Keck OSIRIS AO LIRG Analysis (KOALA): Feedback in the Nuclei of Luminous Infrared Galaxies

The Great Observatories All-sky LIRGs Survey (GOALS; Armus et al. 2009) consists of 201 of the brightest and closest (U)LIRGs in the local universe, a complete subset of the flux-limited IRAS Revised Bright Galaxy Sample (f_{60 μm} > 5.24 Jy and galactic latitude $| b| \gt 5^\circ$; Sanders et al. 2003). The survey incorporates a wealth of ancillary data spanning the entire electromagnetic spectrum from Chandra X-ray to the Very Large Array radio regime. Our near-infrared Keck sample is limited to selecting from among the most luminous (${L}_{\mathrm{IR}}\gt {10}^{11.4}$) 88 objects with coverage by HST-ACS observations that have suitable guide stars for the AO system. The HST-ACS criterion ensures that we have high spatial resolution (005) images with precision astrometry for planning the AO observations, which is critical given the small field of view (FOV) of OSIRIS. At the distance of our sample (z < 0.08, though mostly z < 0.05), we resolve the inner kiloparsec region of each source with ∼20–80 pc per resolution element.

Incorporating observing feasibility (e.g., observable from Maunakea and with available guide stars that fulfill the tip-tilt requirements of the Keck LGS/NGS AO systems; Wizinowich et al. 2000, 2006; van Dam et al. 2006), approximately half the HST-GOALS sample may be followed up with the Keck AO system. Since one of our program's goals is to trace feedback properties along the merging sequence, our observing priorities took into account the merger classification scheme for the GOALS sample as adopted from Haan et al. (2011) and Kim et al. (2013): (0) a single galaxy with no obvious major merging companion, (1) separate galaxies with symmetric disks and no tidal tails, (2) distinguishable progenitor galaxies with asymmetric disks and/or tidal tails, (3) two distinct nuclei engulfed in a common envelope within the merger body, (4) double nuclei with visible tidal tails, (5) a single or obscured nucleus with prominent tails, and (6) a single or obscured nucleus but with disturbed morphology and short faint tails signifying a postmerger remnant.

We initially prioritized late-stage mergers with one nucleus or two close nuclei on the verge of coalescence (Papers I and II). Our campaign has since been extended to include systems at earlier stages of merging that tend to be at lower luminosities and thus more closely resemble normal star-forming galaxies in other nearby studies (e.g., Kennicutt et al. 2003). Our current study comprises 21 interacting systems (22 nuclei; see Table 1) for which K-band data targeting atomic and molecular hydrogen transitions have been gathered. Comparisons of our current KOALA sample with the entire HST-GOALS sample in infrared luminosity and merger stage are shown in Figure 1. The KOALA sample as presented here is not yet complete with respect to the limitations imposed by the Keck AO system requirements. However, it has achieved a representative range in L_IR and merger stage relative to the HST-GOALS sample, which is complete in infrared luminosity within GOALS down to ${L}_{\mathrm{IR}}\geqslant {10}^{11.4}$.

Zoom In Zoom Out Reset image size

The observations were taken with OSIRIS (Larkin et al. 2006) on the Keck II and Keck I telescopes before and after 2012 January 3, respectively. The weather conditions varied from run to run but were sufficiently good (with typical FWHM ∼ 006 estimated from imaging the tip-tilt star). We employed AO corrections with both natural and laser guide stars, the tip-tilt requirements for which are R magnitude <15.5 and separation r < 35'' and R magnitude <18.5 and r < 65'', respectively.

For this study, we focused on the observations obtained with the broad K-band filter "Kbb" (and "Kcb," which has identical spectral specifications as "Kbb" but with a pupil paired to the 100 mas plate scale) covering wavelengths 1965 nm < λ < 2381 nm with spectral resolution R ∼ 3800. This wavelength range enables us to primarily target the molecular hydrogen rovibrational transitions and atomic hydrogen recombination lines as tracers for shock excitation and star formation, respectively. Our choice of observing mode, balancing the size of the FOV and the angular resolution, depended upon each target's redshift and spatial features. The FOV is 056 × 224 for the 35 mas plate scale, 08 × 32 for the 50 mas plate scale, and 16 × 64 for the 100 mas plate scale. We aimed to center the observations at our best guess of the kinematic nucleus or nuclei based on high-resolution HST-ACS F435W and F814W observations (Kim et al. 2013, A. Evans et al. 2019, in preparation) aligned at the position angles that would best capture multiple nuclear star clusters and/or the kinematics of the nuclear region (see Paper I for  both the gaseous and stellar distributions). Standard A0V stars were imaged throughout the observing runs for telluric corrections. A summary of the observation details for the individual galaxy systems can be found in Table 1.

Most of the data sets were processed using the standard OSIRIS pipeline (Krabbe et al. 2004) that incorporates dark-frame subtraction, channel level adjustment, cross-talk removal, glitch identification, cosmic-ray cleaning, data-cube assembly, dispersion correction, scaled sky subtraction for enhanced OH-line suppression, and telluric correction. The extracted spectra were subsequently processed  to remove bad pixels. More details regarding the reduction process for specific galaxy systems may be found in U et al. (2013), Medling et al. (2015b), Davies et al. (2016), Paper I, and Paper II.

Five H₂ emission lines in our spectra trace the warm molecular gas. These lines can be thermally excited through collisions with the surrounding atomic gas, itself heated by shocks, turbulence, UV photons from OB stars, or even X-rays from the central AGN (Mouri 1994). On the other hand, nonthermal excitation processes include absorption of UV photons (resonance fluorescence) and collisions with high-energy electrons (Burton 1987). The flux ratios between the transitions can partially distinguish between thermal and nonthermal excitation mechanisms, thanks to the difference in their efficiencies in populating the various vibrational levels. Much effort has been put into developing models to predict the values of these line intensity ratios for the scenarios of nonthermal excitation (Black & van Dishoeck 1987), shock heating (Brand et al. 1989), and thermal UV excitation (Sternberg & Dalgarno 1989).  Detailed studies of the near-infrared H₂ line ratios in Seyferts (Müller-Sánchez et al. 2018), star-forming galaxies (Riffel et al. 2013), LIRGs (Bedregal et al. 2009; Emonts et al. 2014; Väisänen et al. 2017), and nearby spirals (Mazzalay et al. 2013; Smajić et al. 2015; Busch et al. 2017) have been enabled within the past decade by a suite of near-infrared integral field instruments. Several of the results have indicated that the warm molecular gas in these systems is consistent with being thermally excited by a combination of X-ray radiation and slow shocks.

