The search for high-mass protostars with ALMA revealed up to kilo-parsec scales (SPARKS): I. Indication for a centrifugal barrier in the environment of a single high-mass envelope
T. Csengeri, S. Bontemps, F. Wyrowski, A. Belloche, K. M. Menten, S. Leurini, H. Beuther, L. Bronfman, B. Commerccon, E. Chapillon, S. Longmore, A. Palau, J. C. Tan, J. S. Urquhart
aa r X i v : . [ a s t r o - ph . GA ] A p r Astronomy & Astrophysicsmanuscript no. G328p25_14032018_1strevision c (cid:13)
ESO 2018April 19, 2018
The search for high-mass protostars with ALMA revealed up tokilo-parsec scales (SPARKS)
I. Indication for a centrifugal barrier in the environment of a single high-massenvelope
T. Csengeri , S. Bontemps , F. Wyrowski , A. Belloche , K. M. Menten , S. Leurini , H. Beuther , L. Bronfman , B.Commerçon , E. Chapillon , , S. Longmore , A. Palau , J. C. Tan , , and J. S. Urquhart (A ffi liations can be found after the references) Received , 2017; accepted , 2017
ABSTRACT
The conditions leading to the formation of the most massive O-type stars, are still an enigma in modern astrophysics. To assess the physicalconditions of high-mass protostars in their main accretion phase, here we present a case study of a young massive clump selected from theATLASGAL survey, G328.2551-0.5321. The source exhibits a bolometric luminosity of 1 . × L ⊙ , which allows us to estimate its currentprotostellar mass to be between ∼
11 and 16 M ⊙ . We show high angular-resolution observations with ALMA reaching a physical scale of ∼
400 au.To reveal the structure of this high-mass protostellar envelope in detail at a ∼ ′′ resolution, we use the thermal dust continuum emission andspectroscopic information, amongst others from the CO ( J = J = ( J = , − , ) lines tracing shocks along the outflow, as well as several CH OH and HC N lines that probe the gas of the inner envelopein the closest vicinity of the protostar. The dust continuum emission reveals a single high-mass protostellar envelope, down to our resolution limit.We find evidence for a compact, marginally resolved continuum source, which is surrounded by azimuthal elongations that could be consistentwith a spiral pattern. We also report on the detection of a rotational line of CH OH within its t = ff set from the dust continuum peak, and exhibiting a distinct velocity component ± − o ff set comparedto the source v lsr . Rotational diagram analysis and models based on local thermodynamic equilibrium (LTE) assumption require high CH OHcolumn densities reaching N (CH OH) = . − × cm − , and kinetic temperatures of the order of 160-200 K at the position of these peaks.A comparison of their morphology and kinematics with those of the outflow component of the CO line, and the SO line suggests that the highexcitation CH OH spots are associated with the innermost regions of the envelope. While the HC N = J = N = e ( J = OH spots correspond well to the expected Keplerian velocity around acentral object with 15 M ⊙ consistent with the mass estimate based on the source’s bolometric luminosity. We propose a picture where the CH OHemission peaks trace the accretion shocks around the centrifugal barrier, pinpointing the interaction region between the collapsing envelope andan accretion disk. The physical properties of the accretion disk inferred from these observations suggest a specific angular momentum severaltimes larger than typically observed towards low-mass protostars. This is consistent with a scenario of global collapse setting on at larger scalesthat could carry a more significant amount of kinetic energy compared to the core collapse models of low-mass star formation. Furthermore, ourresults suggest that vibrationally exited HC N emission could be a new tracer for compact accretion disks around high-mass protostars.
Key words. stars: massive – stars: formation – submillimeter: ISM
1. Introduction
Whether high-mass star formation proceeds as a scaled-up version of low-mass star formation is an open questionin today’s astrophysics. Signatures of infall and accretionprocesses associated with the formation of high-mass starsare frequently observed: ejection of material (Beuther et al.2002; Zhang et al. 2005; Beltrán et al. 2011; Duarte-Cabral et al.2013) with powerful jets (Guzmán et al. 2010; Moscadelli et al.2016; Purser et al. 2016) and the existence of (massive) rotatingstructures, such as toroids and disks has been reported towardsmassive young stellar objects (MYSOs) (Beltrán et al. 2005;Sanna et al. 2015; Cesaroni et al. 2017). Most of these studiesfocus, however, on sources with high luminosities ( L bol > × L ⊙ ), and are frequently associated with at least one em-bedded UC H II region (see also Mottram et al. 2011). Someof them harbour already formed O-type YSOs typically ac-companied by radio emission and surrounded by hot molec- ular, as well as ionised gas. Some examples are G23.01-00.41 (Sanna et al. 2015), G35.20-0.74N (Sánchez-Monge et al.2013a), and G345.4938 + II region, is characterised by typi-cally lower bolometric luminosity (e.g. Molinari et al. 2000;Sridharan et al. 2002; Motte et al. 2007). This stage can be con-sidered as an analog of the Class 0 stage of low-mass protostars(Duarte-Cabral et al. 2013), which could be the main accretionphase, dominated by the cold and dusty envelope, and is accom-panied by powerful ejection of material (Bontemps et al. 1996;André et al. 2000). Due to the high column densities, extinction Article number, page 1 of 18 & Aproofs: manuscript no. G328p25_14032018_1strevision is very high towards these objects, and therefore high-mass pro-tostars in this early stage are elusive (e.g. Bontemps et al. 2010;Motte et al. 2017). Rare examples of them are found to be sin-gle down to ∼
500 au scales. They typically only probe a lim-ited mass range: one of the best studied sources is the protostarCygX-N63 with a current envelope mass of ∼ M ⊙ potentiallyforming a star with a mass of . M ⊙ (Bontemps et al. 2010;Duarte-Cabral et al. 2013).We report here on the discovery of a single high-mass proto-stellar envelope with the largest mass observed so far, reachingthe 100 M ⊙ mass range on 0.06 pc scale, and potentially forminga star with a mass of ∼ M ⊙ , corresponding to an O5-O4 typestar. We resolve its immediate surroundings using high angularresolution observations reaching ∼
400 au scale with the AtacamaLarge Millimeter / submillimeter Array (ALMA), which show ev-idence for a flattened, rotating envelope, and for accretion shocksimplying the presence of a disk at a few hundred au scales.
