EDP Sciences
Free Access
Issue
A&A
Volume 616, August 2018
Article Number L1
Number of page(s) 7
Section Letters to the Editor
DOI https://doi.org/10.1051/0004-6361/201833684
Published online 13 August 2018

© ESO 2018

1. Introduction

The Galactic center (GC), in particular the neighborhood of the central supermassive black hole (SMBH) at Sgr A* (Schödel et al. 2002; Ghez et al. 2008; Gillessen et al. 2009), represents a unique laboratory to understand the properties of interstellar gas in extreme conditions: intense UV fields and powerful winds from massive stars, shocks, strong gravitational shears, and complicated orbital motions (Morris & Serabyn 1996; Genzel et al. 2010). The interstellar material in the innermost region of our Galaxy consists of a cavity of radius ~1.5 pc containing warm dust and gas, ionized by UV radiation from about 100 OB-type massive stars of the central cluster (e.g., Krabbe et al. 1995). Some of the ionized gas streamers (the mini-spiral) seem to bring material close to the very center (e.g., Zhao et al. 2009; Irons et al. 2012; Lau et al. 2013). Between ~1.5 pc and ~2.5 pc, a ring of denser molecular gas and cooler dust exists (the CND, e.g., Guesten et al. 1987; Bradford et al. 2005; Montero-Castaño et al. 2009; Requena-Torres et al. 2012; Mills et al. 2013).

Owing to hostile conditions inside the cavity of the CND, the presence of molecular gas was originally not expected. Recent studies, however, point toward its existence (e.g., Herrnstein & Ho 2002; Goicoechea et al. 2013; Ciurlo et al. 2016; Moser et al. 2017; Mills et al. 2017; Yusef-Zadeh et al. 2017). Indeed, obtaining high-resolution spectral images of the molecular gas emission close to Sgr A* is currently feasible with radio interferometers such as ALMA (e.g., Moser et al. 2017; Yusef-Zadeh et al. 2017). In particular, Goicoechea et al. (2018) have detected a collection of 12CO (J = 3−2) cloudlets emitting at higher positive velocities than reported before (up to υLSR ≃ +300 km−1). Owing to the small cloudlet sizes (≲20 000 AU), broad line profiles (Δυ ≃ 25−40 km s−1), but similar line-velocity centroids over the cavity, these authors have suggested that these cloudlets are leftovers of disrupted clouds that fall into the central region. Other dissipative effects such as shocks in colliding streams, cloudlet–cloudlet collisions, colliding stellar winds, or advancing photoionization fronts, perhaps triggering gas compression and molecular gas formation, may also explain the presence of molecular gas in the cavity (Goicoechea et al. 2018).

Most of the luminosity emitted by molecular gas heated by shocks or illuminated by strong UV fields (leading to Tk > 100 K) is radiated in mid-J and high-J lines at submillimeter and far-IR wavelengths that are inaccessible to ground-based telescopes. In particular, observations of the CO rotational ladder (e.g., Indriolo et al. 2017; Joblin et al. 2018) are critical tools to determine the heating mechanisms and nature of the molecular gas in the GC and, in a broader context, to better understand the emission from distant active and star-forming extragalactic nuclei (e.g., van der Werf et al. 2010; Rangwala et al. 2011).

2. Observations and data reduction

The spectral scan toward Sgr A* was carried out with the heterodyne receiver HIFI (de Graauw et al. 2010) on board Herschel (Pilbratt et al. 2010). Owing to intense molecular line emission around the CND area, these observations were performed in the Spectral Scan Load Chop mode using an internal cold load as reference. We used the Wide Band Acousto-Optical Spectrometer, which provides a spectral resolution of 1.1 MHz and instantaneous bandwidths of 4 (2.4) GHz in bands 1−5 (6−7). The total observing time was ~40 h. Some of the frequency ranges were affected by isolated spurious signals, together with distorted residual baselines due to imperfect gain drift correction. While these signals were treated by the HIFI pipeline through dedicated line masking, the second artifact was corrected by fitting polynomial baselines on each spectrum prior to deconvolution (Comito & Schilke 2002). Data calibration was performed in HIPE (Ott 2010). Deconvolution allows the recovery of both line and continuum on a single side-band (SSB) intensity and frequency scale, allowing accurate absorption and emission studies. We adopted beam efficiencies and half power beam widths (HPBW) from the latest HIFI beam model1 and used the main-beam temperature scale (Tmb). The rms of the SSB spectra (per 2 km s−1 channel) are ~15, ~25, ~40, ~60, and ~120 mK for bands 1−5, respectively, and ~175 mK for the [C II]158 μm line. Table B.1 lists the total integrated line intensities with global uncertainties up to ≃15%, the mean velocity (<υLSR>), velocity dispersion (<Δυ>), and HPBW at multiple line frequencies.

