EDP Sciences
Free Access
Issue
A&A
Volume 607, November 2017
Article Number L7
Number of page(s) 4
Section Letters
DOI https://doi.org/10.1051/0004-6361/201730910
Published online 15 November 2017

© ESO, 2017

1. Introduction

Andromeda is a galaxy that lies in the green valley (Mutch et al. 2011; Tempel et al. 2011; Jin et al. 2014). According to Belfiore et al. (2016), it is typically a low-ionisation emission-line region (LIER), as first observed by Rubin & Ford (1971) and discussed by Heckman (1996). González-Martín et al. (2015) discussed that the torus is disappearing in LIER: there is indeed little gas in the inner part of M31 (Melchior et al. 2000; Melchior & Combes 2011, 2013). It is the closest external large galaxy in which we can explore the mechanisms that quenched the star formation activity. Optical ionised gas has been observed by Menezes et al. (2013) next to the black hole in a field of view1 of 5′′ × 3.5′′, but this emission is weak. Jacoby et al. (1985) estimated the ionised gas mass in the bulge (10′ × 10′) to be of the order of 1500 M. It also hosts a very massive black hole of 1.4 × 108M (Bender et al. 2005), but as studied by Li et al. (2011a), it is non-active and only murmurs at a level of 10-10LEdd. It hosts very little star formation of the order of 0.250.3 M yr-1, mainly located in the 10 kpc ring of the disc (e.g. Ford et al. 2013; Rahmani et al. 2016). Inside the central region (10′ × 10′), no obvious sign of star formation is detected (e.g. Kang et al. 2012; Azimlu et al. 2011; Amiri & Darling 2016), except for a central cluster of A stars formed 200 Myr ago that is located next to the black hole (within 1′′) (Lauer et al. 2012), designated by P3 by Bender et al. (2005). Viaene et al. (2014) estimated the star formation rate (SFR) on a pixel basis with panchromatic spectral energy distribution modelling. This infrared-based SFR estimated in the central pixel (36′′ × 36′′) is 4 × 10-5M yr-1, while an integration over the central region with a radius of 1 kpc corresponds to 1.25 × 10-3M yr-1. This negligible SFR is much lower than the value predicted by Rimoldi et al. (2016), considering supernovae remnants expected within the sphere of influence of quiescent supermassive black holes. For M31, an SFR of 0.13 Myr-1 is expected in the sphere of influence (RSOI = 14 pc = 3.7”) of its supermassive black hole. A past AGN activity is also expected, and the associated molecular torus, if it survives, should have a radius RMT = 25 pc = 6.7′′. In parallel, Chang et al. (2007) expected next to the black hole an accumulation of molecular gas (about 104M) originating from stellar feed-back. Melchior & Combes (2013) estimated a minimum molecular mass of 4.2 × 104M within 30″ from the centre, while about 106M of gas is expected from stellar feedback (e.g. Gallagher & Hunter 1981).

In this Letter, we present new observations of molecular gas with the IRAM Plateau de Bure interferometer (PdBI). We discuss the implications of the non-detection of gas next to the black hole.

thumbnail Fig. 1

IRAM-PdB CO(10) rms map. The colour-coding corresponds to the noise level (σrms), corrected for the primary beam, of the map obtained at the IRAM-PdB. The blue cross indicates the (optical) centre of Andromeda (Crane et al. 1992). The red crosses correspond to the centre of the four fields observed at PdB with a half-power primary beam of 43.7′′ represented by red circles. The half-power beam is presented in the bottom left corner of the CO(1−0) observations.

