Hidden or missing outflows in highly obscured galaxy nuclei?

TL;DR

This study investigates the relationship between vibrationally excited HCN emission and outflow development in obscured galaxy nuclei, finding that strong HCN-vib emission correlates with a lack of wide-angle outflows, indicating an early evolutionary stage.

Contribution

It provides observational evidence linking high HCN-vib luminosity to an obscured phase before significant nuclear feedback and outflows develop in (U)LIRGs.

Findings

High HCN-vib luminosity systems lack wide-angle outflows.

Bright HCN-vib emission may be destroyed by massive outflows.

HCN-vib traces a heavily obscured, early evolutionary stage.

Abstract

Understanding the nuclear growth and feedback processes in galaxies requires investigating their often obscured central regions. One way to do this is to use (sub)millimeter line emission from vibrationally excited HCN (HCN-vib). It has been suggested that the most intense HCN-vib emission from a galaxy is connected to a phase of nuclear growth that occurs before the nuclear feedback processes have been fully developed. We aim to investigate if there is a connection between the presence of strong HCN-vib emission and the development of feedback in (U)LIRGs. We collected literature and archival data to compare the luminosities of rotational lines of HCN-vib, normalized to the total infrared luminosity, to the median velocities of 119 {\mu}m OH absorption lines, potentially indicating outflows, in a total of 17 (U)LIRGs. The most HCN-vib luminous systems all lack signatures of significant molecular outflows in the far-infrared OH absorption lines. However, at least some of the systems with bright HCN-vib emission do have fast and collimated outflows that can be seen in spectral lines at longer wavelengths. We conclude that the galaxy nuclei with the highest L(HCN-vib)/L(IR) do not drive wide-angle outflows detectable using the median velocities of far-infrared OH absorption lines. It is possible that this is due to an orientation effect where sources which are oriented in such a way that their outflows are not along our line of sight also radiate a smaller proportion of their infrared luminosity in our direction. It could also be that massive wide-angle outflows destroy the deeply embedded regions responsible for bright HCN-vib emission, so that the two phenomena cannot coexist. This would strengthen the idea that vibrationally excited HCN traces a heavily obscured stage of evolution before nuclear feedback mechanisms are fully developed.

Tables

Table 1. Table 1: Sample galaxies.

Name	Typea𝑎aa𝑎a$a$Optical spectral types, H ii, L (for LINER), Sy 1 and Sy 2 (for Seyfert 1 and 2, respectively). When available, we have used the classifications from Veilleux et al. (1995) and Kim et al. (1998), otherwise, we have used classifications from Sanders et al. (1988b), Baan et al. (1998), and Véron-Cetty & Véron (2001).	$z$b𝑏bb𝑏b$b$In priority order, we have used redshifts from Veilleux et al. (2013), Sanders et al. (2003), or Strauss et al. (1992), except for IRAS 05189-2524 where the redshift from Sanders et al. (1991) was used.	$D_L$c𝑐cc𝑐c$c$Luminosity distances were calculated from the redshifts following the same procedure as Sanders et al. (2003).	$L_{\mathrm{IR}}$d𝑑dd𝑑d$d$Infrared luminosities were calculated using the prescription in Table 1 of Sanders & Mirabel (1996) (originally from Pérault (1987)) using IRAS fluxes taken from Sanders et al. (2003), except for IRAS 20414-1651 where fluxes were taken from Moshir et al. (1992), and I Zw 1 where they were taken from Sanders et al. (1989).
			(Mpc)	(${10}^{11}$ L☉)
NGC 4418	Sy 2	$0.0071$	$33.6\pm 2.5$	$1.32\pm 0.19$
IC 860	H ii	$0.0130$	$62.1\pm 4.3$	$1.60\pm 0.22$
IRAS 12224-0624	L	$0.0264$	$119.5\pm 8.2$	$1.98\pm 0.27$
Zw 049.057	H ii	$0.0130$	$62.4\pm 4.3$	$2.04\pm 0.28$
NGC 7469	Sy 1	$0.0164$	$68.2\pm 4.7$	$4.13\pm 0.57$
I Zw 1	Sy 1	$0.0610$	$258.2\pm 17.7$	$8.22\pm 1.13$
UGC 5101	L	$0.0394$	$169.7\pm 11.6$	$9.50\pm 1.30$
IRAS 20551-4250	H ii	$0.0430$	$185.2\pm 12.7$	$10.65\pm 1.46$
IRAS 15250+3609	L	$0.0552$	$243.7\pm 16.7$	$11.08\pm 1.53$
IRAS 08572+3915	L	$0.0584$	$253.1\pm 17.3$	$13.38\pm 1.84$
IRAS 05189-2524	Sy 2	$0.0428$	$180.2\pm 12.4$	$13.50\pm 1.85$
IRAS 22491-1808	H ii	$0.0778$	$337.0\pm 23.1$	$13.88\pm 1.92$
Mrk 273	Sy 2	$0.0378$	$165.6\pm 11.3$	$14.81\pm 2.03$
IRAS 20414-1651	H ii	$0.0871$	$383.4\pm 26.3$	$15.12\pm 2.18$
Arp 220	L	$0.0181$	$84.1\pm 5.8$	$17.25\pm 2.36$
IRAS 13120-5453	Sy 2	$0.0308$	$137.8\pm 9.5$	$19.20\pm 2.65$
IRAS 12112+0305	L	$0.0733$	$326.3\pm 22.4$	$20.35\pm 2.79$
IRAS 17208-0014	H ii	$0.0428$	$189.3\pm 13.0$	$26.21\pm 3.59$
Mrk 231	Sy 1	$0.0422$	$184.3\pm 12.6$	$34.30\pm 4.70$

Table 2. Table 2: L HCN − vib subscript 𝐿 HCN vib L_{\mathrm{HCN-vib}} and outflow properties.

