DISCOVERY OF AN ACTIVE GALACTIC NUCLEUS DRIVEN MOLECULAR OUTFLOW IN THE LOCAL EARLY-TYPE GALAXY NGC 1266

Figure 1 presents the integrated spectra of the ¹²CO data used in this paper, showing good agreement between the CO(1–0) and CO(2–1) fluxes recovered from the single dish and the interferometers. Also of note is that the CO(3–2) line profile from the SMA has the same shape as the CO(2–1), including wing emission. The CARMA observations recover ≈120% of the single-dish flux, and the SMA observations ≈130%. Errors in the single-dish baselines and the intrinsic flux variations of the millimeter calibrators can account for the discrepancy. Therefore, it appears that both CARMA and the SMA recover all of the single-dish flux.

Zoom In Zoom Out Reset image size

With its narrow single-peaked central velocity component (hereafter CVC) and high-velocity wings, the integrated profile shape is more akin to those observed in protostellar outflows (e.g., Bally & Lada 1983) than those of typical external galaxies, where a double-horned profile, is usually seen.

Figure 2 shows the fits and residuals of the single and nested Gaussians. Table 1 gives a summary of the fitted Gaussian parameters. We find that the broad wing and CVC, which we identify respectively with the broad and narrow Gaussian of the nested Gaussian fits, encompass respectively 34% and 66% of the CO(2–1) single-dish integrated flux, which has the best signal-to-noise ratio (S/N). When comparing the line to the escape velocity radial profile measured in the plane of the galaxy (see Figure 3), we find that at least 2.5% of the wing emission is from velocities exceeding the innermost local escape velocity (calculated at 2'') of v_esc ≈ 340 km s⁻¹. The true percentage may be higher, depending on the geometry of the high velocity gas. If the molecular gas is located either out of the plane of the galaxy or at larger radii, the local potential (and thus v_esc) would be much reduced. The v_esc profile was calculated using Jeans Anisotropic Multi-Gaussian Expansion (MGE) modeling (JAM; Section 3.3) to extract the true potential of the galaxy by modeling the observed stellar kinematics (N. Scott et al. 2011a, 2011b, in preparation).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Using the CO (1–0) single-dish flux quoted in Young et al. (2011), a column density of molecular hydrogen N(H₂) = X_CO∫T_mbdv, where the conversion factor X_CO = 2 × 10²⁰ cm⁻² (K km s⁻¹)⁻¹ (Dame et al. 2001), a beam size of 216, and a beam area of 529 square arcsec (which includes a beam shape correction of 1/ln(2); Baars 2007), the total mass of the molecular gas would be 1.7 × 10⁹ M_☉, including a correction factor of 1.36 for He. However, the wing emission is optically thin (see Section 3.4) and contributes little to the total mass. Considering only the CVC, the molecular gas mass is 1.1 × 10⁹ M_☉, including He.

Draine et al. (2007) derive a total dust mass of 1.02 × 10⁷ M_☉ in NGC 1266 (rescaled to our distance of 29.9 Mpc), using Spitzer Infrared Nearby Galaxy Survey (SINGS) data. Since a significant H i reservoir is not observed in NGC 1266 (see below), we may assume that all of the measured dust is associated with the molecular gas. Adopting a gas-to-dust mass ratio of 100 (Savage & Mathis 1979), the dust mass implies a molecular gas mass of 1.0 × 10⁹ M_☉. The derived dust mass is thus in good agreement with the mass derived from the CO measurements and a standard X_CO conversion factor.

Figure 4 shows the full spatial extent of the CARMA CO(1–0) integrated emission (Δv = 390 km s⁻¹), overlaid on an R-band image of the galaxy from SINGS, illustrating that the vast majority of the molecular gas is restricted to the innermost region of NGC 1266. The right-hand panel of the figure shows that the molecular gas is not well described by a single spatial component, but instead as both a diffuse, relatively extended component, which we will call the "envelope," and a highly concentrated nuclear structure, which we will call the "nucleus."