We complemented pointed observation by HIFI with a ~2′ × 2′ map of the 12CO J = 3−2 line (345.796 GHz) at ~7″ angular resolution, observed with the IRAM 30 m telescope (Pico Veleta, Spain) using the EMIR330 receiver and the FFTS backend at 200 kHz resolution. The map was carried by scanning, on the fly, in two orthogonal directions on the sky. Owing to very extended 12CO emission in the GC, the map was taken using a nearby reference position, at (+324″, −124″) with respect to Sgr A* at (0″, 0″), which we successively calibrated using increasingly distant pointed observations ( RA = +1000″, +2000″, +3000″, +4000″, and +5000″). The total integration time was 3.6 h, leading to a rms of ~1.5 K per 2 km s−1 channel. The spectra were baselined and gridded to a data cube through convolution with a Gaussian kernel providing a final resolution of ~10″. The total integrated intensity map is shown in Fig. 1.

thumbnail Fig. 1.

12CO (J = 3−2) integrated intensity map of the inner 2′ of the Galaxy obtained with the IRAM 30 m telescope and dominated by emission from the CND (color scale in K km s−1). The square box shows the field of view observed by ALMA in 12CO (J = 3−2) revealing high positive-velocity cloudlets inside the cavity (Goicoechea et al. 2018). The gray curves show representative orbits CND with 1.5 and 2.3 pc de-projected radii. The black, blue, and gray dashed circles centered at Sgr A* (black star) show HIFI’s HPBWs at multiple frequencies (from ~11″ to ~40″).

Open with DEXTER

3. Results: Detected lines and wing emission

We focus on the brightest detected emission lines. These include 12CO (J = 5−10) rotational lines from 13CO, C18O, HCO+, HCN (J = 6−8), CN, and H2O, and fine structure lines from ionized (C+) and neutral atomic carbon (C). Table B.1 tabulates their spectroscopic and observational parameters. Most of these were previously detected at low spectral resolution with SPIRE and PACS (Goicoechea et al. 2013). The much higher resolution of HIFI allows us to spectrally resolve these complicated line profiles and independently study the various velocity components seen toward the GC.

In addition to emission, H2O 11,0–10,1 and [C II]158 μm lines show absorption features below the continuum level. Compared to the CND, the continuum flux densities measured by HIFI toward Sgr A* are low, i.e., ~50 and ~100 Jy beam−1 at 557 and 809 GHz, respectively. This is due to the lower (and hotter) column densities of dust in the cavity (Etxaluze et al. 2011). The absorption at υLSR ≃ 0 km s−1 is produced by local diffuse clouds of the Galactic arms (Greaves & Williams 1994) and also by clouds in the GC itself (e.g., Goto et al. 2002). In addition, the H2O 11,0–10,1 line shows absorption at high negative velocities, vLSR ≈ −200 to −100 km s−1 (seen also in Sandqvist et al. 2003), generally attributed to high negative velocity gas clouds (−180 km s−1 feature, Guesten & Downes 1981) and to the near side of the, historically called, expanding molecular ring (EMR or −135 km s−1 feature, Scoville 1972). These features are produced by diffuse, low excitation molecular gas in the inner few hundred pc of the Galaxy (Goto et al. 2002) but they are not associated with Sgr A*.