Name	$L_{\mathrm{HCN}-\mathrm{vib}}$	$L_{\mathrm{HCN}-\mathrm{vib}}/L_{\mathrm{IR}}$	$v_{50}$(abs)a𝑎aa𝑎a$a$Median OH outflow velocities taken from Veilleux et al. (2013) or, for sources not in their sample, from archival data following the same procedure. Sources that have not been observed or where the doublet was detected in emission are indicated with not obs. or emi., respectively. The typical uncertainty on the velocities is $50$ km s-1.	Observatory	References
	(${10}^3{L}_{☉}$)	(${10}^{-8}$)	(km s-1)
$J=3\text{–}2$
NGC 4418	$5.0\pm 1.2$	$3.77\pm 0.71$	$111$	SMA	1
IC 860	$4.2\pm 0.8$	$2.65\pm 0.37$	not obs.	PdBI	2
Zw 049.057	$7.3\pm 2.0$	$3.57\pm 0.76$	$37$	PdBI	2
NGC 7469			emi.	ALMA	3
I Zw 1			emi.	ALMA	3
UGC 5101	$20.5\pm 5.5$	$2.15\pm 0.50$	$-9$	NOEMA	4
IRAS 20551-4250	$2.3\pm 0.7$	$0.21\pm 0.06$	$-381$	ALMA	5
IRAS 15250+3609	b𝑏bb𝑏b$b$Imanishi et al. (2016a) interpret an emission feature at the position of the HCN-vib line as a sub-peak of the nearby HCO+ line. Here, we use this value as an upper limit to the HCN-vib luminosity.		$189$	ALMA	3
IRAS 08572+3915 NW			$-489$	ALMA	3
IRAS 22491-1808 E	$13.2\pm 3.9$	$0.95\pm 0.25$	$99$	ALMA	3
Mrk 273			$-201$	NOEMA	6
IRAS 20414-1651			$-32$	ALMA	3
Arp 220 W	$40.9\pm 7.0$	$2.37\pm 0.25$c𝑐cc𝑐c$c$The $L_{\mathrm{HCN}-\mathrm{vib}}/L_{\mathrm{IR}}$ is calculated using the total $L_{\mathrm{IR}}$ of the system, possibly introducing some bias in systems with multiple nuclei.	$21$	ALMA	7
Arp 220 E	$8.9\pm 1.5$	$0.51\pm 0.05$c𝑐cc𝑐c$c$The $L_{\mathrm{HCN}-\mathrm{vib}}/L_{\mathrm{IR}}$ is calculated using the total $L_{\mathrm{IR}}$ of the system, possibly introducing some bias in systems with multiple nuclei.	$21$	ALMA	7
IRAS 12112+0305 NE	$14.3\pm 4.0$	$0.70\pm 0.17$c𝑐cc𝑐c$c$The $L_{\mathrm{HCN}-\mathrm{vib}}/L_{\mathrm{IR}}$ is calculated using the total $L_{\mathrm{IR}}$ of the system, possibly introducing some bias in systems with multiple nuclei.	$-117$	ALMA	3
Mrk 231	$14.5\pm 3.1$	$0.42\pm 0.07$	$-237$	PdBI	8
$J=4\text{–}3$
NGC 4418	$12.4\pm 2.5$	$9.41\pm 1.25$	$111$	SMA	1
IRAS 12224-0624	$16.9\pm 4.5$	$8.52\pm 1.95$	not obs.	ALMA	9
IRAS 20551-4250	$4.8\pm 1.2$	$0.45\pm 0.09$	$-381$	ALMA	10
IRAS 05189-2524	$8.0\pm 2.7$	$0.59\pm 0.18$	$-387$d𝑑dd𝑑d$d$The median velocity of the lines in IRAS 05189-2524 has been adjusted to correct for the different redshift adopted compared to the one used by Veilleux et al. (2013).	ALMA	9
Arp 220 W	$108.1\pm 19.9$	$6.27\pm 0.77$	$21$	ALMA	7
Arp 220 E	$25.8\pm 4.4$	$1.49\pm 0.15$	$21$	ALMA	7
IRAS 13120-5453			$-195$	ALMA	11
IRAS 17208-0014	$101.9\pm 17.5$	$3.89\pm 0.40$	$51$	ALMA	2

Equations

$$L_{\mathrm{HCN}-\mathrm{vib}}=1.04\times 10^{-3}{S}_{\mathrm{HCN}-\mathrm{vib}}\Delta v{\nu }_{\mathrm{rest}}(1+z{)}^{-1}{D}_{\mathrm{L}}^2,$$

Full text

*Aims. *We aim to investigate if there is a connection between the presence of strong HCN-vib emission and the development of feedback in (ultra) luminous infrared galaxies ((U)LIRGs).

*Methods. *We collected literature and archival data in order to compare the luminosities of rotational lines of HCN-vib, normalized to the total infrared luminosity, to the median velocities of $119$ $\mu$m OH absorption lines, potentially indicating outflows, in a total of 17 (U)LIRGs.

*Results. *The most HCN-vib luminous systems all lack signatures of significant molecular outflows in the far-infrared OH absorption lines. However, at least some of the systems with bright HCN-vib emission do have fast and collimated outflows that can be seen in spectral lines at longer wavelengths, including in millimeter emission lines of CO and HCN (in its vibrational ground state) as well as in radio absorption lines of OH.

*Conclusions. *We conclude that the galaxy nuclei with the highest $L_{\mathrm{HCN-vib}}/L_{\mathrm{IR}}$ do not drive wide-angle outflows detectable using the median velocities of far-infrared OH absorption lines. It is possible that this is due to an orientation effect where sources which are oriented in such a way that their outflows are not along our line of sight also radiate a smaller proportion of their infrared luminosity in our direction. It could also be that massive wide-angle outflows destroy the deeply embedded regions responsible for bright HCN-vib emission, so that the two phenomena cannot coexist. This would strengthen the idea that vibrationally excited HCN traces a heavily obscured stage of evolution before nuclear feedback mechanisms are fully developed.