Zoom In Zoom Out Reset image size

In order to separate the nucleus from the envelope in the CARMA map, we take a slice across the integrated intensity map at a position angle of 145°, to trace the elongation of the narrow component, and fit two Gaussians, to the nucleus and the envelope, respectively. Figure 5 illustrates the radial cut as well as the nuclear Gaussian component, which has a characteristic FWHM of 105. We use FWHM/2 to define a characteristic deconvolved radius R_nuc of ≈60 pc (after accounting for the CARMA beam). Using this, the peak intensity of the profile in Figure 5 (24.3 Jy beam⁻¹ km s⁻¹), the pixel size of 008 pixel⁻¹, and X_CO = 2 × 10²⁰ cm⁻² (K  km s⁻¹)⁻¹, we derive a molecular gas mass for the nucleus of $M_{\rm nuc, H_2+He} =$ 4.1 × 10⁸ M_☉, including He. As the spatial distribution of the broad wing emission in the central region of the CARMA CO(1–0) integrated image is unknown, we do not subtract a potential wing contribution from this nucleus molecular gas mass. If we assume that the rest of the mass in the CVC comes from the envelope, we deduce that M_env ≈ 7.0 × 10⁸ M_☉, including He.

Zoom In Zoom Out Reset image size

Figure 6 is a map of the first moment of the CO(1–0) emission and shows a clear velocity gradient across the nucleus. Figure 7 is the position–velocity (PV) diagram at a position angle of 90° (i.e., roughly along the kinematic major axis), suggestive of a rotating disk with a velocity gradient of 1.8 km s⁻¹ pc⁻¹. Rotation can be traced on the PV diagram to a radius of 54 pc, consistent with the radius of the nucleus derived above. At R_nuc ≈ 60 pc, we derive a maximum projected rotation velocity of the nucleus of v_rot ≈ 110 km s⁻¹. The corresponding enclosed dynamical mass, assuming circular rotation and spherical symmetry (M_{dyn, nuc} = R_nucv²_rot/sin ²i_nucG, where i_nuc is the inclination of the nuclear disk and G is the gravitational constant), is uncertain because of the unknown inclination, but the formal lower limit (i_nuc = 90°) is 1.5 × 10⁸ M_☉. A rough inclination estimate based on the axial ratio of the CO(1–0) integrated intensity map shown in Figure 4 and the assumption of a thin disk yields i_nuc ≈ 34°, and thus an enclosed dynamical mass M_{dyn, nuc} ≈ 4.9 × 10⁸ M_☉. The dynamical mass calculated using the rotation of the nucleus is therefore consistent with the ¹²CO-derived molecular gas mass of the nucleus, when taking into account a correction associated with the inclination. The stellar contribution to the nucleus mass is discussed later.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

An independent estimate of the mass of molecular gas in the core was obtained via dynamical modeling of the stellar kinematics. For this we constructed an MGE model (Emsellem et al. 1994) of the stellar mass distribution of NGC 1266 from our own R-band Isaac Newton Telescope (INT) photometry (N. Scott et al. 2011c, in preparation) and the software of Cappellari (2002). We used this MGE model to build an axisymmetric dynamical model of the stellar kinematics based on the JAM formalism (Cappellari 2008). The global anisotropy, inclination, and mass-to-light ratio (M/L) of the galaxy were fitted to the large-scale ATLAS^3D SAURON stellar kinematics (M. Cappellari et al. 2011, in preparation) and will be discussed elsewhere. In this particular case the JAM model also explicitly includes as an extra free parameter: the mass of the molecular gas in the core, modeled as a dark Gaussian with a width of 12. The mass of molecular gas that best fits the SAURON stellar kinematics in the center is M_gas ≈ 8 × 10⁸ M_☉, with a factor ≈3 uncertainty dominated by systematic effects. This value is an order of magnitude larger than the mass in stars within 100 pc of the center, M_{⋆, 100 pc} ≈ 5 × 10⁷ M_☉, as determined from an analogous MGE mass model without the molecular core. This result is thus consistent with the mass in the center of NGC 1266 being dominated by molecular gas.

One final consistency check on the molecular gas mass can be performed using the ¹³CO(1–0) single-dish profile obtained as part of the dense gas census of ATLAS^3D galaxies (A. F. Crocker et al. 2011, in preparation), shown in Figure 8. The total ¹³CO(1–0) line intensity is 0.98 K km s⁻¹ within the 224 beam of the IRAM 30 m telescope. We must first correct I(¹³CO) to reflect a more realistic size for the emitting region and adopt a radius of ≈1''. The correction corresponds to the ratio of the two solid angles subtended: Ω_beam/πR², in this case 567 arcsec²/3.14 arcsec². The resulting corrected I(¹³CO) is 177 K km s⁻¹. Using the ¹³CO to A_V conversion from Lada et al. (1994), I(¹³CO)/(K km s⁻¹) = 1.88 + 0.72 mag⁻¹A_V results in an extinction A_V ≈ 243 mag. Converting A_V to N(H₂) using N(H₂)/A_V = 9 × 10²⁰ cm⁻² mag⁻¹ from Schultz & Wiemer (1975) and Bohlin et al. (1978), and then converting N(H₂) to a molecular gas mass, yields M_nuc ≈ 3.2 × 10⁸ M_☉, corrected for He. Therefore, the ¹³CO-derived mass is in reasonable agreement with the ¹²CO-derived molecular gas mass.

