EDP Sciences
Free Access
Issue
A&A
Volume 606, October 2017
Article Number A52
Number of page(s) 13
Section Interstellar and circumstellar matter
DOI https://doi.org/10.1051/0004-6361/201731056
Published online 09 October 2017

© ESO, 2017

1. Introduction

Massive stars, despite their rarity, are important constituents of galaxies and produce most of their luminosity. The complex and still not clearly defined evolutionary sequence ranges from massive prestellar cores, high-mass protostellar objects (HMPOs), hot molecular cores, and finally to the more evolved ultra compact HII (UCHII) region stage, where the central object begins to ionize the surrounding gas (e.g., Beuther et al. 2007; König et al. 2017). The classification adopted above is of course not unique. The molecular material in star-forming regions has a wide range of temperatures and number densities (10−2000 K and 104−109 cm-3), with different chemical characteristics. Water is a key molecule for determining the physical and chemical structure of star-forming regions because of the large abundance variations between warm and cold regions. In cool molecular clouds, water is mostly found as ice on dust grains, but at temperatures T > 100 K, the ice evaporates, increasing the gas-phase water abundance by several orders of magnitude (Fraser et al. 2001; Aikawa et al. 2008). At temperatures above ~250 K, all gas-phase free oxygen may be driven into H2O, increasing its abundance to ~3 × 10-4 (van Dishoeck et al. 2011).

Major progress in our understanding of interstellar water and star formation has been made with the Herschel Space Observatory. The guaranteed-time key program WISH (Water In Star-forming regions with Herschel, van Dishoeck et al. 2011) probed massive star-forming regions using water observations. To collapse, the gas must be able to release enough thermal energy, in addition to several other condistions; one main WISH goal was to determine how much of the cooling of the warm regions (T > 100 K) is due to H2O. In particular, the dynamics of the central regions has been characterized using the water lines, and the amount of cooling provided has been measured. Herschel observations of many water lines made with the high-spectral resolution Heterodyne Instrument for the Far Infrared (HIFI, de Graauw et al. 2010) allowed observers to perform some type of tomography of the whole protostellar environment, and hence to probe the physical conditions and estimate the water abundance from a few 100 AU to a few 10 000 AU from the star (see Herpin et al. 2016). Nevertheless, observational evidence for the water abundance jump close to the forming stars that heats up their environs is still scarce (e.g., van der Tak et al. 2006; Chavarría et al. 2010; Herpin et al. 2012, 2016). While the outer abundance for the HMPOs studied in the WISH program is well constrained and estimated to be a few 10-8 in all sources, the inner abundance varies from source to source between 0.2 and 14 × 10-5 (Herpin et al. 2016). Observing water lines involving high enough energy levels will allow us to probe the physical conditions and water abundance in the inner layers of the protostellar environment and thus to address this problem. One main difficulty is that up to now it has been impossible to (i) spatially resolve the region where the water jump occurs and (ii) to probe this region with the help of high-excitation optically thin lines.

Since the end of the Herschel mission, only the Stratospheric Observatory for Infrared Astronomy (SOFIA, Young et al. 2012) allows us to observe water in the THz frequency range. Interestingly, some of the THz water lines could even be masing. The H2O excitation model of Neufeld & Melnick (1991) as well as the works of Yates et al. (1997) and more recently, of Daniel & Cernicharo (2013) and Gray et al. (2016), which incorporated the most recent collisional excitation rates of water with ortho- and para-H2, all make important predictions of masing transitions in the supra-THz region. Several transitions could be masing that have frequencies within the L1 and L2 channels of the German Receiver for Astronomy at Terahertz Frequencies (GREAT, Heyminck et al. 2012) on board SOFIA (Heyminck et al. 2012). In particular, three predicted o-H2O masing transitions and one p-H2O masing transition fall in the 1.25–1.50 THz range. Only one o-H2O transition, 82,7−73,4 at 1296.41 GHz, could be observed at the flying altitude of the SOFIA observatory (transmission 62%), the others are absorbed by the remaining atmosphere. Emission from this transition involves an upper energy level at 1274.2 K and can probe the gas deep enough into the inner layers of the hot core. We note that Neufeld et al. (2017) have just reported the first detection of the 82,7−73,4 masing line toward an evolved star.

We have selected NGC 7538-IRS1, the best-studied massive object from our WISH program (Herpin et al. 2016; van der Tak et al. 2013), which has previously been observed in the 22 GHz maser line and exhibits strong thermal water lines. NGC 7538-IRS1 is a relatively nearby (2.65 kpc, Moscadelli et al. 2009) UCHII object in the complex massive star-forming region NGC 7538 surrounded by a molecular hot core. The 22 GHz H2O masers associated with it exhibit a complex spatial distribution, and the strongest and main concentration of the maser features (see Kameya et al. 1990; Surcis et al. 2011) is found in the direction of IRS1. Until recently (see Beuther et al. 2013), the central source was thought to be an O6 star (30 M , L ≃ 8 × 104L ) forming one single high-mass young stellar object (YSO). New interferometric observations of the methanol masers (Moscadelli & Goddi 2014) have demonstrated that NGC 7538-IRS1 consists of three individual high-mass YSOs named IRS1a, IRS1b, and IRS1c within 1600 AU. We note that IRS1a, b, and c are associated with the methanol maser clusters labeled by Minier et al. (2000) B+C, A, and E, respectively. The most massive YSO is IRS1a, with ~25 M and a quasi-Keplerian disk of ~1 M that dominates the bolometric luminosity of the region. Another massive (16 M ) and thick disk orbits the less massive (a few M ) IRS1b object. The third source, IRS1c, is likely to be a massive YSO as well.

