Free Access
Issue
A&A
Volume 541, May 2012
Article Number L7
Number of page(s) 4
Section Letters
DOI https://doi.org/10.1051/0004-6361/201219229
Published online 11 May 2012

© ESO, 2012

1. Introduction

Through jets, winds, and radiation pressure, active galactic nuclei (AGN) can affect the collapse of their surrounding molecular gas and the formation of new stars in their host galaxies. While simulations attribute a considerable role to AGN feedback in shaping galaxy properties (e.g., Croton et al. 2006; Sijacki et al. 2007; Booth & Schaye 2009; Hopkins & Elvis 2010; Debuhr et al. 2012), the question remains open from an observational point of view. Do the AGN-driven outflows that suppress star formation occur frequently enough to affect the observed luminosity and mass functions of galaxies? To answer this question, we not only need to identify objects with massive, AGN-driven outflows of molecular gas, but also to take into account the outflow effects on the different phases of this gas.

The various phases of an AGN-driven molecular gas outflow have been studied in a nearby ultraluminous infrared galaxy (ULIRG), Mrk231. Plateau de Bure interferometric observations showed that the outflowing CO, with a corresponding  H2 mass of 6 × 108M and a flow rate of 700 M yr-1, is capable of suppressing star formation (Feruglio et al. 2010). Aalto et al. (2012) discovered the outflow’s dense (≳104 cm-3) component from broad HCN(1−0) profile wings. On the basis of profile fitting of the 79 μm and 199 μm OH lines seen with Herschel (Fischer et al. 2010), the outflow contains  ≥7 × 108M of molecules. In total, more than 109M of gas can be in the wind.

Another straightforward and quantitative comparison that has yet to be performed is that between the warm and the cold  H2 gas mass in the outflow vs. in the ambient interstellar medium (ISM). A source that is suitable for this analysis is 4C 12.50. It is the only source known to have a massive outflow of gas of few hundred Kelvin, as inferred from its mid-infrared  H2 rotational line profiles (Dasyra & Combes 2011). In this letter, we present evidence of cold molecules in its outflow, and we evaluate the relative fractions of warm and cold gas that the AGN feedback kinematically distorts. We adopt H0 = 70  km s-1 Mpc-1, ΩM = 0.3, and ΩΛ = 0.7 throughout.

thumbnail Fig. 1

Left: FTS and WILMA spectra of 12CO(3−2) in 4C 12.50, binned into 62  km s-1 channels and shifted by a constant c0, when needed. Right: flux-calibrated line profile of 12CO(3−2) , averaged over the WILMA and the FTS data to be least affected by potential artifacts. Gaussian functions with parameters that most closely fit the emission component at the systemic velocity and the absorption component at −950 km s-1 are shown in blue and cyan, respectively. The inverse  H i absorption that is seen in the radio data of 4C 12.50 (Morganti et al. 2004) is overplotted in orange for an arbitrary scale and continuum level.

Open with DEXTER

2. Data acquisition and reduction

The observations were carried out with the 30 m telescope of the Institut de Radioastronomie Millimétrique on December 21–22, 2011, and January 1, 2012 (program 235–11). For the 12CO(1−0) observations, EMIR was tuned to 102.700 GHz. For the 12CO(3−2) observations, the receivers were tuned to 308.086 GHz during the first observing run and to 308.600 GHz during the second observing run, to ensure that the observed line properties were unaffected by possible standing waves, atmospheric line residuals, or bad channels. Parallel to the CO observations, we carried out HCN and HCO+ (2−1) observations with a common tuning of 158.427 GHz. Both the FTS and WILMA backends, with corresponding resolutions of 0.2 MHz and 2 MHz, were simultaneously used to allow consistency checks. The inner and outer parts of the FTS were used, enabling us to better sample the continuum near each line. The wobbler switching mode with a throw of 80′′ and 0.65 s per phase was used to rapidly sample the continuum. The telescope pointing accuracy was 2′′, and the system temperature varied from 310 K to 580 K at 1 mm.

To reduce the 1 mm data, we removed bad channels from all individual scans. The three parts of each FTS scan, corresponding to the output of a different spectrometer, were fitted with linear baselines and brought to the same continuum level, i.e., to the median value of the full scan. This step was performed to ensure that there is no continuum discontinuity between the three parts that could introduce artificial features into the combined spectrum. We averaged the individual scans of each frequency tuning and backend, discarding scans with unstable baselines. The inner and the outer FTS spectra were then stitched together. For each backend, the spectra of the two frequency tunings were brought to a common reference frequency of 308.245 GHz and averaged. The final spectra are shown in Fig. 1. Their total on-source integration time is 2.7 h.