Zoom In Zoom Out Reset image size

With a molecular gas mass of 4.1 × 10⁸ M_☉ (including He), assuming a circular disk, the mean molecular hydrogen surface density in the nucleus is $\Sigma _{\rm H_2} \approx$ 2.7 × 10⁴ M_☉ pc⁻² (corresponding to a mean column density N(H₂) ≈ 1.7 × 10²⁴ cm⁻²) and the mean volume density of H₂ is at least 6.9 × 10³ cm⁻³ (lower limit assuming spherical symmetry). Given that tracers with higher critical densities, such as ¹³CO, CS, HCO⁺, and HCN, are robustly detected in the nucleus (E. Bayet et al. 2011, in preparation), the molecular gas is likely to be clumpy.

The dynamical mass, JAM mass, and ¹³CO-derived mass agree quite well with the ¹²CO(1–0)-derived molecular gas mass of the nucleus, indicating that using the standard Milky Way conversion factor X_CO = 2 × 10²⁰ cm⁻²(K km s⁻¹)⁻¹ is appropriate to determine the mass of the molecular material in the center of NGC 1266. The JAM modeling also indicates that the stellar contribution to the mass of the region with molecular gas is negligible.

We use the CO(2–1) SMA data to resolve the spatial extent of the high-velocity wings of the molecular gas, because of the limited velocity coverage of the correlator when the CARMA data were taken. In order to minimize contamination from the nucleus, we made a map of the emission over the velocity ranges v = 1590–1890 km s⁻¹ and v = 2350–2650 km s⁻¹, eliminating the central 460 km s⁻¹. Figure 9(a) shows the velocity ranges of the CVC and the wings of the line used for mapping. The offsets from v_sys were chosen to achieve equal S/Ns of six for the peaks of the redshifted and blueshifted images in Figure 9(b), which shows that the wings are considerably more spatially extended than the nucleus. If we model the outflow as a bi-spherical structure, similar to what is observed in M82 (Walter et al. 2002), and represent the redshifted and blueshifted lobes by two tangent spheres whose point of contact is the nucleus, the distance the molecular gas has traveled from the center is equal to the width of the lobes. We measure the width of the blueshifted and redshifted lobes to be ≈34 and 42, respectively. Taking the average of the lobe extents and correcting for convolution with the 2'' SMA beam gives us a characteristic extent of ≈32 or 460 pc. The redshifted and blueshifted wings are also spatially offset from one another. The position centroids of the lobes are separated by ≈23, thus each is offset by about 170 pc from the galaxy center. The large spatial extent of the wing emission rules out an interpretation whereby the high-gas velocities are the result of gas having fallen into the central potential well of the galaxy. We thus conclude that the source of the high-velocity wing emission is a kpc-scale molecular outflow, with R_outflow ≈ 460 pc.

Zoom In Zoom Out Reset image size

Further evidence that the wings represent an outflow is that the axis connecting the lobes is intersected by the CVC and is not aligned with the kinematic major axis of the nuclear gas disk. This suggests that the broad wing emission is composed of gas that is being expelled and is not an extension of the nuclear material. If we estimate the inclination of the outflow (with respect to the plane of the sky) using the average offset of the centroids with respect to the nucleus (23) divided by the average extent of the lobes (32) calculated above, we obtain an inclination angle of roughly 20°.

The mass of outflowing gas is somewhat uncertain because the wings are unambiguously detected only in ¹²CO (Figure 1), requiring that the density, optical depth, and temperature of the outflow be determined from the excitation of the three lowest CO rotational transitions. We use the beam-corrected (normalized to the single-dish CO(1–0) beam of 216) intensities of the wing emission from the CO(1–0) and (2–1) IRAM 30 m single-dish spectra and the CO(3–2) SMA spectrum (see Table 1). The beam-corrected values are 11.93 K km s⁻¹ for CO(1–0), 9.16 K km s⁻¹ for CO(2–1), and 7.16 K km s⁻¹ for CO(3–2).