Several bipolar outflows or jets emanating from the IRS1 region in multiple directions have been identified and characterized. A north-south free-free ionized jet with an opening angle 30° has been observed by Sandell et al. (2009), as well as a strong accretion flow toward IRS1 (~2 × 10-3M/yr). Several studies have characterized the NW-SE (PA = −50°, Qiu et al. 2011) CO outflow now thought to originate from IRS1a (Moscadelli & Goddi 2014). In addition, a NE-SW outflow with PA = 40° has been observed by Beuther et al. (2013). An outflow driven by IRS1b that is collimated by its rotating disk has also been observed (Moscadelli & Goddi 2014). The presence of jets or outflows and strong accretion flows makes this shocked region an ideal place for exciting masers.

In this paper we report an observational study of the NGC 7538-IRS1 region. The o-H2O 82,7−73,4 emission was observed with the GREAT instrument on board SOFIA. In addition, the 22 GHz water maser emission was observed using the Effelsberg telescope and the e-MERLIN interferometer. In Sect. 2 we describe the observations, and we present the observational data in Sect. 3. In Sect. 4 we discuss the nature of the detected THz water emission. In Sect. 5 we discuss the region kinematics and physical conditions, and in Sect. 6 we summarize the results of our study.

2. Observations and data reduction

2.1. SOFIA observations

Observations in single-point chopping mode of the 82,7−73,4 line of ortho-HO were carried out toward NGC 7538-IRS1 (αJ2000 = 23h13m45.3s,δJ2000 = + 61°2810.0′′) as part of the SOFIA Cycle 3 project on 2015 December 9, using GREAT. At the systemic velocity of NGC 7538-IRS1, the system temperature was typically 2000 K (SSB) and signal-band zenith opacity 0.08. The on-source integration time was 20 min. The chop throw was 100′′ to either side of the on-source position (chop-nod method). One channel (L1) of GREAT was tuned to the 1296.41106 GHz water line frequency (lower sideband LSB), the other channel, the Low Frequency Array (LFA) seven-pixel array, was tuned to the [CII] 158 μm line. The [CII] line was detected but a discussion of these results will be presented elsewhere. We employed the digital 4GFFT spectrometer (Klein et al. 2012) that analyzed a bandwidth of 1.5 GHz with a spectral resolution of 0.283 MHz (0.056 kms-1). Data have later been smoothed to 1.129 kms-1. The rms of the spectra is 90 mK at this latter spectral resolution.

The data were processed with the latest version of the GREAT calibrator and converted from the scale (ηf = 0.97) into main-beam temperature units, TMB, by applying the main-beam coupling efficiency ηmb = 0.69 for the L1 channel. The conversion factor Jy/K we applied on the data is 971 Jy/K. The half-power beam width is 20.6′′ at this frequency. All spectra were calibrated for the transmission in the signal band, and the continuum-level correction for double-sideband reception was applied. Further analysis was made within the CLASS1 package. The continuum level (SSB) is TMB = 2.0 K (at 1296.4 GHz).

2.2. Effelsberg observations

In order to constrain the maser models and because of the variability of the maser emission, nearly contemporaneous observations of the 616−523 transition of ortho-H2O (rest frequency 22.23508 GHz) were carried out on 2015 December 11 (integration time on-source of 15 min) with the MPIfR 100 m telescope at Effelsberg, Germany. The half-power beam width was 39′′ and the pointing accuracy was, on average, 4′′. We used the 1.3 cm double-beam secondary focus receiver (K/Jy conversion factor of 1.6) with the XFFT backend in high-resolution mode (32768 channels with a spectral resolution of ~0.04 kms-1). Data were smoothed to 0.1 kms-1 and the rms noise at this resolution is 37 mJy. The system temperature was 190 K. All data were calibrated in Jy; the calibration parameters were derived by continuum observations of suitable flux density calibrators. The main-beam efficiency is 0.64. The calibration uncertainty is about 10–15%.

2.3. e-MERLIN observations

Interferometric observations of the same source of the ortho-H2O 616−523 line emission were performed with e-MERLIN (commissioning observations) on 2016 April 20. These were the very first 22 GHz images made after the e-MERLIN upgrade. Four telescopes were used, the longest baseline being 217 km, with the 25 m Mark 2, Darnhall, and Pickmere telescopes and the 32 m Cambridge telescope. At the assumed distance of 2.7 kpc, the synthesized beam of 20 mas corresponds to a spatial resolution element of approximately 54 AU, but the positions of individual bright maser components can be fit with a relative accuracy ~1 AU.

The observations were made in full Stokes parameters (although only the parallel bands were processed, giving total intensity images). One spectral window (spw) of 4 MHz was centered on the 22.23508 GHz maser line (adjusted for the source velocity immediately before observations), divided into 512 channels, giving a spectral resolution of 0.105 km s-1 and a total useful span ~50 km s-1. Two similar spw were placed either side (but they did not contain any emission). A single 125 MHz spectral window (also with 512 channels) was placed to overlap them for calibration. The pointing position of αJ2000 = 23h14m01.749s,δJ2000 = + 61°2719.80′′ was used for NGC 7538, which was approximately observed 10 h on target, in 6-min scans interleaved with the phase reference source J2302+6405. The quasar 3C 84 was used to provide the bandpass correction and flux scale, with a flux density in 2016 April of 37 Jy (measurements kindly provided by Anne Lähteenmäki, Metsahovi, priv. comm.). The local standard of rest (LSR) correction was applied to the line data before self-calibration and imaging.