In addition to emission at |υLSR| < 150 km−1 (typical of the CND), the mid-J12CO lines display a high positive-velocity wing (HPVW) and have emission up to υLSR ≃ +270 km s−1 (zooms in Fig. 2). The HPVW emission is faint (~500 mK) and is not seen in 13CO (J = 5−4), implying optically thin mid-J12CO emission. This component has not been discussed in previous observations of the CND (e.g., Bradford et al. 2005; Requena-Torres et al. 2012) but pioneering detection of broad 12CO J = 7−6 line widths toward Sgr A* suggested the presence of molecular gas within the cavity (Harris et al. 1985). Goicoechea et al. (2013) also reported broad [O I] 63 μm emission wings toward Sgr A* (up to ≃300 km s−1).

thumbnail Fig. 2.

Mid-J12CO and 13CO lines observed with HIFI toward Sgr A* (baseline subtracted). The two red vertical lines at υLSR ≃+120 and +270 km−1 delineate the HPVW emission.

Open with DEXTER

Central region ALMA images unveil the presence of smallscale 12CO (J = 3−2) cloudlets inside the cavity (shown in Fig. 1). These cloudlets emit in the same HPVW range, showing independent spectral features at υLSR = +150−200 km s−1 (high positive-velocity gas; HPVG) and at υLSR = +200−300 km s−1 (extreme positive velocity gas; EPVG). From the CO J = 3−2 maps, we determine an interferometer-alone to single-dish line flux ratio of ≃0.2 (HPVW range). As we expect the unresolved cloudlets to be hot (Goicoechea et al. 2013; Mills et al. 2017) the contrast between emission from the extended background and these cloudlets is likely to increase for the more excited CO lines, leading to larger flux ratios. In the following we attribute the origin of the mid-J 12CO HPVW emission to the same cloudlets. Their filling factor in the HIFI beams is also about 0.2.

H2O and C+ also show HPVW components (Fig. 3). Owing to its 11.3 eV ionization potential, the detection of [C II]158 μm implies that the high-velocity gas is irradiated by stellar far-ultraviolet photons (FUV; 5 ≲ E < 13.6 eV). The presence of H2O 11,0–10,1 HPVW emission, and of several rotationally excited H2O lines (Goicoechea et al. 2013) suggests elevated temperatures, such as those in shocks; the gas-phase routes leading to abundant water vapor are endothermic or possess energy barriers (van Dishoeck et al. 2013). Interestingly, the atomic carbon fine-structure lines [C I]492, 809 GHz do not show HPVW. Only the ground-state line at ~492 GHz displays an independent feature at υLSR ≃ +165 km s−1; the [C I]809 GHz line is barely detected in this velocity range. This feature is traditionally associated with the back side of the EMR. Indeed, the very low [C I]809/492 line intensity ratio ≲1 of this feature (with I492 and I809 in W m−2 sr−1) implies cold gas temperatures and thus is not related to the source of HPVW emission.

thumbnail Fig. 3.

HIFI multi-line detection toward Sgr A*. The two red vertical lines at υLSR ≃ +120 and +270 km−1 delineate the HPVW emission. The intensity scale is in Tmb units (baseline subtracted) except for H2O 11,0–10,1 that is in Tmb/TC (divided by the continuum).

Open with DEXTER

The HIFI scan also shows mid-J HCO+ and HCN emission lines (up to J = 8−7). These lines do not display HPVW emission component either (Fig. B.1). Their lower velocity emission range (|υLSR| < 150 km s−1) and high HCN/HCO+ intensity ratios (1.6–1.7; likely enhanced by IR pumping of vibrationally excited HCN lines, see e.g., Mills et al. 2013) agree with previous maps of the CND in the same species (Montero-Castaño et al. 2009; Mills et al. 2013). Given the large size of HIFI beams at these frequencies, very likely the HCN and HCO+ emission toward Sgr A* is beam-peaked emission from the CND.