At 3 mm, we averaged all useful WILMA or FTS scans, after shifting them to a reference frequency of 102.756 GHz and removing bad channels from them. In contrast to the 1 mm data, the 3 mm data had signatures of standing waves owing to the strong synchrotron continuum radiation, which increases with λ. We iteratively fitted the baseline of the WILMA data with sinusoidal functions of periods that are integer multiples of a fixed period, representing different harmonics of the same standing wave. The best-fit solution that was indicated by the CLASS baseline routine corresponded to functions with frequencies of 0.28, 0.84, 1.68, and 2.52 GHz, and an amplitude of 1 mK. These functions were removed from both the WILMA spectrum and the central part of the inner FTS spectrum. The final 12CO(1−0) spectra, with a total on-source integration time of 1.2 h, are shown in Fig. 2. A similar approach was taken for the 2 mm data. A main-beam-temperature-to-flux conversion factor of 5.0 Jy K-1 was used to flux calibrate the final spectra at all wavelengths.

thumbnail Fig. 2

WILMA (left) and FTS (right) 12CO(1−0) spectrum of 4C 12.50, binned to 93  km s-1 channels. The dashed histogram corresponds to the average spectrum prior to the removal of sinusoidal baselines, which represent the harmonics of a standing wave (shown in orange). The final spectrum is shown as a solid, filled histogram.

Open with DEXTER

3. Results: CO in emission and in absorption

The 12CO(3−2) profile of 4C 12.50 consists of an emission component at the galaxy systemic velocity, i.e., at z = 0.1218, and of an absorption feature at approximately −1000 km s-1 from it. The CO(2 → 3) absorption is seen in all four frequency tuning and backend combinations, indicating that the feature is intrinsic to the source. Its signal-to-noise ratio (S/N) is  ~6 in both the WILMA and the FTS data, which have a corresponding root-mean-square noise of 1.7 mK and 1.8 mK, respectively. The outer FTS data provide further evidence that the absorption is real by ruling out the existence of standing waves that could create a feature of this depth. The observed optical depth at the center of the line, τobs,0, is 0.22 (±0.05), with an absorption minimum at 30 (±2) mJy below the local 150 (±20) mJy continuum. Averaging the best-fit Gaussian parameters of the data for each backend indicates that the minimum is at −950 (±90) km s-1. The CO absorption profile remarkably resembles that of the  H i absorption seen in radio data (Fig. 1; Morganti et al. 2004). Its resolution-corrected width is 250 (±80) km s-1, reflecting the collective motions of multiple clouds along the line of sight.

An absorption line is also seen at −1100 km s-1 away from 12CO(1−0) after the progressive subtraction of standing wave harmonics from the 3 mm baseline. However, the wave amplitude is comparable to the absorption minimum (Fig. 2), impeding its exact profile study. We thus simply constrain τobs,0(0 → 1) towards the 0.51 (±0.06) Jy continuum background to an upper limit of 0.03.

The 12CO(1−0) emission peaks at a main beam temperature of 9.2 (±1.2) mK above its underlying continuum. Its resolution-corrected width is 400 (±95) km s-1, and its intensity, ICO(1−0), is 4.0 (±1.3) K km s-1 (see also Evans et al. 1999, 2005). From this, we deduce that the cold  H2 gas reservoir of 4C 12.50 is 1.0 (±0.3)  ×  1010M, assuming that the gas mass is given by the product (Solomon et al. 1997), where ΩB is the telescope main beam area in arcsec2, DL is the source luminosity distance in Mpc, and α is the CO luminosity to  H2 mass conversion factor. We used a 103 GHz beam of 28′′ (61 kpc), and α = 0.8 M (K km s-1 pc2)-1 (Downes & Solomon 1998) because 4C 12.50 has a ULIRG-like infrared luminosity (2.5 × 1012L) based on its 12, 25, 60, and 100 μm flux (Golombek et al. 1988; Sanders & Mirabel 1996; Guillard et al. 2012).

The 12CO(3−2) emission is detected with a S/N of 25 and 21 in the WILMA and the FTS data, respectively. Its main beam temperature peaks at 17.5 (±1.0) mK above the local continuum. Its resolution-corrected width is 510 (±65) km s-1, and its intensity is 9.9 (±1.6) K km s-1.

