MASSIVE MOLECULAR GAS FLOWS IN THE A1664 BRIGHTEST CLUSTER GALAXY

We detected and imaged both CO(1–0) and CO(3–2) rotational transition lines in the BCG of A1664. Figure 1 shows the continuum-subtracted total spectral line profiles extracted from a 6'' × 6'' region at CO(1–0) and a 3'' × 3'' region at CO(3–2). A larger 6'' × 6'' region at CO(3–2) produces a consistent total flux but a significantly noisier spectrum. Both the CO(1–0) and CO(3–2) line profiles show two distinct components: a broad component centered on the BCG's systemic velocity and a narrower high-velocity system (HVS) blueshifted to −570 km s⁻¹. These spectral profiles were each fit with two Gaussian components using the package mpfit (Markwardt 2009) and the best-fit results are shown in Table 1. Although the BCG's systemic component is significantly broader, the peak flux for the HVS is roughly a factor of two higher, producing a similar integrated intensity for each component. The total CO(1–0) integrated intensity of $3.2\pm 0.4\,{\rm Jy\rm \; km\rm \; s^{-1}}$ is roughly half of the IRAM single-dish signal found by Edge (2001). This may be due to uncertainties in the continuum baseline subtraction or may indicate more extended emission filtered out by the interferometer or lying below our sensitivity. Missing short spacings will filter out extended emission at scales larger than ∼9'' at CO(1–0) and ∼3'' at CO(3–2).

Zoom In Zoom Out Reset image size

Integrated intensity maps of the CO(1–0) and CO(3–2) emission are displayed in Figures 2 and 3 with no threshold value adopted on the intensities. This combination gives a higher resolution image of the molecular gas near the nucleus at CO(3–2) and sensitivity over more extended scales at CO(1–0). The bulk of the molecular gas is concentrated in the galaxy center within a ∼3'' radius at CO(1–0) and a ∼1.5'' radius at CO(3–2). We have produced contours from integrated intensity maps covering the velocity range of the BCG's systemic component (red) and the HVS (blue and green) to show their spatial extent. Both maps show a similar morphology with these two components separated spatially as well as spectrally. The kinematics of these two components broadly match those found in Hαあるふぁ, Paαあるふぁ, and ro-vibrational H₂ integral field unit (IFU) observations of the BCG (Wilman et al. 2006, 2009).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In CO(1–0), the BCG's systemic component extends ∼5'' (11 kpc) from northeast to southwest, covering a velocity range from −250 km s⁻¹ to +250 km s⁻¹ and roughly peaking on the continuum position. The line emission from this component is highly asymmetric about the nucleus with the greatest extent toward the northeast. There is approximately twice as much CO(1–0) flux to the northeast of the nucleus compared with the southwest.

The HVS extends ∼4'' (9 kpc) from northwest to southeast and covers a velocity range from −700 to −450 km s⁻¹. It appears marginally to the north and east of the nucleus and, given the uncertainty in its position along the line of sight through the galaxy, it is not clear if the HVS physically interacts with the BCG's systemic component.

The CO(3–2) map shows a morphology broadly similar to the CO(1–0) but reveals more complex structure on smaller scales. There is an apparent drop in the intensity of the BCG's systemic component where it overlaps with the HVS, possibly suggesting a physical interaction between these components. The southwest blob of the BCG's systemic component is roughly spatially coincident with the AGN. The HVS may separate into two gas blobs, each contained within a projected diameter of ∼2 kpc. One clump is projected immediately to the north of the nucleus, where the deficit of emission is seen in the BCG's systemic component. The second high-velocity clump appears ∼3 kpc in projection to the southest and extends toward the northern clump. It could be one extended filament.

In Figures 2 and 3, the HVS has been subdivided into higher (blue) and lower (green) velocity contours to show that the lower velocity gas in this component tends to lie at larger distances from the nucleus. There is a shift in the velocity of the gas in the HVS at CO(3–2) from −510 ± 20 km s⁻¹ in the southeast blob to −590 ± 10 km s⁻¹ in the northern clump. The FWHM of the HVS at CO(3–2) is ∼230 ± 30 km s⁻¹ in the southeast blob but drops to 130 ± 10 km s⁻¹ in the clump to the north of the nucleus. Although the separate clumps of the HVS are spatially unresolved at CO(1–0), there is a shift in the velocity from −545 ± 9 km s⁻¹ to −600 ± 10 km s⁻¹ along the HVS filament toward the nucleus and a corresponding decrease in the FWHM from 200 ± 20 km s⁻¹ to 130 ± 30 km s⁻¹, consistent with the CO(3–2) results.