The calibration was performed in CASA, and final imaging and component fitting in AIPS. The position of the brightest maser was established as a reference position before self-calibration. Test image cubes were made at coarse resolution to look for masers within ~1.5 arcmin radius of the pointing center (within the half-power point of the primary beam). Four fields were imaged at full resolution, covering all the emission we identified, see Table 1.

Table 1

e-MERLIN 22 GHz imaging parameters.

The e-MERLIN beam at 22 GHz is not yet fully characterized. We made images without primary beam correction, and then fit 2D Gaussian components to each patch of emission above 4σrms (rejecting components that were isolated in fewer than three channels or any obvious sidelobes of brighter components). We estimated the primary beam corrections using a scaled VLA beam, and divided the measured fluxes by the factors given in Table 1, with uncertainties increasing with distance from the pointing center. The most distant fields were also slightly affected (<10%) by time smearing.

The noise σrms is ~20 mJy in quiet channels for fields close to the pointing center (IRS 1–3), rising to twice that in the most distant fields. The sparse visibility plane coverage means that the brighter channels are severely dynamic-range limited, with σrms typically 10% of peaks, or even more in more remote fields. The astrometric position accuracy is a few tens of milliarcseconds (mas); the relative accuracy depends on the signal to noise ratio and is <1 mas for all but the faintest or most remote components.

2.4. HIFI observations

The source NGC 7538-IRS1 was part of the WISH GT-KP sample. Fifteen water lines (see Table A.1) were observed with the HIFI spectrometer on board the Herschel Space Observatory in the pointed (or mapping) mode at frequencies of between 547 and 1670 GHz in 2010 and 2011 (the observation identification numbers, obsids, are listed in Table A.1) toward NGC 7538-IRS1 (same coordinates as for SOFIA observations). For the pointed observations, the double-beam switch observing mode with a throw of 3′ was used. The off positions were inspected and do not show any emission. The frequencies, energy of the upper levels, system temperatures, integration times, and rms noise levels at a given spectral resolution for each of the lines are provided in Table A.1.

Data were taken simultaneously in H and V polarizations using both the acousto-optical Wide-Band Spectrometer (WBS) with 1.1 MHz resolution and the digital autocorrelator or High-Resolution Spectrometer (HRS). Calibration of the raw data into the TA scale was performed by the in-orbit system (Roelfsema et al. 2012). Conversion into Tmb was made using the beam efficiency2 given in Table A.1 and a forward efficiency of 0.96. The flux scale accuracy is estimated to be between 10% for bands 1 and 2, 15% for bands 3 and 4, and 20 % in bands 6 and 71. Data calibration was performed in the Herschel Interactive Processing Environment (HIPE, Ott 2010) version 14. Further analysis was made within CLASS. These lines are not expected to be polarized. Thus, after inspection, data from the two polarizations were averaged together. Because HIFI is operating in double-sideband mode, the measured continuum level was divided by a factor of 2 in the figures and tables (this is justified because the sideband gain ratio is close to 1). We note that the p-H2O 524–431 line is affected by strong baseline and ripple problems.

3. Analysis

3.1. THz water emission

The ortho-HO 82,7−73,4 line emission is detected toward NGC 7538-IRS1 with SOFIA. As shown in Fig. 1, the line profile exhibits two features at − 57.7 ± 0.6 and − 48.4 ± 0.5 kms-1 where 22 GHz emission is also observed (see Fig. 2). From Gaussian fitting, the S/N ratio of the integrated lines at –57.7 and –48.4 kms-1 is 5.5 and 3.2, respectively. Hereafter the –57.7 and –48.4 components are called features (1) and (2), respectively. At the frequency of the observed emissions, there is no contamination by other molecular species according to Splatalogue catalog3. The peak intensities and line widths are 133.1 Jy and 4.7 (±1.3) kms-1 for feature (1), and 139.4 Jy and 2.6 (±1.3) kms-1 for feature (2).

The velocity and line width of feature (1) are comparable to what has been observed from the CS thermal lines (vLSR = −57.4 kms-1, δv = 5.5 ± 1 kms-1, van der Tak et al. 2000) or OH with SOFIA (Csengeri et al. 2012). Conversely, feature (2) is centered at a velocity different from the generally adopted source velocity. In addition, the line is narrower by a factor of 2. The nature of these water components is discussed in Sect. 4.

thumbnail Fig. 1

Continuum-subtracted SOFIA spectra of the o-H2O 82,7−73,4 line at 1296.41106 GHz toward NGC 7538-IRS1. The vertical dotted line indicates VLSR at –57.4 kms-1. The spectral resolution is 1.129 kms-1. The Gaussian fit is shown in red. The small insert shows the same spectra on a zoomed velocity scale.

Open with DEXTER

3.2. 22 GHz single-dish observation

Several velocity components of the 22 GHz water maser line are seen in the Effelsberg spectra (see Fig. 2). All lines are narrow with a width on the order of 1 kms-1 or less. Two main groups of lines can be distinguished: one made of three strong lines close to the source velocity (~− (56−60) kms-1)), and another group with two lines centered at –45.1 and –48 kms-1. These lines are highly variable with time (on a timescale of several months), as shown by Felli et al. (2007) from their long-term monitoring observations. Nevertheless, most of the time, the ~− 60 kms-1 feature is the strongest maser line.

According to the existing 22 GHz maser maps and our e-MERLIN map (see next section), all of these maser components are associated with IRS1-3. The non-detections of more distant masers are explained by the Effelsberg telescope’s primary beam, which is smaller by one-third than that that of e-MERLIN.