Here we employ two particular H₂ line ratio diagnostics, 2−1 S(1)/1−0 S(1) versus 1−0 S(3)/1−0 S(1) and versus 1−0 S(2)/1−0 S(0), to gain insight into the excitation mechanisms of our galaxies by extracting their integrated values over the high-S/N, H₂ emission–dominated "nuclear" region within each galactic nucleus (see Figure 3 and Table 3) and comparing them to various theoretical models (Figure 4). First, we note that the locus of the galaxies lies closer to the various thermal models on the left side of the plot, suggesting that nonthermal UV fluorescence, though plausibly present, is unlikely to contribute to the dominant excitation mechanism. The exception is NGC 7469N, which, interestingly, features offset ($\sim 0\buildrel{\prime\prime}\over{.} 2$) Brγ emission relative to the continuum peak. We further note that several of the systems where we found shocked gas based on enhanced H₂/Brγ ratios (e.g., Mrk 273 in U et al. 2013; IRAS F17207−0014 in Medling et al. 2015b; III Zw 035; filled cyan circles in the figure) reside near the region on the diagnostic plot occupied by predictions from shock models. See Section 4.4 for further analysis of the H₂/Brγ of this sample.

Zoom In Zoom Out Reset image size

We also compare our H₂ line ratios with those of several LIRGs found in the literature in Figure 4. For both NGC 5135 (Bedregal et al. 2009; $\mathrm{log}{L}_{\mathrm{IR}}/{L}_{\odot }=11.2$) and NGC 3256 (Emonts et al. 2014; $\mathrm{log}{L}_{\mathrm{IR}}/{L}_{\odot }=11.6$), the H₂ line ratios within their respective nuclei are represented as single components. In the case of NGC 6240 (Draine & Woods 1990; $\mathrm{log}{L}_{\mathrm{IR}}/{L}_{\odot }=11.9$), the H₂ line ratios were extracted from the inner 3'' region. These sources are broadly consistent with the (U)LIRGs in our sample, where the H₂ gas is likely excited by X-ray irradiation and/or shocks from supernova remnants. We also show the spatially resolved H₂ line ratios extracted from different regions within IRAS F19115–2124, where the aperture corresponding to the circumnuclear region is shown to be dominated by nonthermal emission and that corresponding to the strongest star-forming areas appears to be dominated by thermal excitation (Väisänen et al. 2017).

Given the H₂ line ratios, we make use of the equations from Reunanen et al. (2002) and Rodríguez-Ardila et al. (2004, 2005) to compute the rotational and vibrational temperatures within the nuclei of our sample:

$$\begin{eqnarray}&&{T}_{\mathrm{vib}}\cong 5600/\mathrm{ln}\left(1.355\times \displaystyle \frac{1-0{\rm{S}}(1)}{2-1{\rm{S}}(1)}\right)\end{eqnarray} \tag{ 1 }$$

$$\begin{eqnarray}&&\mathrm{and}\ \quad {T}_{\mathrm{rot}}\cong -1113/\mathrm{ln}\left(0.323\times \displaystyle \frac{1-0{\rm{S}}(2)}{1-0{\rm{S}}(0)}\right).\end{eqnarray} \tag{ 2 }$$

The derived vibrational and rotational temperatures are listed in Table 3 and illustrated in Figure 5. The mean vibrational temperature for our sample is $\langle {T}_{\mathrm{vib}}\rangle =3240\pm 810\,{\rm{K}}$, whereas the mean rotational temperature is $\langle {T}_{\mathrm{rot}}\rangle =1360\pm 390\,{\rm{K}}$. The vibrational temperatures for our sample tend to be above those measured for Seyferts (T_vib ≲ 2600 K; Reunanen et al. 2002). We note that the mean values for T_vib and T_rot differ by a factor of ∼2 for most of the sample, but the difference is particularly appreciable for NGC 7469N. A large difference between T_rot and T_vib is usually explained by the presence of fluorescently excited H₂, while purely thermally excited H₂ would exhibit much more similar T_rot and T_vib values (e.g., Black & van Dishoeck 1987; Draine & Bertoldi 1996; Martini et al. 1999).

Zoom In Zoom Out Reset image size

Given the expected level of heavy dust extinction in the nuclei of (U)LIRGs (A_V ∼ 3−17 mag; Piqueras López et al. 2013), star formation rates (SFRs) as computed from optical or UV tracers may not offer the complete picture in this domain. The hydrogen recombination lines in the near-infrared, such as the Brackett or Paschen series, may offer a truer measure of the nuclear SFR, as they trace ionizing photons from young stars. However, if the dust extinction in the nuclear region were patchy or had an unusual wavelength dependence, even global SFRs determined from these near-infrared lines may be biased. Thus, we compute the spatially resolved extinction map for each source and determine the nuclear SFR using the dust-corrected Brγ line flux as a proxy for recent star formation. Following, e.g., Calzetti et al. (1994) and Domínguez et al. (2013), we compute the intrinsic luminosities L_int as follows:

$$\begin{eqnarray}&&{L}_{\mathrm{int}}(\lambda )={L}_{\mathrm{obs}}(\lambda ){10}^{0.4{A}_{\lambda }},\end{eqnarray} \tag{ 3 }$$

where L_obs is the observed luminosity and A_λ is the extinction at wavelength λ, ${A}_{\lambda }=k(\lambda )E(B-V)$. We determine the color excess E(B − V) using the relationship between the nebular emission-line color excess and the Brackett decrement (Brγ and Brδ; e.g., Momcheva et al. 2013),

$$\begin{eqnarray*}\begin{array}{rcl}E(B-V) & = & \displaystyle \frac{E(\mathrm{Br}\delta -\mathrm{Br}\gamma )}{k(\mathrm{Br}\delta )-k(\mathrm{Br}\gamma )}\\ & = & \ \displaystyle \frac{2.5}{k(\mathrm{Br}\delta )-k(\mathrm{Br}\gamma )}{\mathrm{log}}_{10}\left[\displaystyle \frac{{\left(\mathrm{Br}\gamma /\mathrm{Br}\delta \right)}_{\mathrm{obs}}}{{\left(\mathrm{Br}\gamma /\mathrm{Br}\delta \right)}_{\mathrm{int}}}\right],\end{array}\end{eqnarray*}$$

