ESCAPE, ACCRETION, OR STAR FORMATION? THE COMPETING DEPLETERS OF GAS IN THE QUASAR MARKARIAN 231

Using the local thermal equilibrium ${{F}_{{\rm CS}(2-1)}}$–${{M}_{{{{\rm H}}_{2}},{\rm dense}}}$ relation discussed in Alatalo et al. (2015) of

$$\begin{eqnarray*}&&{{N}_{{\rm CS}}}=1.90\times {{10}^{11}}{{T}_{{\rm ex}}}{{e}^{7.05/{{T}_{{\rm ex}}}}}{{I}_{{\rm CS}(2-1)}},\end{eqnarray*}$$

where ${{I}_{{\rm CS}(2-1)}}$ is the integrated intensity of the CS(2–1) line in K km s^2–1, assuming ${{T}_{{\rm ex}}}\approx 70$ K (van der Werf et al. 2010) and a CS abundance of ≈10⁻⁹, we find a total dense gas mass of $\approx 1.8\pm 0.3\times {{10}^{10}}\;{{M}_{\odot }}$. This dense gas mass is consistent with the total dense gas mass found with other tracers (Solomon et al. 1992; Solomon & Vanden Bout 2005; Feruglio et al. 2010), given the uncertainty in the CS/H₂ abundance, which is at least a factor of 2. Our dense gas mass estimate is lower than the dynamical mass derived within 1100 pc of the center³ of Mrk 231 is $3.15\times {{10}^{10}}$ ${{M}_{\odot }}$ (Downes & Solomon 1998), necessitating that the majority of the central mass in Mrk 231 is in the form of dense molecular gas. The signal-to-noise ratio of the CS data and resolution of these data are not able to confirm whether the CS is part of the face-on warped disk reported in Davies et al. (2004) and in dense gas reported in Aalto et al. (2015). Despite the limitation in resolution, the gradient seen in the CS moment1 map also appears consistent with the HCN kinematics seen by Aalto et al. (2012a). The agreement of the molecular gas mass also appears consistent with the idea that the majority of the molecular gas in Mrk 231 is in a dense form.

Our integrations were not sufficiently sensitive to detect CS(2–1) emission associated with broad wings (as is the case in other dense gas tracers, including HCN, reported by Aalto et al. 2012a), and warrants deeper observations to determine whether gas traced by CS (suggested to be a tracer of some of the densest molecular cores; Baan et al. 2008) is also taking part in the molecular outflow.

To derive the dense gas column along the line of sight (LOS) to the AGN in Mrk 231 (white cross in Figure 3 moment maps), we find that the pixel associated with the radio point source contains an intensity of 51.8 ± 2.0 K km s⁻¹ (using a Kelvin per Jansky factor of 78.35). Converting this intensity into a column density of H₂ (assuming an excitation temperature of 70 K and a CS/H₂ abundance of 10⁻⁹), we find a CS-derived LOS column to the AGN of $1/2\times {{N}_{H}}=N({{{\rm H}}_{2}})\approx 1.1\times {{10}^{24}}$ cm⁻² (if the AGN were sitting beneath the maximum value in the CS(2–1) map, the obscuring column would be $N({{{\rm H}}_{2}})\approx 1.5\times {{10}^{24}}$ cm⁻²). While this value was consistent with a previous estimate from X-ray absorption models (Braito et al. 2004), newer results from NuSTAR show N_H to be $1.2\times {{10}^{23}}$ cm⁻² (Teng et al. 2014). The discrepancy between the CS-derived column might suggest that the obscuring medium in the nucleus of Mrk 231 is quite clumpy (at a scale much smaller than the CARMA resolution), which is consistent with the suggestion by Teng et al. (2014) that the wind punching out of the system has created a preferential LOS to the AGN.