2. Observations and data reduction
We study at high angular-resolution the mid-infrared quietmassive clump , G328.2551-0.5321, selected from the com-plete sample of such sources identified from the APEXTelescope Large Area Survey of the Galaxy (ATLASGAL)(Schuller et al. 2009; Csengeri et al. 2014, 2017a). The SPARKSproject (Search for High-mass Protostars with ALMA up to kilo-parsec scales, Csengeri et al., in prep, a) targets 35 of thesesources corresponding to the early evolutionary phase of high-mass star formation (Csengeri et al. 2017b). Located at a dis-tance of 2.5 + . − . kpc , our target is embedded in the MSXDCG328.25-00.51 dark cloud (Csengeri et al. 2017a). We show anoverview of the region in Fig. 1, left panel.G328.2551-0.5321 has been observed with ALMA in Cy-cle 2, and the phase center was ( α, δ ) J2000 = (15 h m . s , − ◦ ′ ′′ λ ) to 1574 m (1809 k λ ). The total time onsource was 7.4 minutes, and the system temperature ( T sys ) variesbetween 120 and 200 K.The spectral setups used for the 7 m and 12 m array ob-servations are identical, and the signal was correlated in low-resolution wide-band mode in Band 7, yielding 4 × .
75 GHze ff ective bandwidth with a spectral resolution of 0.977 MHzwhich corresponds to ∼ − velocity resolution. The fourbasebands were centred on 347.331, 345.796, 337.061, and333.900 GHz, respectively.The data have been calibrated in CASA 4.3.1 with thepipeline (version 34044). For the imaging, we used Briggsweighting with a robust parameter of −
2, corresponding to uni-form weighting, favouring a smaller beam size, and used theCLEAN algorithm for deconvolution. We created line-free con-tinuum maps by excluding channels with line emission above3 σ rms per channel determined on the brightest continuum sourceon the cleaned datacubes in an iterative process. The synthesisedbeam is 0.22 ′′ × . ′′ with 86 ◦ position angle measured from Mid-infrared quiet massive clumps are defined by weak or no emis-sion in the 21 − µ m wavelength range. We follow here the definitionof Csengeri et al. (2017a) which is based on Motte et al. (2007). As dis-cussed there, our mid-infrared flux limit corresponds to that of an em-bedded star with 10 L ⊙ . north to east, which is the convention we follow from here on.The geometric mean of the major and minor axes correspondsto a beam size of 0 . ′′ ( ∼
400 au at the adopted distance for thesource).To create cubes of molecular line emission we subtractedthe continuum determined in emission free channels aroundthe selected line. To favour sensitivity, we used here a robustparameter of 0 . ′′ × . ′′ , corresponding, on average, to a 0 . ′′ resolution ( ∼
650 au). For the SiO (8–7) datacube we loweredthe weight on the longest baselines with a tapering functionto gain in signal to noise. The resulting synthesised beam is0.69 ′′ × ′′ (corresponding to ∼ ∼ ′′ , where the sensitivity of our 12 m array observationsdrops. The interferometric filtering has, however, a significantimpact on both the continuum and the CO (3–2) emission, whichare considerably more extended than the largest angular scalesprobed by the ALMA configuration used for these observations.For these two datasets, we therefore combine the 12 m arraydata with the data taken in the same setup with the 7 m array(Csengeri et al. 2017b). For this purpose we used the standardprocedures in CASA for a joint deconvolution, and imaged thedata with a robust parameter of − . × . ′′ with 86 ◦ positionangle, with a geometric mean of 0 . ′′ for the continuum maps,and 0 . × . ′′ with 88 ◦ position angle, with a geometric meanof 0 . ′′ for the CO (3–2) line. We measure an rms noise levelin the final continuum image of ∼ . / beam in an emissionfree region close to the centre on the image corrected for primarybeam attenuation. The primary beam of the 12m array at this fre-quency is 18. ′′
5. We measure a 3 σ rms column density sensitivityof 2 − × cm − for a physically motivated range of plausi-ble temperatures corresponding to T d = −
100 K, respectively,and within a beam size of 0.16 ′′ , and a 5 σ mass sensitivity of0.03–0.13 M ⊙ for the same temperature range.
3. Results
We discuss here the dust continuum image obtained withALMA, which reveals a protostellar envelope that stays sin-gle down to our resolution limit of 400 au in physical scale(Sect. 3.1). We then report the detection of a high-energy ro-tational line of CH OH within its t = OH lines within its t = OH emitting gas(Sect. 3.3). Finally, to constrain the origin of the CH OH emis-sion, we compare it to other molecules, such as CO, SO andHC N tracing outflowing gas associated with the protostellar ac-tivity (Sect. 3.4). The full view of the molecular complexity ofthis source will be discussed in a forthcoming paper (Csengeri etal. in prep, b). We use N (H ) = F ν RB ν ( T d ) Ω κ ν µ H2 m H [cm − ], where F ν is the 3 σ rms fluxdensity, B ν ( T ) is the Planck-function, Ω is the solid angle of the beamcalculated by Ω = . × Θ , where Θ is the geometric mean of thebeam major and minor axes; κ ν = . − including the gas-to-dust ratio, R, of 100; µ H is the mean molecular weight per hydrogenmolecule and is equal to 2.8; and m H is the mass of a hydrogen atom.Article number, page 2 of 18. Csengeri et al.: The search for high-mass protostars with ALMA revealed up to kilo-parsec scales (SPARKS) Fig. 1.
Overview of the region centred on ℓ =+ . b = − . Spitzer / GLIMPSE (Benjamin et al. 2003) and MIPSGAL (Carey et al. 2009) surveys (blue: 4.5 µ m, green: 8 µ m, red: 24 µ m) and is shownin Galactic coordinates. The contours show the 870 µ m continuum emission from the ATLASGAL survey (Schuller et al. 2009; Csengeri et al.2014). The arrow marks the dust continuum peak of the targeted clump. Right:Line-free continuum emission imaged at 345 GHz with the ALMA12m array. The color scale is linear from − σ to 120 σ . The contour shows the 7 σ level. The FWHM size of the synthesised beam is shown in thelower left corner. Fig. 2.