where k(Brδ) = 0.43 and k(Brγ) = 0.36 are the reddening curves evaluated at Brδ and Brγ wavelengths, respectively (e.g., Cardelli et al. 1989; Calzetti 2001, D. Calzetti 2018, private communication). We obtain (Brγ/Brδ)_int = 3.0/2.1 = 1.4 from Osterbrock (1989), which, in combination with the extinction curves and observed Brγ/Brδ maps, provides the color excess map for each nucleus and, subsequently, the dust-corrected Brγ maps. We then convert the luminosity of Brγ, L(Brγ), to SFR using

$$\begin{eqnarray}&&\mathrm{SFR}\,({M}_{\odot }\,{\mathrm{yr}}^{-1})=8.2\times {10}^{-40}\,L(\mathrm{Br}\gamma )\,(\mathrm{erg}\,{{\rm{s}}}^{-1}),\end{eqnarray} \tag{ 4 }$$

as adopted from Kennicutt (1998). In order to compare SFRs across galaxies and data sets covering different physical scales of the nuclear regions, we further compute the surface density of the SFR, Σ_SFR (in M_⊙ yr⁻¹ kpc⁻²), by taking into account only the subregion within the OSIRIS FOV where reliable H₂ and Brγ fluxes can be measured. The observed and extinction-corrected total nuclear Brγ luminosities, along with their corresponding SFR surface densities, are listed in Table 3.

The total warm molecular hydrogen mass in the galactic nuclei can be computed from the flux-calibrated H₂ 1−0 S(1) emission line. Using the prescriptions from, e.g., Mazzalay et al. (2013),

$$\begin{eqnarray}&&{M}_{{{\rm{H}}}_{2}}\simeq 5.0875\times {10}^{13}{\left(\displaystyle \frac{{D}_{L}}{\mathrm{Mpc}}\right)}^{2}\,\left(\displaystyle \frac{{F}_{1-0{\rm{S}}(1)}}{\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}}\right){10}^{0.4{A}_{2.2}},\end{eqnarray} \tag{ 5 }$$

where ${M}_{{{\rm{H}}}_{2}}$ is the mass of the warm H₂ at T ≃ 2000 K in M_⊙, D_L is the luminosity distance in Mpc, F_1−0S(1) is the flux of the H₂ 1−0 S(1) line, and A_2.2 is the extinction at 2.2 μm (e.g., Scoville et al. 1982). The total flux of H₂ within the same subregion of reliable measurements and the corresponding derived molecular gas mass are listed in Table 3.

Given that the warm (T ∼ 2000 K) molecular gas is most likely thermally excited due to shocks or, perhaps, X-ray radiation (Figure 4), and that the Brγ line traces UV-ionizing radiation from young stars, the H₂ 1−0 S(1)/Brγ (hereafter H₂/Brγ) ratio quantifies the relative contributions from these energy sources. Specifically, a high H₂/Brγ ratio signals a contribution to the H₂ heating that is above and beyond that contributed by the source of the radiation that produces the H ii regions. That excess heating may be due to shocks or X-ray radiation.  The H₂/Brγ ratio  is arguably better at separating starbursts, Seyferts, and composite galaxies than its optical counterparts,  because H₂ traces the warm molecular gas at ∼100 K rather than 10⁴ K ionized atomic gas and is therefore less closely coupled with high-ionization lines (e.g., [O iii]). The H₂/Brγ ratio divides starburst galaxies and Seyferts at 0.6, while H₂/Brγ ≳ 2−3 further indicates shock heating or photoionization by the central AGN (e.g., Larkin et al. 1998; Rodríguez-Ardila et al. 2005; Riffel et al. 2008).

In our sample of 21 interacting systems, six sources have regions with H₂/Brγ > 2 (UGC 08058, IRAS F17207−0014, UGC 08696, IRAS F22491−1808, UGC 05101, and III Zw 035). These shocked candidates consist of five ULIRGs and one LIRG within our sample (see Figure 6). The shocked nature of the outflowing molecular gas in UGC 08696 and IRAS F17207−0014 has been discussed extensively in U et al. (2013) and Medling et al. (2015b), respectively. UGC 08696 features a dual AGN system where enhanced H₂/Brγ ratios are found both at the site of a hard X-ray AGN in the southwestern region and near the obscured AGN in the north. Given that the northern AGN is very obscured, its surrounding H₂ gas is likely to be primarily heated by shocks, despite the presence of an AGN. The shocked molecular gas in IRAS F17207−0014 is spatially coincident with the base of a nuclear superwind.

Zoom In Zoom Out Reset image size

The strongest new shocked outflow candidate  is the OH megamaser host galaxy III Zw 035, whose H₂/Brγ line ratio and H₂ kinematic maps point to a clump of shocked, outflowing H₂ just south of the continuum peak (see Figure 3). This shocked gas is seen emanating from the minor axis of the disk in a fan shape. There is a hint of biconical signature on the opposite side of the disk, but the spatial coverage of the OSIRIS FOV is too limiting to confirm its full extent. The lack of AGN signature from hard X-ray or [Ne v] detection in our multiwavelength ancillary data set (Iwasawa et al. 2011; Petric et al. 2011; Inami et al. 2013) would suggest that these shocks  are not being driven by an AGN. IRAS F22491−1808 hosts two kinematically distinct nuclei that are 2.2 kpc apart (see Paper I and the Appendix for details), and a molecular outflow has been detected in H₂ (Emonts et al. 2017) but not in OH (Veilleux et al. 2013). Our OSIRIS data show that the H₂ outflow entrains shocked gas with H₂/Brγ > 2 in the eastern nucleus. Both UGC 08058 and UGC 05101 feature a bright AGN in the center, so the gas is difficult to see at  the nucleus given the bright continuum source, but excited H₂ gas is detected in the circumnuclear region that may be caused by shocks or AGN photoionization.

For the remaining sources, the H₂/Brγ ratio reaches no higher than ∼1.5. In several cases, H₂ is preferentially located on the outskirts (∼150–250 pc) of the nucleus relative to Brγ (e.g., UGC 08387, NGC 2623, and IRAS F03359+1523). Here the spatial coincidence of Brγ relative to the K-band continuum indicates the sites of nuclear star formation, whereas the off-nuclear H₂ gas may be expelled winds from the nearby central sources based on kinematics analysis (to be presented in a forthcoming paper). For CGCG 436–030, IRAS F20351+2521, and NGC 6090, the H₂ flux is weak but cospatial relative to that of Brγ, resulting in low H₂/Brγ values across the nuclear region.