Open with DEXTER

2. Observations

The 3 mm observations were carried out at IRAM PdBI with the five-antenna configuration between 3 September and 24 December 2012 in C-array and D-array. The receivers were tuned to a frequency of 115.386 GHz, to account for the systemic velocity of M31 (300 km s-1). We used the WideX correlator, which provides a broad frequency range of 3.6 GHz and 2 MHz spectral resolution. We thus probe a velocity range of 9000 km s-1. The integration time is about six hours per field, but for the field at the offset (16.8, 21.5), which has been integrated, 34% lower: as shown in Fig. 1, the level of noise is higher in the south-western part of the data cube. After calibration within the GILDAS reduction package, the visibility set was processed with the MAPPING software. A one-iteration CLEAN deconvolution was applied to recover a primary-beam-corrected data cube close to the signal. The beam size of the final data was 3.37′′ × 2.44′′ (corresponding to 12.8 pc×9.3 pc) with a position angle PA = 70deg, while the data are sampled with a pixel size of 0.61. At the black hole position, we reach a sensitivity of 3.2 mJy/beam at 1σ with Δv = 5.1 km s-1, and 0.6 mJy/beam at 1σ with Δv = 304 km s-1, as displayed in Fig. 2. The extremities of the band are more noisy, and hence, a more optimistic value is found using the radiometer formula with 0.4 mJy/beam at 1 σ with Δv = 304 km s-1. We subsequently restrict our analysis to the velocity range (300, +300) km s-1. Figure 1 displays the rms map obtained for each channel, where the signal has been removed.

3. Analysis

Table 1

Detection of one small clump.

3.1. No accumulation of gas next to the black hole

If there were a gaseous disc/ring surrounding the black hole, as observed in the Milky Way (e.g. Dahmen et al. 1998; Kauffmann et al. 2017) and in external galaxies (Ricci et al. 2015), given the black hole mass and the stellar velocity dispersion, one would expect a molecular gas signal of 2 mJy with a line width of about 1000 km s-1 according to Chang et al. (2007). This corresponds to a 104M accumulation of molecular gas next to the black hole due to stellar feedback from the eccentric inner stellar disc.

Following Solomon & Vanden Bout (2005), we can derive the molecular hydrogen mass MH2 = αCO × LCO with αCO = 4.36 M (K km s-1 pc2)-1, and the line luminosity, in K km s-1 pc2: (1)with SCOΔV in Jy km s-1, DL in Mpc and νrest in GHz. Relying on the level of noise we have reached, we estimate an (3σ) upper limit of the molecular gas mass in the beam of 4300 M (2400 M) for a line width of 1000 km s-1 (300 km s-1 ). As discussed above, these values can be increased by 40% (30%) with a direct smoothing of the large band data.

thumbnail Fig. 2

Spectra of the central pixel smoothed at 304 km s-1. The dashed (dotted) lines correspond to the 1σ (3σ) interval, with σ = 0.6 mJy. The systemic velocity of M31 (300 km s-1) has been removed.

Open with DEXTER

thumbnail Fig. 3

Position of CO excess superimposed on 100 μm Herschel/PACS map. The contours (1.5, 2.6, 3.7, 4.9, and 6 MJy/sr) correspond to the Spitzer 8 μm map where the stellar continuum has been substracted by the adjustment of elliptical annuli using the standard surface photometry algorithm developed for IRAF (Jedrzejewski 1987). The projected field of view is 230 pc ×230 pc. Following Rimoldi et al. (2016), the white circle plotted with the dashed line corresponds to the expected extent of a molecular torus (RMT = 25 pc ~ 6.7″), while the white circle plotted with the full line corresponds to the radius of the sphere of influence of the black hole (RSOI = 14 pc ~ 3.7″), as displayed in Fig. 1.

Open with DEXTER

3.2. Small gas clumps: residuals of a past molecular torus?

We investigated the possibility that the gas might be present in the form of small clumps. Given the good velocity resolution and the quality of the interferometric data, we looked for statistically significant weak signals, higher than 3σ over a beam area, with the procedure described in Dassa-Terrier et al. (in prep.). The cores of the detected clumps are plotted in Fig. 3 as blue crosses, while the beam associated with each detected pixel is represented in grey.