A zoom on Fig. 1 of the protostar centred on ∆ α = − . ′′ , and ∆ δ = − . ′′ o ff set from the phase center. a)Continuum emission smoothedto a resolution of 0.27 ′′ showing the extended emission at the scale of the MDC. The color scale is linear between − / beam, andthe contour displays the 1.4 mJy / beam level corresponding to ∼ σ . Dashed ellipse shows the FWHM of the MDC from Table 1 adopted fromCsengeri et al. (2017b). The region of the inner envelope shown in panels b to c is outlined by a dashed box. The
FWHM beam is shown in thelower left corner of all panels. b) Line-free continuum emission in the original, unsmoothed 12m and 7m array combined map where the beam hasa geometric mean of 0.17 ′′ . The color scale is linear between − σ and 120 σ, contours start at 7 σ and increase on a logarithmic scale up to 120 σ by a factor of 1.37. The red and blue dashed lines show the direction of the CO outflow (see Fig. 7 and Sect. 3.4). The dark gray dashed ellipseshows 2 × the major and minor axes of the fitted 2D Gaussian. c) Residual continuum emission after removing a 2D Gaussian with a fixed lowerpeak intensity from the envelope component in order to enhance the contrast of the inner envelope. The color scale is linear from 0 to 50 σ . Thecontours start at 3 σ and increase by 6 σ . White ellipse shows the FWHM of the fitted 2D Gaussian to the residual from panel b, and the green lineoutlines the azimuthal elongations. The black dotted line marks the direction perpendicular to the outflow. d) Residual continuum emission afterremoving the Gaussian fit to the envelope component (see the text for details). The color scale is the same as in panel c. Contours start at 5 σ andincrease by 10 σ . Green contours show 80% of the peak of the velocity integrated emission of the 334.436 GHz t = OH line shown in Fig. 4.
We show the line-free continuum emission of G328.2551-0.5321in Fig. 1, right panel, which reveals a single compact object downto 400 au physical scales. The source drives a prominent bipolaroutflow (Sect. 3.4) suggesting that it hosts a protostar undergoingits main accretion phase. We resolve well the structure of theenvelope, and show a zoom on the brightest region in Fig. 2a.To extract the properties of the bulk emission of the dust, weuse here a 2D Gaussian fit in the image plane as a first approach. This reveals the position of the continuum peak at ( α, δ ) J2000 = (15 h m . s , − ◦ ′ ′′ − . ′′ , − . ′′ o ff setfrom the phase centre, and gives a full-width at half-maximum( FWHM ) of 0.96 ′′ × ′′ with a position angle of 74 ◦ . We cal-culate the envelope radius as R env90% = . × Θ / √ Θ is the beam deconvolved geometric mean of the major and minoraxes FWHM . This gives a radius, R env90% , of 0.59 ′′ correspondingto 1500 au. In the following we refer to this component, thus thestructure within 1500 au, as the inner envelope (Fig. 2b).In order to investigate the structure of the inner envelope, weremove a low-intensity 2D Gaussian fit, i.e. allowing only pos-itive residuals for the brightest, central structure (Fig. 2c). This Article number, page 3 of 18 & Aproofs: manuscript no. G328p25_14032018_1strevision reveals azimuthal elongations exhibiting a warped S-shape con-necting to the continuum peak.
Fig. 3.
Real part of the visibility measurements versus uv -distanceshown for the ALMA 12 m array data. The data points show an averageof line-free channels. The red and blue lines show fits to the envelope,an elliptic Gaussian, and a single component power-law fit, respectively.The green line shows a two component fit with a power-law and a com-pact disk source (see the text for more details). When removing the bulk of the dust emission, i.e. our first 2DGaussian fit to the envelope shown in Fig. 2b, we find a compactresidual source with a peak intensity larger than 10 σ in Fig. 2d.We determine its properties with another 2D Gaussian fit in theimage plane data and identify a compact source that is resolvedonly along its major axis. The beam deconvolved R disk90% radius is ∼ ′′ , corresponding to an extent of 250 au. The properties andthe corresponding flux densities of the inner envelope and thecompact component based on our analysis in the image planeare summarised in Table 1.These results suggest that the envelope itself is not welldescribed by a Gaussian flux density distribution, and to testwhether the residual source is better represented by a di ff erentflux density profile, we also perform a fitting procedure in the uv -domain. We fit the visibilities averaged over line-free channels asa function of uv -distance for the ALMA 12m array data (Fig. 3).We find that a single power-law component provides a relativelygood fit to the data up to ∼
300 m long baselines. The visibilitypoints show,however, a significant residual on the longer base-lines further suggesting the presence of a compact component(see inset of Fig. 3).We used various geometries to fit the compact component,however, we can only constrain that it is unresolved along itsminor axis. Our models show that the compact source is con-sistent with a disk component with a flux density of 43 mJy,and a 0.2 ′′ major axis FWHM . As a comparison, our analysisin the image plane attributes a somewhat larger flux density tothis compact component, but finds a similar spatial extent. Laterwe argue that this component, that is significantly more compactthan the in the inner envelope, likely corresponds to a compactaccretion disk around the central protostar (Sect. 4.3).
To understand the distribution of dust emission on various scales,we compare the recovered emission from the 12 m array obser- vations, the 12 m and 7 m array combined data, and the total fluxdensity from the single dish observations from ATLASGAL. Werecover a total flux density of 2.9 Jy in the field with the 12 m ar-ray observations only, which is a significant fraction (73%) of the ∼ µ m single-dish peak intensity measured on the ATLASGALemission map is 10.26 Jy at this position, from which 8.32 Jyhas been assigned to the clump in the catalog of Csengeri et al.(2014). This means that the 12m and 7m array combined obser-vations recover ∼
50% of the total dust emission from the clump.Clearly, there is a large concentration of emission on the small-est scales which agrees with our previous results comparing theclump and the core scale properties in Csengeri et al. (2017a).Based on this information, we describe the structure of thesource in the following. We attribute the large scale emissionto the clump, whose parameters are obtained from the ATLAS-GAL data at 0 .