1 Introduction

Luminous and ultraluminous infrared galaxies ((U)LIRGs) are gas-rich systems that radiate intensely in the infrared portion of the electromagnetic spectrum, with infrared luminosities ($L_{\mathrm{IR}}=L(8\text{--}1000$ $\mathrm{\mu m})$) in excess of $10^{11}L_{\sun}$ and $10^{12}L_{\sun}$, respectively (e.g., Sanders & Mirabel, 1996). These large luminosities are due to dust-reprocessed radiation from intense star formation, an active galactic nucleus (AGN), or both. In the local Universe, (U)LIRGs are relatively rare compared to less luminous objects, but surveys at millimeter and submillimeter wavelengths have shown that they are much more numerous at high redshifts (e.g., Smail et al., 1997; Hughes et al., 1998), indicating that they play an important role in the evolution of galaxies. Two important questions in this context are: how did the supermassive black holes (SMBHs) found in the nuclei of most galaxies grow, and how are they related to the evolution of their host galaxies? From observations, we know that tight relations between the mass of the central SMBH and various properties of the host galaxy exist in large ellipticals (e.g., Magorrian et al., 1998; Ferrarese & Merritt, 2000; Gebhardt et al., 2000; Kormendy & Ho, 2013). It has been suggested that these relations were partly established through a process, mostly taking place in the early Universe, in which galaxies of similar masses collide and merge (see, e.g., Kormendy & Ho, 2013, for a review). In such mergers, large amounts of gas are funneled into the new nucleus, giving rise to intense starbursts, AGN activity, or both. The exact physics of how the black hole-host relations would be established are still obscure, but several suggested galaxy merger scenarios include a starburst dominated phase followed by a phase of obscured AGN activity. Eventually, the AGN becomes strong enough to drive outflows that clear the surrounding material and reveal the central activity (e.g., Sanders et al., 1988a; Hopkins et al., 2006; Urrutia et al., 2008; Simpson et al., 2012). Thus, studies of the most dust-embedded phase of evolution and the early stages of feedback are essential in order to understand how SMBHs grow together with their host galaxies.

However, observations of the most enshrouded galaxy nuclei are hampered by the large amounts of obscuring material that surround them. One solution to this problem is to observe at millimeter and radio wavelengths, where the dust is less optically thick (e.g., Barcos-Muñoz et al., 2015). Some common tracers of obscured galaxy nuclei at these wavelengths are the low-$J$ rotational lines of the HCN and HCO*+* molecules. Recently, however, it has been shown that, in the most obscured systems, even these dense-gas tracers are heavily affected by continuum- and/or self-absorption by cooler foreground gas (e.g., Sakamoto et al., 2009; Aalto et al., 2015b; Martín et al., 2016). A better suited tracer of the most obscured systems is offered by HCN in its first vibrationally excited state, $v_{2}=1$ (hereafter HCN-vib). Strong emission lines from rotational transitions within the $v_{2}=1$ state have been detected in external galaxies at (sub)millimeter wavelengths, first by Sakamoto et al. (2010) and then in many subsequent observations. With energy levels that lie more than $1000$ K above the ground state, HCN-vib is mainly excited by absorption of mid-infrared radiation. However, to be efficiently populated it requires brightness temperatures in excess of $100$ K at $14$ $\mu$m, the wavelength of the transition between the ground and first vibrationally excited states of HCN, translating into a H2 column density larger than $2\times 10^{23}$ cm*-2* for a dust temperature of $100$ K (Aalto et al., 2015b). This makes bright HCN-vib emission an excellent unabsorbed (Martín et al., 2016) tracer of high column density gas with high mid-infrared surface brightness. While the direct mid-infrared radiation from such regions is often hidden from us by the large quantities of obscuring material, the rotational transitions inside the vibrationally excited states occur at millimeter and submillimeter wavelengths where the dust opacity is lower. The low-J rotational lines of HCN-vib are thus useful probes of the nuclear mid-infrared source in heavily obscured objects. Since the first extragalactic (sub)millimeter detection of HCN-vib rotational line emission by Sakamoto et al. (2010), the field has developed rapidly (e.g., Imanishi & Nakanishi, 2013; Aalto et al., 2015a, b; Imanishi et al., 2016a, b, 2018; Martín et al., 2016), notably thanks to the Atacama Large Millimeter/submillimeter Array (ALMA).

In their study of HCN-vib in obscured galaxies, Aalto et al. (2015b) found that all galaxies with detected HCN-vib emission show evidence of inflows, outflows, or both. In addition, they found a tentative trend that galaxies with fast molecular outflows have fainter HCN-vib emission relative to their total infrared luminosity. The trend is especially striking when comparing the strength of the HCN-vib emission to the velocity of the wide-angle OH outflows studied by, for example, Veilleux et al. (2013). Based on a relatively small sample of nine galaxies, Aalto et al. (2015b) tentatively suggest that strong HCN-vib emission is connected to a rapid phase of nuclear growth that occurs right before the onset of strong feedback.

In this paper, we present the results of a study employing a sample of $19$ galaxies including data from the literature as well as previously unpublished data, in order to further investigate the connection, or lack thereof, between strong HCN-vib emission and molecular outflows. The sample used in this study is described in Sect. 2 and in Sect. 3 we present the results which are then discussed in Sect. 4. Finally, the conclusions that we draw from the analysis are summarized in Sect. 5. Throughout the paper, a $H_{0}=73$ km s*-1* Mpc*-1*, $\Omega_{\mathrm{m}}=0.27$, and $\Omega_{\Lambda}=0.73$ cosmology is adopted.