We use the RADEX large velocity gradient software (van der Tak et al. 2007) to determine volume densities and column densities in the wind. With RADEX, assuming a line width v_fwhm = 353 km s⁻¹, we find that the conditions required to reproduce the beam-corrected flux ratios above are an H₂ density n(H₂) ≈ 10³ cm⁻³, a kinetic temperature T_kin ≈ 100 K, and a CO column density N(CO) ≈ 1.0 × 10¹⁶ cm⁻². The N(CO) was also constrained using on the optically thin estimate of Knapp & Jura (1976) assuming subthermally excited CO, and the result is consistent with that from RADEX. Assuming a CO/H₂ abundance ratio of 10⁻⁴, the derived N(CO) and our adopted beam size of 216 imply a molecular outflow mass of 2.4 × 10⁷ M_☉, including He. It is however possible that the outflowing molecular mass is higher. The optically thin approximation (i.e., that we are seeing all CO molecules) provides us with a lower limit to the total outflowing molecular mass. If the state of the molecular gas in the outflow of NGC 1266 is more akin to gas in either ultra-luminous infrared galaxies (ULIRGs) or giant molecular clouds (GMCs) in the Milky Way, where a conversion factor (X_CO) is used, the total outflowing mass could be a factor of ≈10–20 higher.

While no H i emission is detected in NGC 1266, H i appears in absorption against the 1.4 GHz continuum source. The H i observations have thus far only been taken at low resolution, so are presently unresolved. The top panel of Figure 10 shows the H i absorption profile, which exhibits a broad blueshifted velocity component, and the fitted Gaussians describing the absorption profile. The bottom panel of Figure 10 shows a good correspondence between the blueshifted absorption feature in the H i and the high-velocity wing of the CO(2–1) emission line profile. This identifies the blueshifted molecular gas as moving out (as opposed to infalling from behind) and implies that the outflow in NGC 1266 has multiple phases.

Zoom In Zoom Out Reset image size

Calculating the column density from the absorption feature, assuming T_spin = 100 K, reveals that the H i column in front of the continuum source is N_H = 2.1 × 10²¹ cm⁻², with the outflowing component contributing 8.9 × 10²⁰ cm⁻² and the systemic component contributing 1.2 × 10²¹ cm⁻². If we assume that the H i and CO are cospatial (R_HI = R_outflow = 460 pc) and that the total mass of outflowing H i is twice what we calculate (since we are only able to detect the blueshifted lobe), we derive the total H i mass in the outflow to be 4.8 × 10⁶ M_☉ per lobe, totaling 9.5 × 10⁶ M_☉. If we include the H i contribution, the total neutral gas mass (H i + H₂ + He) of the outflow is thus 3.3 × 10⁷ M_☉.

The molecular gas (as traced by CO) and atomic gas are likely not the only constituents of the mass in the outflow. By not accounting for other states of the gas, we are underestimating the true outflow mass. For example, Roussel et al. (2007) detect a large reservoir of warm H₂ in NGC 1266 using Spitzer, with a total mass of $M_{\rm H_2,warm} \approx 1.3\times 10^7$ M_☉. It is very likely that this mass of warm H₂ belongs to the outflow rather than the nuclear disk. The total warm H₂ luminosity is $L_{\rm H_2,warm} \approx 1\times 10^{41}$ erg s⁻¹. For the nuclear region to sustain this luminosity of warm H₂, the active galactic nucleus (AGN) would be required to have a bolometric luminosity of ≳ 2 × 10⁴⁴ erg s⁻¹, assuming first that $L_{X{\rm, AGN}} \gtrsim 100 L_{\rm H_2, warm}$, Ogle et al. 2010; and L_{bol, AGN} ≳ 16 L_X, Ho 2008), a factor of three larger than L_FIR (7 × 10⁴³ erg s⁻¹; Gil de Paz et al. 2007). If this warm H₂ is indeed part of the outflow, our estimate of the total mass of the outflow should increase by 40%. Adding in the contribution of the ionized and hot gas would increase the mass of the outflow further, thus we consider our estimate of 3.3 × 10⁷ M_☉ to be a conservative lower limit.