32 pc scales (Fig. 1a). The smaller scale structureis attributed to a Massive Dense Core (MDC) forming a sin-gle protostellar envelope which has been first identified basedon the ALMA 7m array observations in Csengeri et al. (2017b)at ∼ N (H ) ∼ . − × cm − for T = −
100 K, we smoothed the data to illustrate the extent of theMDC. The original, not smoothed ALMA 12 m and 7 m arraycombined data reveals the brightest emission with a 1500 au ra-dius corresponding to the inner regions of the envelope showingthe highest column densities.Here, we attempt to provide a more robust mass estimate onthe available mass reservoir for accretion based on the MDCproperties within a radius of ∼ T d ) from a two component (coldand warm) greybody fit to the far-infrared spectral energy distri-bution (SED) (see App. A), where in particular the wavelengthsshorter than 70 µ m are a sensitive probe to the amount of heateddust in the vicinity of the protostar. From the fit to the SED weobtain a cold component at 22 K , and a warm component at ∼
48 K that we assign to the inner envelope. These models show,that the amount of warm gas is only a small fraction ( < T d =
22 Kfor the MDC we obtain M MDC = M ⊙ . Using a di ff erent dustemissivity (e.g. Peretto et al. 2013), and / or a gas-to-dust ratio of150 would increase this value by 50-100%. The uncertainty inthe distance estimation of this source would either decrease thisvalue by 36%, or increase it by roughly a factor of three. On themeasured physical sizes the e ff ect of distance uncertainty is lessdramatic, it would either decrease the inner envelope radius by20% or increase it by 70%, if the source is located at the farthestlikely distance.Since the MDC is gravitationally bound, its mass should beavailable for accretion onto the central protostar. Assuming ane ffi ciency of 20-40% (Tanaka et al. 2017), we can expect that anadditional 24 − M ⊙ could still be accreted on the protostar.Therefore, this makes our target one of the most massive proto-stellar envelopes known to date, which is likely in the process offorming an O-type star. In Csengeri et al. (2017a) we used T d =
25 K for all cores. We used Eq. 2 from Csengeri et al. (2017a) with the same pa-rameters for the dust (dust emissivity, κ ν = . g − fromOssenkopf & Henning 1994 accounting for a gas-to-dust ratio, R , of100.)Article number, page 4 of 18. Csengeri et al.: The search for high-mass protostars with ALMA revealed up to kilo-parsec scales (SPARKS) Table 1.
Dust continuum measurements.
Component Peak Intensity Integrated flux d.
FWHM p.a. a Physical size Used data[Jy / beam] [Jy] [ ′′ × ′′ ]Clump 8.32 14 .
95 28.3 ′′ × ′′ − ◦ .
32 pc APEX / LABOCA b MDC 2.10 4.0 6.1 ′′ × ′′ ◦ .
06 pc ALMA 7m array c Inner envelope d ′′ × ′′ ◦ e ALMA 7m +
12m arrayResidual (disk) d ′′ × ′′ ◦
250 au e , f ALMA 7m +
12m array
Notes. ( a ) The position angle of the fitted Gaussian is measured from north to east. ( b ) The corresponding parameters are extracted from the catalog of Csengeri et al. (2014) based on the ATLASGAL data. ( c ) The listed parameters are from the ALMA 7 m array from Csengeri et al. (2017b). ( d ) Parameters obtained with a 2D Gaussian fit in the image plane in this work. ( e ) Corresponds to the beam deconvolved R , see the text for details. ( f ) This estimate is based on the resolved major axis.
Table 2.
Summary of the molecular transitions studied in this work.
Molecule Quantum number Frequency Log Aij E up / k n a cr Database[GHz] [s − ] [K] [cm − ]CH OH − A t = − − − − .
57 45 1 . × CDMSCH OH − A t = − − .
29 79 3 . × CDMSCH OH − A t = − − .
25 488 6 . × CDMSCH OH − A t = − − − .
84 197 1 . × CDMSCH OH − E t = − − .
61 62 6 . × CDMSCH OH − E t = –2 − .
26 315 CDMSCH OH − A t = − − .
79 747 JPL CH OH − A t = c − –12 − .
40 193 1 . × CDMS CH OH − A t = c − –14 − .
39 254 6 . × CDMSHC N = b − .
52 307 CDMSHC N = e − .
48 645 CDMSSO v = , − , − .
26 43 7 . × JPLCO 3 − − . × CDMS
Notes. ( a ) Calculated at T =
100 K using collisional rate coe ffi cients from the LAMDA database (Schöier et al. 2005) where available. ( b ) For HC N the molecular datafile lists cross sections up to J up =
21, and T =
80 K. ( c ) Calculated from the collisional rate coe ffi cients of the main isotopologue. Fig. 4. a) Color scale shows the integrated intensity map of the t = OH line at 334.4 GHz. The green triangles indicate the positions wherethe spectrum has been extracted for the rotational diagram analysis on the CH OH spots, and are labeled as A and B components. The black crossmarks the position of the dust continuum peak. The beam is shown in the lower left corner. b) The color scale shows the continuum emission fromFig. 2b, contours and markers are the same as on panel a. c) Integrated spectrum of the torsionally excited CH OH transition at 334.4 GHz overthe area shown in panel a. The green lines show the two component Gaussian fit to the spectrum. The blue dashed line shows the v lsr of the source.d) Position-velocity diagram along the ∆ α axis and averaged over the shown extent of the cube corresponding to ∼ ′′ . The dotted lines mark theposition of the dust peak and the v lsr of the source. The total observed bandwidth of 7.5 GHz reveals emission fromseveral molecular species. In this study we focus on a selectedlist of molecules summarised in Table 2. We first discuss thetorsionally excited CH OH emission in Sect. 3.2.1, and to bet- ter constrain its physical origin in Sect. 3.2.2 we discuss all un-blended torsional ground state transitions of CH OH.