3.3. Map of the 22 GHz maser emission

The entire e-MERLIN map, shown in Fig. 3, encompasses the highly luminous infrared sources IRS1-3, IRS9, and IRS11 which were first identified by Werner et al. (1979). We have identified 286 individual features that can be gathered into 109 groups based on their peak velocity and coordinates. All these groups are reported in Table B.1, where their position, peak flux density Sν, peak velocity, and velocity range are listed. We have compared our maser positions with the work of Galván-Madrid et al. (2010), taking into account the proper motions from Moscadelli & Goddi (2014), μRA = −2.45 mas/yr and μDec = −2.45 mas/yr. Both maps are consistent. Our Figs. 3 and 4 around IRS1-3 and IRS11 show maser positions very close to those labeled M and S in Table 2 of Galván-Madrid et al. (2010) and are shown here as green crosses. Moreover, the total size of the image around IRS1-3 (12′′, a size comparable to the SOFIA beam at 1.3 THz) was used to synthesize the e-MERLIN 22 GHz spectrum, which we compare in Fig. 2 with the line profile obtained at Effelsberg in order to verify the flux density scale (e-MERLIN data are commissioning data), and to check that the interferometer is detecting the entire flux. Both line profiles are very similar, except for the apparently negative feature in the –63.5 to –64 kms-1 velocity range, which is due to residual sidelobes arising from strong IRS11 emission in this velocity range. The maser variability over this short time-period (four months) is thus not significant and the entire flux is detected.

thumbnail Fig. 2

Effelsberg (in black, spectral resolution δv of 0.1 kms-1) and e-Merlin (in blue, δv = 0.1 kms-1, the apparent absorption near –63.5 kms-1 is attributed to e-MERLIN residual sidelobes, see Sect. 3.3) spectra of the 616−523 transition of ortho-H2O at 22 GHz toward NGC 7538-IRS1 plotted over the SOFIA (δv = 1.1 kms-1) spectrum of the o-H2O 82,7−73,4 line at 1296.41106 GHz (in red).

Open with DEXTER

thumbnail Fig. 3

Entire e-MERLIN map of 22 GHz maser emission from NGC 7538, including IRS1-3, IRS9, and IRS11, of the 22 GHz water maser line. Colored circles show the relative position of individual maser features, with color denoting the maser VLSR, according to the color-velocity conversion code reported on the right side of the panel. The water maser positions reported by Galván-Madrid et al. (2010), corrected for proper motion (Moscadelli & Goddi 2014), are shown as crosses (same color-velocity conversion code) for comparison. The beam of the SOFIA observation is overlaid in black for comparison.

Open with DEXTER

thumbnail Fig. 4

As in Fig. 3, but zoomed on IRS1-3.

Open with DEXTER

thumbnail Fig. 5

Zoom toward IRS1 north of the e-MERLIN map shown in Fig. 3. Colored empty and filled circles show the relative position of individual 22 GHz water masers (this paper) and 6.7 GHz methanol maser features (Moscadelli & Goddi 2014), respectively. VLSR is according to the color-velocity conversion code shown on the right side of the panel.

Open with DEXTER

We detect 68 maser groups for IRS11, some of them very bright with Sν well above 100 Jy, spanning a velocity range from –44.75 to –65.08 kms-1. The maximum flux is observed at –60 kms-1 (964 Jy) in IRS1. Only three main groups of masers are observed toward IRS9, and they are blueshifted (~− (69−79) kms-1) compared to IRS1 source velocity. The strongest maser group peaks at 39 Jy in IRS9.

Approximately 25 arcsec southwest from IRS1 (offsets ~–(12–14) and –(21–24)), a new source has been found that exhibits powerful maser emission (up to ~200 Jy for group 5) and velocities from –45 to –63 kms-1 (see Fig. 3 and Table B.1). The feature “E” observed by Kameya et al. (1990) with VLSR = −63.3 kms-1 at αJ2000 = 23h13m42.51s,δJ2000 = + 61°2745.1′′, that is at offset ~− 20″ and − 25″, is much farther away from this new spot.

A first zoom of the IRS1-3 region is shown in Fig. 4. The IRS3 source lies to the NW with offsets from IRS1 of –5.5 and 2.5 in RA–Dec, respectively. Two main groups are detected for IRS3, they are moderately bright (14–88 Jy) and span a small velocity range (from ~–56 to –58 kms-1). At this scale, IRS1 is made of what Surcis et al. (2011) called the north source (the main source) and the nouth source, 1.2″ south. In addition to the maser group at ~− 70 kms-1 for IRS1, which has also been observed by these authors, we detect maser spots at redder velocities of ~− 46 kms-1.

The close-up view of HO maser features around NGC 7538-IRS1 North (see Fig. 5) reveals 26 maser groups that can be split into two distinct velocity ranges: a first large group of masers within ±3 kms-1 from the main source LSR velocity, represented in blue in Fig. 5, and a second group of redshifted masers whose LSR velocity is between –45 and –48 kms-1. The “blue” group is made of numbers of features that are concentrated very close to IRS1a (as defined by Moscadelli & Goddi 2014) plus several spots spread in the northwest. The “red” group masers lying north of IRS1a are less well aligned along a NE–SW direction. The maser spatial distribution is discussed in more detail in Sect. 4.1.