Combined with the molecular emission and H i absorption data, X-ray, Hα, and radio continuum data further elucidate the nature of the outflow. Figure 11 shows a comparison of the Hα narrowband image from the SINGS survey (Kennicutt et al. 2003), a 1.4 GHz radio continuum map (Baan & Klöckner 2006), and the unsmoothed X-ray map from the Chandra X-ray Telescope Advanced CCD Imaging Spectrometer (ACIS) on the same spatial scale (K. Alatalo et al. 2011, in preparation). A spectral fit to the Chandra data co-added over the extent of the emission shows that the X-ray emission is dominated by thermal bremsstrahlung. The 1.4 GHz continuum shows both a centrally peaked, unresolved core (possibly from a radio jet), as well as two more diffuse spurs running from the southeast to the northwest. The lobed structure first apparent in the molecular gas (Figure 9) is seen in all these tracers. In fact, the X-rays and the 1.4 GHz spurs, which do not suffer significantly from extinction, show an asymmetry between the foreground and background lobes, relative to the CO at v_sys. This likely means that the relative faintness of the Hα flux from the redshifted lobe as compared to the blueshifted lobe is not primarily due to extinction. The spatial relationship between the molecular gas and the X-ray, 1.4 GHz and Hα emitting material supports this picture. The Hα, radio spurs, and X-rays likely are emitted from the interface where the outflowing material collides with the ISM of NGC 1266. This interpretation is supported by the thermal bremsstrahlung emission in the Chandra spectrum, although the radio spur emission could also possibly come from a radio jet.

Zoom In Zoom Out Reset image size

The geometry of the CO outflow can be explained using two possible scenarios. First, it could be due to a molecular outflow launched deep in the nucleus of the galaxy, confined azimuthally by the massive molecular disk. Second, it could directly originate from the molecular disk and be entrained by hot winds launched by the AGN, surviving only because of the extreme density of the molecular disk. In either of these scenarios, at least some of the molecular material escapes from the galaxy still intact, and will energize the intergalactic medium (IGM). This is especially true if the outflow continues to accelerate from its launch point.

A summary of the derived properties of NGC 1266 is presented in Table 2. The mean molecular hydrogen surface density of the nucleus is 2.7 × 10⁴ M_☉ pc⁻², corresponding to a mean column density N(H₂) ≈ 1.7 × 10²⁴ cm⁻², two orders of magnitude higher than the surface density of individual GMCs in the Milky Way, and comparable to what is seen in some of the most luminous starburst galaxies (Kennicutt 1998).

Table 2. NGC 1266 Characteristic Properties

Derived Property	Value
Mnucleus	4.1 × 108 M☉
Rnucleus	60 pc
MCVC	1.1 × 109 M☉
〈Σ(H2)nucleus〉	2.7 × 104 M☉ pc−2
〈N(H2)nucleus〉	1.7 × 1024 cm−2
〈n(H2)nucleus〉	6.9 × 103 cm−3
$M_{\rm H I+H_2, outflow}$	3.3 × 107 M☉
$M_{\rm H_2, outflow}$	2.4 × 107 M☉
Routflow	450 pc
voutflow	177 km s−1
τdyn	2.6 Myr
$\dot{M}$	13 M☉ yr−1
${\rm KE_{{\rm outflow}}}$	1.0 × 1055 erg
Loutflow	1.3 × 1041 erg s−1
τdep	85 Myr

Download table as:  ASCIITypeset image

For an outflow with a mass of 3.3 × 10⁷ M_☉ and an outflow velocity of half the full width at half-maximum (v_fwhm) of the broad wing component (177 km s⁻¹), the kinetic energy associated with the outflow is 1.0 × 10⁵⁵ erg, equivalent to the total kinetic energy expelled from 10⁴ supernova explosions. The dynamical time, τ_dyn, for the outflow to reach its maximum radius of 460 pc at this velocity is only 2.6 Myr. If the edge of the CO emission in the lobes is produced from only the highest velocity gas, the relevant timescale is even shorter, τ_dyn ∼ 1 Myr.

From the total kinetic energy and τ_dyn, the mechanical luminosity of the outflow L_outflow = 1.3 × 10⁴¹ erg s⁻¹. The mass outflow rate is $\dot{M} = M_{\rm outflow}/\tau _{\rm dyn}$ = 13 M_☉ yr⁻¹. At this rate, if the source of the gas in the outflow is the nuclear molecular disk, the nucleus of NGC 1266 will be completely depleted in τ_dep = 32 Myr or 85 Myr for all of the gas in the center (nucleus + envelope), both very short timescales.

NGC 1266 is unusual in several respects. There is a large reservoir of molecular gas concentrated within the central 100 pc. The concentration of H₂ in the nucleus implies that the gas must have lost nearly all of its angular momentum. Although such high central concentrations of H₂ are observed in interacting galaxies, NGC 1266 is an isolated galaxy with no sign of interaction or merger.