We can note that there is an overall correlation with the dust emission detected at 100 μm detected with Herschel/PACS, while CO exhibits a better correlation with the 8 μm dust emission map from Spitzer, displayed as contours in Fig. 3. Next to the centre, the region of interest in this Letter, there is no dust emission at 100 μm. Within 12, there is only one clump with a peak flux above 6σ, which we discussed here. The other detections and their reliabilities will be discussed elsewhere (Dassa-Terrier et al., in prep.).

We extracted the spectrum of the remaining clump in GO VIEW in MAPPING/Gildas. We then analysed this spectrum with CLASS in Gildas. The results are provided in Table 1. We estimate a velocity dispersion σV< 5.9 ± 1.3 km s-1, which corresponds to an estimated mass of the clump of 2000 M. We checked that the signal is not resolved spatially as the integrated line in a single pixel or in the region of interest is about the same within the error bars. We thus assume that they have a typical projected size of Rapp< 5.2 pc (FWHM/2) corresponding to the beam size or a root-mean-squared (RMS) spatial size of σr< 4.6 pc.Fig. 4 displays the channel map corresponding to this signal, while Fig. 5 displays its spectra. In the following, we consider these upper values to discuss qualitative properties of this clump.

According to the Galactic velocity dispersion-size relation of Solomon et al. (1987), this clump lies above the σ/(1 km s-1)= (R/ (1 pc))0.5 relation. It has an apparent mass surface brightness 21 M/pc2 (computed within the FWHM area). It also lies well above the correlation between σv/R1/2 and Σ discussed by Heyer et al. (2009).With these characteristics, one can question if a 2000 M clump is gravitationally bound at a distance D = 9.1 pc from the black hole. It should have a density  cm-3 to resist the tidal force from the supermassive black hole (Chen et al. 2016). This is much higher than the apparent mean density napp ~ (MH2/mP)/ [4/3πRapp3] ~ 150 cm-3. This would correspond to a volume filling factor smaller than 3.5 × 10-5, which is well below the volume filling factor (~1%) estimated in Melchior & Combes (2016) in the gas present along the minor axis in the inner bulge. This suggests that the gas is clumped at a smaller scale, as observed in the Galaxy (e.g. Combes 2001; Heyer & Dame 2015). This would correspond to a clump size smaller than pc. In addition, one can note that the CO luminosity and derived mass are significantly lower than the main clumps observed in the Central Molecular Zone (e.g. Oka et al. 2001).

Last, one can consider constraints from the kinematics. A cloud in the gravitational field of the black hole would have a circular velocity of 360 km s-1: if one considers an inclination of 55 deg (Bacon et al. 2001), one would expect a velocity of 200 km s-1. If there are non-circular motions, this is compatible with the amplitude of our value of 176 km s-1. However, we would expect a velocity gradient within the beam of about 120 km s-1, while we detect a velocity dispersion of 13 km s-1. Figure 6 shows the location of the clump discussed here superimposed on an HST colour image (336275 nm) with respect to the position of the so-called P1, P2, and P3 components.

These are convincing signs that this clump is probably not exactly in the mid-plane: if it were at 100 pc along the line of sight, this gas clump would have the CO critical density and a filling factor of 1%. Then, it would not be in the sphere of influence of the black hole, and its real size and velocity dispersion are probably below our spectral and spatial resolutions.

Rosolowsky & Leroy (2006) discussed that diffuse emission can be missed as a result of interferometric filtering, but here we expect clumpy gas in this region with little filtering. In addition, these authors discussed that stable recovery of cloud properties can be achieved with a signal-to-noise ratio of at least 10, while we reach 6.6. It is thus difficult to push this type of analysis further.