The gas clumps labeled A and B in Figure 2 appear to be separate structures with velocities of −400 ± 20 km s⁻¹ and −370 ± 40 km s⁻¹, respectively. Using a single Gaussian component fit to each spectrum, the integrated intensity was $0.17\pm 0.07\,{\rm Jy\rm \; km\rm \; s^{-1}}$ for gas clump A and $0.3\pm 0.1\,{\rm Jy\rm \; km\rm \; s^{-1}}$ for gas clump B. Gas clump A is also observed at CO(3–2) with an integrated intensity of $0.8\pm 0.2\,{\rm Jy\rm \; km\rm \; s^{-1}}$ but gas clump B is not detected.

Figure 4 shows the position–velocity (PV) cuts along P.A. =77° (defined east from north) across the BCG's systemic component from northeast to southwest at CO(1–0) and CO(3–2). A two-dimensional Gaussian was fit to the integrated intensity maps of the BCG systemic component to determine this best-fit P.A. and this was found to be consistent at both CO(1–0) and CO(3–2) within the error. The PV slices summed the line emission across the width of the BCG's systemic component. The CO(1–0) PV diagram shows a smooth velocity gradient from −250 km s⁻¹ to +250 km s⁻¹, with no visible flattening of the gradient at large radii. The CO(3–2) emission also traces this steady velocity gradient. This appears consistent with a flattened disk-like structure rotating in the rest frame of the BCG but the radial velocity profile is not clear. The velocity structure is similar to that of the rotating gas disk in Hydra A (e.g., Simkin 1979; Ekers & Simkin 1983; Wilman et al. 2005; Hamer et al. 2013). The AGN appears offset from the gas dynamical center by ∼0.5'' but this may instead be due to the uncertainty in the systemic velocity. The emission is clumpy and strongly asymmetric about the nucleus at both CO(1–0) and CO(3–2), suggesting the disk may be only partially filled (e.g., Wilman et al. 2005).

Zoom In Zoom Out Reset image size

Figure 5 shows a PV slice along the HVS from southeast to northwest with a best-fit P.A. =120° at CO(3–2). The HVS may separate into two gas blobs, each with a very broad velocity dispersion of close to 300 km s⁻¹. The velocity at the peak intensity shifts from ∼ − 450 km s⁻¹ to ∼ − 600 km s⁻¹ from the southeast to the northwest blob but there is not a clear velocity gradient for the intervening gas, as found for the BCG's systemic component. The PV diagram for the HVS at CO(1–0) shows a similar velocity structure but at lower spatial resolution.

Zoom In Zoom Out Reset image size

From the integrated CO(1–0) intensities, S_COΔでるたv, in Table 1, we calculated the molecular gas mass:

$$\begin{equation} M_{\rm mol} \,{=}\, 1.05\times 10^4 X_{\rm CO}\left(\frac{1}{1+z}\right) \left(\frac{S_{\rm CO}\Delta v}{\,{\rm Jy\rm \; km\rm \; s^{-1}\,}}\right) \left(\frac{D_{\rm L}}{{\rm Mpc}}\right)^2{\, M_{\odot }}, \end{equation} \tag{ 1 }$$

where X_CO is the CO-to-H₂ conversion factor (${{\rm cm^{-2}}\, (\rm K\,{\rm km\rm \; s^{-1}\,})^{-1}}$), D_L is the luminosity distance, and z is the BCG redshift (e.g., Solomon et al. 1987; Solomon & Vanden Bout 2005; Bolatto et al. 2013). Under the assumption of the Galactic value $\rm X_{CO}=2\times 10^{20}\,{\,{\rm cm^{-2}\,}(\rm \; K\,{\rm km\rm \; s^{-1}\,})^{-1}}$, we find a molecular gas mass for the BCG's systemic component of M_disk = (5.7 ± 1.0) × 10⁹ M_☉ and a mass for the HVS of M_HVS = (5.0  ±  0.7) × 10⁹ M_☉. The total molecular gas mass in the BCG is then (1.1  ±  0.1) × 10¹⁰ M_☉. The two separate gas clumps, A and B (Figure 2), have molecular gas masses of (6  ±  2) × 10⁸ M_☉ and (1.0  ±  0.3) × 10⁹ M_☉, respectively. The true value of X_CO may depend on environmental factors such as the gas phase metal abundance, temperature, density, and dynamics.

