DETECTION OF A METHANOL MEGAMASER IN A MAJOR-MERGER GALAXY

The isotropic luminosity of molecular megamasers (MM) is more than one million times greater than that of masers observed toward Galactic star forming regions. MM emission from hydroxyl (OH) and formaldehyde (H₂CO) is observed to trace the enhanced star formation regions in the central molecular zones (CMZ) of luminous infrared galaxies (LIRGs; Baan 1989; Darling & Giovanelli 2002; Baan et al. 1993). In contrast, H₂O MM are exclusively associated either with circumnuclear accretion disks or the jets of active galaxies (Miyoshi et al. 1995) rather than mergers or LIRGs, although kilomasers, which have luminosities of the order of 1000 times greater than typical Galactic masers can be found in LIRGs (e.g., Darling et al. 2008).

The intense maser emission observed from many transitions of methanol (CH₃OH) within the Milky Way has been widely used to study the formation and evolution of the most massive stars (e.g., Green et al. 2009; Chen et al. 2014). Galactic methanol maser transitions are empirically divided into two classes (Menten 1991). Class I methanol masers are produced in regions where molecular gas is mildly shocked by processes such as outflows (e.g., Kurtz et al. 2004; Chen et al. 2009, 2011) or expanding Hii regions (Voronkov et al. 2010) . They are pumped through collisional processes. Class II methanol masers are found in the molecular gas close to young high-mass stars and are pumped by far-infrared (FIR) radiation (Cragg et al. 2005).

Attempts to detect methanol masers in galaxies beyond the Milky Way have had limited success. Extragalactic methanol masers have only been detected in four galaxies, three of which are nearby galaxies: the Large Magellanic Cloud (LMC) and M31 for the 6.7 and/or 12.2 GHz class II transition detections (Green et al. 2008; Ellingsen et al. 2010; Sjouwerman et al. 2010), and NGC 253 for the 36.2 GHz class I transition detection (Ellingsen et al. 2014). The isotropic luminosity from these detected transitions is at most tens of thousands of times higher than observed for typical Galactic masers. Hence, they fall several orders of magnitude short of being megamasers. To date, more than 100 galaxies exhibiting OH or H₂O megamaser emission have been searched for methanol megamaser emission from the class II 6.7 GHz transition, but there have been no reported detections (Ellingsen et al. 1994; Phillips et al. 1998; Darling et al. 2003). However, recent detections of methanol maser emission from the class I 36.2 GHz transition toward the center of the Milky Way (Yusef-Zadeh et al. 2013) and starburst galaxy NGC 253 (Ellingsen et al. 2014) strongly suggest that methanol megamaser emission from this transition may be detectable in the central regions of galaxies. This is further supported by the recent report of methanol megamaser emission from the ${{5}_{-1}}\to {{4}_{0}}$ E (84.5 GHz) transition toward the Seyfert galaxy NGC 1068 (Wang et al. 2014).

In this paper, we report the first detection of both class I and class II methanol megamaser emission from the 36.2 and 37.7 GHz transitions toward the central region of Arp 220, the closest ultra-luminous infrared galaxy (adopted distance 77.6 Mpc; 1'' = 376 pc) and an advanced merger system.

The observations were made using ATCA in the H168 array configuration (baseline lengths between 61 and 192 m) on 2014 March 26. The synthesized beam width for the observations at 36 GHz was approximately $\sim 9^{\prime\prime} \times 6\buildrel{\prime\prime}\over{.} 6$. The data were collected using the Compact Array Broadband Backend (CABB; Wilson et al. 2011) configured with $2\times 2.048$ GHz bands, centered on frequencies of 35.3 and 37.3 GHz, respectively. The two bands cover the rest frequencies of the ${{4}_{-1}}\to {{3}_{0}}$ E class I and ${{7}_{-2}}\to {{8}_{-1}}$ E class II transitions of methanol at frequencies of 36.169265 and 37.703700 GHz, respectively (Müller et al. 2004). Figure 1 shows a rotational energy level diagram of E-type methanol with the targeted transitions marked. Each of the CABB bands has 2048 spectral channels, each of 1 MHz bandwidth, corresponding to a velocity width of 8.3 km s⁻¹. The observing strategy involved three cycles of 10 minutes onsource on Arp 220 (pointing center $\alpha \;=\;$ 15:34:57.2 $\beta \;=\;$+23:30:12 (J2000)) interleaved with 2 minute observations of a phase calibrator (1548 + 056). Thus a total onsource time of 30 minutes on Arp 220 was obtained. PKS B1938-634 and PKS B1921-293 were observed as the amplitude and bandpass calibrators, respectively.