Statistical results of the  H₂/Brγ line ratio for our sample (for individual galaxies as well as for ULIRGs and LIRGs as two populations) are presented in Table 4. We note that the median value for our LIRGs, 0.77, is consistent with the VLT/SINFONI results presented for LIRGs (median = 0.77 with 5th and 95th percentiles of 0.29 and 2.30, respectively) in Colina et al. (2015), which consists of a sample of LIRGs with $\mathrm{log}{L}_{\mathrm{IR}}/{L}_{\odot }=11.1\mbox{--}11.7$, reaching a lower luminosity range than our LIRGs. The VLT/SINFONI observations span a  larger FOV and coarser  resolution (average FWHM ∼ 06) than our OSIRIS observations. While the sample in Colina et al. (2015) is  more nearby ($z\lt 0.018$), their IFS coverage includes the inner ∼2.5−5 kpc regions, more extended than ours in general. The statistical results for our ULIRGs (median H₂/Brγ = 1.43) are similar to those obtained from Gemini/NIFS observations of Seyferts (median = 1.41 with 5th and 95th percentiles = 0.40 and 4.00, respectively) as reported in Colina et al. (2015), even though not all of our ULIRGs host an AGN. The similarity in the high H₂/Brγ ratio between the ULIRGs and the Seyferts suggests that the two populations both have significant excesses of warm molecular gas compared to what is seen in the larger population of LIRGs.

The effect of angular resolution on line ratios was considered significant in previous studies, particularly in the case of LIRGs (e.g., Colina et al. 2015). Given that lower-resolution observations are commonplace, whether due to seeing limitations or different instrument capabilities, we explored how differences in spatial resolution might affect our measurements in these exotic environments within the nuclei of (U)LIRGs. High-resolution data, such as those taken with OSIRIS, present the opportunity to distinguish resolved sources from diffuse emission at small scales and quantify the dilution of the emission-line signal, if any.

First, it is worth reiterating the findings from Colina et al. (2015) that are relevant to the discussion at hand. In assessing the effect of spatial resolution, they compared the integrated (e.g., flux-weighted) and median [Fe ii] (1.64 μm)/Brγ and H₂/Brγ measurements and found that, with the exception of a few objects, most LIRGs show large differences in the [Fe ii]/Brγ ratio with integrated over median values ∼1.5–2.2 times higher. They reasoned that integrated values are more affected by the locus of the nuclei and might therefore appear as higher excitation than the median ones, but they questioned whether this was a general behavior in LIRGs or galaxies in general if larger volumes of galaxies with IFS data were present.

We investigated whether a similar effect in line ratios might be seen in our data. Within our sample, four sources were taken in both the 35 and 100 mas plate scales: III Zw 035, IRAS F01364−1042, IRAS F03359+1523, and MCG +08–11–002. Since all but the two observations for MCG +08–11–002 were obtained on different nights, line ratios provide a more robust test of consistency than calibrated line fluxes, which may incorporate photometric errors from different observing conditions. We extract the H₂/Brγ statistics from Table 4 for these objects and compare the flux-weighted mean over median values at different plate scales. We do find a slight enhancement in the mean over median values of ∼1.1–1.3 times, but we see no systematic differences between those measured from the 100 and 35 mas scales.

There may be several reasons for the apparent discrepancy from the Colina et al. (2015) claim. The OSIRIS area coverage is more limited to the central kpc region, which is dominated by the nuclear source. Thus, a median measurement representative of the nuclear region is less susceptible to the diffuse emission on broader physical scales. Even in the case of our coarser resolution, the nuclear components are resolved. In addition, Colina et al. (2015) illustrated that the disparity between the median and integrated line ratios is negligible in Seyferts but more prominent in LIRGs and star-forming galaxies. There may be a range of the expected divergence among different types of galaxies, and our sample lies in the higher-luminosity end of such a spectrum. Lastly, the Colina et al. (2015) conclusion was based on the integrated and median values of the [Fe ii]/Brγ ratio. They showed the H₂/Brγ measurements in their Figure 7, where the differences appear to be smaller.

In Figure 7, we show the histograms of our derived SFR and Σ_SFR for the sample both before and after extinction correction. In both cases, it is clear that there is an ∼1–1.5 dex enhancement in the SFR and SFR surface density that was obscured by the dust in the nuclear regions of our galaxies.  This correction corresponds to an extinction up to ∼2.5–4 mag in the K band, or ∼25–40 in the V band, significantly higher than the ${A}_{V}\sim 3\mbox{--}17$ mag quoted from Piqueras López et al. (2013) but consistent with the A_V ∼ 5–40 mag found in other AO studies of LIRG nuclei (Mattila et al. 2007; Väisänen et al. 2017). If we take a closer look at the VLT/SINFONI results, the A_V values of individual spaxels reach upward of 20–30 mag. In some cases, the extinction at the very center of the K-band continuum peak is not reported due to insufficient S/N in the Brackett line fluxes. The lack of signal at the center, coupled with the integrated nature of the statistics reported in Piqueras López et al. (2013), could explain the discrepancy in extinction corrections between this work and their VLT/SINFONI results.

Zoom In Zoom Out Reset image size

For comparison, observed SFRs have been derived for both nuclear and extranuclear star-forming clumps identified in high-resolution HST/WFC3 narrowband images of Paα and Paβ emission of the LIRGs in GOALS. The nuclear SFR ranges from 0.02 to 10 M_⊙ yr⁻¹, while the nuclear Σ_SFR spans ∼4–20 M_⊙ yr⁻¹ kpc⁻² (K. Larson et al. 2018, in preparation). These measurements are broadly consistent with the median values of our uncorrected SFR and Σ_SFR distributions. Dust corrections computed from Paschen lines in the H band and Brackett lines in the K band should be applied accordingly, as different extinctions would be determined depending on the lines used and, hence, on the depth probed through the dust (e.g., Väisänen et al. 2017).

Furthermore, we break down the distribution of dust-corrected SFR and Σ_SFR by infrared luminosity. The bottom panels of Figure 7 show the histograms of the corrected SFR and Σ_SFR for LIRGs and ULIRGs, respectively. In terms of total SFRs, ULIRGs do, on average, make more stars than their lower-luminosity counterparts in the resolved nuclear regions. However, we see here that some LIRGs match their ultraluminous counterparts in SFR. Also, the ULIRGs appear indistinguishable from the LIRGs in terms of SFR surface densities, suggesting that they are similarly efficient at producing stars at these scales from near-infrared star formation indicators. It is plausible that the ULIRG nuclei are still optically thick and thus undercorrected at these wavelengths; future resolved far-infrared observations could confirm these findings. We note that these observed SFR and SFR surface density values within our S/N-defined nuclear regions are not consistent in physical scales across our sample that spans a range in redshifts, such that even SFR surface density could be affected if a radial gradient were present. The fact that the LIRGs and ULIRGs exhibit similar distributions in Σ_SFR is consistent with what was observed with VLT/SINFONI of the inner regions of nearby southern (U)LIRGs (Piqueras López et al. 2016).