Article number, page 5 of 18 & Aproofs: manuscript no. G328p25_14032018_1strevision
Fig. 5. CH OH 0th moment map and position velocity diagrams for the transitions listed in Table 2. The top row shows the 0th moment mapcalculated over a velocity range of [ − −
30] km s − , contours start at 30% of the peak and increase by 15%. White cross marks the position ofthe continuum peak. The beam is shown in the lower left corner. The symmetry of the methanol molecule is labeled on each panel, as well as theupper level energy of the transition. The subsequent row shows the position velocity map along the ∆ α axis and averaged over the shown extent ofthe cube corresponding to ∼ ′′ . The right-ascension o ff set of the continuum peak and the v lsr velocity of the source are marked as white dottedline. Dashed lines show the v lsr ± . − corresponding to the peak velocity of the CH OH spots.
Our continuous frequency coverage between 333.2 and337.2 GHz, as well as between 345.2 and 349.2 GHz, includesseveral transitions of CH OH and its C isotopologue. Most in-terestingly, towards the inner envelope, we detect and spatiallyresolve emission from a rotational transition of CH OH, fromits first torsionally excited, t =
1, state at 334.42 GHz withan upper energy level of 315 K (Fig. 4a). Its spatial morphol-ogy shows two prominent emission peaks (marked as A and Bin Fig. 4a), which spatially coincide with the azimuthal elonga- tions within the envelope. The emission drops, however, signifi-cantly towards the continuum peak, i.e. the protostar (Fig. 4, b).We find that the observed morphology is dominated by two ve-locity components, which show an o ff set, on average, of − . + . − compared to the v lsr of the source (Table 3). Asdiscussed in Sect. 3.2.2, optical depth e ff ects are unlikely to be atthe origin of the observed velocity pattern. We spatially resolvethe emission from these spots, and estimate the peak of its dis- We adopt the v lsr of the dense gas seen on the clump / core scale usedin Csengeri et al. (2017a).Article number, page 6 of 18. Csengeri et al.: The search for high-mass protostars with ALMA revealed up to kilo-parsec scales (SPARKS) Table 3.
Observational parameters for the CH OH t = OH.
Observed parameters LTE fit parameters v lsr a ∆ v N size T ex [km s − ] [km s − ] [cm − ] [ ′′ ] [K]A − . ± − .
6) 4.5 ± . × − . ± + .
6) 5.6 ± × Notes. ( a ) The number given in parenthesis corresponds to the di ff erencebetween the line velocity and the − . − v lsr of the source. tribution to fall between a projected distance of 300 and 800 ausymmetric from the protostar.The pv -velocity diagram along the ∆ α axis and averaged per-pendicularly to this axis over the shown extent of ∼ . ′′ revealsthat the emission is dominated by the two velocity components,and shows a pattern consistent with rotational motions (Fig. 4d).We compare these observations to simple models of the gas kine-matics in more detail in Sect. 4.3. To further investigate the origin of the 334.426 GHz t = OH and its C isotopologue in the torsionalground state, t =
0. They probe a range of upper energy lev-els between 45 K and 488 K, and based on our LTE modelling(Sect. 3.3.2) they are unlikely to be blended with emission fromother species. We use these lines, in particular, to test whether ahigh optical depth toward the protostar could mimic the observedvelocity pattern and morphology of the torsionally excited stateline.These maps reveal three transitions of the CH OH − A sym-metry state with upper level energies of E up <
200 K that peak onthe continuum source, while its higher energy transitions showthe two prominent peaks like the t = OH line . Our LTEmodelling in Sect. 3.3.2, indeed shows that the three lowest en-ergy transitions of CH OH − A have high optical depths.The other transitions are, however, optically thin and theyshow two peaks of emission o ff set from the protostar similarlyto the t = OH line, while the emission drops towardsthe position of the protostar. In addition, their pv -diagrams arealso similar to the t = OH line revealing the two velocitycomponents. Among these lines, we have the two CH OH tran-sitions detected with a high signal-to-noise ratio which are theleast a ff ected by optical depth e ff ects. This leads us to concludethat the two prominent spots traced by the t = OHline cannot be a result of a large optical depth of CH OH towardsthe continuum peak.We calculate the critical densities for these CH OH transi-tions in Table 2, and find that they all trace high density gas (ifthermalised), strictly above 10 cm − , but typically on the orderof 10 cm − . We notice that the di ff erent transitions have a vary-ing contribution as a function of upper energy level from thecentral source, which suggests that they may trace two physi-cal components, one associated with the inner envelope showingthe bulk emission of the gas likely at lower temperatures, andanother, warmer and denser component associated with the twopeaks of the CH OH t = The CH OH − E transitions in the band have an order of magnitudesmaller Einstein coe ffi cients which can explain why despite their lowerenergy level, all the CH OH − E lines show two peaks of emission. We use here two methods to measure the physical conditions to-wards the methanol spots, and the position of the protostar aswell. For this we extracted the spectrum covering the entire ob-served 7.5 GHz. To convert it from Jy / beam to K scales we useda factor of 198 K / Jy calculated for the 0.23 ′′ averaged beam sizeat 335.2 GHz, and 347.2 GHz. Taking a mean conversion factorfor the entire bandwidth adds less than 10% inaccuracy in themeasurement of the brightness temperatures. OH transitions
We perform a rotational diagram analysis (Garay et al. 2010;Gómez et al. 2011) to estimate the rotational temperature of theCH OH emission, and its column density, N (CH OH) at theCH OH peak positions indicated in Fig. 4, and towards the posi-tion of the protostar. We extracted the integrated intensities us-ing a Gaussian fit to the CH OH transitions listed in Table 2.We include the CH OH lines with an isotopic ratio of 60(Langer & Penzias 1990; Milam et al. 2005), and exclude the335.582 GHz and 336.865 GHz transitions from the fit. This isbecause as a combination of their large optical depths and inter-ferometric filtering due to their spatial extent (see in Fig. 5) weobserve considerably lower fluxes than expected if the CH OHemission is thermalised. For the rest of the transitions we assumeoptically thin emission. We show the rotational diagram in Fig. 6,where the error bars show the measured error on the Gaussian fitto the spectral lines, and are on the order of 10%. We obtain sim-ilar values for the two methanol peaks of T rot = −
175 K,and N (CH OH) = ∼ × cm − . The individual measurementshave relatively large uncertainties, however, our results suggestthat the two methanol spots have on the order of magnitude sim-ilar temperatures and column densities. Ignoring the e ff ect ofpotentially more severe blending, we performed the same mea-surement on a spectrum extracted towards the continuum peak,which suggests similarly low rotational temperature as towardsthe brightest CH OH spot, and shows a somewhat lower columndensity of N (CH OH) = ∼ × cm − . While Fig. 5 suggestssystematic di ff erences in the CH OH emission between the highexcitation CH OH spots and the continuum peak, the populationdiagram analysis shows that the three positions have similar col-umn densities and rotational temperatures of methanol. This isparticularly interesting since the radiation field, and hence thetemperature is expected to be the strongest at the position of theprotostar.