The ICO(3−2)/ICO(1−0) ratio of the ambient gas, 2.5, is below its theoretically predicted value of 9 for the case of optically thick gas that is thermalized to a single  Tex. It is also below that of the central line of sight of M 82, of several local AGN, ULIRGs, and radio galaxies, but above that of the outflow of M 82, and similar to that of the Milky Way. This comparison is visualized in Fig. 3 for the line luminosities. Even though this ratio is indicative of large reservoirs of cold and diffuse gas, it does not rule out the presence of a warm and dense, highly excited CO component, which could be detected through high J emission lines. The non-detection of HCN (or HCO+) at the 3σ level of 0.86 K km s-1 for a width of 460 km s-1 cannot be used to argue for or against either interpretation. It leads to ICO/IHCN ≥ 8 (Gao & Solomon 2004) for a linearly interpolated CO(2−1) intensity of 7 K km s-1.

4. Discussion: on the properties of a multi-phase AGN-driven outflow

The rapid molecular gas motions that we detected provide direct evidence of an outflow in 4C 12.50 because none of its merging components (Axon et al. 2000) is moving faster than  ±250 km s-1 from its systemic velocity (Holt et al. 2003; Rodríguez Zaurín et al. 2007). The maximum gas velocity reaches −1500 km s-1, which is atypical of supernova-driven outflows (e.g., Heckman et al. 2000; Rupke et al. 2005), but characteristic of AGN-driven outflows (e.g., Rupke & Veilleux 2011; Sturm et al. 2011). AGN radiation pressure and winds can affect the gas kinematics in 4C 12.50: the AGN radiation does reach and ionize part of the outflowing gas, which contains both Ne V and O IV ions (Spoon & Holt 2009; Dasyra et al. 2011). The gas can also be pushed by shocks created as the radio jet of 4C 12.50 (Stanghellini et al. 1997) propagated through the ISM. An  H i outflow in front of a bright radio knot (Morganti et al. 2004) makes this scenario plausible for the neutral gas, and even for the molecular gas. A jet-ISM interaction can be responsible for the irregular  H2 kinematics (Dasyra & Combes 2011; Nesvadba et al. 2011), the strong  H2 emission (Ogle et al. 2010), and the high CO excitation (Papadopoulos et al. 2008) of radio galaxies.

To constrain the line-of-sight CO column density,  NCO, from the observed absorption, we assume that the gas is homogeneously distributed and in local thermodynamic equilibrium (LTE). We can then relate  NCO to τ through (1)(e.g., Wiklind & Combes 1995), where c is the lightspeed, h is the Planck constant, k is the Boltzmann constant, EJ and gJ are the energy and the statistical weight of the state J, respectively, AJ + 1,J is the Einstein coefficient for spontaneous emission, ν is the frequency of the emitted line, and Q is the partition function at the excitation temperature  Tex. The ratio ∫τ(2 → 3)   dV/∫τ(0 → 1)   dV agrees with its observed limit,  >6.4, if    Tex ≥ 65 K. Turbulence and shocks can easily bring the gas to this  Tex through collisions. For    Tex = 65 K,  NCO is equal to 1.8 × 1018 cm-2.

This number will increase with increasing  Tex, but will remain on the same order of magnitude up to 200 K. The filling factor of the background source by the molecular clouds,  fc, can play an important role in the  Tex and  NCO determinations. Partial coverage of the background source (fc < 1) can cause the observed (beam-averaged) optical depth to deviate from its actual value, which is then equal to −ln [1 − (1 − eτobs)/fc]  (Wiklind & Combes 1994). For gas in LTE, this relation is computed from the radiative transfer equation, assuming that    Tex ≪ Tbg and that the cosmic microwave background contribution to the gas excitation is negligible (e.g., Staguhn et al. 1997). Requiring the gas to be optically thick limits  fc to values above 0.2. Data of multiple transitions are needed to measure fc, further constrain  NCO, and examine any large velocity gradients (Scoville & Solomon 1974) and different energy level populations (Israel et al. 1991).

For the present assumptions and a CO-to-H2 abundance ratio of 10-4, the  H2 column density towards the outflow,  NH2, is  ≳2 × 1022 cm-2. The  H i column density is 3 × 1021 cm-2, based on both optical  Na I data (Rupke et al. 2005) and radio  H i data (Morganti et al. 2005, for Tspin = 1000 K). For the ambient  H i gas, which is located 100 pc northwest of the radio core, the column density is 6 × 1020–1022 cm-2 (Morganti et al. 2004; Curran & Whiting 2010, for Tspin = 100 K). The largest uncertainty in this comparison is the wavelength-dependent spatial distribution of the background and the foreground source. The radio core emission typically dominates the millimeter and even the submillimeter (850 μm; Clements et al. 2010) continuum, because the emission from diffuse jet knots is described by a steep power law that declines rapidly with frequency (Krichbaum et al. 1998, 2008). We thus assume that the millimeter background source area is contained within the 2 cm beam.