2 Sample

We have searched for (U)LIRGs with existing ALMA observations of HCN-vib that are either already published or that have the data publicly available in the the ALMA science archive111The ALMA science archive is available at https://almascience.nrao.edu/alma-data/archive.. In addition, we have included published HCN-vib observations taken with the Plateau de Bure Interferometer (PdBI), the Northern Extended Millimeter Array (NOEMA), or the Submillimeter Array (SMA). We have not included non-detections, unless they have existing observations of the OH doublet at $119$ $\mu$m, which we use to determine the outflow velocity. Non-detections have also been excluded when a meaningful upper limit to the HCN-vib flux could not be determined, either due to low sensitivity spectra or due to significant flux from the HCO*+* line which is separated from HCN-vib by ${\sim}400$ km s*-1*. This is a possible source of bias, as HCN-vib emission may be hidden by the HCO*+* line in sources with outflowing gas. In practice, most sources that have been excluded either do not have any outflow signatures in the far-infrared OH lines that we use to trace outflows (see below), or have not been observed in these lines. The two exceptions are IRAS 19254-7245 (Imanishi et al., 2016a) and NGC 6240 (ALMA project 2015.1.01448.S, PI: R. Tunnard), in which the median velocities of the OH $119$ $\mu$m lines are ${\sim}-250$ and ${\sim}-200$ km s*-1*, respectively. In total, $10$ sources have been excluded, most of them due to lack of OH measurements.

For consistency of the sample, we ensure that outflow velocities were measured using the same common tracers and following the same procedures. All sources with HCN-vib observations have therefore been checked for observations of the far-infrared OH doublet at $119$ $\mu$m with the Herschel Space Observatory (Pilbratt et al., 2010) as this is a known tracer of molecular outflows on subkiloparsec scales, which has been observed in a large number of (U)LIRGs (e.g., Fischer et al., 2010; Veilleux et al., 2013; González-Alfonso et al., 2017b). The OH molecule has high abundances in photodissociation regions as well as in X-ray dominated regions (e.g., Maloney et al., 1996; Goicoechea et al., 2011; Meijerink et al., 2011). Unlike other common outflow tracers like the millimeter rotational transitions of CO and HCN, the ground state OH $119$ $\mu$m doublet is often detected in absorption towards the far-infrared background (e.g., Veilleux et al., 2013), providing an unambiguous tracer of the gas motion. The transitions between excited states (for example the doublets at $65$ and $84$ $\mu$m) are generally weaker and the line wings are not always detected (González-Alfonso et al., 2017b). Recently, the $119$ $\mu$m doublet was also used to detect a fast outflow ($\lesssim 800$ km s*-1*) in a galaxy at a redshift of $5.3$ (Spilker et al., 2018). Systematic searches of outflow signatures in the far-infrared OH lines have revealed evidence of molecular outflows in approximately two thirds of the observed (U)LIRG samples (Veilleux et al., 2013; Spoon et al., 2013). From this detection rate, Veilleux et al. (2013) inferred a wide average opening angle of ${\sim}145\degr$, assuming that all objects in their sample have outflows. In fact, based on radiative transfer models as well as spatially resolved observations of two outflows, González-Alfonso et al. (2017b) argue that the $119$ $\mu$m OH absorptions are primarily sensitive to wide-angle outflows. On the other hand, unambiguous signatures of inflows are only found in one tenth of the sources in the sample of Veilleux et al. (2013), suggesting that OH inflows are either rare or have, on average, smaller opening angles.

In particular, the 43 sources in the far-infrared OH sample of Veilleux et al. (2013) have been checked for observations of the (sub)millimeter HCN-vib $J=3\text{--}2$ or $J=4\text{--}3$ lines. Fourteen of these had HCN-vib detections or data that allowed for meaningful upper limits to be estimated. One of them, IRAS 05189-2524 (see Appendix B), has previously unpublished data in the ALMA science archive while data for the remaining thirteen were taken from Sakamoto et al. (2010); Aalto et al. (2015a, b); Imanishi et al. (2016a, b); Martín et al. (2016); Privon et al. (2017); Aladro et al. (2018); and König et al. (in prep.). In addition, five sources not included in the sample of Veilleux et al. (2013) were found to have usable observations of HCN-vib. One, IRAS 12224-0624 (see Appendix B), has previously unpublished data in the ALMA science archive while the other four were taken from Aalto et al. (2015b) and Imanishi et al. (2016a). Three of these five sources, Zw 049.057, IRAS 20414-1651, and NGC 7469, also have existing observations of the OH $119$ $\mu$m doublet (see Appendix C).

In total, 19 (U)LIRGs with existing HCN-vib observations were found. Of these, 17 also have observations of the OH $119$ $\mu$m doublet. The final sample, with the adopted redshift and distance of each source, is presented in Table 1.

3 Results

Vibrationally excited HCN luminosities, $L_{\mathrm{HCN-vib}}/L_{\mathrm{IR}}$ ratios, and OH median velocities are presented in Table 2. We adopt the $L_{\mathrm{HCN-vib}}/L_{\mathrm{IR}}$ ratio as a parameter to describe the strength of HCN-vib in a normalized fashion. However, we are aware that some part of the total $L_{\mathrm{IR}}$ may be unrelated to the dusty nucleus, and instead come from, for example, an extended starburst. In particular, in systems with multiple nuclei, for example Arp 220 and IRAS 12112+0305, $L_{\mathrm{IR}}$ will have contributions from both nuclei. In our sample, we are not aware of any sources were these effects would affect the $L_{\mathrm{HCN-vib}}/L_{\mathrm{IR}}$ ratio enough to significantly change our results. We use the median velocity of the OH lines instead of the terminal outflow velocity as the former provides more robust values (Veilleux et al., 2013). Velocities are only given for those sources with absorption in the OH $119$ $\mu$m doublet, as this means that the gas is in the foreground and that a positive or negative velocity shift can be interpreted as evidence of gas moving towards or away from the background nucleus, respectively. We note that this does not necessarily indicate in- or outflowing gas in all cases as, for example, interactions between two nuclei can also affect the OH kinematics. In addition, it is important to take the uncertainties in the velocity determination into account. For example, Veilleux et al. (2013) define an outflow as having an OH absorption feature with a median velocity more negative than $-50$ km s*-1*, which is the typical uncertainty on the velocities. The requirement to detect the OH doublet in absorption is a possible source of bias as this requires a high enough column density of OH in front of a strong enough far-infrared background source. However, we note that all galaxies with an HCN-vib detection, if observed, have also been detected in absorption in the OH $119$ $\mu$m doublet. Therefore, no sources with a combination of high HCN-vib luminosity and a fast outflow have been missed due to this effect.