Our observations cover a 7.5 GHz bandwidth, and the spectraextracted towards the CH OH peak positions show line forests ofother molecular species, typically COMs. Therefore, to analysethe CH OH emission, we modelled the entire spectrum using theWEEDs package (Maret et al. 2011) assuming LTE conditions,which are likely to apply due to the high volume densities. Themolecular composition of the gas towards the CH OH peaks willbe described in a forthcoming paper, together with the detailedanalysis.In short, we performed the modelling in an iterative process,and started first with the CH OH lines. The input parameters arethe molecular column density, kinetic temperature, source size, v lsr , and line-width. From these parameters we fix the source sizeto 0.4 ′′ , which means that the emission is resolved, as it is sug-gested by the data (Fig. 5). The modelled transitions may have Article number, page 7 of 18 & Aproofs: manuscript no. G328p25_14032018_1strevision
Fig. 6.
Rotational diagram of the CH OH (and isotopologue) transitionsfrom Table 2. The colours correspond to the measurements on di ff erentpositions; green shows the position marked as A, red shows the positionmarked as B, and blue corresponds to the central position marked bya cross in Fig. 4. The filled circles show the CH OH lines, squares the CH OH lines, and triangles the optically thick lines that are not usedfor the fit. The error bars show the linearly propagated errors from theGaussian fit to the integrated intensities. di ff erent source sizes, however, as long as they are resolved byour observations, the actual source size does not influence the re-sult. The line-width of the CH OH line is obtained by the Gaus-sian fits to the extracted spectra. After obtaining a first, reason-ably good fit to the listed CH OH transitions, we started to sub-sequently add other molecular species, mainly COMs that are re-sponsible for the lower intensity lines (Csengeri et al in prep,b).We created a grid of models for the CH OH column density, N (CH OH), between 10 and 10 cm − , and kinetic tempera-tures between 50 and 300 K. We sampled by 25 linearly spacedvalues both parameter ranges, and then visually assessed the re-sults. These models show that the detection or non detection ofcertain transitions allows us to put a rather strict upper limit onthe kinetic temperature. Above T kin >
200 K, our models predictthat other transitions of the t = J = , , − , , ), and 334.680 GHz ( J = − , , − − , , ).Although these frequency ranges are a ff ected by blending withCOMs, models with T kin >
200 K predict these rotational tran-sitions within its t = × cm − , and a kinetic temperature around 160–200 K forthe two positions. We find that the peak brightness temperaturesfor the transitions observed across the band are more sensitive tothe methanol column density than to the variation in kinetic tem-perature. While this modelling takes into account blending withother transitions, our results are consistent with the estimates ofcolumn density and rotational temperature obtained from the ro-tational diagram analysis (Sect. 3.3.1). J =3–2) line The protostellar activity of the compact continuum source is re-vealed by a single bipolar molecular outflow shown in Fig. 7.The CO (3–2) line shows emission over a broad velocity range, ∆ v , of ±
50 km s − with respect to the source rest velocity ( v lsr )(Fig. 7a). Imaging the highest velocities of this gas reveals a sin-gle and confined bipolar molecular outflow. The orientation of the high-velocity CO emission clearly outlines the axis of ma-terial ejection. The high velocity CO (3–2) emission coincideswell with the integrated emission from the shock tracer SiO(8–7) (Fig. 7b), which likely traces the bow shocks along theoutflow-axis (Fig. 7c).In particular the brightest CO emission of the red-shiftednorthern lobe appears to be confined (Fig. 7c). Coinciding withthe terminal position of the northern lobe, we detect emission atthe source rest velocity of the shocked gas tracer, SiO (8–7), thatis analogous to the bow-shocks observed in the vicinity of low-mass protostars. These indicate the shock front of the outflowinggas impacting the ambient medium.Assuming that the maximum velocity observed in the CO(3–2) line corresponds to the speed at which the material hasbeen ejected from the vicinity of the protostar, we can estimatewhen this material has been ejected. We measure an angular sep-aration of ∼ ′′ between the central object and the bow-shockseen in the SiO (8–7) line corresponding to a projected physi-cal distance of 25000 au. Based on the observed line-wings ofthe CO (3–2) line, the measured velocity extent of the flow is ∼
50 km s − . We estimate an inclination angle, i , of 56 ◦ , where i = ◦ describes a face-on geometry, and i = ◦ corresponds toan edge-on view. This is obtained from the axis ratio of the mea-sured envelope size from the dust emission assuming that it has acircular morphology . After correcting for the projection e ff ects,we obtain a dynamical age estimate of t dyn ∼ . × yr for theprotostar. This estimate is, however, a ff ected by the uncertaintyof the inclination angle, and that of the jet velocity creating thebow shocks compared to the high velocity entrained gas seen byCO. While the jet velocity could be higher than traced by theentrained CO emitting gas leading to an even shorter time-scale,the highest velocities seen in CO may not reflect the expansionspeeds of the outflow lobes as material accelerates. Consideringthe mass of the central object (Sect. 4.1) and the typically ob-served infall rates of the order of 10 − M ⊙ / yr (Wyrowski et al.2012, 2016), the larger values of the age estimate, of the orderof a few times 10 years to 10 years, seem the most plausible.This estimate, at the order of magnitude, supports the picture ofthe protostar being very young. To understand the origin of the CH OH t = and cyanoacety-lene, HC N, in Fig. 8. Transitions from the latter species are de-tected both from the vibrational ground and excited states. Theselines are not a ff ected by blending, and probe various excitationconditions (see Table 2). It is clear that from the investigatedmolecules, the methanol emission corresponds the best to thedistribution of the dust continuum; both SO and HC N show adi ff erent morphology.We show the SO (8 , –7 , ) line at 334.673 GHz whichpeaks on the position of the protostar and shows the most ex-tended emission among the species discussed here, with a north-south elongation spatially coinciding with the outflow axis. Sim-ilarly, the HC N lines peak on the protostar. The = J = The outflow is rather confined with a small opening angle of ∼ ◦ . Simple geometric considerations based on the outflow orientation,opening angle and the fact that there is no significant overlap along theline-of-sight between the blue and red shifted emission, we can excludean inclination range between i < ◦ , and i > ◦ . An inclination anglerange between 15 ◦ and 75 ◦ would result in t dyn = . × − . × yr.Article number, page 8 of 18. Csengeri et al.: The search for high-mass protostars with ALMA revealed up to kilo-parsec scales (SPARKS) Fig. 7. a) Color scale shows the continuum emission from Fig. 1 panel b. The contours show the CO (3–2) integrated emission between −
80 and −
65 km s − for the blue, and between −
30 and +
36 km s − for the red, respectively. White cross marks the position of the dust continuum peak.The inset shows a spectrum of the integrated emission over the area of the lowest contours. b)The contours show the velocity integrated SiO (8–7)emission starting from 4 σ (1 σ = / beam km s − ) , and increase by 2 σ levels. c) Overlay of the CO (3–2) contours on the velocity integratedSiO (8–7) emission shown in panel b. The beam is shown in the lower left corner of each panel. On panel c it corresponds to that of the CO (3–2)map. Fig. 8.
