Star formation histories in mergers: The spatially resolved properties of the early-stage merger LIRGs IC 1623 and NGC 6090
The role of major mergers in galaxy evolution is investigated through a detailed characterization of the stellar populations, ionized gas properties, and star formation rates (SFR) in the early-stage merger LIRGs IC 1623 W and NGC 6090, by analysing optical Integral Field Spectroscopy (IFS) and high resolution HST imaging. The spectra were processed with the starlight full spectral fitting code, and the emission lines measured in the residual spectra. The results are compared with control non-interacting spiral galaxies from the CALIFA survey. Merger-induced star formation is extended and recent, as revealed by the young ages (50–80 Myr) and high contributions to light of young stellar populations (50–90), in agreement with merger simulations in the literature. These early-stage mergers have positive central gradients of the stellar metallicity, with an average 0.6 Z. Compared to non-interacting spirals, they have lower central nebular metallicity, and flatter profiles, in agreement with the gas inflow scenario. We find that they are dominated by star formation, although shock excitation cannot be discarded in some regions, where high velocity dispersion is found (170–200 km s). The average SFR in these early-stage mergers (23–32 M yr) is enhanced with respect to main-sequence Sbc galaxies by factors of 6–9, slightly above the predictions from classical merger simulations, but still possible in about 15 of major galaxy mergers, where U/LIRGs belong.
keywords:galaxies: interactions – galaxies: evolution – galaxies: stellar content – galaxies: ISM – galaxies: star formation – techniques: spectroscopic
Merger-induced star formation was first predicted by Toomre & Toomre (1972), and confirmed by Larson & Tinsley (1978) in a study of optical colours of Arp galaxies. However, not all merging galaxies have significantly enhanced star formation rates (SFR). Recent results show that the enhancement of the star formation depends on multiple factors, such as the merger geometry and the properties of the progenitor galaxies (Cox et al., 2008; Di Matteo et al., 2007, 2008). For example, gas-rich (S+S) merging pairs show significant enhancements in the specific star formation rates (sSFR) that are not found in S+E pairs (Cao et al., 2016). Moreover, a clear anti-correlation between the sSFR and the pair separation has been established. In particular, close pairs with projected separation 20 h kpc have sSFR enhancement 2 (Ellison et al., 2008), while the sSFR enhancement of wider pairs is significantly less (Patton et al., 2013).
Besides enhanced SFR, the spatial extent of the star formation is other important factor that must be taken into account when comparing the star formation in mergers with that in non-interacting control galaxies. Extended star formation is observed in many galaxy mergers from early-stage merger or separated progenitors (Wang et al., 2004; Elmegreen et al., 2006) to advanced or post-coalescence systems (Evans et al., 2008; Wilson et al., 2006). Recent high resolution models determine the SFR efficiency in mergers by resolving parsec scale physical processes (Teyssier et al., 2010; Hopkins et al., 2013; Renaud et al., 2015). These simulations find that extended starbursts arise spontaneously during the first two pericentre passages, due to fragmentation of the gas clouds produced by an increase of the ISM supersonic turbulence as a consequence of the tidal interaction itseft (Renaud et al., 2014). In these models, a merger-induced nuclear starburst is also present, but it starts later in the merger, during the final passage. They conclude that extended star formation is important in the early stages of the merger, while the nuclear starbursts will occur in advanced stages.
Mergers of gas rich discs, such as those seen in luminous and ultra-luminous infrared galaxies (LIRGs and ULIRGs, Surace et al. 1998, 2000; Veilleux et al. 2002; Kim et al. 2013) are therefore extreme examples, where star formation is caught at its peak. They are unique laboratories offering insight into the physical processes triggered by galaxy mergers. The observational characterization of star formation in mergers at different stages is necessary to test the validity, and to put contraints on merger simulations.
To understand the role of major mergers in galaxy evolution we have undertaken an integral field spectroscopy study of galaxies in different stages across the merger sequence. In this paper we focus on the detailed study of two LIRGs at an early merger stage: IC 1623 and NGC 6090. Both are classified as type IIIb systems according to the morphological classification of Veilleux et al. (2002). We aim to fully characterize their spatially resolved stellar population, and ionized gas properties. In a future work we plan to extend this study to merging systems observed in the Calar Alto Legacy Integral Field Area (CALIFA) survey (Sánchez et al., 2012), in order to cover different merger stages and to statistically study the evolution of merger-induced star formation.
The paper is organized as follows: Sect. 2 describes the observations and data reduction process. In Sect. 3 we summarize the results from the photometric analysis of the star clusters detected in HST images. In Sect. 4 we apply the fossil record method to analize the stellar continuum and derive the spatially resolved stellar population properties: stellar mass and stellar mass surface density, ; stellar extinction, A; luminosity weighted mean age, ; mass weighted mean metallicity, ; and the contributions to luminosity and mass of young, intermediate, and old stellar populations. Section 5 presents results on the ionized gas emission, focusing on the morphology, nebular A, and the ionization conditions. We discuss the results in Sect. 6, and Sect. 7 presents the conclusions.
Throughout the paper we assume a flat cosmology with = 0.272, = 0.728, and = 70.4 km s Mpc (WMAP, seven years results). For IC 1623 redshift (z=0.020067) this results in a distance of 86.8 Mpc. At this distance 1 corresponds to 0.421 kpc. For NGC 6090 (z=0.029304), the distance is 127.7 Mpc, and 1 corresponds to 0.619 kpc.
1.1 Ic 1623
IC 1623 (also commonly referred to as VV 114 = Mrk 236; L = 4.5 10 L; e.g., Sanders et al. 2003) is a pair of luminous infrared interacting galaxies at 86.8 Mpc, with a projected separation between the two nuclei of 6 kpc. The western progenitor, IC 1623 W, is very bright in the UV and optical, with many star clusters detected in HST images. Its properties are very similar to Lyman Break galaxies, and it is considered as the nearest Lyman Break Analog (Grimes et al., 2006). In this paper we are concentrating on the western progenitor, which is covered by our IFS data. The top left panel of Figure 1 shows the HST F435W image of IC 1623, with rectangles indicating the positions where we have obtained IFS observations. In contrast, IC 1623 E is optically obscured but very infrared-bright. About half of the warmer dust traced in the mid-IR is associated with the eastern galaxy, where both compact (nuclear region) and extended emission is found (Le Floc’h et al., 2002). Evidence for wide-spread star formation activity and shocks across the entire system is found in the UV, optical, and mid-IR (Knop et al., 1994; Goldader et al., 2002; Rich et al., 2011). Analysing far-UV and X-ray data, Grimes et al. (2006) found a galactic superwind in IC 1623 moving at 300–400 kms. Assuming a stellar origin for the hard X-ray emission (i.e., no contamination by an AGN), they estimate a similar SFR for both progenitors: 28 M yr for IC 1623 E and 33 M yr for IC 1623 W. This is probably an upper limit in the case of IC 1623 E, as a highly obscured AGN was identified in sub-arcsecond resolution ALMA observations of HCN (4-3) and HCO (4-3) (Iono et al., 2013).
1.2 Ngc 6090
NGC 6090 (Mrk 496 = UGC 10267; L = 3.2 10 L; e.g., Sanders et al. 2003) is another pair of luminous infrared interacting galaxies at 127.7 Mpc. The bottom left panel of Figure 1 shows the HST F435W image of NGC 6090. At optical and NIR wavelengths NGC 6090 appears as a double nucleus system separated by 3.2 kpc in projection (Dinshaw et al., 1999). The two galaxies are of roughly equal size, with a pair of tidal tails that have a full extent of 50 kpc. The northeastern galaxy (NGC 6090 NE) has a clear distorted spiral structure viewed face-on, while the southwestern nucleus (NGC 6090 SW) is seen edge-on, with an amorphous morphology and no spiral arms visible. The most striking feature of NGC 6090 is the abundance of star forming knots in the overlapping region, on the western side of NGC 6090 NE, and the paucity of similar clusters in its eastern spiral arms. Moreover, far- and near-UV spectroscopy shows that the NE nucleus is also dominated by a young starburst (González Delgado, 2008). NGC 6090 SW also shows a similar abundance of blue knots on its the eastern, and one extremely luminous knot at the northern end. The molecular gas has a single component that peaks between the two galactic nuclei in the overlap region (Bryant & Scoville, 1999; Wang et al., 2004), suggesting that the gas responds to the interaction faster than the stars. Although the two galaxies show strong H emission in their nuclei, the H emission appears to be brighter in the companion-facing side of each galaxy (Hattori et al., 2004). There is considerable evidence for starburst activity, but no evidence at optical or radio wavelengths for a compact AGN.
2 Observations and Data reduction
2.1 IFS Observations
The IFS observations we report here were taken using the Potsdam Multi-Aperture Spectrophotometer (PMAS) spectrograph (Roth et al., 2005) on the 3.5-m telescope of the Calar Alto Observatory (CAHA), with the Lens Array (LArr) configuration. We have used a spatial magnification of 0.75/lens, covering a 12x 12 field of view (FoV). From the standard stars and also the CAHA seeing monitor observations, we estimate that the seeing was in the range FWHM = 1.3–2.0 for the observations. The V300 grating was used, providing a 3.2 Å/pixel dispersion and covering a wavelength range 3700–7100 Å with a resolution of 7.1 Å FWHM (velocity resolution 170 km s). For each system two pointings were taken, as shown in Figure 1 by black rectangles over the HST images, with a total exposure time of 3h and 1.75h for NGC 6090 and IC 1623 W, respectively.
2.2 IFS Data Reduction
The PMAS LArr data reduction was performed using the R3D package (Sánchez, 2006), in combination with IRAF and our own Python scripts. The reduction consisted of the standard steps for fibre-based integral-field spectroscopy, including bias subtraction, and a flat-field correction that was based on a master CCD flat provided by the CAHA. Calibration frames (continuum and Hg/Ne lamps) were taken before the science frames to correct the for the effects of flexure. The location of the spectra in the CCD was determined using a continuum illuminated exposure. Each spectrum was extracted from the science frames by coadding the flux within an aperture of 5 pixels (3.75) around this location along the cross-dispersion axis for each pixel in the dispersion axis. Wavelength calibration was performed using the Hg and Ne lamp exposures. The final accuracy of the wavelength calibration is 0.38 Å. Differences in the fibre-to-fibre transmission throughput were corrected by comparing the wavelength-calibrated science frames with twilight-sky exposures. The spectra were then flux calibrated using the spectrophotometric standard stars Feige34 and BD+28D4211, observed in the same night. We corrected the standard stars datacubes from differential atmospheric refraction (DAR, Filippenko 1982). The median sky spectrum was obtained from the sky background frames, and subtracted from the target spectra. Once we had flux-calibrated and sky-subtracted all the individual frames for the same pointing position, they were combined and cosmic rays were rejected using IRAF’s task. Datacubes were created, corrected from DAR, and rotated to the North up, East to the left orientation. The two exposures at different locations were joined together in a mosaic.