We further explore the dust-corrected nuclear SFR and SFR surface density relations with various properties of the (U)LIRG hosts in Figure 8. First, we examine how they may correlate individually with the host's infrared luminosity. Globally, we expect the infrared luminosity to correlate with SFR, particularly in (U)LIRGs, because the light from the formation of young stars is reprocessed through dust in the cooler regime (cf. U et al. 2012). We  saw in Figure 7 that ULIRGs occupy the high end of nuclear SFR; however, the relation between SFR and log L_IR is difficult to quantify (see top left panel in Figure 8). The trend is similarly insignificant in the case of SFR surface density (bottom left panel).

Zoom In Zoom Out Reset image size

Next, we consider the relation of the nuclear SFR with the merger class of the (U)LIRG host system. If the most rapid star formation in (U)LIRGs is triggered by merger activity, we would expect the nuclear SFR to be  most enhanced at coalescence.  We have compiled the merger classification for our sample in our summary table (see Table 5). We have also collated, in the same table, the projected nuclear separation between nuclei, if observed and resolved, from the literature, should it offer additional insights into the accretion of gas onto the galaxy center throughout the merging process.

In the middle panels of Figure 8, we have plotted the dust-corrected nuclear SFR (top) and SFR surface density (bottom) on a logarithmic scale as a function of merger class. We see that the SFR appears to increase with merger class as the interacting galaxy pair coalesces at merger stages 5–6 (τ = 0.44 ± 0.004). This increase is consistent with the picture of a merger-induced starburst in the central kiloparsec region, as seen in simulations (e.g., Moreno et al. 2015), as well as in observations (e.g., Barrera-Ballesteros et al. 2015).

We consider the relation between the nuclear SFR (and Σ_SFR) with the projected nuclear separation between the nuclei, if seen and resolved by observations, in the right panels of Figure 8. An enhancement in the SFR  as the projected separation between the nuclei diminishes is evident in this plot (τ = −0.31 ± 0.04). The scatter in the relation may decrease if the separation is more accurately measured from dynamical modeling of the merger system and does not suffer from projection effects. We also caution that the projected nuclear separation is susceptible to instrument resolution. For instance, NGC 7674W is listed as having a companion ∼20 pc away (Haan et al. 2011), but Kharb et al. (2017) recently reported  high angular resolution observations from the VLBA that resolve the main western nucleus into two radio cores 0.35 pc apart. It is possible that this system has undergone more than one major merger in the past, and it may be difficult to disentangle the effects of the various merging events without detailed dynamical modeling to assess timescales.

From our H₂/Brγ maps, we have identified six of 22 galaxy nuclei that host shocked molecular gas. Five of the six ULIRGs plus one LIRG in our sample host shocked molecular gas in coherent structures (with H₂/Brγ > 2; see Figure 6). The fact that shocks appear to be preferentially found in the more infrared luminous systems may be because ULIRGs have more powerful starbursts and more AGNs to heat the dense gas in the nuclear region. Wide-field optical IFS studies have found that the galaxy-wide shock fraction increases with merger stage (Monreal-Ibero et al. 2010; Rich et al. 2015), which also correlates with infrared luminosity. It is thus not surprising to find more shocked gas among the ULIRG hosts than in their less-luminous counterparts. However,  our current study using a high angular resolution instrument such as OSIRIS has  the unique ability to peer through the dust into the inner kiloparsec of the merging system. The shocked gas detected in this work may represent the smoking-gun signature of the more widespread shocks seen in larger IFS surveys (e.g., Rich et al. 2015; Ho et al. 2016), the connection between which is muddled by drastic differences across existing instrument resolutions and is yet to be thoroughly investigated.

The less-luminous exception, and arguably the most unequivocal case of shocked molecular gas in this sample, was found in the LIRG III Zw 035. As the only LIRG that shows spatially coherent, definitively shock-excited H₂ gas, III Zw 035 has an unremarkable infrared luminosity of $\mathrm{log}{L}_{\mathrm{IR}}/{L}_{\odot }=11.62$ at a redshift of 0.0278. It is classified as  merger stage 3, with a nuclear separation of 5 kpc. The fact that it has a nondetection in the Swift-BAT hard X-ray bands (Koss et al. 2013) but exhibits Compton-thick AGN qualities in the Spitzer-IRS data (González-Martín et al. 2015) means that its center must be heavily obscured. It also has the most compact 33 GHz continuum emission (with a nuclear half-light radius of only 30 pc) among a sample of 22 local (U)LIRGs from the GOALS sample (Barcos-Muñoz et al. 2017). 

Further, we want to investigate whether the detected shocks may be a manifestation of AGN- or starburst-driven feedback. We assemble AGN signatures from across the electromagnetic spectrum in Table 5. Among this sample, 12 of 21 systems have been observed by the Chandra X-ray Observatory either as part of GOALS or by other programs. Four are classified as AGN hosts from X-ray observations (UGC 05101, UGC 08058, UGC 08696, and VV 340a; Iwasawa et al. 2011; Ricci et al. 2017). Four additional galaxies are identified as AGN hosts via the detection of the [Ne v] 14.3 μm line in the mid-infrared Spitzer-IRS spectra (UGC 08387, NGC 2623, NGC 7469N, and NGC 7674W; Petric et al. 2011). Among these AGN hosts, UGC 05101, UGC 08696, and UGC 08058 have  shocked molecular gas identified in this work. The proximity of the shocked gas to the central AGN may indicate AGN feedback as the source for the shock heating in those three cases.

On the other hand, VV 340a, UGC 8387, and NGC 7469N are three AGN hosts that do not show any signs of shocks or feedback. At an early stage of merging (stage 1; Haan et al. 2011), VV 340a hosts a Compton-thick AGN as detected by Chandra (Iwasawa et al. 2011). If the presence of shocks in (U)LIRGs is plausibly induced by galaxy interaction, the lack of a shock signature in VV 340a   could suggest an upper limit on how early shock heating might take place in merging progenitors. In the case of UGC 8387, the OSIRIS data are incomplete in the coverage of the nucleus due to unfortunate observing conditions. A hint of low-velocity, excited H₂  is present on the outskirts of the molecular gas disk, so more follow-up observations at the center of the Seyfert nucleus may be worthwhile. As for NGC 7469, the  northern nucleus we show here is 26 kpc away from the Seyfert 1 source in the system, which does indeed  feature biconical outflows  traced by  coronal lines (Müller-Sánchez et al. 2011). The remaining two AGN hosts (NGC 2623 and NGC 7674W)  feature high-dispersion (≳150–200 km s⁻¹) H₂ gas that may be heated by AGN photoionization; a detailed study of the kinematics will be presented in a future paper.