If we were to use the CS to investigate the star formation efficiency within Mrk 231, the total dense gas-traced surface density is ${{{\Sigma }}_{{{{\rm H}}_{2}}}}\approx 6450\;{{M}_{\odot }}$ pc⁻². Comparing this to the star formation surface density (assuming the dense gas and star formation are co-spatial) of $61\;{{M}_{\odot }}$ yr⁻¹ kpc⁻², we find that this matches the expectation from the Kennicutt–Schmidt relation (Kennicutt 1998).

Accounting for the distance to Mrk 231, $0\buildrel{\prime\prime}\over{.} 49$ corresponds to a projected separation between the CO(1–0) centroids of 415 pc. The CARMA data also successfully resolve the lobes, with sizes of 2 2 and 2 3 of the red- and blueshifted wings, respectively. To deproject this distance, an accurate inclination angle is needed. Inclination angles derived for the molecular disk (which we assume is perpendicular to the outflow) has derived inclinations angles ranging from $i=10{}^\circ$ (Downes & Solomon 1998; Davies et al. 2004) to $i=45{}^\circ$ (Richards et al. 2005), meaning that the deprojected distance ranges between 570 and 2390 pc. If we assume that the outflow is conical with a large opening angle (Rupke & Veilleux 2011; Cicone et al. 2012), then the distance between the wings is likely similar to the diameter of the lobes, which we estimate to be 1800 pc (after de-convolving the lobes). This size is slightly larger than the CO(1–0) separation found by Cicone et al. (2012), although we see extension in the east–west direction and opposite of the rotation of the CS disk, rather than the north–south direction as was observed in HCN by Aalto et al. (2012a). While this is consistent with the angle of rotation of the molecular disk, the velocities probed by the CARMA observations exceed the escape velocity, and therefore are unlikely to be due to rotation.

We use 600 km s⁻¹ for the characteristic velocity of each wing (as well as the deprojected radius, 900 pc) to calculate the dynamical timescale of the outflow of ${{\tau }_{{\rm dyn}}}\approx 1.5$ Myr, about twice as long as the timescale originally reported by Feruglio et al. (2010). If we use this dynamical timescale in conjunction with the outflow mass from Feruglio et al. (2010), and supported by Aalto et al. (2012a), of 5.8 $\times {{10}^{8}}\;{{M}_{\odot }}$, we find that the total mass outflow rate is $390\;{{M}_{\odot }}$ yr⁻¹, a factor of 2 smaller than Feruglio et al. (2010), and a factor of $\approx 2$ larger than the star formation rate in the system (Veilleux et al. 2009). If all $390\;{{M}_{\odot }}$ yr⁻¹ taking part in the molecular outflow were to escape the galaxy, we would expect the molecular gas to be depleted in $\approx 46$ Myr through the action of the molecular outflow alone, and 32 Myr if both outflow and the star formation gas consumptions are considered. This depletion timescale assumes that all molecular gas in the thin disk is intercepted by the molecular outflow, which is assumed to be traveling perpendicular to the disk. If the molecular outflow is unable to interact with most of the gas in the molecular disk, this could increase the molecular outflow depletion timescale considerably.

In Mrk 231, the molecular gas is being depleted by three competing mechanisms: consumption through star formation, escape from the galactic potential via a molecular outflow and accretion onto the supermassive black hole. The timescale for star formation to completely consume the central gas is 110 Myr. The molecular outflow appears to be the dominant depleter within the Mrk 231 system (assuming that the outflowing molecular mass is being completely expelled).