Finally, we flux-recalibrated our LArr data using broad-band imaging from HST WFC435W in the case of IC 1623, and HST WFC435W together with SDSS g and r filters for NGC 6090. From the deviation between the photometry obtained from broad-band imaging (in a 4.5 arcsec aperture) and the photometry extracted from the IFS in the same aperture, we found, in both systems, an average accuracy of the spectrophotometric calibration of 15 across the wavelength range covered by our data set.
2.3 HST imaging data
In addition to our IFS data, we downloaded from the Hubble
Legacy Archive (HLA)
2.4 Overview of the data
In the right panels of Figure 1 we show the PMAS LArr continuum flux at 5110 Å rest-frame (obtained by averaging between 5050 to 5170 Å). We have superimposed in contour form the HST F435W image, smoothed to match the spatial resolution of the IFS ( 1.7 arcsec). To better identify the main regions, in the left panels of Fig. 2 we show the HST F435W image in greyscale with the IFS continuum superimposed as red contours, along with some spectra extracted in circular apertures of radius.
In the case of IC 1623 W, the nuclear region, the emission peak C1 (which corresponds to a giant star forming region), and the distorted spiral arm are labelled. The C1 spectrum is dominated by the emission lines and has a blue continuum. We detect broad He II 4686 emission, which is a signature of Wolf Rayet stars in this system (Crowther, 2007). In the nucleus the line emission is much weaker, and the high order Balmer lines are visible in absorption, characteristic of intermediate age (100 Myr to 1 Gyr) stellar populations.
In the case of NGC 6090, the NGC 6090 NE nucleus is labelled in the bottom left panel of Fig. 2 (and marked with a black dot in the bottom right panel of Fig. 1). To its west, there is the region of multiple star clusters (MSC). In NGC 6090 SW we considered the position of the peak of the IFS continuum to represent the nucleus (marked with a black dot in right panel of Figure 1). The spectra in the inner 2 of both nuclei are dominated by line emission, with the SW nucleus having a bluer continuum.
We also note that the brightest point source in the HST images (both optical and NIR) of NGC 6090 SW, is the knot located to the northwest of the progenitor, as previously reported by Dinshaw et al. (1999). These authors also found a mismatch between the radio emission peak and this knot, and suggested that it may not be the nucleus of the SW progenitor but a foreground star instead. The IFS has allowed us to confirm that this knot is not a foreground star, because the main absorption lines are at the redshift of the galaxy. That is the reason why we cannot discard it as the true nucleus of NGC 6090 SW, and its position will henceforth be shown in the maps with a star symbol. We discuss the minor effect this misidentification would have on our results in Appendix B.
3 Star cluster photometry
Using the HST images from FUV to NIR, we performed aperture photometry of the super star clusters in IC 1623 and NGC 6090. This was to make a rough estimation of their stellar population properties. Clusters were detected with IRAF’s DAOFIND task, and photometry performed using the phot task in APPHOT.
According to Alvensleben (2004), a long-wavelength baseline from U-band through to the NIR is necessary to determine the properties of star clusters in a reliable way, where the availability of U-band photometry is crucial. In fact, we find that the best colour-colour diagrams to break the age-extinction degeneracy are FUV-F435W vs. F435W-F814W and F435W-F814W vs. F814W-NIC160W. A total of 228 and 156 clusters were detected in IC 1623 and NGC 6090, respectively, with a signal-to-noise ratios SNR 5 in FUV, F435W, and F814W filters. Analogously, 225 and 117 clusters were detected in IC 1623 and NGC 6090 with a SNR 5 in F435W, F814W, and F160W filters.
All the clusters with positive detections it at least one of the diagrams are shown in Figure 3. In the case of IC 1623, the obscured population of clusters in IC 1623 E is shown in red, while that of IC 1623 W is shown in blue.
The star clusters properties (ages, masses, and the amount of dust obscuration) are estimated by comparing their colours and magnitudes with solar metallicity Charlot & Bruzual (2007, unpublished) simple stellar population (SSP) models. Model colours have been computed using STSDAS.SYNPHOT software.
Details of the methodology, together with the colour-colour, colour-magnitude diagrams are presented in the Appendix A. Here we summarize the main results.
IC 1623. Most IC 1623 W clusters are very young ( 200 Myr)
and affected by low to almost no foreground extinction ( 1 mag).
In contrast, the clusters in IC 1623 E are affected by 2–6 mag
For the only IC 1623 E cluster emitting in FUV, we derive an age below 10 Myr,
and A = 2–3 mag. This suggests that, although affected
by large amounts of dust extinction, some clusters in IC 1623 E are as
young as or younger than those in IC 1623 W.
Moreover, the clusters in IC 1623 W have masses between
10 and 10 M, while those in IC 1623 E
they are more massive
NGC 6090. Most of the clusters in NGC 6090 are also young ( 300 Myr) and affected by 1 mag extinction, but some of them could have A up to 2 mag. In general, NGC 6090 clusters have masses between 10 and 10 M. We find that the two “clusters” with higher masses (around 10–10 M) are NGC 6090 NE nucleus and NGC 6090 SW nucleus. The total stellar mass in clusters in NGC 6090 is M.
As noted in Section A.5, star cluster photometry allow us to trace the star formation which occurred in the last 500 Myr – 1 Gyr (at best, for no extinction), but not in previous epochs. Although spectral synthesis reveals that intermediate-age stellar populations are not important contributors in these galaxies (Fig. 8, Section 4.7), the star cluster detection limits are relevant to be known when studying more advanced merger stages, where a significant intermediate-age population could be present.
4 Stellar populations
In this section we use the stellar continuum shape to characterize the spatially resolved stellar population properties and constrain the star formation histories of IC 1623 and NGC 6090.
In order to quantify the effects of mergers on the properties of the stellar populations, the results of our LIRGs are compared to control samples of CALIFA spirals from González Delgado et al. (2015), which have been calibrated with version 1.5 of the reduction pipeline, explained in detail in the Data Release 2 article (García-Benito et al., 2015). The comparisons are made with Sbc/Sc spirals because the progenitor galaxies of LIRGs are gas-rich spirals without significant bulges (Sanders & Mirabel, 1996; Kim et al., 2013), and because they have stellar masses similar to IC 1623 and NGC 6090. We have defined two possible control samples. On the one hand, the average of all the Sbc galaxies in CALIFA, because they approximately cover the same mass range as our LIRGs, with M(M) = 10.70 0.34. On the other hand, using later Sc galaxies, but averaging only those in a mass range compatible with our LIRGs, M(M) = 10.45–10.85. There are 14 Sc galaxies in the CALIFA sample meeting this requirement, and 70 Sbc galaxies.
Our method to extract the stellar population information from the IFS data cubes is based on the full spectral synthesis approach. This technique has been extensively tested and applied (Cid Fernandes & González Delgado, 2010; González Delgado & Cid Fernandes, 2010; Cid Fernandes et al., 2013; Pérez et al., 2013; Cid Fernandes et al., 2014; González Delgado et al., 2014), and it has been proven to reduce the age-metallicity degeneracy (Sánchez-Blázquez et al., 2011). It is also notable that, despite the diversity of spectral synthesis methods, substantial changes in the results are more likely to come from revisions in the input data and from updates in the base models, the single most important ingredient in any spectral synthesis analysis (Cid Fernandes et al., 2013). The general consensus in the field is that uncertainties in the results for individual objects average out for large statistical samples (Panter et al., 2007). With IFS, each galaxy is a statistical sample per se, and even if the results for single spaxels (or zones) are uncertain, the overall trends should be robust.
We have analyzed the stellar population properties with the starlight code (Cid Fernandes et al., 2005), which fits an observed spectrum (O) in terms of a model (M) built by a non-parametric linear combination of N SSPs from a base spanning different ages () and metallicities (). Dust effects are modeled as a foreground screen with a Calzetti et al. (2000) reddening law with R = 4.5, assuming the same reddening for all the SSP components. Kinematical effects are also accounted for by assuming a Gaussian line-of-sight velocity distribution. The fits were carried out in the rest-frame 3650–6950 Å interval. Windows around the [OI], [OII], [OIII], [NII], HeI, and Balmer series from H to H emission lines were masked in all fits. Because of its interstellar absorption component, the NaI D doublet was also masked. The process is similar to that previously applied in CALIFA data (Cid Fernandes et al., 2013, 2014).
The results reported in this paper rely on the GM base (Cid Fernandes et al. 2013, CF2013 hereafter) of SSP spectra, that combines the granada models of González Delgado et al. (2005) for Myr with those of Vazdekis et al. (2010). They are based on the Salpeter IMF. It contains 156 SSPs, comprising four metallicities, , 0.4, 1, and 1.6 Z, and 39 ages between and yr. For comparison purposes, we have also performed the spectral synthesis using the CB model base, which is built from an update of the Bruzual & Charlot (2003) models, replacing STELIB by a combination of the MILES (Sánchez-Blázquez et al., 2006) and granada (Martins et al., 2005) spectral libraries (the same ones used in base GM). The IMF is that of Chabrier (2003) and it comprises 160 SSPs of four metallicities, , 0.4, 1, and 2.5 Z, and 40 ages between and yr.
We discarded from the analysis all the spaxels with SNR 5 in the continuum, measured in the rest-framed window between 5590 and 5680 Å. Also, the index
proposed by CF2013 has been used to quantify the quality of the spectral fits (see their Section 5.1). Again, O and M are the observed and model spectra, respectively, and the sum is carried over the N wavelengths actually used in the fit, i.e., discarding masked, flagged, and clipped pixels. = 10 is a reasonable quality-control limit. Typical median values of 3 and 5 are found in IC 1623 W and NGC 6090, respectively. From now on, only the spaxels with below the control limit are shown in the maps, and considered in the analysis.