Is the outflow in NGC 1266 a common evolutionary stage through which most ETGs pass, or is NGC 1266 a pathological phenomenon? The short gas depletion time of ≈85 Myr is reasonably consistent with finding only one such case so far in the 60 ATLAS^3D CO detections (Young et al. 2011). The expectation would be to find three, assuming that the typical molecular gas depletion time due to star formation in ETGs is the same as the 2 Gyr found in late-type galaxies (Leroy et al. 2008; Bigiel et al. 2008). However, the sheer strength of the CO line in NGC 1266 certainly aided the outflow detection, and it is likely that, given the lower S/Ns present in the rest of the sample, the low surface brightness features at high velocities would generally be lost in the noise.

Optically, NGC 1266 does not stand out within the ATLAS^3D sample. It is a fast rotator with regular stellar kinematics (Krajnović et al. 2011) and is of normal metallicity. Its absolute magnitude, M_K = −22.9, corresponds to a stellar mass just below the median of the sample. On the other hand, its radio flux at 1.4 GHz places NGC 1266 in the top 10% of radio emitters within the sample, although this emission could be explained by the FIR-radio SF correlation based on the amount of molecular gas detected in the source (Murgia et al. 2005). Among the ∼100 known ETGs with detected molecular gas (Wiklind et al. 1995; Welch & Sage 2003; Sage et al. 2007; Combes et al. 2007; Young et al. 2011), ≈20 have been previously mapped interferometrically (see Crocker et al. 2011, and references therein). Compared to those, the molecular gas in NGC 1266 stands out both in terms of its compactness and its large mass. The star-forming ETGs seem to exhibit only a narrow range of mean molecular gas surface densities, 10–500 M_☉ pc⁻², similar to the range of Milky Way values. The surface density of the nucleus of NGC 1266 (≈2.7 × 10⁴ M_☉ pc⁻²) is two orders of magnitude larger than this.

NGC 1266 should perhaps be compared to radio galaxies, 1/4–1/3 of which have been shown to contain significant reservoirs of molecular gas (Evans et al. 2005; Ogle et al. 2010). The H i absorption profile seen in NGC 1266 is akin to that in the handful of known radio galaxies with outflowing H i: IC 5063, 3C 305 (Morganti et al. 1998; Oosterloo et al. 2000; Morganti et al. 2005a, 2005b), 4C 12.50 (Morganti et al. 2004) and 3C 293 (Morganti et al. 2003; Emonts et al. 2005), with mass outflow rates ∼50 M_☉ yr⁻¹ at high (≳ 1000 km s⁻¹) velocities. The H i in each of these galaxies is detected in absorption against a radio source and covers hundreds of parsecs to kiloparsecs. While the outflowing H i mass in NGC 1266 is orders of magnitude lower than in these galaxies, it is possible that both types of systems share a common origin. However, although the luminosity and structure of the H i absorption do mirror those in radio galaxies, the samples described by Evans et al. (2005) and Ogle et al. (2010) and the outflowing systems all show signs of interactions, companions, and tidal features, which are all markedly absent from NGC 1266.

Given the high surface density of the molecular gas, NGC 1266 ought to be forming stars. Based on the Kennicutt–Schmidt (K-S) relation (Kennicutt 1998), $\Sigma _{\rm H_2} \approx$ 2.7 × 10⁴ M_☉ pc⁻², and a radius of 60 pc, the expected star formation rate (SFR) of the nucleus alone is 3.1 M_☉ yr⁻¹, adjusted for a Kroupa initial mass function (IMF). In order to understand the star formation efficiency of the gas, we use multiple tracers to place constraints on the SFR, including the total far-infrared luminosity (L_FIR), the 24 μm emission, and the 8 μm emission from polycyclic aromatic hydrocarbons (PAHs). Those three tracers are systematically affected by the presence of an AGN, but in different ways, so we use all three to define a range of possible SFRs. Both the 24 μm emission and L_FIR suffer from contamination by a buried AGN and will thus systematically overestimate the SFR. PAH molecules are destroyed by the harsh ionization fields associated with a strong AGN and will thus possibly underestimate the true SFR (Voit 1992).