A zoom on the envelope showing the CO (3–2) outflow lobes indicated in blue and red contours, the CH OH t = −
55 to −
35 km s − of SO , HC N ( J = N = J = J = axis, and is more compact compared to the SO line. The higherexcitation state = J = pv -diagrams). Due to the rela-tively simple source geometry, we show the averages along ∆ α and ∆ δ axes. For the CO emission we use the cube covering theprimary beam, while for the other species we only use the regionshown in Fig. 8. The high-velocity outflowing gas is clearly vis-ible in the CO (3–2) line, and the kinematic pattern of both theSO , and HC N = N = OH lines; it is, however not as broad as the SO , and HC N = v lsr ∼ −
33 km s − close tothe HC N =
4. Discussion
We identify a single high-mass protostellar envelope withinthe ∼ M ⊙ mid-infrared quiet clump, G328.2551-0.5321. InSect. 4.1 we argue that the protostar is still in its main accre- tion phase, resembling the Class 0 phase of low-mass protostars(c.f. Duarte-Cabral et al. 2013). We investigate the origin of theCH OH emission in Sect. 4.2 and propose that in particular thetorsionally excited state line traces shocks due to the infall fromthe envelope to an accretion disk. In Sect. 4.3 we compare itskinematics with a vibrationally excited state transition of HC N,that we propose as a potential new tracer for the accretion disk.
A luminosity of L bol = . × L ⊙ (see App. A) associated witha massive envelope, and a strong outflow point to a still stronglyaccreting, young massive protostar. Due to the high extinctiontowards the protostar, we use evolutionary diagrams (Fig. 10)based on protostellar evolution models (Hosokawa & Omukai2009) to estimate the mass of the central object.According to models of protostellar evolution the ob-served luminosity is too high to originate only from accretion(Hosokawa et al. 2010), instead it is rather dominated by theKelvin-Helmholtz gravitational contraction of the protostellarcore. Using the typical accretion rate observed towards high-mass protostars and YSOs, i.e. of the order of 10 − M ⊙ yr − (Wyrowski et al. 2016), the Hosokawa tracks indicate that theprotostar has to be bloated, and close to the maximum of Article number, page 9 of 18 & Aproofs: manuscript no. G328p25_14032018_1strevision
Fig. 9.
Position-velocity ( pv ) diagrams along the ∆ δ axis and averaged over the shown extent of the cube (top row) and along the ∆ α axis (bottomrow) of the CO (3–2), SO , HC N ( J = N = J = lsr of the source, and the position of the dust continuum peak. The dashed lines indicate velocities of v lsr ± − roughly corresponding tothe velocities of the CH OH peaks. The red rectangle on the lower left panel corresponds to the region shown in the other panels. The contoursstart at 20% of the peak and increase by 10%, except for the panel of the CO and SO lines, where the lowest contours start at 5% of the peak. its radius during its protostellar evolution, suggesting that itis at the onset of the Kelvin-Helmholtz contraction phase(Hosokawa et al. 2010). During this regime, the luminosity doesnot depend strongly on the actual accretion rate and mainly de-pends on the mass of the protostar.Using these models for the typically expected accretion rateof ˙ M acc = − M ⊙ yr − , we find a central mass around 11 M ⊙ for L bol = . × L ⊙ . Evolutionary tracks with accretion ratesof ˙ M acc = − × − M ⊙ yr − give a similar protostellarmass range, between 11.1 and 15.2 M ⊙ , respectively. The proto-stellar radius adjusts, roughly in a proportional way, to the av-erage accretion rate with a range of radius from 8 to 260 R ⊙ for˙ M acc = − × − M ⊙ yr − . We can therefore assume that thecurrent protostellar mass is relatively well determined, and lies inthe range between 11 and 16 M ⊙ . Taking the dynamical age esti-mate from Sect. 3.4.1 and the current propostellar mass, we canput an upper limit on the accretion rate by ˙ M acc = M proto /τ dyn ,which is between 3 . × − and 4 . × − M ⊙ / yr. This estimateis, however considerably a ff ected by the uncertainties in the dy-namical age estimate (see Sect. 3.4.1).Several types of maser emission also support the presenceof an already high-mass protostellar object; the evolutionarymodels indicate, however, that due to the bloating of the pro-tostar no strong ionising emission is expected despite its highmass. This is consistent with the lack of radio continuum de-tection towards this object. The field has been covered at 3.6and 6 cm by a radio survey of southern Red MSX Sources(RMS) (Urquhart et al. 2007; Lumsden et al. 2013) that targeteda nearby MYSO / UCHII region only showing radio emission1.1 ′ o ff set compared to our position. Similarly, radio contin-uum observations at 8.4 GHz only report a 4 σ upper limit of0.6 mJy (Phillips et al. 1998). Based on this upper limit andadopting a spherical model of ionised plasma with typical val- ues of T e = K electron temperature, EM = − pc cm − emission measure, we find that only a very compact H II regionwith a radius of 100 au could have remained undetected in theseobservations. This is an independent confirmation for the youngnature of the protostar.Based on the estimated current core mass of 120 M ⊙ , we mayexpect a final stellar mass of ∼ M ⊙ (Fig. 10). The protostar ofG328.2551-0.5321 is therefore a particularly interesting object,and we suggest that it is one of the rare examples of a bloated,high-mass protostar, precursor of a potentially O4-O5 stellartype star. There are only very few candidates of such bloatedprotostars in the literature (e. g. Palau et al. 2013, and referencestherein), and most of them correspond to objects detected in theoptical, and the source presented here is much more embedded.Throughout this work we refer to the continuum peak asa single high-mass protostellar envelope. However, we can notexclude that the source would be fragmented at smaller, i.e. <
400 au scales which would lead to the formation of a close bi-nary from a single collapsing envelope. While O-type stars havea high multiplicity (e. g. Sana 2017), we do not find any clearevidence at the observed scales that would hint to multiplicityon smaller scales. For example, the small outflow opening anglecould suggest that either the system is still young, or there is asingle source driving the outflowing gas. Alternatively, gravita-tional fragmentation of the massive inner envelope could lead tothe formation of companions at a subsequent evolutionary stage.