4.2 Stellar mass
An important starlight output is the stellar mass, which is calculated taking into account the spatial variations of the mass to light ratio. The total stellar mass obtained from the sum of all spaxels is M = 3.910 M and 6.810 M for IC 1623 W and NGC 6090 respectively (for Salpeter IMF). This is the mass locked in stars today. Also counting the mass returned by stars to the interstellar medium M’ = 5.510M and 9.610M were involved in star formation for IC 1623 W and NGC 6090, respectively.
Since in LIRGs the optical light is dominated by young stellar populations and the extinction may be important in some regions, we could be underestimating the mass of the old stars. The old stellar populations emit the bulk of their light in the 1–3 m near-IR window, and assuming that they are the principal contributors to the IR light, we have determined an alternative estimate of the stellar mass using MIR data from Spitzer at 3.6 and 4.5 m, following the approach proposed by Jarrett et al. (2013). The global mass to light ratio at 3.6m () can be expressed as:
where [3.6] and [4.5] are the magnitudes in the 3.6 and 4.5 m bands. This expression was calculated under several assumptions that we have to take into account. On the one hand, it considers the SSPs of Bruzual & Charlot (2003) with a Chabrier IMF. On the other hand, it considers that the average mass-to-light ratio (at 3.6 m) for strictly old stars is log dex, as determined by Meidt et al. (2012, 2014), and valid in the range of metallicities covered by our models. With this procedure we found that the stellar mass derived from the MIR is 2.810M and 4.210M for IC 1623 W and NGC 6090, respectively (Chabrier IMF), in good agreement with our stellar masses derived from the spectral synthesis, if MIR masses are scaled by a factor 1.8 as expected due to the change in the IMF.
This agreement between the mass estimations derived from optical spectra and MIR imaging confirms that for IC 1623 W and NGC 6090 we are not underestimating the stellar masses with spectral synthesis, and also indicates that the extinction estimates are approximately correct.
4.3 Stellar mass surface density
In Figure 4 we show the stellar mass surface
density () maps
for IC 1623 W (upper left panel) and NGC 6090 (upper right).
The black dashed circles indicate the position of 1 half-light
The 2D maps of the stellar population properties were azimuthally averaged to study their radial variations using PyCASSO (CF2013). Radial apertures 0.1 HLR in width were used to extract the profiles. Expressing radial distances in units of HLR allows the profiles of the early-stage mergers to be compared on a common metric with the control samples of Sbc and Sc galaxies. The profiles are shown in the lower panel, in red for NGC 6090 NE, NGC 6090 SW in green, and IC 1623 W in blue. The uncertainties are shaded in red, green, and blue, respectively. They represent the standard error of the mean, calculated as the standard deviation divided by the square root of the number of points in each distance bin, with 0.2 HLR of width. As it is normalized by the square root of the number of points/spaxels in each HLR bin, the differences in the uncertainties between the inner and outer regions are visually not so strong, as the larger the radius, the larger is N. In terms of dispersion (=standard deviation) we find that with respect to IC 1623 W centre (N=8), the dispersion is a factor 2.3 larger at 0.5 HLR (N=41), 2.4 larger at 1 HLR (N=47), and 1.5 larger at 2 HLR (N=18). In the case of NGC 6090 NE, the dispersion is a factor of 2.3 larger at 0.5 HLR (N=42) than at the centre (N=8), and for NGC 6090 SW is a factor 2.5 larger at 0.5 HLR (N=74) than at the centre (N=12). The black lines are the profiles from Sbc (solid) and Sc (dashed) spiral galaxies from CALIFA (González Delgado et al. 2015, GD2015 hereafter).
The highest values of the mass surface density are reached in IC 1623 W nucleus and in the cluster concentration C1, with a logarithmic value of around 3.7 for in units of pc. From the radial profile we find a negative gradient with distance, going from log = 3.7 at the nucleus to 2.8 at 1.2 HLR, then increasing again up to log = 3.0 at 2 HLR.
In NGC 6090 the stellar mass surface density in the NE progenitor is significantly higher than in the SW. The region with the highest value, log 3.8, is located to the south east of the NGC 6090 NE nucleus. In NGC 6090 SW, log is higher towards the north. From the radial profile, we also found a negative gradient of the stellar mass density with distance: log goes from 3.8 to 3.1 at 0.9 HLR for NGC 6090 NE, and from 3.0 to 2.8 at 0.6 HLR, for NGC 6090 SW.
The central density in IC 1623 W and NGC 6090 NE is similar to the central density of Sbc galaxies, while NGC 6090 SW is more similar to Sc. We have compared the inner gradients of the stellar mass surface density (from 0 to 1 HLR, as defined in equation 6 of GD2015) in early-stage mergers and in the control spirals. We find log = , , , for NGC 6090 NE (in 0.9 HLR), IC 1623 W, Sc, and Sbc, respectively. In the case of IC 1623 W we can also compare the outer gradient (from 1 to 2 HLR, defined as in equation 7 of GD2015), log = 0, , for IC 1623 W, Sc, and Sbc galaxies, respectively. We find that the early-stage merger radial profiles of are flatter than in spirals, but not so significantly if we consider the dispersion of the control samples, between 0.24 and 0.28 dex (GD2015). Moreover, we note that given the recent star formation in IC 1623 W, its HLR is half the HLR of the control samples, and the apparent flattening of the mass density profile may be also due to this, instead of representing a real evolutionary change, at least in the inner 1 HLR. This does not affect NGC 6090 NE SW, as their HLRs are similar to Sbc/Sc galaxies.
4.4 Stellar dust extinction
starlight-based A maps are shown in Figure 5. The scale is the same for both systems to facilitate the comparison.
IC 1623 W (upper left) has little extinction, with A below 0.4 mag nearly everywhere. From the radial profile, we found a slight but nonuniform increase of the average extinction from 0.06 mag in the nucleus to 0.2 mag at 1.5 HLR. The regions most affected by extinction are located near areas where HII regions accumulate, such as C1, where the extinction reaches values up to 1.2 mag. As an explanation we suggest that the enhancement of A could be due to the accumulation of dust swept by the winds and supernovae of the star forming regions, at least in some locations. Another possibility is that the interaction itself has led to the accumulation of dust in these regions.
In NGC 6090 NE (upper right), A ranges from 1.2 to 1.6 mag, while in NGC 6090 SW it runs from from 0.6 mag in the North to about 0.2 mag in the South. To the West of NGC 6090 NE and in the bridge between the two galaxies the dust extinction has an intermediate value of 0.8–0.9 mag. There is approximately 1 mag difference in the stellar dust extinction between the two progenitors. When averaging (lower panel), we do not find significant gradients of the dust distribution across the main bodies of the individual galaxies, reflected in the nearly flat radial profiles. We note that in these two early-stage merger systems one of the progenitors is significantly more obscured than the other. In these cases it happens that the eastern progenitor is more obscured, although the difference is higher for IC 1623 than for NGC 6090.
The radial profiles of A in IC 1623 and NGC 6090, are flat. This is why the profiles shown in the previous section resemble the surface brightness profiles (not shown). In contrast, typical spiral galaxies show a negative radial gradient independently of the morphology (GD2015).
The simplest way to quantify the star formation history (SFH) of a system is to compress the age distribution encoded in the light population vectors to their first moments. For this purpose we will use the following definition for the mean stellar age:
where is the fraction of light at the normalization wavelength (5635 Å) due to the base population with age and metallicity .
The mean stellar age maps are shown in Figure 6. The scale is the same for both maps to facilitate the comparison and goes from = 6.0 to 9.0.
In IC 1623 W the nuclear regions are clearly older than the outer parts. From the radial profile we find a negative trend of the mean age with distance: from 300 Myr in the nucleus to 30 Myr 1.2 HLR away. From 1.2 to 2 HLR the age remains approximately constant at 15–40 Myr, indicating the IC 1623 W disc is still forming stars.
In NGC 6090 the older spaxels are located in the outskirts for both progenitors, but especially for the SW galaxy, where they reach ages up to 300 Myr. The youngest spaxels ( 50 Myr) are located in the bridge. When averaging we found that both progenitors have the same mean ages, –100 Myr.
In both early-stage merger systems the age profiles are significantly flatter than in Sb to Sc galaxies (GD2015). Even for IC 1623 W, where we find a negative gradient of -280 Myr in the inner 1 HLR, it is still much lower than the -2.4 Gyr(-620 Myr) gradient in the inner 1 HLR of Sbc(Sc) galaxies.
The mean metallicity has been obtained in a similar way to the mean light weighted ages, but weighing by the mass:
where is the mass fraction in the base population with age and metallicity . This is the same definition adopted in GD2015. The mass-weigthed metallicity maps are shown in Figure 7. The scale range in both maps is the same, and spans the full range of the base (from to in log-solar units).
As can be seen in Figure 7, the scatter is large, maybe due to the weakness of the metallic absorption lines in the spectra. Moreover, our results could be affected by the age-metallicity degeneracy, although the spectral synthesis has been proven to be better than the index method to break it (Sánchez-Blázquez et al., 2011).
Focusing on the radial averages, it seems that IC 1623 W, NGC 6090 NE, and SW, have global mean metallicities around 0.6 Z (but ranging from 0.4 to 0.8 Z). We find that beyond 0.7 HLR in IC 1623 W the metallicities are similar to those of Sbc galaxies, but in the central regions (in the inner 0.5 HLR) the metallicities are smaller, more similar to Sc galaxies. Something similar happens in NGC 6090 NE, where a decrease in the metallicity below 0.4 HLR is found. The metallicity radial profiles in the inner 1 HLR of these three galaxies have flat or even slightly positive gradients, contrary to what happens in spirals earlier than Sc, which show a negative gradient (GD2015). However, we cannot claim a dilution or interpret this further, given the large dispersion of up to 0.25 dex in the metallicity of the control samples.