Zoom In Zoom Out Reset image size

The data reduction was performed with MIRIAD using the standard techniques for ATCA observations. Atmospheric opacity correction was undertaken and the accuracy of the absolute flux density calibration is estimated to be ∼30%. Continuum emission from the central region of Arp 220 was used to produce a model for a number of iterations of phase-only self-calibration. The data for the 36.2 and 37.7 GHz methanol transitions were extracted from the self-calibrated data set through continuum subtraction and imaged at a spectral resolution of 20 km s⁻¹. The resulting rms noise in a single spectral channel for the 36.2 and 37.7 GHz methanol transitions was ∼4 mJy beam⁻¹.

In addition, observations which adopted the same CABB configuration setup and observing strategy used in the March observations were also made in a Directors' time allocation on 2014 May 16 with the 1.5D array configuration (baseline lengths between 107 and 1469 m; corresponding to a synthesized beam width of $\sim 4^{\prime\prime} \times 0\buildrel{\prime\prime}\over{.} 5$). A further 70 minutes on source for Arp 220 were obtained.

Continuum emission was detected toward the center of Arp 220 as a point source with an integrated intensity of 33.3 mJy at H168 array observations. This measured intensity is similar to that (∼60 mJy) recently measured at 32 GHz with the JVLA (Barcos-Muñoz et al. 2014) within a factor of ∼2. Using the data collected from this array, we imaged the barycentric velocity range 4600–6200 km s⁻¹ for both the 36.2 and 37.7 GHz methanol transitions with a velocity resolution of 20 km s⁻¹. Other molecular species within the central region of Arp 220 (e.g., CO and HCN) show emission over a velocity range from ∼5000–5900 km s⁻¹ (e.g., Martín et al. 2011; Greve et al. 2009). We summed the spectral channels within this velocity range to construct the integrated emission in the two methanol transitions. Figure 2 shows both the 36 GHz continuum emission and the integrated methanol emission of two methanol transitions with the optical Hα+[N ii] $\lambda \lambda$6548, 6583 intensity map (Taniguchi et al. 2012) as the background. The optical Hα emission is attributed to the ongoing starburst activity. From Figure 2, it can be seen that the majority of the emission from both methanol transitions is consistent with the distribution of the optical Hα emission, suggesting that the advanced merger provides the physical conditions required to excite maser emission for the two methanol transitions. Notably, strong Hα emission is present in the two starburst nuclei of Arp 220, however, no significant methanol emission was detected toward the central region of Arp 220 wherein more active starburst activity occurs. We discuss possible reasons for this in Section 4.2.

Zoom In Zoom Out Reset image size

At the angular resolution ($\sim 7^{\prime\prime}$) of the ATCA H168 array observations, two significant regions of 36.2 GHz methanol emission were detected. They lie southeast and northwest of the Arp 220 central region. The total extent of the methanol emission region is approximately 20'', which corresponds to 7.5 kpc for an assumed distance of 77.6 Mpc. Three significant regions of 37.7 GHz methanol emission were detected with offsets from the center of Arp 220 ranging from 7–15 '' (corresponding to 3–6 kpc).

We extracted the integrated spectrum of the two methanol transitions toward the regions where significant emission is observed in the integrated intensity images and show them in the left and right panels of Figure 2. The integrated spectra for these transitions exhibit spectral features with linewidths of several tens of km s⁻¹ toward each of these significant methanol emission regions. The integrated spectrum over all these emission regions for the 36.2 GHz transition (presented in Figure 3) shows two main broad (110–180 km s⁻¹) spectral components. The spectrum of the 37.7 GHz transition integrated over all of the emission regions (Figure 3) shows six narrow (30–60 km s⁻¹) features covering the velocity range of 5000–5900 km s⁻¹ (Greve et al. 2009). An additional redshifted feature for the 37.7 GHz transition at a velocity of ∼6000 km s⁻¹ may correspond to a separate outflow component. We performed Gaussian fitting of these spectral components and the parameters of the fits obtained are listed in Table 1.

Zoom In Zoom Out Reset image size

For comparison, the spectra of the two transitions measured from the higher angular ($\sim 1^{\prime\prime}$) and spectral resolution observations taken in the 1.5D array configuration (see Section 2) are also shown in Figure 3. Their spectra at a velocity resolution of 10 km s⁻¹ were extracted from the uv data by vector averaging toward the regions where methanol emission was observed in the images obtained from the H168 array observations. The higher angular resolution observations show the presence of narrower spectral components with line widths of less than about 10 km s⁻¹ for both methanol transitions as they were only detected in 1–2 channels. The majority of the methanol emission detected for the two transitions in the H168 array observations was not detected in the 1.5D array data. However, because of the limited uv coverage of the 1.5D array data, we cannot obtain useful images and cannot determine if the non-detection is because the emission is resolved by the longer baseline, or if it is due to atmospheric decorrelation.