4. Physical conditions of high-velocity CO gas

From the detection of a few low-energy rotational lines (e.g., with ALMA) it is not easy to constrain the physical conditions of the high-velocity cloudlets seen inside the cavity. With Herschel/HIFI we spectrally resolve the HPVW emission in several rotationally excited CO lines. From the observed lines, we determine the beam-averaged 12CO column density (N) and rotational temperature (Trot) associated with the HPVW component, which has line intensities integrated from +120 to +270 km s−1 (see Table B.2). Table 1 summarizes the inferred N(12CO) and Trot(mid-J 12CO) values based on rotational diagram fits (Fig. 4). The HPVW component shows a higher degree of excitation (Trot ≃ 100 K versus 70 K) and more pronounced curvature than the main spectral component, i.e., total integrated line intensity minus HPVW. As for HCN and HCO+, signal from the mid-J 12CO main component is likely dominated by emission from the CND and surroundings (Fig. 1).

Table 1.

Beam-averaged rotational temperatures, column densities, and correlation coefficients obtained from rotational diagram fits.

thumbnail Fig. 4.

Observed CO intensities of the main and HPVW components toward Sgr A* translated to rotational population diagrams. The green curve shows the best isothermal non-LTE model. The red and blue dashed curves show predictions of PDR models with G0 = 104.9 and ζCR = 10−15 s−1 (line intensities multiplied by 0.04).

Open with DEXTER

Assuming a 12CO abundance of 2 × 10−5 with respect to H nuclei (Goicoechea et al. 2018), the 12CO column density inferred from the HPVW rotational diagram is equivalent to AV ≃ 0.5 mag of visual extinction, or 2.5 mag if we assume a uniform beam-filling factor of 0.2 for the cloudlet ensemble. These values agree with the average extinction estimated from ALMA images (Goicoechea et al. 2018) and translate to a hot molecular gas mass of ~10–60 M in the inner 1.5 pc of the Galaxy. This mass is comparable to that of neutral atomic (~300 M, Jackson et al. 1993) and ionized gas (~60 M, e.g., Lo & Claussen 1983) inside the central cavity.

Given the strong stellar FUV radiation field in the region (G0 ≃ 105.3−104.3 in Habing units, Goicoechea et al. 2018), we might expect a dominant role of radiative heating. However, state-of-the-art photodissociation region (PDR) models run with the Meudon code (Le Petit et al. 2006) for G0 ≃ 104.9, a cosmic-ray ionization rate of ζCR = 10−15 s−1, and constant thermal pressures Pth/k = 108−109 K cm−3 do not reproduce the shape of the HPVW CO rotational ladder, especially as J increases (see Fig. 4). Similar conclusions were anticipated by Goicoechea et al. (2013) from (velocity unresolved) observations of higher J CO lines (up to J = 24−23). These authors suggested that only a very small filling factor (<0.05) ensemble of FUV-irradiated dense (nH ≈ 107 cm−3) clumps would be compatible with a dominant role of radiative heating (PDRs). However, the filling factor of the HPVW cloudlets determined from ALMA is larger and gas densities seem much lower; otherwise the cloudlets would be prominent in IR extinction maps.

In order to constrain the range of average physical conditions that reproduce the observed mid-J CO lines (HPVW component), we have run a grid of non-local, non-LTE excitation models (see Appendix A). While only for very high densities (nH > 107 cm−3) the gas is close to thermalization (TrotTk ≃ 100 K), at lower densities the excitation becomes subthermal (Trot = 100 K≪Tk), the implied gas temperature is higher, and the rotational population diagram shows curvature (Neufeld 2012). Our best fit is for Tk ≃ 1200 K and nH ≃ 104.5 cm−3 (green curve in Fig. 4). This model also fits the single-dish CO J = 3−2 emission fraction expected from the spatially unresolved high-velocity cloudlets (filling factor of ≃0.2). We note, however, that a more accurate solution will require us to observe and spectrally resolve higher J CO lines.