3.3. Comparison with properties derived from infrared data

Figure 3 shows that there is a deficit of 100 μm emission next to the centre, in the region surrounding the black hole. Given the lack of resolution of infrared data, dust emission, which seems present in some maps in the near- and mid-infrared (Smith et al. 2012; Planck Collaboration Int. XXV 2015), corresponds to the emission in the central region, as displayed in Fig. 3. Viaene et al. (2014) estimated a dust mass of 475 M in a 36′′×36′′  pixel, which is 157 times higher than the PdB beam. The stellar continuum is also very strong next to the centre: Viaene et al. (2014) computed a stellar mass of 3.71 × 108M in the same region. This central stellar component accounts for the central dust heating, as discussed by Groves et al. (2012). This stellar mass corresponds to old stars and should produce some supernova Ia (SNIa) every 100 yr (Li et al. 2011b). Sjouwerman & Dickel (2001) have detected four supernovae remnants. These supernovae might maintain supersonic turbulence (Mac Low & Klessen 2004) and have contributed to expel the gas from the central region.

4. Discussion

We have detected one molecular gas clump with an estimated mass of 2000 M within 2.4 (9 pc) from the centre. Figure 6 displays a superposition of the central CO clump intensity on the 336275 nm PHAT/HST image of the central region. While our molecular gas is at the position angle of the P1-P2 disc, it is blueshifted, while the stellar disc is redshifted. As discussed in Melchior & Combes (2011), it could be rotating in a different orbit, while it could also be not exactly in the mid-plane and belong to a more distant component like the inner disc or inner ring. In addition, as argued in Sect. 3.2, the detected velocity dispersion is incompatible with the velocity gradient expected in the gravitation field of a large black hole. We conclude that this clump is outside the sphere of influence of the black hole.

thumbnail Fig. 4

Adjacent channel maps of the clump detected at 2.4′′ (9.1 pc) from the centre. The contours are represented with a step of 3.2 mJy/beam, namely 3.2, 6.4 9.6, 12.8, and 16 mJy/beam. The cross indicates the position of the optical centre. If it is not an outlier, seen close to the centre only in projection, it lies inside the sphere of influence of the black hole.

Open with DEXTER

thumbnail Fig. 5

Spectrum of the possible detection at 9.1 pc, potentially within the sphere of influence of the black hole. The systemic velocity of M31 (300 km s-1) has been removed.

Open with DEXTER

The upper limit we place on the mass of the molecular gas next to the black hole is not compatible with the expectations from stellar mass loss that is due to the inner eccentric stellar disc. Chang et al. (2007) predicted a gas mass of 104M as a result of stellar mass loss and a CO(10) signal of 2 mJy and a line width of 1000 km s-1 within 1 pc of the black hole. Our observations exclude it at the level of 8.8σ. Given the simultaneous absence of dust emission (Viaene et al. 2014) and of molecular gas (this work), we can argue that the stellar feed-back produced by the inner eccentric stellar disc population does not accumulate next to the black hole. Their model expects a compact source, which should have been detected by our interferometric observations. We can then question whether their proposed mechanism to produce the P3 star cluster is valid. Our result confirms that the current absence of star formation next to the black hole (e.g. Rosenfield et al. 2012; Groves et al. 2012) is due to the lack of gas. As discussed in Melchior & Combes (2011), the scenario of a frontal collision with M32 as proposed by Block et al. (2006) could account for the two-ring morphology as well as for the absence of gas in the centre. In addition, the presence of a bipolar outflow of soft X-ray emission, detected along the minor axis by Li & Wang (2007), Bogdán & Gilfanov (2008) and possibly triggered by supernova explosions, could account for part of the missing gas.

thumbnail Fig. 6

336 nm 275 nm PHAT map (Dalcanton et al. 2012): the blue crosses correspond to the position of the so-called P1 and P2, the two maxima of this region (Bender et al. 2005). The black cross corresponds to our reference point, quoted by Crane et al. (1992) as the “optical nucleus” from de Vaucouleurs & Corwin (1985). The star refers to the radio nucleus detected by Crane et al. (1992): it is in good agreement with the P2 position. In the bottom left corner, we show the IRAM-PdB beam of our observations. The black contours correspond to our detection within the sphere of influence of the black hole (red circle): we summed the three main channels and display the contours above 450 mJy km s-1.