4. Nature of the THz water emission

4.1. Spatial origin

Considering the SOFIA beam size (21′′) and the approximate 12′′ square size of the image made by e-MERLIN (primary beam, see Table 1) toward IRS1-3 (see Figs. 3 and 4), the detected THz water emission can only originate from IRS1-3. Then, comparing the velocity range of features (1) and (2) with the velocity of maser spots detected by e-MERLIN (Figs. 2 and 4 and Table B.1), we can infer the following:

  • water in IRS1 North or IRS3 can give rise to feature (1),

  • only IRS1 North exhibits 22 GHz maser emission at velocities (~− 48.4 kms-1) similar to feature (2) THz emission.

At the scale of Fig. 5, we can see that the sizes of the respective emitting regions for maser features around –48 and –58 kms-1 are comparable, but not spatially coincident. Maser spots at velocities similar to THz feature (2) emission are located 0.2′′ northeast of the high-mass YSO IRS1a and even more than 0.1′′ east of the massive protostar IRS1b. We also conclude that a similar beam dilution of the two maser features is obtained with the Effelsberg 22 GHz beam (39′′).

Table 2

Observed line emission parameters for the detected lines with HIFI toward NGC 7538-IRS1.

4.2. Thermal or maser?

4.2.1. Thermal modeling

All water lines observed toward NGC 7538-IRS1 with HIFI are shown in Fig. 6. A detailed analysis of three low-J HO line profiles has been made by van der Tak et al. (2013) in several HMPOs, including NGC 7538-IRS1. Several lines from the rare isotopologue HO and the para-ground-state line of HO were detected in addition to the HO lines toward NGC 7538-IRS1 (see Fig. 6, left).

thumbnail Fig. 6

HIFI spectra of HO/HO (left) and HO (right) lines (in black), with the continuum for NGC 7538-IRS1 pointed position. The best-fit model is shown as a red line over the spectra (χin(H2O) = 8 × 10-6 and χout(H2O) = 4 × 10-8). The model adopting the SOFIA water inner abundance (5.2 × 10-5) is shown as a blue line above the spectra. Vertical dotted lines indicate the VLSR (–57.4 kms-1 from the line modeling). The spectra have been smoothed to 0.2 kms-1, and the continuum is divided by a factor of two.

Open with DEXTER

A Gaussian fit to these line profiles (see Table 2) indicates, depending on the line, three velocity components: a narrow (3.2–4.6 kms-1), a medium (5.1–9.3 kms-1), and a broad (>12.2 kms-1) velocity component. The physical origin of these components (attributed to a dense core, an envelope, and an outflow) has been discussed in several papers (e.g., Herpin et al. 2016). All of these components in NGC 7538-IRS1 are globally centered at a velocity similar to that of the THz feature (1), that is, ~− 57 kms-1. No emission is detected at –48 kms-1, which is thus a first indication of the nonthermal origin of this emission.

Following the method of Herpin et al. (2012, 2016), we modeled all water line profiles in a single spherically symmetrical model using the 1D radiative transfer code RATRAN (Hogerheijde & van der Tak 2000). The envelope temperature (20–1130 K) and number density (1.1 × 105−3.8 × 108 cm-3) structure for the hot core is taken from van der Tak et al. (2013). This analysis assumes a single source within the HIFI beam, hence encompassing the IRS1-3 substructure. The outflow parameters, intensity and width, come from our Gaussian fit (see Table 2) for the broad component. The envelope contribution is parametrized with the water abundance (outer χout for T < 100 K, inner χin for T > 100 K, assuming a jump in the abundance in the inner envelope at 100 K due to the evaporation of ice mantles), the turbulent velocity (vturb), and the infall velocity (vinf). We adopt the following standard abundance ratios for all the lines: 4.5 for HO/HO (Thomas & Fuller 2008), 614 for HO/HO (based on Wilson & Rood 1994), and 3 for ortho/para H2O.

Considering that the width of the velocity components is not the same for all lines (see Table 2), a model with an equally turbulent velocity for all lines does not fit the data well. The best result (see Fig. 6) is obtained with a turbulent velocity of 1.5–2.5 kms-1 (5.5 kms-1 is even needed for the doubtful p-H2O 524–431 line) for HO, depending on the modeled line, and 2.5 kms-1 for the rare isotopologues lines. No infall is needed at the scale probed by the HIFI lines. All modeled lines are centered at roughly − 57.4 ± 0.5 kms-1.

Modeling of the entire set of observed lines constrains the water abundance, but only a few lines are optically thin enough to probe the inner part of the envelope. We hence derive HO abundances relative to H2 of 8 × 10-6 in the inner part and of 2 × 10-8 in the outer part.

We then applied the results of the above thermal model to feature (1) of the THz o-H2O 82,7−73,4 line (we do not see any thermal line corresponding to feature 2), assuming the same abundance and vturb = 2.5 kms-1. Figure 7 shows that we do not reproduce the observed line, the intensity being only half what is observed. We can perfectly reproduce feature (1) with an increased water inner abundance of 5.2 × 10-5 (see blue line in Fig. 7). We also successfully applied this new model to the HO 313−220 line (Eup ≃ 200 K), reproducing the line intensity and width observed by van der Tak et al. (2006).

Adopting now the water inner abundance deduced from our modeling of the o-H2O 82,7−73,4 line, we again modeled the HIFI water thermal lines (in blue in Fig. 6). The result is less satisfactory, especially for the p-H2O 524–431 (but this observation is affected by line ripples, and the baseline determination is also affected by methanol line blending) and rare isotopologue lines. This might reflect the limitations of our symmetrical 1D model. It is also possible that the 82,7−73,4 line detected by SOFIA emanates from the inner part of the hot core, while the water lines observed with HIFI are from somewhat cooler regions farther out. Thus, a temperature gradient could explain our results without requiring a higher water abundance throughout the region. On the other hand, feature (1) of the 1296 GHz water line and the HO 313−220 line can be well matched by a thermal emission profile if the water abundance is increased. Nevertheless, nonthermal effects cannot be excluded: our RATRAN model could miss a nonthermal contribution at 1296 GHz, which, in that case, would be on the order of 50%.