We explore the relation between nuclear shocks and AGN strength more quantitatively in Figure 9. We adopt the average mid-infrared and bolometric AGN fractions computed from various Spitzer-IRS and other diagnostics in Díaz-Santos et al. (2017); see references therein. While the signature from shocked H₂ is diluted in the mean H₂/Brγ ratio relative to the corresponding maximum value, weak correlations between shocked gas and AGN strength measured in the mid-infrared and bolometrically are seen (Kendall's correlation coefficient τ ∼ 0.3, though generally insignificantly), suggesting that some of the observed shocks may be powered, at least in part, by AGNs in these nuclei. The scatter in these weak correlations can be attributed to patchy dust obscuration in the near-infrared wavelengths or the presence of other mechanisms at work to drive shock excitation.

Zoom In Zoom Out Reset image size

To further verify the impact of spatial resolution on our results, we rebinned all the data to the lowest physical resolution present in the sample, e.g., that of IRAS F22491−1808 at 146.7 pc spaxel^–1. The most important effect is the loss of the ability to visually identify resolved structures originally present in the sample but also to identify potential shocked gas, given how its peak signal is smoothed out. The quantitative influence from the degradation of angular resolution on peak H₂/Brγ values is further shown as gray stars in Figure 9.

Our finding that nearly half of the AGN hosts exhibit shocked H₂ suggests that AGNs may shock-heat the molecular gas, but it is not ubiquitous, and photoionization may also be important. Further, we cannot rule out the possibility that other mechanisms, such as cloud–cloud collisions or mechanical perturbations of the ISM, might also contribute to the shocks in these AGN hosts. We also find shocked gas in four non-AGN hosts that must have been excited by other ionizing sources. We will further discuss the impact of AGN- and starburst-driven winds on the ISM in a future paper.

In Table 5, we record the detection of feedback signatures in other wavelength regimes in the literature. Molecular outflows in galaxy mergers have previously been identified in the longer far-infrared and millimeter wavelengths, for instance, using the OH 119 μm feature from observations taken with the Herschel Space Observatory (Veilleux et al. 2013). Due to differences in the target selections, there are only six ULIRGs in our overlapping sample. Of these, two sources (UGC 08696 and UGC 08058) exhibit OH outflows with median velocities of ∼−200 km s⁻¹. These velocities agree very well with those of our molecular outflows seen in the near-infrared H₂ transitions (U et al. 2013; Medling et al. 2015b). We do see outflowing gas in UGC 08696, though the S/N of our data cubes renders the case of UGC 08058 less conclusive.

Of the remaining four sources in our overlapping sample with Herschel observations, three have detected inflows based on redshifted OH absorption features with median velocity v₅₀(abs) ≥ 50 km s⁻¹: IRAS F22491−1808, IRAS F15250+3608, and IRAS F17207−0014. However, the spectral resolution of OSIRIS (Δv ∼ 80 km s⁻¹) hinders our ability to detect  slow, inflowing gas. As for IRAS F17207−0014, our in-depth analysis of the OSIRIS data set along with large-scale optical IFS data has revealed molecular gas dynamics that are more consistent with outflows (Medling et al. 2015b). The other discrepancy between our analysis and that of Veilleux et al. (2013) rests with UGC 05101, where we see shocked, highly turbulent molecular gas well in excess of 200 km s⁻¹ emanating from the Seyfert nucleus, but only v₅₀(abs) = −9 km s⁻¹ was detected of the OH 119 μm feature. It is reasonable to explain this difference in terms of physical scales and the multiphase nature of outflows: it would be normal to have fast, warm outflows closer to the nuclei and slower, cool outflows spatially averaged over kiloparsec scales, which is where most of the outflowing mass would be. We note, however, that a maximum velocity v_max(abs) of −1200 km s⁻¹ was reported in the Herschel work, which is among the fastest outflow velocities in that analysis.

In order to get a sense of how the ionized gas and  dust content might correlate with the warm molecular gas and star formation in these dusty galaxies, we also compare the near-infrared  diagnostic H₂/Brγ with the mid-infrared line ratio H₂/PAH  extracted from Spitzer-IRS (Stierwalt et al. 2014) in Figure 10 for the 18 galaxies in our overlapping sample. (All of the galaxies in our sample have been observed with Spitzer-IRS, but only 18 of them, two with two separate observations at different plate scales, have detected the 7.7 μm PAH feature enabling the ensuing analysis. The mid-infrared numerator H₂ incorporates the total line flux summed from the mid-infrared S0–S2 transitions. In cases where one of the H₂ lines is a mere upper-limit detection, its error has been conservatively estimated at a level equal to its corresponding flux measurement and subsequently propagated into the error of the total H₂ flux.) The H₂/Brγ ratios are the clipped mean and maximum values (excluding outliers) within the central ≲400 pc region. The adopted H₂/PAH values incorporate the H₂ (S0–S2) transitions and the 7.7 μm PAH feature in the larger IRS slit (≲2 kpc).

Zoom In Zoom Out Reset image size

Overall, the sample shows a correlation between the two diagnostic line ratios (as indicated by the dotted  line in Figure 10, with a correlation coefficient τ = 0.40 and p-value of 0.0135); a correlation is expected because the H₂ is likely to be excited by the same processes across the two wavelength regimes. The nonparametric generalized Kendall's τ and the Theil–Sen median-based linear model (implemented in the mblm R statistical package) were used for the correlation measurement and linear fitting, given their robust treatment of potential outliers. The scatter about the linear fit may be due to differences in patchy extinction in the nuclear regions and in the mismatched physical scales. Some scatter is also likely due to Brγ and PAH spatial variations, since the ionized gas and dust emission regions need not correlate on small scales, as was seen in Díaz-Santos et al. (2008, 2010) using high-resolution (04) mid-infrared Gemini/T-ReCS data. 