If we assume that all of the far-infrared (FIR) luminosity (7 × 10⁴³ erg s⁻¹; Gil de Paz et al. 2007; Combes et al. 2007) is from star formation (an upper limit since the nucleus of NGC 1266 contains an AGN), we calculate a global obscured SFR_FIR of 2.2 M_☉ yr⁻¹ using Kennicutt (1998), re-calibrated for a Kroupa IMF. The unobscured SFR from the measured far-ultraviolet (FUV) contributes negligibly to the overall SFR (SFR_FUV = 0.003 M_☉ yr⁻¹; Gil de Paz et al. 2007; Leroy et al. 2008). We also calculate the SFR using the 24 μm and PAH emission, measured by Spitzer as part of the SINGs survey (Kennicutt et al. 2003). The SFR is related to the 24 μm luminosity (L₂₄) following Calzetti et al. (2007): SFR₂₄/( M_☉ yr⁻¹) =1.27 × 10⁻³⁸(L₂₄/erg s⁻¹)^0.885. For NGC 1266, L₂₄ = 1.06 × 10⁴³ erg s⁻¹ once adjusted to our adopted distance (Temi et al. 2009), so the associated SFR₂₄ ≈ 1.5 M_☉ yr⁻¹ (the stellar continuum contribution to the 24 μm emission is ≪1%), within 33% of the SFR_FIR derived from the FIR emission. To measure the SFR from the PAH emission, we use the mid-IR fluxes from J. Falcón-Barroso et al. (2011, in preparation). We subtract off the stellar contribution from the 8.0 μm flux by scaling the 3.6 μm image; the scale factor of the 3.6 μm image can range from 0.22 to 0.29 (Calzetti et al. 2007), with elliptical galaxies having typical values of 0.26 (Wu et al. 2005; Shapiro et al. 2010). From the corrected PAH emission, we measure SFR_PAH ≈ 0.57 M_☉ yr⁻¹ (J. Falcón-Barroso et al. 2011, in preparation) a factor of four lower than the SFR derived using the FIR emission.

Comparing the SFR predicted from the molecular gas via the K-S relation to the SFRs derived from the 8 μm, 24 μm, and FIR emission, we find that the predicted SFR is a factor of 1.4–5 times larger than the measured SFR, depending on which SFR tracer is used. However, NGC 1266 is well within the scatter of the K-S relation (Leroy et al. 2008), as well as the range of SFRs observed in the SAURON and ATLAS^3D sample galaxies (Shapiro et al. 2010; J. Falcón-Barroso et al. 2011, in preparation).

The star formation-driven wind model of Murray et al. (2005) requires that the SFR be equal to or greater than the mass outflow rate for radiation-driven outflows, which we refer to as the Murray criterion. Our calculations for NGC 1266 show that even if we consider the higher SFR estimate (SFR ≈ 2 M_☉ yr⁻¹), our lower limit on the mass outflow rate ($\dot{M} \approx$ 13 M_☉ yr⁻¹) still exceeds it by a factor of at least a few. This will only increase with more accurate accounting of the total mass of the outflow. Therefore, the NGC 1266 molecular outflow does not fulfill the Murray criterion, and momentum coupling (i.e., radiation-driving) and star formation are unlikely to be the drivers of the outflow.

NGC 1266 is classified as a low-ionization nuclear emission-line region (Moustakas & Kennicutt 2006). The 1.4 and 5 GHz radio continuum images reveal a prominent and unresolved point source in the nucleus (e.g., Figures 11 and 12). Chandra observations reveal hard X-ray emission centered on the radio point source, consistent with an obscured AGN (see Figure 12). Given the overwhelming evidence for the presence of an AGN in NGC 1266, and that star formation is much less intense than in the few other galaxies exhibiting molecular outflows (see Section 4.3), an AGN-driven outflow seems a reasonable possibility.

Zoom In Zoom Out Reset image size

To investigate whether the AGN is any more capable, we use the relationship between the 1.4 GHz luminosity and total jet power from Bîrzan et al. (2008): log (P_jet/10⁴² erg s⁻¹) = 0.35log (P_1.4GHz/10²⁴WHz⁻¹) + 1.85. The total radio flux at 1.4 GHz is 9.3 × 10²⁰ W Hz⁻¹, resulting in a total jet power of 6.1 × 10⁴² erg s⁻¹. Comparing this to the L_mech of the outflow of 1.3 × 10⁴¹ erg s⁻¹, a coupling of ∼2% is required to drive the outflow with the AGN, a rather modest value. Therefore, the AGN does indeed generate sufficient energy to power the outflow through the mechanical work of the radio jet.

There is a small sample of galaxies that have been shown to exhibit molecular outflows: M82, Arp 220, Mrk 231, and M51. Arp 220 is a starbursting major merger and has a molecular gas surface density of 6 × 10⁴ M_☉ pc⁻² (Mauersberger et al. 1996), similar to that of the molecular nucleus of NGC 1266. Sakamoto et al. (2009) detected an outflow that they estimate has a mass of 5 × 10⁷ M_☉ and a mass outflow rate of 100 M_☉ yr⁻¹, with an outflow velocity of ∼100 km s⁻¹. These numbers, though uncertain, are also comparable to what is observed in NGC 1266. However, the SFR in Arp 220 is ≳ 200 M_☉ yr⁻¹(Soifer et al. 1987), almost two orders of magnitude greater than in NGC 1266, providing a ready source of energy for the outflow and meaning that the Murray criterion is met.