4.2.2. Maser modeling

We have made quasi-contemporaneous observations of the potentially masing o-H2O 82,7−73,4 transition at 1296 GHz and of the 22 GHz maser line to provide line intensity ratios enabling us to estimate the physical conditions leading to these maser emissions. Line intensity ratios are less dependent on the cloud geometry, and in the saturated regime, such ratios tend to be independent of the exact ratio of the beaming angles, which should then be close to one.

We estimate the 1296/22 opacity ratio from the recently published models of Gray et al. (2016). These models incorporate 411 levels and 7597 radiative transitions, the most recent collision rates, and line overlap effects. Even if these models have been developed primarily for evolved stars, the broad physical parameter space (including Tdust rising from 50 K to 350 K) that has been explored can be used for the conditions considered here. There is a significant overlap between conditions supporting the 22 GHz and 1296 GHz masers, for instance, both could come from gas above 1000 K with o-H2O number densities between 104 and 2 × 105 cm-3 (i.e., nH2 ~ 3 × 109−6 × 1010 cm-3, taking 3 × 10-5 as standard conversion factor from n(H2O) to n(H2), Gray et al. 2016). However, strong 1296 GHz maser emission is obtained for a relatively narrow range of physical conditions, typically TK around 1000 to 3000 K and nH2O around 104−105 cm-3, while 22 GHz masers are excited within 500 to 3000 K and for a broader range of H2O densities up to several 105 cm-3. Compared to the 22 GHz inversion, the 1296 GHz inversion is biased toward higher values of TK and it is lost with increasing dust temperatures, namely a low value of Tdust, around 50 K, is preferred. The dust temperature is unlikely to be much higher than 50 K in shocked gas. Assuming Tdust = 50 K, Gray et al. (2016) predicted τ(1296) /τ(22) ~ 1.9. This ratio does not result from specific conditions representing the observed source. It is, instead, the ratio of the maximum 1296 GHz depth to the maximum found at 22 GHz from models covering a large parameter space. Saturation is also not accounted for in this ratio. Nevertheless, the 1.9 ratio suggests that the flux density and brightness temperature at 1296 GHz can be several times that at 22 GHz. To compare features (1) and (2) to our model predictions, we derive the line peak ratios from Fig. 8 where all spectra have been smoothed to the same spectral resolution. We derive S(1296)/S(22) = 1.2 and 7.4 for features (1) and (2), respectively, but cannot directly estimate the brightness temperature of the 1296 GHz emission since its spatial extent is unknown. At –48 kms-1, feature (2) is definitely much brighter in the 1296 GHz line than at 22 GHz. This is in agreement with maser conditions suggested from the 1.9 opacity ratio above. We further note that feature (2) is nearly 2.5 times narrower than feature (1) suggesting again that the former is not thermal. Assuming that the 1296 and 22 GHz emissions have a similar spatial extent, we derive from the observed S(1296)/S(22) = 7.4 at –48 km s-1, Tb(1296) = Tb(22) × 2.2 × 10-3 and, because the condition Tb(22) on the order of or greater than 106 K is easily met at 22 GHz (e.g., Elitzur et al. 1992), we get Tb(1296) greater than 2200 K, that is, suprathermal emission. We stress that deriving an exact value for Tb(1296) is impossible since we have no spatial information. When we use the e-MERLIN resolution of 20 mas as an upper limit to the beamed size of maser spots, however, we obtain Tb in the range >106 to >109 K, with a mean value of >3 × 107 K. The beamed area of maser spots is likely to be at least two orders of magnitude smaller, giving an average Tb at least 109 K. This lower limit to Tb strongly suggests, but does not prove, the maser nature of this emission.

Our conclusion is thus that the feature (2) water emission is likely a maser. The absence of the typical narrow substructure in the line profile could just be due to the low S/N (and spectral resolution) that hides it. Observations with some instrument that has much higher spatial resolution at 1296 GHz would be the only way to unambiguously prove it is a maser.

thumbnail Fig. 7

SOFIA spectra of the o-H2O 82,7−73,4 line showing for feature (1) the best-fit thermal model in red from our HIFI data (see Fig. 6) and a model adopting a water inner abundance of 5.2 × 10-5 (blue profile).

Open with DEXTER

thumbnail Fig. 8

As in Fig. 2, with all spectra smoothed to a spectral resolution of 1.1 kms-1. The SOFIA fluxes have been divided by a factor 7.4.

Open with DEXTER

5. Discussion

5.1. Water content

Depending on the excitation of the o-H2O 82,7−73,4 line, purely thermal or not, we have shown that the water inner abundance might differ by more than one order of magnitude in the hot core of NGC 7538-IRS1. From the integrated fluxes measured for the lines (considered in this study) that are at least partly in emission, we derived the water luminosities and then estimated, assuming isotropic radiation, that the minimum total HIFI water luminosity is 0.6 L (equal to the sum of all individual observed luminosities). Even though the true water emission from the inner parts might be much greater, the cool envelope absorbs much of the emission. This confirms the low contribution of water cooling to the total far-IR gas cooling compared to the cooling from other species (Karska et al. 2014; Herpin et al. 2016). Moreover, from the modeling, and assuming that the feature (1) water emission is purely thermal, we estimate the total water mass in the envelope to be 10-3M  and that 93% of this mass resides in the inner parts, to be compared with 2 × 10-4M  and 69% in the case of a nonthermal contribution.