4.7 Contribution of young, intermediate and old populations
Population synthesis studies in the past found that a useful way to summarize the SFH is to condense the age distribution encoded in the population vector into age ranges. This strategy originated in methods based on fitting equivalent spectral indices (Bica et al., 1988; Cid Fernandes et al., 2003), but it was also applied to full spectral fits (González Delgado et al., 2004). We follow this strategy by dividing into: young stellar populations (YSP, Gyr), intermediate age stellar populations (ISP, Gyr), and old stellar populations (OSP, Gyr). We select these three age bins to correspond to the lifetimes of populations with distinctly different optical line and continuum features.
Figure 8 presents the maps (left) and radial profiles (right) of the light fractions due to YSP, ISP, and OSP (x, x, and x). IC 1623 W is shown in the top panels, and NGC 6090 in the bottom panels. We find that both early-stage merger systems are dominated by young stellar populations over their full observed extents.
In IC 1623 W we find that the contribution of the YSP to light is around 90 everywhere, except in the centre, where it is around 50. In contrast, in Sbc galaxies the YSP contribution is much lower, 10 in the centre, and 20 in the disc. In IC 1623 W the ISP has a significant contribution only in the inner 0.3 HLR, with contributions of 30–40, higher but still similar to Sbc galaxies, . The contribution of OSP is everywhere below in IC 1623 W, while in Sbc galaxies it ranges from in the nucleus to at 2 HLR.
In NGC 6090 we find that almost all spaxels have YSP contribution above (much higher than the 10– in Sbc galaxies), except for some spaxels located at the easternmost end of NGC 6090 NE and at the south edge of NGC 6090 SW, where the contribution of ISP is also significant. However, the average contribution of ISP is small, below 10 for both progenitors, smaller than the 20–40 typical of Sbc galaxies. The OSP contributes around of the light of both progenitors over their full extents, much smaller than its contribution in Sbc galaxies.
Analogously, in Fig. 9 we present the maps (left) and radial profiles (right) of the mass contribution by percentage of the YSP, ISP, and OSP (m, m, and m). For both IC 1623 W (top panels) and NGC 6090 (bottom) most of the mass is in old stellar populations.
In IC 1623 W this is true everywhere, except for some spaxels in the outer regions which are completely dominated by young components. The radial profiles of the OSP and YSP mass contributions evolve inversely to each other. In a non-uniform way, the OSP contribution decreases from in the nucleus to at 2 HLR. Conversely, the mass in YSP grows from in the nucleus to at 2 HLR. In terms of the mass in ISP, they range from in the nucleus to almost 0 beyond HLR.
The same happens with NGC 6090, with the OSP dominating the mass everywhere except in the easternmost end of NGC 6090 NE, in some spaxels to the north west of NGC 6090 NE nucleus, and to the south of NGC 6090 SW, where the contributions of ISPs are also large (50–, and up to in some spaxels). From the radial profile, we see that for both progenitors the OSP is responsible for more than of the mass, and this is true over the full observed extent of the system. The fractions of stellar mass in YSP and ISP are comparatively low.
We have also computed the total masses in YSP, ISP, and OSP. From the total stellar mass in IC 1623 W ( M), the mass in young, intermediate age, and old populations are (), (), and M (), respectively. In NGC 6090 ( M), the masses in YSP, ISP, and OSP are (), (), and M (), respectively.
In addition, we have compared the mass in clusters, as derived in Section 3, with the mass in YSP derived using the model base CB, which was also used in the photometric study. On the one hand, we found that the mass in star clusters represents only 3 and 4 of the total stellar mass of IC 1623 W and NGC 6090, respectively. On the other hand, in IC 1623 W, the mass in clusters and the mass in YSP from the spectroscopy are comparable to each other within a factor of 1.3, in agreement with the idea that the vast majority of stars forms in clusters rather than in isolation (Chandar et al., 2015). However, in NGC 6090, the mass in star clusters is a factor 4 higher than the mass in SSPs with ages 300 Myr from the spectroscopy. This is within the error expected from the uncertainties in the exact ages of the clusters. From the photometry we found an upper limit for the ages of 300 Myr, while from the spectroscopy we found that the young components are in fact younger than 30 Myr. The ratio of the mass-to-light ratios for these populations at Z is approximately 3.6. Therefore, the age uncertainty in the photometric results could lead to mass uncertainties of up to a factor .
5 Ionized gas emission
The starlight best-fittings were subtracted from the observed spectra to obtain 3D cubes with the pure nebular emission line spectra. These emission spectra contain information about the ionized gas distribution and properties. Using the ratio of H to H intensities, we have determined the ionized gas dust attenuation (Section 5.3), which is necessary to compute the total deredened H luminosity. The H luminosity is a commonly used SFR tracer, but can be affected by shock emission. Following Rich et al. (2011), we have analyze the emission line ratios (Section 5.4) and velocity dispersions (Section 5.5) to confirm which spatial regions are indeed dominated by star formation.
We have measured the intensities of some of the most prominent emission lines – [OII]3727, H, [OIII]5007, [OI]6300, H, [NII]6584, and [SII]6716,6731 – when present in the spectra.
First, a local continuum was estimated using narrow windows located on either side of each line. This local continuum was subtracted to account for small mismatches between the starlight fit and the actual galaxy continuum. The emission line intensities in each spectrum were then obtained by fitting Gaussians. Nearby, blended lines (e.g. [NII] and H) were fitted simultaneously using multiple Gaussian profiles. In this way, emission-line intensity maps were created for each individual line.
The statistical uncertainty in the measurement of the line fluxes was calculated from the product of the standard deviation of the adjacent continuum, , and the FWHM of the emission line. In order to avoid spurious weak line measurements, we have only considered spaxels where the lines have SNR 3.
5.2 Ionized gas morphology
IC 1623 W. Figure 10 shows the observed H (left panel) emission flux for IC 1623 W on a logarithmic scale to enhance the contrast. As this maps shows, the ionized gas extends across the whole FoV. The brightest regions are those that correspond to aggregates of SSCs and star forming regions along the distorted spiral arm. To get a better idea of the overall morphology of ionized gas in IC 1623 W, in the right panel of Figure 10 we have superimposed as red contours the LArr H emission over the HST F435W continuum image (greyscale). We can see that H emission is concentrated in the distorted spiral arm and SSCs aggregates (which we have labelled C1 and C2 in the image). Furthermore, in the nuclear region we detect an elongated structure, C3, which appears to leave or enter the nucleus. Due to its morphology, at first we thought it might be a gas inflow/outflow. Yet we see in the HST image that this emission also a stellar counterpart with several SSCs along it. Moreover, we will see in Section 5.4 that the gas ionization is due to star formation in C3.
NGC 6090. Analogously, in Figure 11 we show the observed H (left panel) map for NGC 6090. The ionized gas distribution is extended across both progenitors in a more uniform and less clumpy way than in IC 1623 W. The NE progenitor is more luminous than the SW one. In the right panel of Figure 11 we show the HST F435W continuum image of NGC 6090 with the H emission superimposed as red contours. In the NE progenitor, the H peak is offset to the west of the nucleus, spatially coincident with the MSC region. For the SW progenitor, H peaks in the northern half where most of the star forming regions accumulate.
5.3 Ionized gas dust attenuation
The measured ratio of H to H intensities provides an estimate of the nebular dust attenuation, , along the line of sight. We assume an intrinsic ratio of I(H)/I(H)=2.86, valid for case B recombination with T = 10000 K and electron density 10 cm (Osterbrock & Ferland, 2006), which are typical of star forming regions. We have used the Calzetti et al. (2000) reddening function, as it has been shown to work better for starburst galaxies.
In Figure 12 we show the maps for IC 1623 W (top) and NGC 6090 (bottom).
In IC 1623 W the mean nebular extinction is 0.8 mag. Following the trend found for the stellar extinction, the regions most affected by extinction are located near areas where HII regions accumulate. Near C1, reaches its maximum value of around 2.5 mag. The same happens around C2, with values up to 1.6 mag. Also, values between 1.3 and 1.6 mag are found in the distorted spiral arm, and between 1.0 and 1.3 mag to the south-east of the nucleus and along the elongated structure C3. Everywhere else, the extinction is below 1 mag.
In NGC 6090, the average extinction of the NE progenitor is 1.8 mag. ranges from 1.5 mag in the nucleus of the galaxy, to values around 2 mag in the north, and up to 2.5 mag in a few spaxels in the south. For the SW progenitor we found an average value around 0.8 mag, but it also varies across the galaxy from 0.5 mag in the south to 1 mag in the north. We found an average difference of 1 mag between the two progenitors of the system, as also happens with the stellar extinction (see upper right panel of Figure 5). In the overlapping bridge region –1.4 mag.
There is a good spatial correlation between the extended H emission and the regions where the YSP are contributing with 70 to the total light. As we will see in the next Section, the ionization of the warm gas in these early-stage merger LIRGs is indeed due to star formation, as this correlation suggests.
Moreover, we have compared the ratio between the ionized gas extinction to the stellar extinction, using only the spaxels where we are able to determine the ionized gas extinction, that is, where the SNR of both H and H are above 3. More than 95 of the spaxels have values of the ratio between 0 and 7. Values above 7 are considered unreliable, as they correspond to spaxels with below the uncertainty with which we can determine it (of 0.1 mag, Cid Fernandes et al. 2013). For IC 1623 we found (median and standard deviation), while for NGC 6090 . Analysing UV and optical spectra of 39 starburst galaxies, Calzetti et al. (1994) found = 2.0 0.4, which is compatible with the results from our early-stage merger LIRGs. The fact that in starbursts and interacting galaxies is larger than is attributed to the fact that emission line sources are preferentially located within dusty birth clouds, whereas the stellar light mainly samples the diffuse dust in the ISM. Kreckel et al. (2013) also find a similar result analyzing IFS data from 8 nearby galaxies, but noting a more complex behaviour of the nebular to stellar extinction ratio, which is sensitive to the flux.
5.4 Ionisation mechanisms: Star-formation and shocks
Diagnostic diagrams comparing emission line flux ratios are the most widely used method for identifying the nature of the dominant ionisation mechanisms in galaxies. Fig. 13 presents two such diagrams: [NII]6584/H vs. [OIII]5007/H (left panel, henceforth the “BPT diagram”, after Baldwin et al. 1981), and [OI]6300/H vs. [OIII]5007/H (middle panel, following Veilleux & Osterbrock 1987) for both IC 1623 (top) and NGC 6090 (bottom). Panels on the right show the H velocity dispersion () maps.