Also shown on the plot are the outlier-excluded maximum H₂/Brγ values for each galaxy; the variance in how much they are offset with respect to their average counterpart is challenging to predict based on integrated properties, but it highlights the importance of high angular resolution observations. The flatter fit with a negligible slope (τ = 0.21 and p-value of 0.194) between the H₂/PAH values from integrated light through the IRS slit and the maximally enhanced excited OSIRIS spaxels suggests that shocked regions may be small and unresolved at 10'' (∼10 kpc) scales. Spatially resolving and quantifying the cooler gas and dust will require mid-infrared imaging or IFS observations using the upcoming JWST. Our OSIRIS shock candidates  span a range in H₂/PAH ratios, likely due to the fact that significant star formation, which could drive shocks and feedback, contributes to the PAH emission. We note that all of the identified shock candidates lie above the upper limit of H₂/PAH  expected from photodissociation region (PDR) models (see Stierwalt et al. 2014 for the adoption of this value in these H₂ and PAH transitions).

UGC 08058, or Mrk 231, is the most luminous ULIRG in GOALS and, hence, in this KOALA sample. As a bona fide QSO, it has been found to host molecular outflows as an indicator of AGN feedback in previous infrared and submillimeter work (Alatalo 2015; González-Alfonso et al. 2017). With velocities well exceeding 500 km s⁻¹ ∼1 kpc away from the center, one might expect to see high-velocity gas close to the ionizing source in the central kpc region. However, the S/Ns in our emission lines are weak, as the near-infrared spectrum is dominated by the AGN continuum. Thus, we were not able to directly detect outflow signatures in the H₂ or Brγ line kinematics close to the center, though a coherent region ∼250 pc SE of the nucleus is seen to feature elevated H₂/Brγ ratios.

IRAS F17207−0014 is a late-stage merging ULIRG with two kinematically distinct nuclei, the collision of whose ISM has likely induced shocks tracking the base of a collimated outflow (see details in our previous work in Medling et al. 2015b). High-resolution Plateau de Bure interferometer data also found a CO molecular outflow plausibly associated with a hidden AGN in the western nucleus, as suggested by García-Burillo et al. (2015). From the SFR map, we find that the sites for star formation as traced by Brγ are to the west of the outflow base.

UGC 08696, or Mrk 273, is another ULIRG in the late-stage merging phases analyzed in detail in our previous work (U et al. 2013). We have found evidence for a molecular outflow originating from a plausible obscured AGN nucleus in the north (instead of from the X-ray-bright AGN nucleus in the southwest). The multiphase and multiscale nature of its outflow has been detailed in other studies of the warm and cold ionized and molecular gases (e.g., Rupke & Veilleux 2013; Veilleux et al. 2013; Cicone et al. 2014; Aladro et al. 2018). Here we can see that the outflow is traced by shocked H₂ gas in the northern nucleus.

IRAS F22491−1808 is a ULIRG system with two kinematically distinct nuclei in the K-band continuum and Paα emission approximately 2.2 kpc apart (see Paper I for a detailed flux and kinematics map of H₂ and Paα). Since the H₂ and Paα do not share the same kinematics, the presence of a strongly streaming or outflowing shocked gas has been conjectured. Indeed, H₂ outflows have been detected by Emonts et al. (2017) from the eastern galaxy, which is the fainter component in the system and cospatial with our maximum H₂/Brγ spaxels.

IRAS F15250+3608 is the only galaxy in Paper I for which we did not see rotation in either stars or gas.  There is potentially shocked H₂ gas near the nucleus reaching a velocity dispersion upward of 150 km s⁻¹, which could be worth follow-up observations to confirm.

UGC 5101 features a Compton-thick AGN with strong mid-infrared (Armus et al. 2004, 2007) and X-ray signature (Imanishi et al. 2003; Ptak et al. 2003; González-Martín et al. 2009) but lacks optical signatures (Yuan et al. 2010). In our OSIRIS maps, the AGN continuum dominates, though H₂ and Brγ emission  both  peak at the same location. The H₂ gas is particularly bright near the nucleus, such that enhanced H₂/Brγ spaxels can be seen coming out of the center and may be interpreted as a shock candidate.

VV 340a is at an early stage of merging as categorized in Haan et al. (2011). It hosts a Compton-thick AGN that was detected by Chandra but not by Swift-BAT (Iwasawa et al. 2011; Koss et al. 2013). Its large-scale disk features a solid dust lane, while its nuclear disk is seen in the continuum and H₂ emission (see nuclear disk properties in Paper I). We see no coherent structure in the H₂/Brγ map that could be associated with outflows, which, if warranted by the depth of the data, could place an upper limit on how early outflows are triggered in the nuclei of merging progenitors.

IRAS F01364−1042 is a LIRG with a clear rotating gas disk detected in H₂ and Paα as detailed in Paper I. In the 100 mas scale data, extended H₂ to the south between the major and minor axis has been seen, lending potential support to an outflowing structure traced by H₂. In the 35 mas scale data, the continuum is resolved into east–west extended emission with the enhanced H₂/Brγ spaxels originating from the center. The spatial correlation of this excited molecular gas with prominent non-Keplerian motion and elevated velocity dispersion as mentioned in the text lends support that there could be molecular outflows present, to be confirmed by follow-up high angular resolution observations.

UGC 08387, or IC 883, has been identified as an AGN host by the detection of the [Ne v] line from IRS spectra (Petric et al. 2011). A parsec-scale radio jet detected using VLBI further suggests that the nucleus plays host to AGN activity (Romero-Cañizales et al. 2017). Though the mosaicking of the nucleus of this galaxy is unfortunately incomplete  due to observing conditions, we included the data analysis here for completeness. Even though the central AGN is not pictured within the frame, we see a hint of what may be low-velocity, excited H₂ on the outskirts of the molecular gas disk. Due possibly to heating by AGN photoionization, the H₂/Brγ map may indicate a layer of H₂ gas heated by the Seyfert nucleus worthy of future follow-up observations perpendicular to the gas disk.

CGCG 436−030 is an LIRG with a star-forming clump to the northwest of the nucleus, as demonstrated in its Brγ flux map well modeled in Paper I. In this galaxy, the H₂ gas is relatively weak and does not appear to be shock-heated by the nearby star-forming clump. Both H₂ and Brγ display rotational kinematic signatures that follow each other fairly consistently.

NGC 6670E is the eastern component of a merging galaxy pair with a companion ∼17 kpc away (see resolved X-ray emission in Mudd et al. 2014), but the measured separation between the two nuclei as detected in H-band HST-NICMOS images within the eastern galaxy is ∼1 kpc (Haan et al. 2011). There are two caveats regarding the analysis of this galaxy: (1) the OSIRIS data analyzed in this work were taken after Paper I was published, and thus we do not yet have the quantitative properties of the nuclear disks; and (2) this was one of three nuclei for which we did not have proper calibration frames of our standard stars, and thus we were unable to flux-calibrate this galaxy in the consistent manner we did the others. For this reason, the units on the continuum, H₂, and Brγ maps are not properly displayed, and the intensity of the line and continuum should be treated in relative units. We were also unable to produce an SFR map, since we did not have a properly calibrated map of the Brγ luminosity.