5.2. Kinematics and geometry of the region

The previous section suggests that the THz water feature (1) is consistent with thermal excitation (even if a nonthermal contribution is possible), while feature (2) could be masing. Different physical conditions, that is, a different spatial origin, could explain these different behaviors. According to Gray et al. (2016), this could for instance be explained by a higher dust temperature in the region where the –57 kms-1 water is excited compared to the –48 kms-1 region (the o-H2O 82,7−73,4 inversion is lost with increasing dust temperature).

When we plot the 6.7 GHz methanol maser spots and the location of high-mass YSO IRS1a and b from Moscadelli & Goddi (2014) over our 22 GHz e-MERLIN map, the –48 kms-1 22 GHz maser spots appear located close to the IRS1b source, while the –57 kms-1 spots are associated with the IRS1a source. Interestingly, Moscadelli & Goddi (2014), based on NH3 maps, explain that the gas surrounding IRS1b (a less evolved source) has a lower temperature than the gas observed toward IRS1a, which is a more massive and evolved YSO, with T lower than 250 K. Hence, the higher temperature in IRS1a might be less suitable for water maser emission at 1296 GHz. It is more likely, however, that IRS1a could collisionally quench the maser if the number density is too high. We estimate the critical density (at T = 50 K) corresponding to the 1296 GHz transition to be 5 × 107 cm-3. The gas in the core of IRS1a is thus so dense (number densities of up to a few 108 cm-3 can be reached in the inner part, see Sect. 4.2.1) that it begins to quench the maser action.

The maser spot distribution as derived from our e-MERLIN map provides new information when compared to previous published works. Figure 5 shows the spatial location of the 22 GHz maser features and the 6.7 GHz methanol masers of Moscadelli & Goddi (2014). According to these authors, the two individual high-mass YSOs, IRS1a and b, lie at the center of a line of methanol masers tracing disks with position angles of PA = + 71° and − 73°, respectively. The strongest “blue” (i.e., v ≤ −56 kms-1) maser spots we detect can be associated with IRS1a, others are distributed roughly along a line with PA = −25° and another with PA = −50°. Some similar linear distribution with PA = −52° was seen by Surcis et al. (2011), who suggested that these water masers were almost aligned with the CO NW-SE outflow, elongated 0.3 pc from IRS1a with PA = −40° (Kameya et al. 1990; Gaume et al. 1995; Qiu et al. 2011). They proposed that the water masers are pumped by a shock caused by the interaction of the outflow with the infalling gas (observed at scales 1000 AU, Beuther et al. 2013). The e-MERLIN map shows water masers following this NW-SE axis of the CO outflow, but at slower velocities compared to what Qiu et al. (2011) observed in CO (–78; –64 kms-1). These H2O maser spots might either trace the cavity of the outflow, that is, a cone with an opening angle of ~25° and PA ≃ −40°, or trace two different outflows originating from IRS1a, the outflow at PA = −25° being almost perpendicular to the disk.

Maser spots whose velocity is close to the THz feature (2) (i.e., –44; –50 kms-1) are located NW from IRS1a and W from IRS1b. They are also distributed along a line with PA ≃ + 28°, that is, NE-SW, which can be associated with the outflow observed by Beuther et al. (2013) in HCO+(4-3) with PA ≃ 40°. Moreover, the NH3 (Moscadelli & Goddi 2014), OCS, CH3CN, and 13CO (Zhu et al. 2013) observations show a velocity gradient in the same direction (PA ~ 30−40°) with line emission at velocities similar to our feature (2) emission in this NE region. Moscadelli & Goddi (2014) explain these “redshifted” NH3 features toward the NE by an outflow driven by IRS1b that is collimated by its rotating disk. Hence, as proposed above on the basis of a too high temperature toward IRS1a, we propose that the THz feature (2) is a maser, not associated with IRS1a, and is pumped by shocks that are driven by the IRS1b outflow.

6. Conclusions

SOFIA observations toward NGC 7538-IRS1 of the o-H2O 82,7−73,4 line emission were presented. Two separate velocity features were detected: one associated with the source velocity (–57.7 kms-1), and another one lies at –48.4 kms-1. By combining these observations with near-simultaneous observations of the 61,6−52,3 masing transition of ortho-H2O at 22 GHz with the Effelsberg telescope and with the e-MERLIN interferometer, we discussed the nature of these THz emission features.

A thermal water model based on HIFI observations can reproduce the 82,7−73,4 line component at the source velocity if the water inner abundance is increased by more than an order of magnitude to 5.2 × 10-5 compared to what is estimated by the “HIFI alone” model. In addition, the observed brightness ratio (1296/22) for both features was compared to the maser predictions and led us to conclude that while the THz emission feature at the systemic velocity is mostly thermal, the –48.4 kms-1 feature is likely masing.

We argue that the two line components do not arise from the same location, meaning that different physical conditions could explain these different natures. The thermal emission is excited in the innermost part of the IRS1a protostellar massive object and is an excellent probe of the water reservoir in the inner part of the hot core. We suggest that the maser emission is associated with shocks driven by the IRS1b outflow.