While ${{\dot{M}}_{{\rm out}}}$ of Mrk 231, updated in the prior section is indeed larger than ${{\dot{M}}_{{\rm SFR}}}$, ${{\dot{M}}_{{\rm out}}}$ might not be an accurate representation of the mass of molecular gas that is currently being expelled from the system. Mrk 231 is a massive system, with an estimated stellar mass of $\approx 3\times {{10}^{11}}\;{{M}_{\odot }}$ (U et al. 2012), and thus much of the outflowing molecular gas will not escape the potential well of Mrk 231, instead falling back and "stirring up" the massive molecular disk in the center. Using a dynamical mass within 1100 pc of the center of Mrk 231 of $3.15\times {{10}^{10}}\;{{M}_{\odot }}$ (Downes & Solomon 1998; Davies et al. 2004), and derive an escape velocity from the central kiloparsec of ∼500 km s⁻¹. If we assume that all gas that is traveling above ${{v}_{{\rm esc}}}$ will be able to successfully escape the system, then 15% of the total outflowing molecular gas (from the Gaussian fit of Feruglio et al. 2010) escapes.⁴ We use our updated maps to calculate the total escaping mass from the blue- and redshifted wing fluxes reported in Section 2 and Table 1, using a conservative merger-based conversion factor (Narayanan et al. 2011), and find ${{M}_{{\rm esc}}}=(2.93\pm 0.49)\times {{10}^{8}}\;{{M}_{\odot }}$. Dividing by the dynamical time, we find ${{\dot{M}}_{{\rm esc}}}\approx 195\;{{M}_{\odot }}$ yr⁻¹, which is quite comparable to the current star formation rate of 170 ${{M}_{\odot }}$ yr⁻¹ than were we to assume all outflowing mass is escaping. This updates the depletion timescale due to the escape of molecular gas, ${{\tau }_{{\rm esc}}}$, to 92 Myr. The combined depletion timescale ${{\tau }_{{\rm SF}+{\rm esc}}}$ is then 49 Myr.

The accretion onto the black hole can be derived using the bolometric luminosity of the AGN ($\approx 2.8\times {{10}^{12}}\;{{L}_{\odot }}$; Veilleux et al. 2009, 2013), and assuming an efficiency $\eta =0.17$, to be ${{\dot{M}}_{{\rm acc}}}\approx 1\;{{M}_{\odot }}$ yr⁻¹, or at a rate $\sim {{10}^{-2}}$ other depletion rates, a minor constituent to the depletion of the central molecular gas, and requiring a Hubble time to completely remove the molecular gas from the system.

Without a constant driving mechanism, the majority of the molecular outflow will eventually fall back into the center, either to be accreted, formed into stars, or re-launched. One would expect that this gas would fall back and re-inject energy into the central molecular disk, possibly acting to inhibit star formation (Alatalo et al. 2015; Guillard et al. 2015) and therefore extend the molecular gas consumption timescale. An extension of the depletion timescale expected for Mrk 231 from ∼20 Myr (the typical lifetime for O-stars) to ∼50 Myr, appears more consistent with the poststarburst population found in Mrk 231 (Canalizo & Stockton 2000), and common in other quasars (Cales et al. 2011; Canalizo & Stockton 2013), as well as other hosts of molecular outflows (such as NGC 1266; Alatalo et al. 2011, 2014).

The rapidity at which Mrk 231 is currently expelling its molecular material compared with weaker AGNs might be due to the different circumstances that created this system and outflow. A major merger occurred in a system already sufficiently massive to host a massive black hole, which is thus energetically capable of driving the bulk of the molecular gas from the system. Sub-${{v}_{{\rm esc}}}$ gas would be able to fall back into the system and inject turbulence, in effect temporarily stalling efficient star formation and possibly extending the depletion time just slightly. This would account for the mismatch in depletion timescales between ∼30 Myr (the depletion timescale should all outflowing gas escape and with the current star formation rate) and ∼50 Myr (the stellar population age seen in Mrk 231; Canalizo & Stockton 2000).

Deeper observations of the molecular outflow in Mrk 231, focusing on spatial resolution on at different distances away from the AGN will be able to determine what fraction of the molecular gas is escaping the gravitational potential, and provide a more complete understanding of the driving mechanism as well as the energy injection rate. Low frequency imaging could pinpoint the location of fossil shells (Schoenmakers et al. 2000), as Mrk 231 has already been detected at low frequency (Cohen et al. 2007). Combining these observations with the multiwavelength suite available for Mrk 231 will provide additional constraints, such as whether the molecular outflow changes phase as it accelerated, and whether there is any evidence that gas ultimately falls back into the molecular disk of Mrk 231.