The central gas surface density of ∼200  M_☉ pc⁻² is not unusual for a star-forming disk (e.g., Bolatto et al. 2013). The star-forming regions produce U-band continuum emission from which Kirkpatrick et al. (2009) determine a star formation rate of ∼20  M_☉ yr⁻¹. There is no bright point source associated with the AGN. For a radius of ∼3 kpc, the surface densities of star formation and molecular gas are ${\rm log}\Sigma _{\rm SFR}=-0.2\,{\,{{\, M_{\odot }}\rm \; yr^{-1}\,}\rm \; kpc^{-2}}$ and logμみゅー_CO = 2.3  M_☉ pc⁻², respectively. The A1664 BCG lies with circumnuclear starbursts and central regions of normal disks on the Schmidt–Kennicutt relation (Kennicutt 1998), showing that despite the massive gas flows it is similar to other normal galaxies. The surrounding X-ray atmosphere, from which the molecular gas presumably cooled, also has a subsolar metallicity (0.5–1  Z_☉, Kirkpatrick et al. 2009). Therefore, we argue that the Galactic X_CO value is likely to be reasonable. Even if X_CO lies a factor of a few below the Galactic value, the molecular gas flows reported here would still exceed 10⁹ M_☉, which would not qualitatively alter our conclusions (e.g., McNamara et al. 2013).

Kirkpatrick et al. (2009) noted that at least four galaxies are projected against the envelope of the BCG, raising the possibility of a merger. Although no galaxy is observed coincident with the HVS, the central optical morphology is complicated by bright knots of star formation and dust regions (Figure 6; O'Dea et al. 2010). It is therefore difficult to rule out a collision but there are several factors weighing against this hypothesis. The BCG resides at the center of a dense cluster atmosphere. A galaxy plunging into the center would likely be stripped of most of its gas before it reached the BCG (e.g., Vollmer et al. 2008; Kirkpatrick et al. 2009; Jablonka et al. 2013). Although molecular gas is more resilient to ram pressure stripping than atomic gas, the molecular gas masses in the BCG and the HVS are unusually large and there are few galaxies in nearby clusters harboring this much gas (e.g., Combes et al. 2007; Young et al. 2011). In the context of a merger scenario, the different velocities of the BCG's systemic component and the HVS may require two unlikely merger events. These interactions would also have to be close to direct hits to the BCG center, which again would be unlikely (e.g., Benson 2005).

Zoom In Zoom Out Reset image size

It is more likely that such large quantities of molecular gas originated in gas cooling from the hot cluster atmosphere (Edge 2001; Salomé & Combes 2003). In BCGs, there are clear associations between the shortest radiative X-ray cooling times, the strength of optical line emission, CO and H₂ emission, and star formation (Cowie et al. 1983; Crawford et al. 1999; Peres et al. 1998; Rafferty et al. 2008). The BCG in A1664 is among the brightest Hαあるふぁ emitters in the Crawford et al. (1999) ROSAT brightest cluster sample, the Hαあるふぁ emission aligns with the soft X-ray emission (Allen et al. 1995), and the X-ray radiative cooling time in the center is short (0.4 Gyr; Kirkpatrick et al. 2009). In the nearby BCG NGC 1275, the CO filaments are co-spatial with the coolest X-ray gas and the outer two filaments have radial velocity gradients with larger blueshifted velocities at smaller radii (Salomé et al. 2006; Lim et al. 2008). The molecular gas kinematics are consistent with free fall in the gravitational potential of NGC 1275, as expected if they originated in cooling from the X-ray atmosphere.