Acknowledgments

We would like to thank Anne Lähteenmäki (Metsähovi Radio Observatory, Aalto University, Finland) for the flux measurements of 3C 84 and Alex Kraus (MPIfR-Bonn, Germany) for having performed and reduced the Effelsberg observations. The Effelsberg 100-m telescope is a facility of the MPIfR (Max-Planck-Institut für Radioastronomie) in Bonn. e-MERLIN is a national facility operated by The University of Manchester on behalf of the Science and Technology Facilities Council (STFC). We thank the SOFIA operations and the GREAT instrument teams, whose support has been essential for the GREAT accomplishments, and the DSI telescope engineering team. Based [in part] on observations made with the NASA/DLR Stratospheric Observatory for Infrared Astronomy. Sofia Science Mission operations are conducted jointly by the Universitis Space Research Association, Inc., under NASA contract NAS297001, and the Deutsches SOFIA Institut, under DLR contract 50 OK 0901. GREAT is a development by the MPI für Radioastronomie and the KOSMA/Universität zu Köln, in cooperation with the MPI für Sonnensystemforschung and the DLR Institut für Planetenforschung. Herschel is an ESA space observatory with science instruments provided by European-led Principal Investigator consortia and with important participation from NASA. HIFI has been designed and built by a consortium of institutes and university departments from across Europe, Canada and the United States under the leadership of SRON Netherlands Institute for Space Research, Groningen, The Netherlands and with major contributions from Germany, France and the US. Consortium members are: Canada: CSA, U.Waterloo; France: CESR, LAB, LERMA, IRAM; Germany: KOSMA, MPIfR, MPS; Ireland, NUI Maynooth; Italy: ASI, IFSI-INAF, Osservatorio Astrofisico di Arcetri- INAF; Netherlands: SRON, TUD; Poland: CAMK, CBK; Spain: Observatorio Astronómico Nacional (IGN), Centro de Astrobiología (CSIC-INTA). Sweden: Chalmers University of Technology – MC2, RSS & GARD; Onsala Space Observatory; Swedish National Space Board, Stockholm University – Stockholm Observatory; Switzerland: ETH Zurich, FHNW; USA: Caltech, JPL, NHSC.).

References

Appendix A: Water line transitions observed with Herschel/HIFI

Table A.1

Water line transitions observed with Herschel/HIFI.

Appendix B: Water maser positions

Table B.1

Water maser positions and velocities at 22 GHz (e-MERLIN observations of April 2016).

All Tables

Table 1

e-MERLIN 22 GHz imaging parameters.

Table 2

Observed line emission parameters for the detected lines with HIFI toward NGC 7538-IRS1.

Table A.1

Water line transitions observed with Herschel/HIFI.

Table B.1

Water maser positions and velocities at 22 GHz (e-MERLIN observations of April 2016).

All Figures

thumbnail Fig. 1

Continuum-subtracted SOFIA spectra of the o-H2O 82,7−73,4 line at 1296.41106 GHz toward NGC 7538-IRS1. The vertical dotted line indicates VLSR at –57.4 kms-1. The spectral resolution is 1.129 kms-1. The Gaussian fit is shown in red. The small insert shows the same spectra on a zoomed velocity scale.

Open with DEXTER
In the text
thumbnail Fig. 2

Effelsberg (in black, spectral resolution δv of 0.1 kms-1) and e-Merlin (in blue, δv = 0.1 kms-1, the apparent absorption near –63.5 kms-1 is attributed to e-MERLIN residual sidelobes, see Sect. 3.3) spectra of the 616−523 transition of ortho-H2O at 22 GHz toward NGC 7538-IRS1 plotted over the SOFIA (δv = 1.1 kms-1) spectrum of the o-H2O 82,7−73,4 line at 1296.41106 GHz (in red).

Open with DEXTER
In the text
thumbnail Fig. 3

Entire e-MERLIN map of 22 GHz maser emission from NGC 7538, including IRS1-3, IRS9, and IRS11, of the 22 GHz water maser line. Colored circles show the relative position of individual maser features, with color denoting the maser VLSR, according to the color-velocity conversion code reported on the right side of the panel. The water maser positions reported by Galván-Madrid et al. (2010), corrected for proper motion (Moscadelli & Goddi 2014), are shown as crosses (same color-velocity conversion code) for comparison. The beam of the SOFIA observation is overlaid in black for comparison.

Open with DEXTER
In the text
thumbnail Fig. 4

As in Fig. 3, but zoomed on IRS1-3.

Open with DEXTER
In the text
thumbnail Fig. 5

Zoom toward IRS1 north of the e-MERLIN map shown in Fig. 3. Colored empty and filled circles show the relative position of individual 22 GHz water masers (this paper) and 6.7 GHz methanol maser features (Moscadelli & Goddi 2014), respectively. VLSR is according to the color-velocity conversion code shown on the right side of the panel.

Open with DEXTER
In the text
thumbnail Fig. 6

HIFI spectra of HO/HO (left) and HO (right) lines (in black), with the continuum for NGC 7538-IRS1 pointed position. The best-fit model is shown as a red line over the spectra (χin(H2O) = 8 × 10-6 and χout(H2O) = 4 × 10-8). The model adopting the SOFIA water inner abundance (5.2 × 10-5) is shown as a blue line above the spectra. Vertical dotted lines indicate the VLSR (–57.4 kms-1 from the line modeling). The spectra have been smoothed to 0.2 kms-1, and the continuum is divided by a factor of two.

Open with DEXTER
In the text
thumbnail Fig. 7

SOFIA spectra of the o-H2O 82,7−73,4 line showing for feature (1) the best-fit thermal model in red from our HIFI data (see Fig. 6) and a model adopting a water inner abundance of 5.2 × 10-5 (blue profile).

Open with DEXTER
In the text
thumbnail Fig. 8

As in Fig. 2, with all spectra smoothed to a spectral resolution of 1.1 kms-1. The SOFIA fluxes have been divided by a factor 7.4.

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.