hayashi et al. (2009) Hayashi, M., et al. 2009, ApJ, 691, 140
- Heckman et al. (1981) Heckman, T. M., Miley, G. K., van Breugel, W. J. M., & Butcher, H. R. 1981, ApJ, 247, 403
- Heckman et al. (1990) Heckman, T. M., Armus, L., & Miley, G. K. 1990, ApJS, 74, 833
- Heckman et al. (1998) Heckman, T. M., Robert, C., Leitherer, C., Garnett, D. R., & van der Rydt, F. 1998, ApJ, 503, 646
- Heckman et al. (2005) Heckman, T. M., et al. 2005, ApJ, 619, L35
- Hoopes et al. (2007) Hoopes, C., et al. 2007, Accepted for publication in the GALEX special issue of ApJS (astro-ph/0609415)
- Hopkins et al. (2002) Hopkins, A. M., Schulte-Ladbeck, R. E., & Drozdovsky, I. O. 2002, AJ, 124, 862
- Hopkins et al. (2008) Hopkins, P. F., Hernquist, L., Cox, T. J., Dutta, S. N., & Rothberg, B. 2008, ApJ, 679, 156
- Hopkins et al. (2009) Hopkins, P. F., Cox, T. J., Dutta, S. N., Hernquist, L., Kormendy, J., & Lauer, T. R. 2009, ApJS, 181, 135
- Huang et al. (2005) Huang, J.-S., et al. 2005, ApJ, 634, 137
- Hughes et al. (1998) Hughes, D. H., et al. 1998, Nature, 394, 241
- Immeli et al. (2004) Immeli, A., Samland, M., Westera, P., & Gerhard, O. 2004, ApJ, 611, 20
- Iwata et al. (2009) Iwata, I., et al. 2009, ApJ, 692, 1287
- Johnson et al. (2007) Johnson, B. D., et al. 2007, ApJS, 173, 392
- Kauffmann et al. (2003) Kauffmann, G., et al. 2003, MNRAS, 346, 1055
- Kauffmann & Heckman (2009) Kauffmann, G., & Heckman, T. M. 2008, MNRAS, submitted (arXiv:0812.1224)
- Kennicutt et al. (1998) Kennicutt, R. C., Jr., Schweizer, F., Barnes, J. E., Friedli, D., Martinet, L., & Pfenniger, D. 1998, Saas-Fee Advanced Course 26: Galaxies: Interactions and Induced Star Formation,
- Kereš et al. (2009) Kereš, D., Katz, N., Fardal, M., Davé, R., & Weinberg, D. H. 2009, MNRAS, 395, 160
- Kewley et al. (2001) Kewley, L. J., Heisler, C. A., Dopita, M. A., & Lumsden, S. 2001, ApJS, 132, 37
- Kewley et al. (2006) Kewley, L. J., Groves, B., Kauffmann, G., & Heckman, T. 2006, MNRAS, 372, 961
- Kissler-Patig et al. (2006) Kissler-Patig, M., Jordán, A., & Bastian, N. 2006, A&A, 448, 1031
- Kriek et al. (2007) Kriek, M., et al. 2007, ApJ, 669, 776
- Koekemoer et al. (2002) Koekemoer, A. M., Fruchter, A. S., Hook, R. N., & Hack, W. 2002, The 2002 HST Calibration Workshop : Hubble after the Installation of the ACS and the NICMOS Cooling System, Proceedings of a Workshop held at the Space Telescope Science Institute, Baltimore, Maryland, October 17 and 18, 2002. Edited by Santiago Arribas, Anton Koekemoer, and Brad Whitmore. Baltimore, MD: Space Telescope Science Institute, 2002., p.337, 337
- Kong et al. (2004) Kong, X., Charlot, S., Brinchmann, J., & Fall, S. M. 2004, MNRAS, 349, 769
- Kormendy et al. (2008) Kormendy, J., Fisher, D. B., Cornell, M. E., & Bender, R. 2008, ApJS, in press (arXiv:0810.1681)
- Krist (1995) Krist, J. 1995, Astronomical Data Analysis Software and Systems IV, 77, 349
- Kroupa (2008) Kroupa, P. 2008, Pathways Through an Eclectic Universe, 390, 3
- Lambas et al. (2003) Lambas, D. G., Tissera, P. B., Alonso, M. S., & Coldwell, G. 2003, MNRAS, 346, 1189