Solid lines in both diagnostic diagrams mark the “maximum starburst lines” obtained by Kewley et al. (2001) by means of photoionization models of synthetic starbursts. These lines are often interpreted as delimiting the region occupied by “pure AGN”, even though this is not what they were designed for (see discussion on Cid Fernandes et al. 2011). The dotted line in the left panels marks the frontier semi-empirically drawn by Kauffmann et al. (2003) to separate SDSS star-forming galaxies from those hosting an AGN. Systems between these two lines are often called “composite”, a nomenclature that betrays the traditional interpretation of such systems as a mixture of star-forming and AGN.
None of the spaxels in either IC 1623 or NGC 6090 lies above the K01 line in the BPT diagram (left panels) and only a few do so in the [OI]/H one (middle), indicating that AGN, even if present, contribute little to the emission lines. In fact, most of the points lie below the K03 line, confirming that photoionization by young stars is the main mechanism of line production in these sources. Black stars in these panels mark the line ratios of the spatially integrated spectra, and their position in the diagnostic diagrams further confirms the dominance of star formation as the main ionization source.
Interestingly, however, in both cases a significant number of spaxels occupy the intermediate zone between K03 and K01 in BPT-coordinates. Moreover, the colour-coding makes it evident to the eye that the points which depart the most from the star forming sequence are those with high (yellow-red colours). In IC 1623 these large gas velocity dispersion regions are mostly located towards the NW of the map (top right panel), while in NGC 6090 (bottom right) they are concentrated in the South of NGC 6090 SW.
These are signs of tidally induced shocks, associated with the interaction process, contributing to the ionization of the gas (Colina et al., 2005). Extended shock ionisation has been previously reported in local U/LIRGs, taking advantage of IFS studies (Monreal-Ibero et al., 2006; Rich et al., 2011, 2014). In all these cases, shock ionisation exhibits characteristics of extended LINER-like emission with broadened line profiles. In the case of our IFS data for IC 1623 W and NGC 6090, we find broadened line profiles falling in the so-called “composite” region. The “composite” systems are conventionally associated with a superposition of normal HII regions and the narrow line region of AGN. However, the location of these shocked regions in BPT coordinates, together with the spatial coherence of the line-ratios and kinematical data revealed in our IFS, allow us to circumvent these limitations and correctly identify the gas ionization mechanism in our early-stage merger systems.
It is further useful to note that the misidentification of shocked regions with a mixture of star-formation and AGN ionization is not the only caveat in a naive reading of our diagnostic diagrams. In the case of NGC 6090 some low spaxels populate the zone between the K03 and K01 demarkation lines (blue points in the bottom left panel of Fig. 13). Their off-nuclear locations and closeness to the K03 line argue in favour of them being pure HII regions, and not “star-forming AGN composite” systems, which their nominal locations in the BPT diagram would indicate. This caveat has been noted before by Sánchez et al. (2015), who find that some of the bona fide HII regions in the CALIFA sample lie in the (ill-named) “composite” region.
5.5 Ionized gas kinematics: velocity dispersion
We trace the gas kinematics of NGC 6090 and IC 1623 W using the results of Gaussian fits to their emission lines. To obtain the final velocity dispersion, we have subtracted the instrumental resolution () in quadrature, where at H is 140 km s.
In the right panel of Figure 13 we show the ionized gas velocity dispersion maps for IC 1623 W (top) and NGC 6090 (bottom).
In IC 1623 W the velocity dispersion is low in the star forming regions and in the distorted spiral arms, with 120 km s. However, 120 km s in several outermost spaxels, with a region to the NW of the nucleus where reaches up to 170 km s.
In NGC 6090 there exists a gradual increase in the velocity dispersion from –150 km s to the east side of NE progenitor, to 170 km s in the bridge between progenitors, and up to 190 km s to the SW of the SW progenitor.
We interpret the high velocity dispersion regions in these early-stage mergers as a consequence of shocks producing a broadening of the emission lines. Moreover, in Section 5.4 we saw that there exists a correspondence between these regions and “composite” excitation conditions in the BPT diagram. IFS data for IC 1623 were previously analyzed by Rich et al. (2011), who found widespread shocks (LINER-like ionization) in the outskirts of the system, beyond the region traced by our data. In the region in common (the core of IC 1623 W) our results are in total agreement with them, finding similar values of the lines ratios [NII]/H 0.9 – , [OI]/H 1.7 – , consistent with star-formation ionization. While Rich et al. (2011) are able to further decouple the contribution arising from shocks and star formation, we cannot follow a similar approach due to our lower spectral resolution (140 km s vs. 40 km s in H).
5.6 Gas metallicities
We estimate the nebular metallicity of the star forming regions using the O3N2 index, as calibrated by Marino et al. (2013). To exclude shocked regions from this analysis in an approximate way we remove spaxels with km s in IC 1623 W and km s in NGC 6090, corresponding to the points marked in yellow–red in Fig 13.
The resulting values of 12 + log (O/H) range from 8.20 to 8.40 in IC 1623 W, and from 8.40 to 8.52 in NGC 6090. For a solar value of 12 + log(O/H) = 8.69 (Asplund et al., 2009) these values correspond to and 0.6 solar (on average) for IC 1623 W and NGC 6090, respectively.
Also, in Figure 14 we compare the nebular oxygen metallicity in red, with the oxygen abundance of the Milky Way (MW) in blue, obtained using the calibration by Afflerbach et al. (1997), and with the stellar metallicities in black. The early-stage mergers have lower central oxygen metallicities in comparison with the MW, and flatter profiles. The nebular and stellar metallicities are similar, both in IC 1623 W and NGC 6090, but the stellar metallicities show slightly more positive gradients, while the nebular profiles are flat. However, a larger sample and more spatial coverage would be necessary to assess these results in a statistically significant way. In Section 6.4 our results are discussed in the context of the literature.
In this section we discuss our main findings in the context of the merger-driven evolution of galaxies. The observational characterization of star formation in mergers at different stages is necessary to test the validity and put contrains on merger simulations. Merging LIRGs, such as the ones presented in this paper, are unique laboratories to study this in detail. First, we calculate and compare the current Star Formation Rate (SFR) of IC 1623 W and NGC 6090, using the spectral synthesis, H, and MIR data.
6.1 Multiwavelength star formation rates
We explore two ways to estimate SFRs from our IFS data. First, we compute the SFR from the dereddened H luminosity considering that shocks are not important, that is, assuming that the gas is photoionized by massive stars only. We use the conversion to star formation rate provided by E. Lacerda (private communication), calculated for the average metallicity of our galaxies 0.7 Z and for Bruzual & Charlot (2003) models. The calibration with a Salpeter IMF is very similar to the standard calibration of Kennicutt (1998), calculated for solar metallicity models.
Secondly, we use our stellar continuum decomposition to estimate the SFR by summing all the stellar mass formed since a lookback time and performing a mass-over-time average, following equation 11 in Cid Fernandes et al. (2013), that gives the mean SFR since . One can tune to reach different depths in the past. For comparison purposes we estimate it for Myr (González Delgado et al., 2016; Asari et al., 2007), which is of the same order of the lifetime of the ionizing populations also emitting in H, that is O and early B stars.
Additionally, we have used a hybrid calibration, including the Spitzer MIPS 24 m flux to check whether there is any star formation completely obscured from view in the optical, which is not accounted for by the Balmer decrement attenuation correction. We have followed the calibration given by Kennicutt et al. (2009).
The three different approaches lead to very similar SFR results, with an average and dispersion of 23 4 M yr for IC 1623 W, and 32 4 M yr for NGC 6090, for Salpeter IMF, and a factor 1.6–1.7 lower with Chabrier IMF models. The consistency between the hybrid calibration including MIR and the optical ones, indicates that, for IC 1623 W and NGC 6090, there is no substantial star formation which is not accounted for by the optical data.
6.2 Merger-induced star formation in mergers
Global enhancement of star formation
In both IC 1623 W and NGC 6090 we find that the SFR is increased with respect to the so called main sequence star forming galaxies (MSSF, Renzini & Peng 2015). Using the González Delgado et al. (2016) calibration for the MSSF in the same mass range as our early-stage mergers ( Sbc), we obtain values of 3.0–4.4 M yr, a factor 6–9 lower than the SFRs of our two early-stage mergers. This increase is larger than the predictions from classical merger simulations by Di Matteo et al. (2008), who found that boosts in the SFR greater than 5 are rare, but still possible in about 15 of major galaxy interactions and mergers. We suggest that U/LIRG population could belong to this 15 in the sense that star formation in them is “caught in the act”.
As the SFR depends on the total amount of mass,
we have also calculated the average specific star formation
rate, sSFR = SFR / M, which is sSFR 0.60 Gyr
for IC 1623 W, and 0.47 Gyr for NGC 6090,
these are significantly larger,
a factor 5–6, than the sSFR of the Sbc/Sc control galaxies
measured at 1 HLR ( 0.1 Gyr).
From the IFS and star cluster photometry we find that merger-induced star formation in IC 1623 is both widespread (out to 1.5 HLR), and recent, with an average light weighted age of 50 Myr in IC 1623 W. The centre of IC 1623 W (0.5 HLR) is older, 300 Myr, than the surrounding area, 30 Myr (Fig. 6), due to the presence of a significant contribution of intermediate-age stellar populations (Fig. 8) probably induced in a previous stage of the merger. However, the light is everywhere dominated by young stellar populations, with contributions in the range 50–90% across IC 1623 W compared with 10–20% in the Sbc/Sc control galaxies. The extinction in this progenitor is low 0.2 mag, and has a flat radial profile in comparison with Sbc/Sc control spirals (Fig. 5). However, we note that IC 1623 E is affected by much higher levels of extinction (up to 2–6 mag) according to the star cluster photometry (Fig. 3 and A2). IC 1623 W has a mean stellar metallicity of 0.6 Z, with a flat or slightly positive radial profile (Fig. 7). The nebular metallicity profile is also flatter than in non-interacting control spiral galaxies and consistent with the existence of gas inflows of metal poor gas to the centre of the system, as explained below in Section 6.4.