Goicoechea et al. (2018) estimated (assuming Trot(CO 3−2) = 100 K precisely) density peaks around nH ≃ (0.2−1.0) × 105 cm−3 for the cloudlets resolved by ALMA. For these densities, our non-LTE excitation models shown in Figure A.1 imply temperatures ranging from Tk ≃ 400−2000 K. Both the elevated Tk and low nH (compared to the local Roche density) are compatible with the non-thermal H2 level populations inferred from IR observations with the VLT (Ciurlo et al. 2016). These Tk and nH also agree with H2 multi-line detections (Tex(H2) ~ 1100 K, Mills et al. 2017) with ISO. Infrared observations provide lower spectral resolution, thus it is difficult to extract the different velocity components independently. In summary, our observations add more quantitative evidence to the existence of high-velocity, hot molecular cloudlets inside the cavity of the CND and close to Sgr A* (<1 pc). Although the detection of [C II]158 μm wing emission demonstrates the presence of FUV illumination, PDRs alone cannot drive the heating of the HPVW component. Instead, FUV-irradiated shocks, which have G0 values that are much higher than in the models of Lesaffre et al. (2013) and Melnick & Kaufman (2015), likely dominate the heating of the hot molecular gas.

Acknowledgments

We are indebted to Tom Bell for initiating this project and for his help in the early data reduction. We thank our anonymous referee for useful comments. We also thank the Spanish MCIU and the ERC for funding support under grants AYA2017-85111-P and ERC-2013-Syg-610256-NANOCOSMOS.


2

At low densities, nHncr(J), CO rotational level populations are such that plotted in a rotational diagram they show considerable curvature (Neufeld 2012). We also fit these models with a straight line.

References

Appendix A

Non-local, non-LTE excitation isothermal models

To constrain the range of plausible physical conditions that reproduce the mid-J CO line excitation and inferred Trot, HPVW values, we run a grid of non-local and non-LTE models (Monte Carlo model in Appendix of Goicoechea et al. 2006). We included 12CO-H2 and 12CO-H rotationally inelastic collisions (Yang et al. 2010; Balakrishnan et al. 2002) and assumed a molecular gas fraction of 90%, n(H) = 0.25 n(H2); this value is consistent with a strongly FUV-irradiated gas in which not all hydrogen is in molecular form. We adopted a non-thermal velocity dispersion of Δυ = 35 km s−1 (from turbulence and macroscopic gas motions). This value is the observed average linewidth of the cloudlets detected by ALMA. We take the same N(12CO) obtained from the rotational diagram and search for the range of Tk and nH = n(H) + 2n(H2) values that reproduce the derived Trot, HPVW ≃ 100 K and observed line intensities.

Figure A.1 shows isothermal model results in the form of iso- contours. For each model we created rotational diagrams and determined a rotational temperature () by fitting a straight line2 to the synthetic CO line intensities in the Jup = 5−10 range. To estimate the fit goodness we calculated the minimum rms value of log10(IHIFI/Imod). Figure A.1 shows rmsmin/rms contours of 1, 0.95 and 0.9 in magenta, red, and blue, respectively. The best model (rms = 0.076, indicated with a star in the figure) implies Tk ≃ 1200 K and nH ≃ 104.5 cm−3, but we note that a more accurate solution would require velocity-resolved observations of higher J CO lines.

thumbnail Fig. A.1.

12CO rotational temperatures (, computed by fitting a straight line to synthetic line intensities in the Jup = 5−10 range) obtained from isothermal non-LTE models. The red curve shows the parameter space that reproduces the observed Trot of the HPVW component. The blue-shaded area shows the estimated gas density of the high-velocity cloudlets detected by ALMA (Goicoechea et al. 2018). The magenta, red, and blue curves show rmsmin/rms contours of 1, 0.95, and 0.9 respectively, where rms refers to the rms value of log10 (IHIFI/Imod).

Open with DEXTER

Given the intense IR, visible, and UV radiation fields in the cavity, one may consider whether radiative pumping to CO vibrational and electronic levels can influence the emergent rotationally excited line intensities. For moderate gas densities (>104 cm−3) inelastic collisions are important drivers of CO rotational excitation. Thus very strong mid-IR fields are needed to alter the rotational level populations (e.g., Carroll & Goldsmith 1981; Godard & Cernicharo 2013). The estimated mid-IR color temperatures in the cavity range from ≈200 K (Cotera et al. 1999) to 145–85 K (Lau et al. 2013). These values are around the 160 K threshold estimated by Carroll & Goldsmith (1981) for CO pumping in the no-collision limit. Hence, we suspect that IR pumping can contribute locally, around evolved stars or in protostellar envelopes, but does not drive the excitation of the optically thin mid-J CO lines we detect at high velocities. Indeed, IR observations toward the cavity do not report high positive-velocity features in the wings of the saturated 12CO υ = 1−0 rovibrational lines (Goto et al. 2014).