Nonetheless, it is interesting that the gas peaks for both H₂ and Brγ are displaced from the main continuum peak, residing instead at an off-nucleus star-forming clump ∼200 pc west of the K-band continuum peak. The H₂ does not appear to be excited, though it displays high velocity dispersion that may be traced to the gas peak and could be a candidate for star formation–driven outflows in future studies.

IRAS F06076 is a system of two clearly distinct galaxies with a projected nuclear separation of ∼6.7 kpc according to HST-ACS imaging. Large-scale, optical IFS data from VLT/VIMOS show that at low-intermediate masses and with projected velocities of ∼550 km s⁻¹, the two galaxies are unlikely to ever merge (Arribas et al. 2008). The northern nucleus, which is the one our OSIRIS FOV covers, is dominated by star formation, as indicated by the optical line ratio and velocity dispersion (Rich et al. 2015). Our 35 mas data resolve the K-band continuum emission into two peaks and illustrate that the Brγ gas is extended over this region. The H₂ gas is more concentrated at the continuum peaks, with a velocity dispersion of ∼110 km s⁻¹ in this inner 200 pc region.

IRAS F18090+0130 hosts two progenitor galaxies 49 kpc apart (Haan et al. 2011). At the early stage of merging, this system was observed but not detected in Swift-BAT (Koss et al. 2013). The OSIRIS observations here provide a baseline of what early-stage galaxy mergers may feature before they become morphologically disturbed: the extended continuum is resolved into what appear like spiral arms well traced by the gas in both nuclei. Both the molecular and ionized gas appears extended but not heated, given the lack of significant nuclear activity in the respective regions. Like NGC 6670E, we lacked proper calibration frames for these galaxies and thus were unable to flux-calibrate the OSIRIS data.

The galaxy III Zw 035 is an LIRG that shows strong extension in H₂ emission along the minor axis of the galaxy coupled with non-Keplerian kinematics and enhanced velocity dispersion in the same region. The 100 mas scale data show a hint of this extension in the H₂ maps, but the molecular outflow is best resolved in the 35 mas scale maps in Figure 3. The H₂ peak is not only displaced from that of the K-band continuum but is shown to be shock-excited in a coherent region in the H₂/Brγ map.

IRAS F20351+2521 is an LIRG with galactic-scale spiral arms that feed into the clumpy nature as sampled within the OSIRIS FOV. As was found in Paper I, its H₂ and Brγ emission show similar kinematics, as this early-stage LIRG shows no sign of outflowing or shock-heated gas.

NGC 2623, a well-studied LIRG hosting both off-nuclear star clusters and an AGN, represents a prototypical advanced merger with twin tidal tails extending from a single nucleus (Sanders et al. 2003; Evans et al. 2008). Its K-band IFS data feature smooth flux profiles and kinematically rotating kinematics in all tracers (Paper I). Its H₂ gas is extended relative to the Brγ emission and may have a hint of >150 km s⁻¹ dispersion of H₂ gas along the minor axis of the rotating disk. It is a potential outflow candidate to be resolved by higher-resolution data in the future.

NGC 7469 is a Seyfert 1 system with a disturbed companion ∼26 kpc away. It features an ∼10⁷ M_⊙ black hole measured virially from reverberation mapping (Peterson et al. 2014) surrounded by a star-forming ring (Díaz-Santos et al. 2007). The southern nucleus  features coronal-line biconical outflows (see details in Müller-Sánchez et al. 2011). Here we have imaged the northern component of this merger system with OSIRIS. At 0035 spaxel^–1 resolution, the nucleus appears compact with extended diffuse emission in the K-band continuum. Interestingly, the peak of the Brγ flux is offset (∼02, or 65 pc) from the continuum peak; its rotational kinematics likely signifies an off-nucleus star cluster rather than outflowing ionized gas.

NGC 6090 has a companion nucleus 4 kpc away (Haan et al. 2011). The main nucleus that we targeted with OSIRIS features large-scale spiral arms traced primarily by Brγ flux as star formation sites. Chisholm et al. (2016) detected a galactic outflow with a mass outflow rate of 2.3 M_⊙ yr⁻¹ using the HST Cosmic Origins Spectrograph. As noted in Paper I, extracting kinematic information for the gas has proven to be challenging, given the relative lack of gas in the near-infrared range. The H₂/Brγ line ratio map also provides no evidence for the presence of shocked gas in this nucleus.

NGC 7674W is a famous Seyfert galaxy recently found to be a subparsec binary black hole candidate with a projected nuclear separation of 0.35 pc between radio cores (Kharb et al. 2017). It also has a companion 20 kpc away (Haan et al. 2011). Its K-band continuum emission is compact, while the Brγ gas emission is extended. There is a relative lack of H₂ at the peak of the K-band continuum, possibly masked by strong [Si vi] emission at that location. While the H₂/Brγ ratio does not reach the canonical value of 2 for shocks, it does appear enhanced in a coherent region that also features highly dispersed (≳200 km s⁻¹) gas. This gas is likely heated by photoionization by the AGN in the nucleus; a detailed study of the kinematics will follow.

IRAS F03359+1523 is an early-stage merger with a large, nearly edge-on disk that has yet to be disrupted (as evidenced by the analysis of the 100 mas data in Paper I). Here we zoom into the nucleus of the galaxy with 35 mas data. The distribution of the continuum and gas are consistent with the larger-scale data. However, the higher-resolution data show that enhanced H₂/Brγ may correlate with a region of high H₂ velocity dispersion (∼200 cm s⁻¹), even though the gas has not yet reached our conservative shock threshold.

MCG +08−11−002 is a late-stage merging LIRG with two distinct clumps, with the ages of its stellar clusters thoroughly analyzed in our previous work (Davies et al. 2016). The extended Brγ emission relative to that of the K-band continuum illustrates the strong and clumpy nuclear star formation (Paper I). In the finer-resolution 35 mas data taken since Paper I, the two continuum peaks are better resolved to a separation of 110 pc. The Brγ emission traces the bright but more extended continuum nucleus in the northeastern direction, leaving the compact continuum peak, likely a star cluster, void of ionized gas. The H₂ gas seems to concentrate on the compact star cluster, but we see no evidence for shocked or outflowing molecular gas from the OSIRIS data.