As for IC 1623, the merger-induced star formation in NGC 6090 is widespread, as indicated by the star clusters detected in the HST images (Fig. 3), and recent, with an average light weighted age from the IFS of 80 Myr, with both progenitors, NGC 6090 NE and NGC 6090 SW having similar ages (Fig. 6). As in IC 1623 W, the age profiles are significantly flatter than in Sbc/Sc galaxies, in agreement with a general rejuvenation of the progenitor galaxies due to merger-induced star formation. The light is dominated by young stellar populations in both progenitors, with contributions 70 (Fig. 8). There exist a difference of 1 mag between the stellar extinctions of the two progenitors, with an average of 1.3 mag in NGC 6090 NE, and 0.5 mag in NGC 6090 SW (Fig. 5), and whose radial profiles are flat in comparison with Sbc/Sc control spirals. As in IC 1623 W, NGC 6090 has a mean stellar metallicity of 0.6 Z, with a flat or slightly positive radial profile (Fig. 7), and a flat nebular metallicity profile, also in agreement with the gas inflow scenario.
Based on our results we propose an evolutionary scheme in which the gas components of the progenitors react promptly to the merger process, probably during the first contact (stage II in Veilleux et al. 2002 classification), where the discs overlap but strong bars and tidal tails have not yet formed. During this stage the gas starts inflowing to the dynamical centre of the system, and can trigger some star formation. The intermediate age stellar population in IC 1623 W nucleus (300 Myr) could be a consequence of this, and we note that the star formation triggered during the first contact stage is still not extended, but concentrated in the centre. The early-stage merger (III in Veilleux et al. 2002) starts after the first pericenter passage (in our systems, in the last 50–100 Myr), where most of the extended star formation is due to the fragmentation of gas clouds produced by the increase of the supersonic turbulence of the ISM, as predicted from high resolution models in the literature (Teyssier et al., 2010; Hopkins et al., 2013; Renaud et al., 2015). Therefore, at this stage, the star formation is mainly in the form of star clusters. We found that the mass in the star clusters is similar to that of the 140 Myr age stellar populations derived from the IFS data (within factors of 1.3 and 4 for IC 1623 W and NGC 6090, respectively.) Spatially extended star formation has also been previously observed in other early-stage merger systems (Wang et al., 2004; Elmegreen et al., 2006; Hattori et al., 2004; García-Marín et al., 2009; Arribas et al., 2012). The central nebular metallicities are already diluted at the early merger stage, at least in some cases.
6.3 Ionization mechanism
In both IC 1623 W and NGC 6090 the dominant ionization mechanism is photoionization by young massive stars. For a few spaxels we cannot discard composite photoionization by hot stars and shock ionisation, given also the high velocity dispersion in the north west of IC 1623 W and to the south of NGC 6090 SW. This result is in agreement with previous results from the literature (Rich et al., 2011, 2014).
6.4 Nebular metallicity gradients
Finally, in the literature there exists some controversy about the evolution of the gas metallicity in mergers, with some studies finding lower central metallicity in mergers in comparison to control galaxies, and a flattening of the metallicity profiles, possibly related with gas inflows (Kewley et al., 2010; Rich et al., 2012; Guo et al., 2016), and others finding a similar central metallicity in mergers and control samples, with a decrease compared to the control sample objects at the large radii. This suggests that other processes such as stellar feedback can contribute to the metal enrichment in interacting galaxies (Barrera-Ballesteros et al., 2015).
Nebular metallicities estimated with the O3N2 method are and solar in IC 1623 W and NGC 6090, respectively. In Section 5.6 we showed the oxygen metallicity profiles of our early-stage mergers. In particular, for IC 1623 W we found a gradient of -0.020 0.013 dex/kpc, which is in agreement with the gradient determined by Rich et al. (2012) using Kewley & Dopita (2002) calibration (-0.016 dex/kpc). Rich et al. (2012) also compute a set of merger simulations to trace the oxygen metallicity variation. For close pairs (stage 2) they found a metallicity gradient of -0.025 0.015 dex/kpc, consistent with that found by us in IC 1623 W. In NGC 6090 we find even flatter profiles of -0.012 0.013 dex/kpc in the NE progenitor, and 0.000 0.014 dex/kpc in the SW progenitor, but still within the observational range of close pairs in the Rich et al. (2012) study. All of them are flatter than the MW gradient (-0.060 0.035 dex/kpc), and the control sample of isolated star forming spirals of Rupke et al. (2010) (-0.041 0.009 dex/kpc), in agreement with the gas inflow scenario as the main cause of the dilution.
7 Summary and conclusions
Though Integral Field Spectroscopy PMAS LArr data and multiwavelength HST imaging, and with the starlight full spectral fitting code, in this paper we characterize the star cluster properties, stellar populations, ionized gas properties (ionization mechanism and kinematics), and star formation rates in the early-stage merger LIRGs IC 1623 W and NGC 6090. Early-stage merger results have been compared with control Sbc and Sc galaxies from CALIFA (GD2015) in the same mass range. Our main results are:
From the IFS and star cluster photometry we find that merger-induced star formation in these two LIRGs is widespread (out to 1.5 HLR) and recent, as revealed by the young light-weighted ages (50–80 Myr vs. 1 Gyr in Sbc/Sc control galaxies) and the high contributions to light of young stellar populations (50–90 vs. 10–20 in Sbc/Sc galaxies). These results are in agreement with high resolution merger simulations in the literature.
The A profile in the inner 1 HLR of these two early-stage mergers is flat in comparison with the negative gradient in Sbc/Sc control spirals.
The age profiles are significantly flatter than in Sbc/Sc galaxies, in agreement with a general rejuvenation of the progenitor galaxies due to merger-induced star formation.
The average stellar mass weighted metallicities at 1 HLR are 0.6 Z, similar to Sbc galaxies. Although with a large scatter, the stellar metallicity profiles are comparatively flat or slightly positive as compared to Sbc/Sc galaxies.
Nebular abundances estimated with the O3N2 method are and solar in IC 1623 W and NGC 6090, respectively.
The nebular abundance profiles are flatter (-0.020 0.013 dex/kpc gradient in IC 1623 W, -0.012 0.013 dex/kpc in NGC 6090 NE, and 0.000 0.014 dex/kpc in NGC 6090 SW) than the ones in the Milky Way and in a sample of isolated star forming spirals, in agreement with the scenario where gas inflows are responsible of the dilution.
From the [NII]6583/H vs. [OIII]5007/H, and [OI]6300/H vs. [OIII]5007/H diagnostic diagrams we find that both IC 1623 W and NGC 6090 are dominated by star formation in the region mapped in this work.
In some regions we can not discard a mix of photoionization by hot stars and by shocks, given the high velocity dispersion to the south of NGC 6090 SW, and to the north west of IC 1623 W, consistent with previous results from the literature.
We obtained several estimations of the current SFR using the spectral synthesis (SFR(30 Myr)), the dereddened H luminosity (SFR(H)), and a hybrid calibration using the observed H luminosity and MIR 24 luminosity. All of them lead to very similar values of the SFR, with an average of 23 4 M yr for IC 1623 W, and 32 4 M yr for NGC 6090. From the good consistency between the different estimations we conclude that for IC 1623 W and NGC 6090 there is no substantial star formation which is not accounted for by the optical data.
The average SFR in our early-stage mergers is enhanced by factors of 6 to 9 with respect to main-sequence Sbc star forming galaxies. This is slightly above the predictions from classical merger simulations, which indicate that SFR enhancement factors greater than 5 are rare, but still possible in about 15 of major galaxy interactions and mergers. We suggest that U/LIRGs could belong to this 15 in the sense that star formation in them is caught in the act.
The sSFR in IC 1623 W ( 0.60 Gyr) and NGC 6090 ( 0.47 Gyr) is also enhanced by factors of 5 to 6 with respect to Sbc/Sc galaxies ( 0.1 Gyr).
Interpreting the sSFR as the inverse of a time scale for the SF, our results indicate that, at their present SFR, early-stage mergers build their stellar mass much faster ( 2 Gyr) than Sbc/Sc control galaxies ( 10 Gyr).
Our results are consistent with an evolutionary scheme where some star formation can be initially triggered in the central regions during the first contact stage, followed by extended star formation after the first pericenter passage (early stage mergers).
CALIFA is the first legacy survey carried out at Calar Alto. The CALIFA collaboration would like to thank the IAA-CSIC and MPIA-MPG as major partners of the observatory, and CAHA itself, for the unique access to telescope time and support in manpower and infrastructures. We also thank the CAHA staff for the dedication to this project. Support from the Spanish Ministerio de Economía y Competitividad, through projects AYA2014-57490-P, AYA2010-15081, and Junta de Andalucía FQ1580. ALdA, EADL and RCF thanks the hospitality of the IAA and the support of CAPES and CNPq. RGD acknowledges the support of CNPq (Brazil) through Programa Ciência sem Fronteiras (401452/2012-3).
Appendix A Star cluster photometry
In Section 3 we have summarized the stellar population properties of the super star clusters (SSCs) in IC 1623 W and NGC 6090 using the HST images from FUV to NIR. Their characteristics are summarized in Table A.1 and A.2, respectively.
|Detector/||Plate Scale||Observation||t||Proposal ID,|
|F25SRF2||STIS||FUV-MAMA||0.025||2000-12-25||2371.0||8201, G. Meurer|
|F25QTZ||STIS||NUV-MAMA||0.025||2000-12-25||750.0||8201, G. Meurer|
|F435W||ACS||WFC||0.05||2006-07-12||1260.0||10592, A. Evans|
|F814W||ACS||WFC||0.05||2006-07-12||720.0||10592, A. Evans|
|F110W||NICMOS||NIC2||0.075||1998-08-03||223.8||7219, N. Scoville|
|F160W||NICMOS||NIC2||0.075||1998-08-03||223.8||7219, N. Scoville|
|Detector/||Plate Scale||Observation||t||Proposal ID,|
|F140LP||ACS||SBC||0.025||2008-08-09||2648.0||11110, S. McCandliss|
|F435W||ACS||WFC||0.05||2005-09-18||1380.0||10592, A. Evans|
|F814W||ACS||WFC||0.05||2005-09-18||800.0||10592, A. Evans|
|F110W||NICMOS||NIC2||0.075||1997-11-10||383.6||7219, N. Scoville|
|F160W||NICMOS||NIC2||0.075||2003-12-01||599.4||9726, R. Maiolino|
Here we describe in detail the methodology we have followed to perform the star cluster photometry. The clusters were detected in a combination of the HST optical images in F435W and F814W filters. We registered and resampled all the HST images to a common pixel scale, using GEOMAP/GEOTRAN IRAF tasks. Prior to the detection, we had to remove large-scale background variations in this optical image in two steps. First, it was smoothed with a box median filter, and then, the smoothed image was subtracted from the original. The input object list for photometry was obtained by running DAOFIND task in DAOPHOT package on this background-subtracted image. This procedure is similar to that followed by Larsen & Brodie (2000). As an additional selection criterion, only clusters with S/N 5 in optical filters were considered for subsequent photometry. The detected clusters were shown in Figure 3.