4.1. The Nature of the 36.2 and 37.7 GHz Methanol Emission

4.2. The Pumping Mechanism for the Two Types of Methanol Masers

Essentially all of the molecular emission (e.g., CO) detected in previous observations of Arp 220 originates from the CMZ region, which has an extent of about 1 kpc and encompasses the two merger nuclei (Sakamoto et al. 2009). However,the 36.2 and 37.7 GHz methanol emissions in Arp 220 originate predominantly outside the CMZ, and this is similar to the situation observed for the 36.2 GHz methanol transition in NGC 253. The mechanism proposed by Yusef-Zadeh et al. (2013) to explain the 36.2 GHz methanol emission in the Milky Way CMZ is that high cosmic ray intensities (compared to the Galactic disk) play the key role in releasing methanol into the gas phase from cold dust grains and enhancing the abundance of methanol; however, once released into the gas-phase methanol will be rapidly destroyed if the cosmic ray intensity is too high. Ellingsen et al. (2014) suggested that this mechanism is consistent with the observed distribution of 36.2 GHz methanol emission in NGC 253, with the absence of any emission in the inner 100 pc being due to extreme cosmic ray fluxes. However, there is currently no direct observational evidence to support this speculation.

Figure 4 presents the methanol emission overlaid on the X-ray emission image of McDowell et al. (2003). This figure clearly shows that the 36.2 GHz methanol emission is highly correlated with the diffuse X-ray emission in the northwestern and southeastern regions, referred to as the X-ray "plume" by McDowell et al. (2003). The central nuclear region hosts higher intensity, more compact X-ray emission, but there is no significant 36.2 GHz methanol emission associated with that gas. We expect the X-ray emission to reflect the cosmic ray intensity in Arp 220. Hence, the observed close correspondence between the 36.2 GHz methanol and diffuse X-ray emission and the anti-correlation with high-energy X-ray emission supports the hypothesis that cosmic ray intensity governs the locations where there is a significant abundance of gas-phase methanol. The X-ray plume in Arp 220 is thought to be the result of a starburst-generated superwind (McDowell et al. 2003). The superwind represents regions with both rapidly outflowing and inflowing shock-heated gas, which may provide the collisional pumping agent for the widespread 36.2 GHz class I methanol emission. The methanol is pumped in the expanding shocks surrounding the X-ray cocoon and the distributed emission results from amplification of background or embedded radio continuum. The shocks within Galactic star formation regions produce class I maser emission distributed on scales of around 1 pc, while the plume region in Arp 220 has a scale two to three orders of magnitude larger.

Zoom In Zoom Out Reset image size

Within the Milky Way, class II methanol masers and OH masers are often co-located on arcsecond scales, although they are not cospatial on milliarcsecond scales (Menten et al. 1992; Li et al. 2012), and both species are pumped by emission at FIR wavelengths produced by warm dust (Cragg et al. 2005). In Arp 220, the OH and H₂CO megamaser emission is observed toward the positions of the two nuclei (see Figures 2 and 4), and are thought to result from FIR pumping and amplification of background radio continuum (Baan 1985, 1989; Rovilos et al. 2003). There is one previous detection of a class II methanol transition toward Arp 220, with a detection of the 6.7 GHz transition in absorption (rather than as a maser; Salter et al. 2008). The 37.7 GHz class II methanol emission regions observed in Arp 220 do not spatially correlate with the X-ray structure, but rather surround/straddle the region. As for the 36.2 GHz transition, 37.7 GHz emission arises on much larger scales than the OH and H₂CO masers and is significantly offset from the nucleus. The 37.7 GHz emission regions represent the location of cooler and denser gas at the edge of the diffuse X-ray emission, possibly in regions of triggered star formation, which may host gas with enhanced methanol abundance either produced in situ or driven there by outflows from the plume region. However, existing radio and infrared observations of Arp 220 show no significant counterparts from these regions.

We have detected the first 36.2 and 37.7 GHz methanol megasers toward the major merger galaxy Arp 220. The characteristics of the spectral lines and the spatial distribution of the 36.2 and 37.7 GHz methanol megamaser emission is significantly different from that of the megamasers of other species and the thermal molecular emission in Arp 220. The observed methanol megamasers trace larger-scale feedback (e.g., outflows) and star formation in the circumnuclear starburst regions of a merger system, in contrast to other species of megamaser which originate close to the nucleus. Therefore, methanol megamasers will trace a different physical environment and provide diagnostic and structural information that is complementary to those of other molecular transitions observed in starbursts and the feedback systems in active galaxies.

Our results are based on the observations made using the Australia Telescope Compact Array (ATCA), which is part of the Australia Telescope National Facility funded by the Commonwealth of Australia for operation as a National Facility managed by CSIRO. We thank Dr. Taniguchi Y. for providing optical Hα line data. X.C. acknowledges support from the National Natural Science Foundation of China (11133008 and 11273043), the Strategic Priority Research Program of the Chinese Academy of Sciences (CAS; grant No. XDA04060701), Key Laboratory for Radio Astronomy, CAS. W.A.B. has been supported as a Visiting Professor of the Chinese Academy of Sciences (KJZD-EW-T01) Shari Breen is the recipient of an Australian Research Council DECRA Fellowship (project No. DE130101270).