The line luminosities presented in Table 2 have been calculated following Eq. (1) in Solomon & Vanden Bout (2005), applied to HCN-vib:

[TABLE]

where $L_{\mathrm{HCN-vib}}$ is the HCN-vib luminosity measured in $L_{\odot}$, $S_{\mathrm{HCN-vib}}\,\Delta v$ is the velocity integrated flux in Jy km s*-1*, $\nu_{\mathrm{rest}}$ is the rest frequency in GHz, and $D_{\mathrm{L}}$ is the luminosity distance in Mpc. Due to differences in the adopted redshifts and distances, these values differ slightly from those given in the original reference for some of the sources. In case of non-detections, we have included the $3\sigma$ upper limit to the line luminosity. In the case of IRAS 15250+3609, the HCN-vib line is likely blended with a potential outflow signature in the nearby HCO*+* line. Here, we have attributed the entire emission feature to the HCN-vib line, and the value given should be considered an upper limit to the HCN-vib luminosity. A short discussion of this source as well as descriptions of the analysis of previously unpublished HCN-vib observations and new OH outflow measurements can be found in Appendices A, B, and C.

In Fig. 1, the OH absorption median velocities are plotted against the luminosities of the HCN-vib line, normalized to the total infrared luminosity of each galaxy. When available, the $J=3\text{--}2$ line has been chosen. In the four sources where both transitions have been observed, the luminosity ratio between the $J=4\text{--}3$ and $J=3\text{--}2$ lines lies in the range $2\text{--}3$, with a mean of $2.5$. For sources where only the $J=4\text{--}3$ line has been observed, we have therefore divided its luminosity by $2.5$ and included it in the plot with an extra $20\%$ uncertainty added in quadrature. The two galaxies with only emission in the $119$ $\mu$m doublet, I Zw 1 and NGC 7469, are not included in the plot, but we note that neither of them have an HCN-vib detection.

In the plot in Fig. 1 we see that, in our sample, the sources with high $L_{\mathrm{HCN-vib}}/L_{\mathrm{IR}}$ ratios, in excess of $10^{-8}$, all lack fast outflows as traced by the median velocity of the OH $119$ $\mu$m doublet. However, galaxies without fast outflows do not necessarily have high $L_{\mathrm{HCN-vib}}/L_{\mathrm{IR}}$ ratios. The lack of sources in the upper left portion of the diagram is likely due to selection effects; that area should be populated by less obscured sources of lower luminosity that are not able to drive outflows or provide the necessary conditions for efficient HCN-vib excitation.

4 Discussion

Our comparison of HCN-vib line luminosities and OH $119$ $\mu$m absorption line velocities in nearby (U)LIRGs shows a trend of positive, or low negative, velocities, indicating inflows or slow outflows, in sources with bright HCN-vib emission relative to the total infrared luminosity. This might indicate that regions with the high mid-infrared brightness temperatures required for efficient population of HCN-vib are not affected by inflowing gas, but are destroyed by, or cannot form in the presence of, strong outflows. In this section, we discuss possible reasons for the lack of outflow signatures in the OH $119$ $\mu$m absorptions in HCN-vib luminous sources, and how the HCN-vib lines relate to other tracers of heavily obscured regions.

4.1 Why do the HCN-vib luminous galaxies lack outflow signatures in the far-infrared OH lines?

There are several scenarios that could explain the lack of outflow signatures in the galaxies that have bright HCN-vib emission. One is that there are no outflows, another that the outflows are hidden from detection using the median velocity of the OH $119$ $\mu$m absorption lines, and a third is that the distribution in Fig. 1 is due to orientation effects. These scenarios are discussed in the following sections. We also briefly discuss a possible evolutionary scenario in which the HCN-vib luminous galaxies are in a pre-feedback phase. Admittedly, our sample is quite heterogeneous in that it contains galaxies of different merger types, in different merger stages, of different optical classifications, and with infrared luminosities spanning more than an order of magnitude. It is therefore possible that there are several independent reasons for the lack of far infrared outflow signatures in HCN-vib luminous galaxies.

4.1.1 No outflows

If there are no outflows, it might be due to the HCN-vib luminous galaxies being intrinsically different from the galaxies with weaker HCN-vib emission. For example, Veilleux et al. (2013) find a tendency for systems with dominant AGN to have faster OH outflows; it might then be that the HCN-vib luminous systems, which do not show fast OH outflows, are instead starburst dominated. Indeed, in the two sources with the highest absolute HCN-vib luminosities, IRAS 17208-0014 and Arp 220, less than $10\%$ of the bolometric luminosities are estimated to come from AGN activity, based on the $15$ to $30$ $\mu$m continuum ratio (Veilleux et al., 2013). However, this diagnostic can be misleading in galaxies where the nuclei are optically thick into the far infrared. Furthermore, at least two other sources with strong HCN-vib emission, NGC 4418 and UGC 5101, have AGN fractions $\gtrsim 50\%$. While the uncertainty on the AGN fraction determined in this way is estimated to be $20\%$ on average, it is likely much higher in the strongly buried sources discussed here.

Another possible explanation for a real lack of outflows in some sources is that we are witnessing a pre-feedback phase in young systems that will later evolve into objects with weak HCN-vib emission and fast outflows. This possibility is discussed further in Sect. 4.1.4. As discussed by González-Alfonso et al. (2017b), the lack of outflows could also be due to the fact that, in the most extremely buried sources, it is difficult to find paths with more moderate columns that can be efficiently accelerated to high velocities.

Interestingly, as Aalto et al. (2015b) notes, fast molecular outflows have been found, with other tracers and at longer wavelengths, in some of the HCN-vib luminous systems. For example, a compact $v>800$ km s*-1* CO $J=2\text{--}1$ outflow has been found in the ULIRG IRAS 17208-0014 (García-Burillo et al., 2015) and an equally fast, compact, and collimated, HCN $J=1\text{--}0$ outflow has been detected in the ULIRG Arp 220 (Barcos-Muñoz et al., 2018). Possible outflow signatures are also found in Zw 049.057 (Falstad et al., 2018) as well as in NGC 4418 and IRAS 22491-1808 (Fluetsch et al., 2018), although at lower velocities. If we also consider the fact that UGC 5101 has faint high-velocity wings in its OH $119$ $\mu$m absorption lines (Veilleux et al., 2013), it turns out that all of the six most HCN-vib luminous sources have signatures of molecular outflows.