Photometry was obtained using the phot task in APPHOT. It gives the fluxes of the clusters in all the filters, by specifying the aperture radius and the position and width of the sky ring for background determination. We have followed the recommendations of the reference guide to IRAF/APPHOT package (Lindsey E. Davis), selecting an aperture radius FWHM, and 4 FWHM for the sky ring inner radius and width. The aperture photometry of sources requires the correction for the light loss due to the finite size of the aperture used, that is called the aperture correction. The best way to determine aperture corrections is from real point sources (stars) on the images. However, isolated bright stars to obtain a meaningful correction were only present in optical images, but nor in UV neither in NIR. Thus, aperture corrections were determined from analytically generated point sources using Tiny Tim software (Krist et al., 2011).
We have converted these fluxes into STMAG magnitudes, using equation (6) in Sirianni et al. (2005). The errors have been calculated through error propagation:
where F is the star cluster flux within the aperture, n is the number of pixels in the aperture, sky is the sky background value per pixel, measured in the sky annulus around the cluster, n is the number of pixels in the background annulus, and is the readout noise of the CCD. Errors in the corresponding magnitudes were computed from eq. 5 using .
a.2 Cluster ages
To estimate cluster properties, we have compared the colours and magnitudes of IC 1623 and NGC 6090 clusters with those of Charlot & Bruzual SSPs models (2007, unpublished) in a range of ages between 1 Myr to 13 Gyr. These models are similar to those in Bruzual & Charlot (2003), but replacing STELIB (Le Borgne et al., 2003) by a combination of the MILES (Sánchez-Blázquez et al., 2006) and granada (Martins et al., 2005) libraries. Chabrier IMF and Padova 1994 evolutionary tracks were used. Model colours have been computed using STSDAS.SYNPHOT software. Using the Calzetti et al. (2000) law we have computed also the colours of the models reddened by up to A mag. Despite recent improvements in modelling tecniques, it is very difficult from photometry to estimate metallicity on individual clusters. From the spectroscopy we found that the average value for the metallicity in these systems is 0.6 Z (see Section 4.6). Taking this into account, we have used solar metallicity SSP models (1 Z) in this analysis. However, we have checked that 0.4 Z models predict similar results.
In Figure A1 we show the FUV F435W vs F435W F814W diagram for IC 1623 (top panel) and NGC 6090 (bottom panel), which is the best to break the age-A degeneracy. The colour coding is the same as in Figure 3, but only those clusters with SNR 5 in FUV, F435W, and F814W filters are included in the plots. The black solid line is the path described by SSP models from 1 Myr to 13 Gyr, Z, and A = 0 mag. The greyscale lines are the paths for the same models reddened by 1 to 3 mag, with the lighter shades tracing the more reddened models.
Comparing the position of the clusters in colour-colour diagrams with respect to the model positions we can estimate the range of cluster ages. The main results are summarized in Section 3.
a.3 Cluster extinction
Figure A2 shows the F435W F814W vs. F814W F160W diagram for IC 1623 (top panel) and NGC 6090 (bottom panel). This diagram allows to estimate the range of extiction affecting the clusters which are obscured in the FUV. Only clusters with SNR 5 in F435W, F814W, and F160W filters are shown. Again, the black solid line is the path described by SSPs from 1 Myr to 13 Gyr, Z, and A = 0 mag. The greyscale lines are the paths for the same models reddened by up to 6 mag, with the lighter shades tracing the more extincted models.
We note that, while in NGC 6090 the difference in the values of extinction between the two progenitors is relatively small, 1 mag, for IC 1623 the contrast between both progenitors is much higher, possibly up to 4–6 mag.
a.4 Cluster mass
We found that NICMOS 1.6 band is the most sentitive to mass variations. We estimate the mass range of the star clusters by comparing their 1.6 absolute magnitude with the absolute magnitude of SSPs models of masses ranging from 10 to 10 M. The results are shown in Figure A3 for IC 1623 (top panel) and NGC 6090 (bottom panel). This is a NIR 1.6 absolute magnitude vs F814W - F160W diagram. The paths from models of 10, 10, 10, 10, and 10 M are shown as solid and dashed lines. The models have been cut between 1 - 500 Myr, as we know from the previous section that these are the older ages our clusters can have.
We have performed a rough estimate of the total mass in clusters in these systems by assigning a certain average mass for the clusters in each luminosity L range. In particular, we have separared the clusters in four possible luminosity ranges: from 0.9 log L 1.5 we assign a mass of 10 M, if 1.5 log L 2.65 then 10 M, if 2.65 log L 3.5 then 10 M, and above log L 3.5 we assign 10 M. In the case of IC 1623 W we have 0, 143, 62 and 1 clusters in each range, leading to a clusters mass M 9 10 M. In NGC 6090 we have 0, 19, 79, 19 clusters in each luminosity range, leading to M 2 10 M.
a.5 Cluster detection limits
A major question to be addressed with star cluster photometry is to know if older/more reddened clusters could have been detected. As the star cluster analysis was a complementaty goal, we did not make a proper completeness study as in Davies et al. (2016), where synthetic clusters of different ages/optical depths are introduced in the cluster detection image to quantify which percetange of them could be recovered. However, we followed a more simplistic approach to estimate which clusters could be, in principle, detected, and which not. Using the SSP models of 10 Myr, 100 Myr, 200 Myr, 300 Myr, 500 Myr, and 1 Gyr, and the transmission curves from F435W and F814W filters, we compute the 435 and 814 magnitudes for clusters of different masses (, , , and ) and affected by extinctions of 0 mag, 1 mag, 2 mag, 4 mag, 6 mag and 10 mag. These magnitudes are compared with the 435 and 814 limiting magnitudes expected for point sources (most clusters are not resolved) with SNR=5, obtained from Figure 10.2 and 10.32 of the ACS Instrument Handbook, given the exposure times of the HST observations used in this paper ( s).
The less massive the cluster the more difficult to detect it as it ages/is more reddened. In IC 1623, clusters of up to 300 Myr can be detected for no extinction, for = 1 mag, clusters older than 100 Myr cannot be detected, and for = 2–4 mag, only clusters of 10 Myr can be detected. In NGC 6090, whose distance is larger than IC 1623 one, only clusters up to 100 Myr can be detected for no extinction, while for = 1–2 mag, only 10 Myr clusters can be detected.
From our observations, we found that most clusters in IC 1623 and NGC 6090, have masses between . In principle, clusters up to 1 Gyr (in IC 1623) and 500 Myr (NGC 6090) could be detected up to extinction levels of mag. For mag, only clusters up to 300 Myr (IC 1623) and 100 Myr (NGC 6090) can be detected, and for mag, only 10 Myr clusters can be detected. Above that extinction level, we cannot detect clusters anymore.
The age/extinction limits reach further for the more massive clusters. For example, clusters in IC 1623 E can be detected up to ages of 1 Gyr, and mag.
Because point sources are much harder to detect on top of the diffuse/changing galaxy background than when isolated (which is what the HST manual assumes), we acknowledge that our detection limits are in fact upper limits, as they consider cluster detection under optimal conditions. More realistically, the detection limits become more restrictive depending on the galaxy background.
From this analysis, we conclude that star clusters photometry only allows to trace the recent star formation history of these mergers, 500 Myr–1 Gyr (at best, for no extinction), but not in previous epochs. So if older star formation occurred, complementary methods will be necessary to detect it, as stellar population synthesis. Therefore, star cluster studies are preferably indicated in early-stage mergers where most of the merger-induced star formation is recent, or in more advanced mergers if we are only interested in the youngest populations formed.
Appendix B Alternative radial profiles for NGC 6090 SW
As commented in Section 2.4, the exact position of NGC 6090 SW nucleus is not clear. In the radial profiles shown in previous sections we considered as the nucleus the luminosity peak in the IFS continuum, marked as a black dot in the maps. However, there exists the possibility that the real nucleus is the bright knot located to the North West of this progenitor, which is indicated with a km sbol in the maps. As a sanity check we have repeated the radial profiles of the stellar population properties taking the knot position as origin. The only property for which the radial profile changes is the metallicity. The trend is the same, metallicity grows with distance from the core, but the gradient is steeper from the knot than from our nucleus. From the knot it goes from to , while for our nucleus it goes from to . For the other properties the profiles are very similar and the differences are not significant.
- pubyear: 2015
- pagerange: Star formation histories in mergers: The spatially resolved properties of the early-stage merger LIRGs IC 1623 and NGC 6090–B
- Probably knots with masses higher than 10 M are clumps of several clusters rather than individual unique clusters.
- The HLR is defined as the axis length of the circular aperture which contains half of the total light of the galaxy at the rest-frame wavelength. It was obtained from the continuum flux image at 5635 Å using the PyCASSO package.
- This comparison can be made because we have found that the galaxy-wide average stellar population properties are well represented by their values measured at 1 HLR González Delgado et al. (2015).