M82 has ≈3 × 10⁸ M_☉ of molecular gas entrained in an outflow with velocities up to ≈230 km s⁻¹ (Walter et al. 2002). It is unclear whether this is sufficient for any gas to escape the galaxy. The mass outflow rate is ≈30 M_☉ yr⁻¹, comparable to that estimated for the wind in NGC 1266, and roughly equivalent to the global SFR of M82 (Kennicutt 1998), thus satisfying the Murray criterion. The outflows can thus be driven out via radiation pressure.

Matsushita et al. (2007) have shown recently that M51 has entrained molecular gas in an outflow close to the AGN, with a total molecular mass of 6 × 10⁵ M_☉, two orders of magnitude less than the mass estimate for the NGC 1266 outflow. The derived mass outflow rate of M51 is ≈4 M_☉ yr⁻¹. While this outflow rate is equivalent to the total SFR, the star formation in M51 is known to be distributed throughout the disk (Calzetti et al. 2005) and thus is likely unable to effectively drive much of the nuclear molecular gas out. The AGN luminosity of 2 × 10⁴² erg s⁻¹ (Terashima & Wilson 2001) is also too low to sustain the molecular outflow via photon-driving. Therefore, M51 appears to be a scaled down version of the outflow we observe in NGC 1266, requiring a similar mechanism, such as mechanical work done by a radio jet, to drive out the gas.

Mrk 231, a nearby advanced major merger, shows a molecular outflow similar to that of NGC 1266 (Feruglio et al. 2010). The Mrk 231 CO(1–0) spectrum exhibits a similar profile to that seen in NGC 1266, with broad wings requiring a nested Gaussian fit. This similarity seems to confirm that neither NGC 1266 nor Mrk 231 is a unique case, and AGN feedback may be an efficient means of removing the molecular gas close to the AGN. Feruglio et al. (2010) claim that Mrk 231 has a mass outflow rate of ≈700 M_☉ yr⁻¹ (assuming a density profile of r⁻²), exceeding its SFR of 200 M_☉ yr⁻¹and thus requiring AGN feedback to be powered. If the CO in the outflow is optically thick, as assumed by Feruglio et al. (2010), then the mass outflow rate in Mrk 231 is indeed too high for a star formation-driven wind. On the other hand, if the correct L(CO)-to-M(H₂) mass conversion for both the NGC 1266 and Mrk 231 outflows is the optically thin conversion (as assumed for NGC 1266), then M(H₂) for the Mrk 231 outflow is a factor of ≈5 lower, leading to a mass outflow rate of ∼100 M_☉ yr⁻¹. This is sustainable using the Murray criterion and is similar to Arp 220. Therefore, the case for the molecular outflow in Mrk 231 requiring AGN feedback is uncertain. To obtain a robust estimate of the outflowing mass of Mrk 231, the opacity of the outflowing component must be determined.

New stacked CO(1–0) observations of local ULIRGs by Chung et al. (2011) also show clear evidence that high-velocity broad emission at up to 1000 km s⁻¹ from the systemic velocity is a common feature in these systems. In particular, the authors find that only the starburst-dominated ULIRGs exhibit broad wings, while AGN-dominated ULIRGs (identified by their optical spectra) do not. These results are thus clear evidence that in most observed molecular outflows photon-coupling with the star formation is the principal driver.

While the aforementioned galaxies all exhibit molecular outflows, M51 is the only other bona fide case of a non-photon-driven molecular outflow. However, it has almost two orders of magnitude less molecular gas being driven out than NGC 1266. Mrk 231 might well be an equivalent system, but the mechanisms capable of driving the gas out depend on the CO-luminosity-to-H₂ mass conversion in the outflow. The most acute contrast between NGC 1266 and the other molecular outflow sources is its lack of interaction. All other galaxies are undergoing interactions; major mergers in the case of Arp 220, Mrk 231, and some of the Chung et al. (2011) ULIRGs. NGC 1266 shows no evidence of having undergone an interaction at any time in the recent past, with unperturbed stellar kinematics, no companion, and a lack of H i emission internal or external to the galaxy. This means that while we can point to an interaction as the triggering mechanism in the other known molecular outflows, the striking absence of such a perturbing event makes NGC 1266 even more remarkable.