4.1.2 Obscured, episodic, or collimated outflows

For the HCN-vib bright systems that do have outflows, there must be a reason why we do not see them in the far-infrared OH absorption lines. One possibility is that these objects have extreme optical depths in the far-infrared dust continuum, so that much of the outflowing gas cannot be seen due to obscuration by the dust. In at least some of the sources, high optical depths have indeed been inferred from radiative transfer modeling of Herschel observations (e.g., González-Alfonso et al., 2012; Falstad et al., 2015). If this is combined with young outflows that are still very compact, any signatures of them may be completely hidden in the far-infrared. In such a situation, inflows on larger scales, for example as the one seen in Arp 299A (Falstad et al., 2017), may still be detectable. In fact, at least three of the five most obscured sources (Arp 220, IRAS 17208-0014, and Zw 049.057; Baan et al., 1989; Falstad et al., 2018) have outflows that are seen in the OH lines at centimeter wavelengths but not in the far-infrared. In Zw 049.057, infrared and optical images reveal multiple clouds in a polar dust structure, which might suggest that the outflow in this galaxy is not steady in time (Gallagher et al. in prep.). It is thus possible that, in some galaxies, we are witnessing an episodic process where the nuclear region is alternating between feedback and accretion dominated states.

Looking at the spectra presented by Veilleux et al. (2013), UGC 5101 and IRAS 17208-0014 even exhibit weak blueshifted line wings in the OH $119$ $\mu$m absorption lines. This indicates the presence of gas moving towards us, but as the absorptions at negative velocities are shallow, they do not have a strong effect on the median velocities of the lines. As discussed by González-Alfonso et al. (2017b), this may suggest that we are witnessing collimated outflows in these sources, as opposed to the wide-angle outflows that the OH $119$ $\mu$m lines seem to be primarily sensitive to. With all this in mind, it seems that the regions responsible for the bright HCN-vib emission are not destroyed by (fast) outflows in general, but rather by the kind of wide-angle outflows from the nuclear infrared-emitting region that are able to shift the median velocity of the OH $119$ $\mu$m ground state lines by several hundreds of km s*-1*.

Turning to the two sources that show pure emission in the OH $119$ $\mu$m doublet, I Zw 1 and NGC 7469, we note that they lack HCN-vib detections and that the upper limits to their $L_{\mathrm{HCN-vib}}/L_{\mathrm{IR}}$ ratios are among the lowest in our sample. Furthermore, NGC 7469 has also been observed in the OH $65$ $\mu$m doublet, another tracer of obscured nuclei, but no absorption was detected (González-Alfonso et al., 2015). This is consistent with the results of Veilleux et al. (2013) that the sources with pure OH emission are the ones where an AGN dominates the luminosity and that these objects represent a phase where the feedback has subsided after clearing a path through the dusty surroundings.

4.1.3 Orientation effects

An alternative explanation to the distribution of galaxies in Fig. 1 is that it is produced by a simple orientation effect. For example, using the OH $119$ $\mu$m doublet, Veilleux et al. (2013) found unambiguous evidence of outflows in $70\%$ of their sources, a detection rate which is consistent with all sources having molecular outflows, with an average opening angle of ${\sim}145\degr$. The sources with no outflow signatures would then be those that are oriented approximately edge-on, with outflows perpendicular to our line of sight. It is harder to explain why these would also be the sources with bright HCN-vib emission, as the obscuration in a dusty edge-on disk would rather have the effect of lowering the visible HCN-vib flux, unless the line is masing as seen towards some circumstellar envelopes (e.g., Lucas & Cernicharo, 1989; Bieging, 2001; Menten et al., 2018).

However, the comparison so far has been based on the HCN-vib luminosity divided by the total infrared luminosity which, due to the higher dust opacity in the infrared, is more affected by orientation effects. The infrared luminosity that we see will thus depend on the orientation, with higher values for face-on galaxies. This effect may indeed be significant; for example, Efstathiou et al. (2014) estimated that the true infrared luminosity of IRAS 08572+3915 may be as high as $1.1\times 10^{13}$ L*⊙, more than five times higher than the value inferred assuming the luminosity is isotropic. Another possible example is Arp 220 where the total infrared luminosity estimated from the IRAS fluxes is ${\sim}1.7\times 10^{12}$ L⊙* while Sakamoto et al. (2017) estimate the bolometric luminosity from the western nucleus alone to be ${\sim}3\times 10^{12}$ L*⊙*.

If we do not divide $L_{\mathrm{HCN-vib}}$ by the total infrared luminosity, we see that Mrk 231 has an HCN-vib luminosity comparable to those of the HCN-vib bright galaxies. However, it is still less luminous than in the ULIRGs Arp 220 W and IRAS 17208-0014, and only a factor of two or three brighter than in the LIRGs Zw 049.057, IC 860, and NGC 4418. Furthermore, none of the other sources with $v_{50}\lesssim-200$ km s*-1* have strong HCN-vib emission. Clearly, a larger sample will be required to investigate this issue further. If the distribution of sources in Fig. 1 is indeed due to an orientation effect, more sources with both strong HCN-vib emission and outflow signatures in the far-IR OH lines should show up in an extended sample.