For radial inflow, we expect velocity gradients increasing toward the nucleus with the largest velocities toward the BCG center. Although there is not a clear, smooth velocity gradient across the HVS, the lower velocity gas in this component does tend to lie at a greater distance from the nucleus, suggesting an inflow is plausible. The extent and velocity shifts across the HVS gas clumps and the BCG's systemic component can be used to constrain their dynamics and origin (Figure 3). The velocity shear across the gas clumps in the HVS is ∼250–300 km s⁻¹ along the line of sight (Figure 5). A similar shear in the transverse direction could have separated the clouds on the observed scale in only ∼10⁷ yr. Therefore, the clumps are of order that age or the clouds are moving nearly along the line of sight. In the latter case, the broad velocity shear could signify a velocity gradient with high-velocity infall onto the nucleus.

The BCG's systemic component has a clear and smooth velocity gradient with a velocity increase of ∼500 km s⁻¹ over its ∼11 kpc length. The velocity profile of this component is suggestive of rotation about the BCG center; however, the mass and velocity structure are strongly asymmetric (Section 3.1). For a putative rotating disk, we estimate the Toomre Q criterion for disk stability (Toomre 1964), Q = v_cv_T/πぱいGrΣしぐま, where the circular velocity is v_c ∼ 200 km s⁻¹, the turbulent velocity is v_T ∼ 50–100 km s⁻¹, the disk radius is r ∼ 3 kpc, and the disk surface density is Σしぐま ∼ 200  M_☉ pc⁻². Although highly uncertain, we estimate that the Q parameter is of order one, which is a reasonable value for a normal disk and suggestive that it could be unstable and potentially star forming. However, Figure 6 does not show strong star formation associated with the BCG's systemic component. Star formation could be obscured by dust but the U-band star formation rate, SFR_U ∼ 20  M_☉ yr⁻¹ (Kirkpatrick et al. 2009), and the infrared star formation rate, SFR_IR = 15  M_☉ yr⁻¹ (O'Dea et al. 2008), are comparable, which is inconsistent with buried star formation. This could be a recent infall of material that is in the process of settling into a disk around the nucleus. Its orbital time is ∼9 × 10⁷ yr so a relaxed disk will take at least several 10⁸ yr to form.

An X-ray cooling origin for the gas inflow must explain how molecular hydrogen can form in this environment. It is difficult to form molecular hydrogen rapidly in these systems by a process other than catalytic reactions on dust grain surfaces (e.g., Fabian et al. 1994). Hot ions in the X-ray atmosphere will destroy unshielded dust on much shorter timescales than the radiative cooling time (e.g., Draine & Salpeter 1979). Yet, dust does appear to be a ubiquitous feature of the colder gas phases in these systems (e.g., Egami et al. 2006; Donahue et al. 2007; Quillen et al. 2008; Mittal et al. 2011). Fabian et al. (1994) have speculated that dust may somehow form deep within very cold gas clouds, but ab initio dust formation by coagulation of atoms is likely to be a slow process and difficult in an environment where X-rays can quickly destroy small clusters of atoms (e.g., Voit & Donahue 1995). If dust-rich stellar ejecta from the BCG and ongoing star formation can be mixed into cooling X-ray gas before the dust grains are destroyed by sputtering, then those seed grains may act as sites for further dust nucleation (Voit & Donahue 2011; Panagoulia et al. 2013). Dusty gas clouds could be dragged out from the BCG by buoyantly rising radio bubbles and promote subsequent molecular formation (see Section 4.4.2; e.g., Ferland et al. 2009; Salomé et al. 2011).

Without additional absorption line observations, it is difficult to determine whether the gas flows lie in front or behind the BCG along the line of sight. Therefore, we cannot distinguish between an inflow of gas located behind the BCG and an outflow of gas located in front of the BCG. Figure 6 shows that the molecular gas clumps appear coincident with regions of lower optical–ultraviolet (UV) surface brightness in the BCG, which could indicate either a physical reduction in the emission or dust obscuration. Gas clumps A and B are associated with bright optical–UV knots. The complexity of the structure makes a clear interpretation difficult but it is plausible that the dusty molecular gas lies predominantly in front of the BCG and that the blueshifted velocities indicate an outflow.

4.4.1. Driving an Outflow with Radiation Pressure or Supernovae

4.4.2. Driving an Outflow with a Radio AGN