- Larsen (2002) Larsen, S. S. 2002, AJ, 124, 1393
- Larson & Tinsley (1978) Larson, R. B., & Tinsley, B. M. 1978, ApJ, 219, 46
- Lauer et al. (1995) Lauer, T. R., et al. 1995, AJ, 110, 2622
- Lauer et al. (2007) Lauer, T. R., et al. 2007, ApJ, 664, 226
- Law et al. (2007) Law, D. R., Steidel, C. C., Erb, D. K., Larkin, J. E., Pettini, M., Shapley, A. E., & Wright, S. A. 2007, ApJ, 669, 929
- Law et al. (2009) Law, D. R., Steidel, C. C., Erb, D. K., Larkin, J. E., Pettini, M., Shapley, A. E., & Wright, S. A. 2009, ApJ, submitted (arXiv:0901.2930)
- Lehnert & Heckman (1996) Lehnert, M. D., & Heckman, T. M. 1996, ApJ, 462, 651
- Lehnert et al. (2009) Lehnert, M. D., Nesvadba, N. P. H., Le Tiran, L., Di Matteo, P., van Driel, W., Douglas, L. S., Chemin, L., & Bournaud, F. 2009, arXiv:0902.2784
- Leitherer & Heckman (1995) Leitherer, C., & Heckman, T. M. 1995, ApJS, 96, 9
- Lipari & Macchetto (1992) Lipari, S., & Macchetto, F. 1992, ApJ, 387, 522
- Li et al. (2007) Li, Y., Haiman, Z., & Mac Low, M.-M. 2007, ApJ, 663, 61
- Li et al. (2008) Li, C., Kauffmann, G., Heckman, T. M., Jing, Y. P., & White, S. D. M. 2008, MNRAS, 385, 1903
- Liu et al. (2008) Liu, X., Shapley, A. E., Coil, A. L., Brinchmann, J., & Ma, C.-P. 2008, ApJ, 678, 758
- Lotz et al. (2006) Lotz, J. M., Madau, P., Giavalisco, M., Primack, J., & Ferguson, H. C. 2006, ApJ, 636, 592
- Magorrian et al. (1998) Magorrian, J., et al. 1998, AJ, 115, 2285
- Maraston et al. (2004) Maraston, C., Bastian, N., Saglia, R. P., Kissler-Patig, M., Schweizer, F., & Goudfrooij, P. 2004, A&A, 416, 467
- Marconi & Hunt (2003) Marconi, A., & Hunt, L. K. 2003, ApJ, 589, L21
- Meurer et al. (1995) Meurer, G. R., Heckman, T. M., Leitherer, C., Kinney, A., Robert, C., & Garnett, D. R. 1995, AJ, 110, 2665
- Meurer et al. (1997) Meurer, G. R., Heckman, T. M., Lehnert, M. D., Leitherer, C., & Lowenthal, J. 1997, AJ, 114, 54
- Meurer et al. (1999) Meurer, G. R., Heckman, T. M., & Calzetti, D. 1999, ApJ, 521, 64
- Mieske et al. (2004) Mieske, S., et al. 2004, AJ, 128, 1529
- Mihos & Hernquist (1994a) Mihos, J. C., & Hernquist, L. 1994a, ApJ, 425, L13
- Mihos & Hernquist (1994b) Mihos, J. C., & Hernquist, L. 1994b, ApJ, 431, L9
- Monreal-Ibero et al. (2006) Monreal-Ibero, A., Arribas, S., & Colina, L. 2006, ApJ, 637, 138
- Nikolic et al. (2004) Nikolic, B., Cullen, H., & Alexander, P. 2004, MNRAS, 355, 874
- Noguchi (1999) Noguchi, M. 1999, ApJ, 514, 77
- Norman & Scoville (1988) Norman, C., & Scoville, N. 1988, ApJ, 332, 124
- Osterbrock (1981) Osterbrock, D. E. 1981, ApJ, 249, 462
- Osterbrock & Ferland (2006) Osterbrock, D. E., & Ferland, G. J. 2006, Astrophysics of gaseous nebulae and active galactic nuclei, 2nd. ed. by D.E. Osterbrock and G.J. Ferland. Sausalito, CA: University Science Books, 2006,
- Östlin et al. (2009) Östlin, G., Hayes, M., Kunth, D., Mas-Hesse, J. M., Leitherer, C., Petrosian, A., & Atek, H. 2009, AJ, 138, 923
- Ouchi et al. (2008) Ouchi, M., et al. 2008, ApJS, 176, 301
- Overzier et al. (2008) Overzier, R. A., et al. 2008, ApJ, 677, 37
- Overzier et al. (2009) Overzier, R. A., et al. 2009, in prep.