4.1.4 Evolution

Following Aalto et al. (2015b), we present a possible evolutionary scenario in which bright HCN-vib emission is tracing extremely obscured nuclei in a phase that occurs before the nuclear feedback is able to drive wide-angle outflows, or at least before such outflows can be observed in the far-infrared OH lines. In this scenario, the deeply embedded regions which provide the necessary conditions for HCN-vib excitation can coexist with fast collimated outflows, but are disrupted once nuclear wide-angle outflows have developed. A schematic view of one possible version of the scenario is presented in Fig. 2. Here, the most obscured sources consist of a mid-infrared core, responsible for the HCN-vib emission, surrounded by an obscuring layer of cooler dust that accounts for the far-infrared emission. Outflows in these objects are collimated and often so compact that they are still embedded in the outer layer of dust, and thus obscured from view in the far-infrared. In the next phase, the outflows have broken through the cooler dust layer and widened considerably, making them clearly detectable in most galaxies. As gas and dust is transported from the regions around the nucleus, the conditions in the mid-infrared core change, becoming less favorable for the excitation of HCN-vib. Finally, when a path has been cleared through the surrounding dust, OH is no longer detectable in absorption and the conditions for HCN-vib excitation in the core are no longer met.

A similar evolutionary sequence has been suggested by González-Alfonso et al. (2017a), and we note that, apart from UGC 5101, the HCN-vib luminous sources from our sample coincide with the sources in their extremely buried pre-feedback phase. It should be noted that, even if an evolutionary scenario like this applies in some cases, all sources in our sample may not be on the same evolutionary track. For example, a lot of nuclear power is required in order to push large columns of gas to high velocities, and LIRGs like NGC 4418 and Zw 049.057 may therefore never develop powerful wide-angle outflows detectable in the far-IR OH lines. Instead, they might evolve along the horizontal axis in Fig. 1.

Finally, an interesting comparison can be made with the highly collimated molecular outflows that are found around low-mass protostars in the earliest (class 0) stages of their formation (e.g., Konigl & Pudritz, 2000; Codella et al., 2014). With time, gas at larger and larger angles from the outflow axis may be entrained, eventually sweeping away most of the envelope around the star (e.g., Arce & Sargent, 2006). Furthermore, vibrationally excited HCN has been detected toward molecular hot cores around massive protostars using the direct $l$-type transitions, that is transitions between the $l$-type levels inside a rotational state, at cm-wavelengths (e.g., Rolffs et al., 2011).

4.2 HCN-vib compared to other tracers of obscured nuclear regions.

As discussed in Aalto et al. (2015b), a prerequisite for the excitation of HCN-vib is the presence of warm and dusty regions, making it a good tracer of deeply buried nuclei. Following this logic, our results indicate that the most obscured nuclei lack evolved, wide-angle outflows. Interestingly, the results of González-Alfonso et al. (2017b), who use the equivalent width of the OH $65$ $\mu$m doublet as a measure of the obscuration, suggest that the fastest outflows arise in some of the most obscured nuclei. They do however point out that sources with high equivalent width in OH $65$ $\mu$m do not necessarily have fast outflows. By comparing our Fig. 1 with their Fig. 4, we see that five of the six most HCN-vib luminous sources that are included in both samples also belong to the group of sources with high equivalent width but low outflow velocity, indicating that they are obscured also when using the OH $65$ $\mu$m equivalent width as a measure. On the other hand, five of the sources with low HCN-vib luminosity are also found among the sources with high OH $65$ $\mu$m equivalent width, indicating that they also contain obscured nuclei. It appears that the two measures trace different parts or physics of the galaxy nuclei. Indeed, González-Alfonso et al. (2017b) state that high equivalent widths require dust temperatures in excess of $60$ K, while dust temperatures above $100$ K are required for bright HCN-vib emission (Aalto et al., 2015b). It is thus possible that bright HCN-vib emission exclusively traces more obscured parts of the nuclei, where the dust is hotter.

In a similar manner, Spoon et al. (2013) also find that the fastest outflows are found in those sources that are still deeply embedded as indicated by strong mid-infrared silicate absorptions. However, as already noted by Aalto et al. (2015b), González-Alfonso et al. (2015) find that silicate absorptions are biased toward relatively unobscured mid-infrared emitting regions, again indicating that the HCN-vib is actually tracing a more extreme form of obscuration.

5 Conclusions

We explore a possible correlation between the strength of molecular outflows in the far-infrared OH $119$ $\mu$m lines and the luminosity of rotational lines of vibrationally excited HCN. A simple comparison of HCN-vib line luminosities, normalized to the total infrared luminosity of the host galaxy, and the median velocities of OH $119$ $\mu$m absorption lines shows that galaxies with unusually bright HCN-vib emission tend to lack fast molecular outflows, but that galaxies without fast outflows do not necessarily have bright HCN-vib emission. This may be an orientation effect or something that reflects a true difference between the sources, for example that the most obscured sources cannot drive wide-angle outflows or that their outflows are young and have not yet dispersed the nuclear gas and dust concentrations. Following Aalto et al. (2015b), we present an evolutionary sequence, that may apply to some of the sources, in which bright HCN-vib emission is tracing extremely obscured nuclei in a phase that occurs before the onset of wide-angle outflows. Once these massive outflows have been launched, they quickly disrupt the deeply embedded regions responsible for the bright HCN-vib emission. We note, however, that more studies are needed in order to extend the sample and explore the relation further. As the Herschel Space Observatory, which was used for the OH outflow measurements, is no longer operational, such studies should concentrate on observations of HCN-vib in galaxies which have already been observed in the OH $119$ $\mu$m doublet. Another possibility is to search for hidden outflows using the radio lines of OH, either in its ground state (Baan et al., 1989) or in rotationally excited states (Falstad et al., 2018).

Acknowledgements.

This paper makes use of the following ALMA data: ADS/JAO.ALMA#2015.1.00708.S, ADS/JAO.ALMA#2016.1.00140.S. ALMA is a partnership of ESO (representing its member states), NSF (USA) and NINS (Japan), together with NRC (Canada), MOST and ASIAA (Taiwan), and KASI (Republic of Korea), in cooperation with the Republic of Chile. The Joint ALMA Observatory is operated by ESO, AUI/NRAO and NAOJ. This research has made use of the NASA/IPAC Extragalactic Database (NED), which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration.