Similarly, visible/UV pumping may have some relevance in strongly irradiated environments. Visible/UV pumping (included in the PDR models shown in Fig. 4) can enhance the intensity of the CO rotationally excited lines only by a factor of a few (Godard & Cernicharo 2013). Either way, these radiative excitation processes further decrease the density of the HPVW gas component below our estimation from the grid of non-LTE excitation models. Therefore, the high gas density solutions, nH ≃ 107 cm−3 (shown in Fig. A.1), do not seem favored for the molecular gas in the cavity.

Appendix B

Complementary figures and tables

thumbnail Fig. B.1.

HIFI detection of HCN and HCO+ lines toward Sgr A* showing emission only at velocities typical of the CND (υLSR < 150 km s−1).

Open with DEXTER

Table B.1.

Spectroscopic and observational parameters of the lines discussed in this work.

Table B.2.

Line intensities for the HPVW component.

All Tables

Table 1.

Beam-averaged rotational temperatures, column densities, and correlation coefficients obtained from rotational diagram fits.

Table B.1.

Spectroscopic and observational parameters of the lines discussed in this work.

Table B.2.

Line intensities for the HPVW component.

All Figures

thumbnail Fig. 1.

12CO (J = 3−2) integrated intensity map of the inner 2′ of the Galaxy obtained with the IRAM 30 m telescope and dominated by emission from the CND (color scale in K km s−1). The square box shows the field of view observed by ALMA in 12CO (J = 3−2) revealing high positive-velocity cloudlets inside the cavity (Goicoechea et al. 2018). The gray curves show representative orbits CND with 1.5 and 2.3 pc de-projected radii. The black, blue, and gray dashed circles centered at Sgr A* (black star) show HIFI’s HPBWs at multiple frequencies (from ~11″ to ~40″).

Open with DEXTER
In the text
thumbnail Fig. 2.

Mid-J12CO and 13CO lines observed with HIFI toward Sgr A* (baseline subtracted). The two red vertical lines at υLSR ≃+120 and +270 km−1 delineate the HPVW emission.

Open with DEXTER
In the text
thumbnail Fig. 3.

HIFI multi-line detection toward Sgr A*. The two red vertical lines at υLSR ≃ +120 and +270 km−1 delineate the HPVW emission. The intensity scale is in Tmb units (baseline subtracted) except for H2O 11,0–10,1 that is in Tmb/TC (divided by the continuum).

Open with DEXTER
In the text
thumbnail Fig. 4.

Observed CO intensities of the main and HPVW components toward Sgr A* translated to rotational population diagrams. The green curve shows the best isothermal non-LTE model. The red and blue dashed curves show predictions of PDR models with G0 = 104.9 and ζCR = 10−15 s−1 (line intensities multiplied by 0.04).

Open with DEXTER
In the text
thumbnail Fig. A.1.

12CO rotational temperatures (, computed by fitting a straight line to synthetic line intensities in the Jup = 5−10 range) obtained from isothermal non-LTE models. The red curve shows the parameter space that reproduces the observed Trot of the HPVW component. The blue-shaded area shows the estimated gas density of the high-velocity cloudlets detected by ALMA (Goicoechea et al. 2018). The magenta, red, and blue curves show rmsmin/rms contours of 1, 0.95, and 0.9 respectively, where rms refers to the rms value of log10 (IHIFI/Imod).

Open with DEXTER
In the text
thumbnail Fig. B.1.

HIFI detection of HCN and HCO+ lines toward Sgr A* showing emission only at velocities typical of the CND (υLSR < 150 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.