Open with DEXTER


1

For compatibility reasons, we assume a distance to M31 of 780 kpc throughout, that is, 1 arcsec ~3.8 pc, following Melchior et al. (2000).

Acknowledgments

Based on observations carried out with the IRAM Plateau de Bure Interferometer. IRAM is supported by INSU/CNRS (France), MPG (Germany), and IGN (Spain). We acknowledge the IRAM Plateau de Bure team for the observations. We thank Sabine Kœnig for her support with the data reduction. This project has benefited from support from the Programme National Cosmologie et Galaxies, Specific Action ALMA from INSU and Action Fédératrice Cosmologie et Structuration de l’Univers from the Paris Observatory. We are most grateful to the anonymous referee for the very constructive comments that helped us to substantially improve the manuscript. Special thanks go to Nelson Caldwell for providing a PHAT extract of the central field of Andromeda.

References

All Tables

Table 1

Detection of one small clump.

All Figures

thumbnail Fig. 1

IRAM-PdB CO(10) rms map. The colour-coding corresponds to the noise level (σrms), corrected for the primary beam, of the map obtained at the IRAM-PdB. The blue cross indicates the (optical) centre of Andromeda (Crane et al. 1992). The red crosses correspond to the centre of the four fields observed at PdB with a half-power primary beam of 43.7′′ represented by red circles. The half-power beam is presented in the bottom left corner of the CO(1−0) observations.

Open with DEXTER
In the text
thumbnail Fig. 2

Spectra of the central pixel smoothed at 304 km s-1. The dashed (dotted) lines correspond to the 1σ (3σ) interval, with σ = 0.6 mJy. The systemic velocity of M31 (300 km s-1) has been removed.

Open with DEXTER
In the text
thumbnail Fig. 3

Position of CO excess superimposed on 100 μm Herschel/PACS map. The contours (1.5, 2.6, 3.7, 4.9, and 6 MJy/sr) correspond to the Spitzer 8 μm map where the stellar continuum has been substracted by the adjustment of elliptical annuli using the standard surface photometry algorithm developed for IRAF (Jedrzejewski 1987). The projected field of view is 230 pc ×230 pc. Following Rimoldi et al. (2016), the white circle plotted with the dashed line corresponds to the expected extent of a molecular torus (RMT = 25 pc ~ 6.7″), while the white circle plotted with the full line corresponds to the radius of the sphere of influence of the black hole (RSOI = 14 pc ~ 3.7″), as displayed in Fig. 1.

Open with DEXTER
In the text
thumbnail Fig. 4

Adjacent channel maps of the clump detected at 2.4′′ (9.1 pc) from the centre. The contours are represented with a step of 3.2 mJy/beam, namely 3.2, 6.4 9.6, 12.8, and 16 mJy/beam. The cross indicates the position of the optical centre. If it is not an outlier, seen close to the centre only in projection, it lies inside the sphere of influence of the black hole.

Open with DEXTER
In the text
thumbnail Fig. 5

Spectrum of the possible detection at 9.1 pc, potentially within the sphere of influence of the black hole. The systemic velocity of M31 (300 km s-1) has been removed.

Open with DEXTER
In the text
thumbnail Fig. 6

336 nm 275 nm PHAT map (Dalcanton et al. 2012): the blue crosses correspond to the position of the so-called P1 and P2, the two maxima of this region (Bender et al. 2005). The black cross corresponds to our reference point, quoted by Crane et al. (1992) as the “optical nucleus” from de Vaucouleurs & Corwin (1985). The star refers to the radio nucleus detected by Crane et al. (1992): it is in good agreement with the P2 position. In the bottom left corner, we show the IRAM-PdB beam of our observations. The black contours correspond to our detection within the sphere of influence of the black hole (red circle): we summed the three main channels and display the contours above 450 mJy km s-1.

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.