Appendix A Vibrationally excited HCN in IRAS 15250+3609

The ULIRG IRAS 15250+3609 was observed in the HCO*+* $J=3\text{--}2$ line by Imanishi et al. (2016a). They interpreted a peak close to the expected position of the HCN-vib line as an outflow signature in the HCO*+* line. Their interpretation is supported by the fact that similar features are seen in the HCN $J=3\text{--}2$ and $J=4\text{--}3$ lines (Imanishi et al., 2016a, 2018). However, relative to the main lines, the sub-peak close to the HCO*+* line is stronger than the one close to the HCN line. This could be due to differing abundances or excitation in the outflowing gas but it could also be a contribution from HCN-vib. Furthermore, IRAS 15250+3609 has a high equivalent width in the OH $65$ $\mu$m doublet (González-Alfonso et al., 2015), indicating that the galaxy is highly obscured, as well as an HCN absorption at $14$ $\mu$m (Lahuis et al., 2007). Here, we have attributed the feature as completely due to HCN-vib emission, but it is likely that it is indeed blended with an HCO*+* outflow signature and the value given here should be considered an upper limit to the HCN-vib luminosity.

Appendix B Previously unpublished HCN-vib detections

We found two sources with previously unpublished detections of HCN-vib in the ALMA science archive, IRAS 12224-0624 (project 2015.1.00708.S, PI: L. Armus), and IRAS 05189-2524 (project 2016.1.00140.S, PI: D. Iono). The observations of IRAS 05189-2524 were conducted during two runs on 2016 November 12 and 15 with a total of $42$ antennas with baseline lengths ranging between $15.1$ m and $1.0$ km. A spectral window centered at $341.8$ GHz covered a bandwidth of $1.875$ GHz (${\sim}1650$ km s*-1* at the frequency of HCN-vib $J=4\text{--}3$), with a frequency resolution of $3.9$ MHz. During both runs J0522-3627 was used as bandpass and flux calibrator, and J0457-2324 was used as phase calibrator. The total on-source time of the observations was $3340$ s and the final sensitivity achieved was $0.5$ mJy beam*-1* per $20$ km s*-1* (${\sim}24$ MHz) channel. For the imaging we used Briggs weighting with a robustness factor of 0.5. The resulting beam size is $0.29\arcsec\,\times\,0.22\arcsec$ (PA ${\sim}63\degr$).

The observations of IRAS 12224-0624 were conducted during a single run on 2016 April 27 with a total of $38$ antennas with baseline lengths ranging between $15.1$ m and $452.8$ m. A spectral window centered at $347.4$ GHz covered a bandwidth of $1.875$ GHz (${\sim}1600$ km s*-1* at the frequency of HCN-vib $J=4\text{--}3$), with a frequency resolution of $3.9$ MHz. The calibrators used were J1256-0547, which was used for bandpass and flux calibration, and J1246-0730, which was used for phase calibration. The total on-source time of the observations was $395$ s and the final sensitivity achieved was $1.5$ mJy beam*-1* per $20$ km s*-1* (${\sim}24$ MHz) channel. For the imaging we used Briggs weighting with a robustness factor of 0.5. The resulting beam size is $0.51\arcsec\,\times\,0.44\arcsec$ (PA ${\sim}82\degr$).

For both sources, the Common Astronomy Software Applications (CASA; McMullin et al., 2007) package was used to reapply the calibration and perform imaging. The line fluxes used to calculate the HCN-vib luminosities were then found by fitting Gaussians to the HCO*+* and HCN-vib lines in spectra that were spatially integrated over the sources. These spectra are presented in Fig. 3. We note that the wavelengths of the HCO*+* and HCN-vib lines in IRAS 05189-2524 are offset from those expected when using the redshift of $0.0426$ adopted by Veilleux et al. (2013). They are however consistent with the redshift of ${\sim}0.0428$ found from the CO $J=1\text{--}0$ observations reported by Sanders et al. (1991). Using this redshift instead of the one adopted by Veilleux et al. (2013) to determine the median OH absorption velocity results in an extra velocity shift of $-45$ km s*-1*.

Appendix C New OH outflow measurements

Three of the sources with existing HCN-vib observations also had observations of the OH doublet at $119$ $\mu$m taken with the Photodetector Array Camera and Spectrometer (PACS; Poglitsch et al., 2010) on Herschel but were not included in the sample of Veilleux et al. (2013). The observations of Zw 049.057 (OBSID: 1342248368, PI: E. González-Alfonso) were conducted on 2012 July 20 for a duration of $1478$ s, the observations of IRAS 20414-1651 (OBSID: 1342217908, PI: D. Farrah) were conducted on 2011 April 5 for a duration of $8879$ s, and the observations of NGC 7469 (OBSID: 1342235840, PI: E. González-Alfonso) were conducted on 2011 December 31 for a duration of $1086$ s. All observations were performed in high spectral sampling, range spectroscopy mode. The observations had been processed with version 14.2 of the standard pipeline and were not in need of reprocessing. In all sources, the nuclear far-IR emission is spatially unresolved in the central $9.4$″ (${\sim}2$ kpc) spaxel of the PACS $5$ x $5$ spaxel array. As the central PACS spaxel is smaller than the point spread function of the spectrometer, the spectrum was extracted using the point source correction task in the Herschel interactive processing environment (HIPE; Ott, 2010) version 14.0.1. Before analyzing the absorption lines, polynomials of order two were fitted to the continuum and then subtracted from the spectra. For consistency, the profiles of the $119$ $\mu$m OH doublets were modeled using the same procedure as in Veilleux et al. (2013). Each line was fitted with two Gaussian components characterized by their amplitude, position, and width. The separation between the two lines of the doublet was fixed at $0.208$ $\mu$m in the rest frame and the amplitude and width of the two lines were the same for each component. The median velocities of the absorptions ($v_{\mathrm{50}}$(abs)) were then determined from these fits. The fitting procedure was carried out using the spectroscopic analysis toolkit PySpecKit (Ginsburg & Mirocha, 2011) and the continuum subtracted spectra with the fits overplotted are presented in Fig. 4.