- Owers et al. (2007) Owers, M., Blake, C., Couch, W., Pracy, M., & Bekki, K. 2007, ArXiv e-prints, 707, arXiv:0707.2566
- Papovich et al. (2001) Papovich, C., Dickinson, M., & Ferguson, H. C. 2001, ApJ, 559, 620
- Papovich et al. (2006) 2006, ApJ, 640, 92
- Pettini & Pagel (2004) Pettini, M., & Pagel, B. E. J. 2004, MNRAS, 348, L59
- Reddy et al. (2006) Reddy, N., et al. 2006, ApJ, 644, 792
- Reddy & Steidel (2009) Reddy, N. A., & Steidel, C. C. 2009, ApJ, 692, 778
- Rieke et al. (2009) Rieke, G. H., Alonso-Herrero, A., Weiner, B. J., Pérez-González, P. G., Blaylock, M., Donley, J. L., & Marcillac, D. 2009, ApJ, 692, 556
- Rossa et al. (2006) Rossa, J., van der Marel, R. P., Böker, T., Gerssen, J., Ho, L. C., Rix, H.-W., Shields, J. C., & Walcher, C.-J. 2006, AJ, 132, 1074
- Salim et al. (2007) Salim, S., et al. 2007, ApJS, 173, 267
- Sanders & Mirabel (1996) Sanders, D. B., & Mirabel, I. F. 1996, ARA&A, 34, 749
- Scoville et al. (2000) Scoville, N. Z., et al. 2000, AJ, 119, 991
- Seibert et al. (2005) Seibert, M., et al. 2005, ApJ, 619, L55
- Seth et al. (2008) Seth, A., Agüeros, M., Lee, D., & Basu-Zych, A. 2008, ApJ, 678, 116
- Shankar & Bernardi (2009) Shankar, F., & Bernardi, M. 2009, MNRAS, 396, L76
- Shapiro et al. (2009) Shapiro, K. L., et al. 2009, arXiv:0902.4704
- Shapley et al. (2001) Shapley, A. E., Steidel, C. C., Adelberger, K. L., Dickinson, M., Giavalisco, M., & Pettini, M. 2001, ApJ, 562, 95
- Shapley et al. (2005a) Shapley, A. E., Steidel, C. C., Erb, D. K., Reddy, N. A., Adelberger, K. L., Pettini, M., Barmby, P., & Huang, J. 2005, ApJ, 626, 698
- Shapley et al. (2005b) Shapley, A. E., Coil, A. L., Ma, C.-P., & Bundy, K. 2005, ApJ, 635, 1006
- Shapley et al. (2006) Shapley, A. E., Steidel, C. C., Pettini, M., Adelberger, K. L., & Erb, D. K. 2006, ApJ, 651, 688
- Shaya et al. (1994) Shaya, E. J., Dowling, D. M., Currie, D. G., Faber, S. M., & Groth, E. J. 1994, AJ, 107, 1675
- Siana et al. (2008) Siana, B., Teplitz, H. I., Chary, R.-R., Colbert, J., & Frayer, D. T. 2008, ApJ, 689, 59
- Spinoglio et al. (2002) Spinoglio, L., Andreani, P., & Malkan, M. A. 2002, ApJ, 572, 105
- Springel & Hernquist (2005) Springel, V., & Hernquist, L. 2005, ApJ, 622, L9
- Stark et al. (2008) Stark, D. P., Swinbank, A. M., Ellis, R. S., Dye, S., Smail, I. R., & Richard, J. 2008, Nature, 455, 775
- Stark et al. (2009) Stark, D. P., Ellis, R. S., Bunker, A., Bundy, K., Targett, T., Benson, A., & Lacy, M. 2009, ApJ, 697, 1493
- Steidel et al. (1999) Steidel, C. C., Adelberger, K. L., Giavalisco, M., Dickinson, M., & Pettini, M. 1999, ApJ, 519, 1
- Steidel et al. (2002) Steidel, C. C., Hunt, M. P., Shapley, A. E., Adelberger, K. L., Pettini, M., Dickinson, M., & Giavalisco, M. 2002, ApJ, 576, 653
- Tacconi et al. (2002) Tacconi, L. J., Genzel, R., Lutz, D., Rigopoulou, D., Baker, A. J., Iserlohe, C., & Tecza, M. 2002, ApJ, 580, 73
- Telles & Terlevich (1997) Telles, E., & Terlevich, R. 1997, MNRAS, 286, 183
- Teplitz et al. (2000) Teplitz, H. I., et al. 2000, ApJ, 533, L65
- Tremonti et al. (2004) Tremonti, C. A., et al. 2004, ApJ, 613, 898
- Treyer et al. (2007) Treyer, M., et al. 2007, ApJS, 173, 256
- Trujillo et al. (2006) Trujillo, I., et al. 2006, ApJ, 650, 18
- Vázquez & Leitherer (2005) Vázquez, G. A., & Leitherer, C. 2005, ApJ, 621, 695
- Veilleux & Osterbrock (1987) Veilleux, S., & Osterbrock, D. E. 1987, ApJS, 63, 295
- Walcher et al. (2005) Walcher, C. J., et al. 2005, ApJ, 618, 237
- Weidner et al. (2004) Weidner, C., Kroupa, P., & Larsen, S. S. 2004, MNRAS, 350, 1503
- Wehner & Harris (2006) Wehner, E. H., & Harris, W. E. 2006, ApJ, 644, L17
- Whitmore et al. (2007) Whitmore, B. C., Chandar, R., & Fall, S. M. 2007, AJ, 133, 1067
- Whittle et al. (1988) Whittle, M., Pedlar, A., Meurs, E. J. A., Unger, S. W., Axon, D. J., & Ward, M. J. 1988, ApJ, 326, 125
- Woods et al. (2006) Woods, D. F., Geller, M. J., & Barton, E. J. 2006, AJ, 132, 197
|ID||log||Star Formation Rate||log||log|
|()||( yr)||()||(kpc)||(km s)||()||(Myr)|
|()||(pc)||( kpc)||(km s)||(Myr)||(Myr)||()|