- Afflerbach, A., Churchwell, E., & Werner, M. W. 1997, \apj, 478, 190
- Alvensleben, U. F. 2004, The Formation and Evolution of Massive Young Star Clusters, 322, 91
- Arribas, S., Colina, L., Alonso-Herrero, A., et al. 2012, \aap, 541, A20
- Asari, N. V., Cid Fernandes, R., Stasińska, G., et al. 2007, \mnras, 381, 263
- Asplund, M., Grevesse, N., Sauval, A. J., & Scott, P. 2009, \araa, 47, 481
- Baldwin, J. A., Phillips, M. M., & Terlevich, R. 1981, \pasp, 93, 5
- Barrera-Ballesteros, J. K., Sánchez, S. F., García-Lorenzo, B., et al. 2015, \aap, 579, A45
- Bica, E., Arimoto, N., & Alloin, D. 1988, \aap, 202, 8
- Bruzual, G., & Charlot, S. 2003, \mnras, 344, 1000
- Bryant, P. M., & Scoville, N. Z. 1999, \aj, 117, 2632
- Calzetti, D., Kinney, A. L., & Storchi-Bergmann, T. 1994, \apj, 429, 582
- Calzetti, D., Armus, L., Bohlin, R. C., et al. 2000, \apj, 533, 682
- Cao, C., Xu, C. K., Domingue, D., et al. 2016, \apjs, 222, 16
- Chabrier, G. 2003, \pasp, 115, 763
- Chandar, R., Fall, S. M., & Whitmore, B. C. 2015, \apj, 810, 1
- Cid Fernandes, R., Leão, J. R. S., & Lacerda, R. R. 2003, \mnras, 340, 29
- Cid Fernandes, R., Mateus, A., Sodré, L., Stasińska, G., & Gomes, J. M. 2005, \mnras, 358, 363
- Cid Fernandes, R., Stasińska, G., Mateus, A., & Vale Asari, N. 2011, \mnras, 413, 1687
- Cid Fernandes, R., & González Delgado, R. M. 2010, \mnras, 403, 780
- Cid Fernandes, R., Pérez, E., García Benito, R., et al. 2013, \aap, 557, A86
- Cid Fernandes, R., González Delgado, R. M., García Benito, R., et al. 2014, \aap, 561, A130
- Colina, L., Arribas, S., & Monreal-Ibero, A. 2005, \apj, 621, 725
- Cox, T. J., Jonsson, P., Somerville, R. S., Primack, J. R., & Dekel, A. 2008, \mnras, 384, 386
- Crowther, P. A. 2007, \araa, 45, 177
- Davies, R. L., Medling, A. M., U, V., et al. 2016, \mnras, 458, 158
- Diamond-Stanic, A. M., Moustakas, J., Tremonti, C. A., et al. 2012, \apjl, 755, L26
- Di Matteo, P., Combes, F., Melchior, A.-L., & Semelin, B. 2007, \aap, 468, 61
- Di Matteo, P., Bournaud, F., Martig, M., et al. 2008, \aap, 492, 31
- Dinshaw, N., Evans, A. S., Epps, H., Scoville, N. Z., & Rieke, M. 1999, \apj, 525, 702
- Ellison, S. L., Patton, D. R., Simard, L., & McConnachie, A. W. 2008, \aj, 135, 1877
- Elmegreen, D. M., Elmegreen, B. G., Kaufman, M., et al. 2006, \apj, 642, 158
- Evans, A. S., Vavilkin, T., Pizagno, J., et al. 2008, \apjl, 675, L69
- Filippenko, A. V. 1982, \pasp, 94, 715
- García-Benito, R., Zibetti, S., Sánchez, S. F., et al. 2015, \aap, 576, A135
- García-Marín, M., Colina, L., Arribas, S., & Monreal-Ibero, A. 2009, \aap, 505, 1319
- Goldader, J. D., Meurer, G., Heckman, T. M., et al. 2002, \apj, 568, 651
- González Delgado, R. M., Leitherer, C., Heckman, T., et al. 1998, \apj, 495, 698
- González Delgado, R. M., Cid Fernandes, R., Pérez, E., et al. 2004, \apj, 605, 127
- González Delgado, R. M., Cerviño, M., Martins, L. P., Leitherer, C., & Hauschildt, P. H. 2005, \mnras, 357, 945
- González Delgado, R. M. 2008, Pathways Through an Eclectic Universe, 390, 205
- González Delgado, R. M., & Cid Fernandes, R. 2010, \mnras, 403, 797
- González Delgado, R. M., Pérez, E., Cid Fernandes, R., et al. 2014, \aap, 562, A47
- González Delgado, R. M., García-Benito, R., Pérez, E., et al. 2015, \aap, 581, A103
- González Delgado, R. M., García-Benito, R., Pérez, R., et al. 2015, Highlights of Spanish Astrophysics VIII, 220
- González Delgado, R. M., Cid Fernandes, R., Pérez, E., et al. 2016, \aap, 590, A44
- Grimes, J. P., Heckman, T., Hoopes, C., et al. 2006, \apj, 648, 310
- Guo, R., Hao, C.-N., Xia, X.-Y., Wei, P., & Guo, X. 2016, arXiv:1603.04988
- Hattori, T., Yoshida, M., Ohtani, H., et al. 2004, \aj, 127, 736
- Hopkins, P. F., Cox, T. J., Hernquist, L., et al. 2013, \mnras, 430, 1901
- Iono, D., Saito, T., Yun, M. S., et al. 2013, \pasj, 65, L7
- Jarrett, T. H., Masci, F., Tsai, C. W., et al. 2013, \aj, 145, 6
- Kauffmann, G., Heckman, T. M., Tremonti, C., et al. 2003, \mnras, 346, 1055
- Kennicutt, R. C., Jr. 1998, \araa, 36, 189
- Kennicutt, R. C., Jr., Hao, C.-N., Calzetti, D., et al. 2009, \apj, 703, 1672
- Kewley, L. J., Heisler, C. A., Dopita, M. A., & Lumsden, S. 2001, \apjs, 132, 37
- Kewley, L. J., Rupke, D., Zahid, H. J., Geller, M. J., & Barton, E. J. 2010, \apjl, 721, L48
- Kewley, L. J., & Dopita, M. A. 2002, \apjs, 142, 35
- Kim, D.-C., Evans, A. S., Vavilkin, T., et al. 2013, \apj, 768, 102
- Knop, R. A., Soifer, B. T., Graham, J. R., et al. 1994, \aj, 107, 920
- Kreckel, K., Groves, B., Schinnerer, E., et al. 2013, \apj, 771, 62
- Krist, J. E., Hook, R. N., & Stoehr, F. 2011, \procspie, 8127, 81270J
- Larsen, S. S., & Brodie, J. P. 2000, \aj, 120, 2938
- Larson, R. B., & Tinsley, B. M. 1978, \apj, 219, 46
- Le Borgne, J.-F., Bruzual, G., Pelló, R., et al. 2003, \aap, 402, 433
- Le Floc’h, E., Charmandaris, V., Laurent, O., et al. 2002, \aap, 391, 417
- Marino, R. A., Rosales-Ortega, F. F., Sánchez, S. F., et al. 2013, \aap, 559, A114
- Martins, L. P., González Delgado, R. M., Leitherer, C., Cerviño, M., & Hauschildt, P. 2005, \mnras, 358, 49
- Meidt, S. E., Schinnerer, E., Knapen, J. H., et al. 2012, \apj, 744, 17
- Meidt, S. E., Schinnerer, E., van de Ven, G., et al. 2014, \apj, 788, 144
- Monreal-Ibero, A., Arribas, S., & Colina, L. 2006, \apj, 637, 138
- 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,
- Panter, B., Jimenez, R., Heavens, A. F., & Charlot, S. 2007, \mnras, 378, 1550
- Patton, D. R., Torrey, P., Ellison, S. L., Mendel, J. T., & Scudder, J. M. 2013, \mnras, 433, L59
- Pérez, E., Cid Fernandes, R., González Delgado, R. M., et al. 2013, \apjl, 764, L1
- Renaud, F., Bournaud, F., Kraljic, K., & Duc, P.-A. 2014, \mnras, 442, L33
- Renaud, F., Bournaud, F., & Duc, P.-A. 2015, \mnras, 446, 2038
- Renzini, A., & Peng, Y.-j. 2015, \apjl, 801, L29
- Rich, J. A., Kewley, L. J., & Dopita, M. A. 2011, \apj, 734, 87
- Rich, J. A., Torrey, P., Kewley, L. J., Dopita, M. A., & Rupke, D. S. N. 2012, \apj, 753, 5
- Rich, J. A., Kewley, L. J., & Dopita, M. A. 2014, \apjl, 781, L12
- Roth, M. M., Kelz, A., Fechner, T., et al. 2005, \pasp, 117, 620
- Rupke, D. S. N., Kewley, L. J., & Chien, L.-H. 2010, \apj, 723, 1255
- Sánchez, S. F. 2006, Astronomische Nachrichten, 327, 850
- Sánchez, S. F., Kennicutt, R. C., Gil de Paz, A., et al. 2012, \aap, 538, A8
- Sánchez, S. F., Pérez, E., Rosales-Ortega, F. F., et al. 2015, \aap, 574, A47
- Sánchez-Blázquez, P., Peletier, R. F., Jiménez-Vicente, J., et al. 2006, \mnras, 371, 703
- Sánchez-Blázquez, P., Ocvirk, P., Gibson, B. K., Pérez, I., & Peletier, R. F. 2011, \mnras, 415, 709
- Sanders, D. B., & Mirabel, I. F. 1996, \araa, 34, 749
- Sanders, D. B., Mazzarella, J. M., Kim, D.-C., Surace, J. A., & Soifer, B. T. 2003, \aj, 126, 1607
- Sirianni, M., Jee, M. J., Benítez, N., et al. 2005, \pasp, 117, 1049
- Surace, J. A., Sanders, D. B., Vacca, W. D., Veilleux, S., & Mazzarella, J. M. 1998, \apj, 492, 116
- Surace, J. A., Sanders, D. B., & Evans, A. S. 2000, \apj, 529, 170
- Teyssier, R., Chapon, D., & Bournaud, F. 2010, \apjl, 720, L149
- Toomre, A., & Toomre, J. 1972, \apj, 178, 623
- Vazdekis, A., Sánchez-Blázquez, P., Falcón-Barroso, J., et al. 2010, \mnras, 404, 1639
- Veilleux, S., Kim, D.-C., & Sanders, D. B. 2002, \apjs, 143, 315
- Veilleux, S., Kim, D.-C., Peng, C. Y., et al. 2006, \apj, 643, 707
- Veilleux, S., & Osterbrock, D. E. 1987, \apjs, 63, 295
- Wang, Z., Fazio, G. G., Ashby, M. L. N., et al. 2004, \apjs, 154, 193
- Wilson, C. D., Harris, W. E., Longden, R., & Scoville, N. Z. 2006, \apj, 641, 763