For a beam area of 3 milliarcsec2 (Lister et al. 2003) and  NH2 = 1.8 × 1022 cm-2, the outflow entrains 4.2 × 103M of cold  H2 gas in front of the millimeter core. How this mass compares with the mass of the cold  H2 gas that flows over all lines of sight is unknown. It therefore remains unclear whether the outflow carries less cold than warm H2 gas. However, the outflowing-to-ambient mass ratio of the cold gas will be lower than that of the 400 K gas (37%; Dasyra & Combes 2011), as long as the cold outflowing gas weighs less than 3 × 109M. If this amount of gas existed in the millimeter beam, and without a background source illuminating most of the clouds, the outflow would be seen in emission. Our 12CO(0 → 1) data instead limit any additional outflowing cold gas mass to be  ≤2 × 109M when integrating over a 1000 kpc2 area and a 1500  km s-1 velocity range (Batcheldor et al. 2007).

thumbnail Fig. 3

CO line luminosities in 4C 12.50 and other sources, normalized to LCO(1−0)(4C 12.50)/LCO(1−0). The data are from Fixsen et al. (1999), Dumke et al. (2001), Weiss et al. (2001, 2005), Papadopoulos et al. (2007, 2008), Panuzzo et al. (2010), van der Werf et al. (2010), and Krips et al. (2011). Typical uncertainties are 0.1–0.3 dex.

Open with DEXTER

5. Summary

We have presented deep 30 m observations of 12CO (1−0) and (3−2) in the nearby ULIRG and radio galaxy 4C 12.50, which has an impressive outflow of warm (400 K)  H2 gas (Dasyra & Combes 2011). Our observations led to the discovery of the cold component of the molecular outflow. Absorption from J = 2 → 3 is seen in data obtained with two different frequency tunings and backends. It peaks at −950 km s-1, and its profile resembles that of the  H i absorption component that traces the neutral gas outflow (Morganti et al. 2004). Absorption is also potentially seen for the 0 → 1 transition. Treating the optical depth of the latter as an upper limit and assuming that the clouds fully cover their background source, we have found that the molecular gas has an excitation temperature of at least 65 K in LTE conditions. For this temperature,  NH2 is 1.8 × 1022 cm-2. This suggests that the outflow carries 4.2 × 103M of cold  H2 gas along the nuclear line of sight. Even if the mass of the cold outflowing gas could exceed that of the warm outflowing gas, 5.2 × 107M, for a higher  NCO value or when integrating over several lines of sight,

it would most likely not exceed 3 × 109M, i.e., one third of the ambient cold gas reservoir. The outflowing-to-ambient mass ratio could thus be elevated into the warm gas phase with respect to the cold gas phase.

Acknowledgments

K.D. acknowledges support by the Centre National d’Études Spatiales (CNES), and is thankful to P. Salomé for technical tips and to P. Papadopoulos for useful discussions. Based on data obtained with the 30 m telescope of IRAM, which is supported by INSU/CNRS (France), MPG (Germany), and IGN (Spain).

References

All Figures

thumbnail Fig. 1

Left: FTS and WILMA spectra of 12CO(3−2) in 4C 12.50, binned into 62  km s-1 channels and shifted by a constant c0, when needed. Right: flux-calibrated line profile of 12CO(3−2) , averaged over the WILMA and the FTS data to be least affected by potential artifacts. Gaussian functions with parameters that most closely fit the emission component at the systemic velocity and the absorption component at −950 km s-1 are shown in blue and cyan, respectively. The inverse  H i absorption that is seen in the radio data of 4C 12.50 (Morganti et al. 2004) is overplotted in orange for an arbitrary scale and continuum level.

Open with DEXTER
In the text
thumbnail Fig. 2

WILMA (left) and FTS (right) 12CO(1−0) spectrum of 4C 12.50, binned to 93  km s-1 channels. The dashed histogram corresponds to the average spectrum prior to the removal of sinusoidal baselines, which represent the harmonics of a standing wave (shown in orange). The final spectrum is shown as a solid, filled histogram.

Open with DEXTER
In the text
thumbnail Fig. 3

CO line luminosities in 4C 12.50 and other sources, normalized to LCO(1−0)(4C 12.50)/LCO(1−0). The data are from Fixsen et al. (1999), Dumke et al. (2001), Weiss et al. (2001, 2005), Papadopoulos et al. (2007, 2008), Panuzzo et al. (2010), van der Werf et al. (2010), and Krips et al. (2011). Typical uncertainties are 0.1–0.3 dex.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.