All CH OH lines follow the extension of the envelope, andtheir pv -diagrams are consistent with rotational motions (Fig. 5).Fig. 11 shows the contours of the CH OH t = ff erentvelocity channels, and reveals a clear velocity gradient over the Article number, page 10 of 18. Csengeri et al.: The search for high-mass protostars with ALMA revealed up to kilo-parsec scales (SPARKS)
Fig. 10.
Evolutionary diagram showing M env versus L bol . The color scale shows the predicted distribution of protostars as a fraction relative to1 accounting for the typical stellar initial mass function (Kroupa & Weidner 2003) and for a star formation and accretion history (constant starformation and decreasing accretion rates) as described in Duarte-Cabral et al. (2013). (Figure adapted from Motte et al. 2017). The protostar ofG328.2551-0.5321 is shown in magenta filled circle. bulk of the envelope. When looking at the most extreme velocitychannels, we find that the emission follows well the azimuthalelongations of the envelope. This gradient is spatially resolvedover the blue lobe, that is moving towards our direction and con-nects the envelope to the compact dust component. The recedingarm located on the near side of the envelope is more compact. Al-together this is consistent with a picture of spiral streams devel-oping in the collapsing envelope as the material undergoes infallin a flattened geometry. The development of such spirals is fre-quently seen in numerical simulations of accretion to a central,dominant protostar. They are typically associated with a flattenedstructure (e. g. Krumholz et al. 2007; Hennebelle & Ciardi 2009;Kuiper et al. 2011; Hennebelle et al. 2016b). A similar patternhas been observed towards the low-mass sources BHB07-11,in the B59 core Alves et al. 2017, and Elias 2-27 (Pérez et al.2016); and also on somewhat larger scales of infalling envelopestowards more evolved high-mass star forming regions with anorder of magnitude higher luminosity (Liu et al. 2015).The peak of the torsionally excited state line, together withthe C isotopologue lines, pinpoint the location of the highestmethanol column densities. We explain these observations bytwo physical components for the CH OH emission, the more ex- tended emission associated with the bulk of the inner envelope,and the peak of the t = lsr . In the following we investigatethe physical origin of the t = OH column density upto 2 × cm − towards these positions, which is at least threeorders of magnitude higher than typically observed in the quies-cent gas (e.g. Bachiller et al. 1995; Liechti & Walmsley 1997).Such high CH OH column densities are, however, observed onsmall scales towards high-mass star forming sites, as reportedby e.g. Palau et al. (2017) in the disk component of a high-massprotostar, IRAS20126 + cm − (Bonfand et al. 2017) .In particular, the torsionally excited, t =
1, line with theupper energy level of 315 K, typically requires an infrared ra-diation field at 20–50 µ m in order to populate its upper state.Despite the high average volume densities, assumed to be above n > cm − , radiative excitation could be necessary to populaterotational levels in the t = Article number, page 11 of 18 & Aproofs: manuscript no. G328p25_14032018_1strevision
Fig. 11. a) The color scale is the same as in Fig. 2cshowing the thermaldust emission with the larger scale core emission removed in order toenhance the structure of the inner envelope. The color lines show the50% contour of the peak emission of t = OH line at di ff erentvelocities starting from − . − (blue) to −
38 km s − (red). Thecorresponding velocity of every second contour is shown in the figurelegend. b) Same as the left panel only showing the 90% contours ofmost extreme velocity components in red and blue. The velocities cor-responding to each contour are shown in the figure legend. fore, it is very surprising that this line does not peak on the dustcontinuum at the position of the protostar, instead it peaks onthe inner envelope (Fig 4). This particular pattern could be ex-plained by a decrease of CH OH abundance towards the proto-star, and in fact our results in Sect. 3.3.1 suggest somewhat largerCH OH column densities o ff set from the dust peak. Consider-ing that the H column density increases towards the continuumpeak, and since in Sect. 3.2.2 we rule out optical depth e ff ectsof the discussed transitions, this suggests a decreasing CH OHabundance towards the position of the protostar. The observedhigh CH OH column densities and their emission peak couldalso pinpoint local heating from shocks, which would naturallylead to an increase both in the temperature and in the molecu-lar abundance, especially for CH OH. This is because CH OHhas been found to show an increase in abundance by orders ofmagnitude in various shock conditions as the molecule gets re-leased from the grain surfaces via sputtering (Flower et al. 2010;Flower & Pineau des Forêts 2012).Shocks are produced in a discontinuity in the motion of thegas, and in a collapsing envelope they can emerge in variousconditions. One possibility is an origin associated with the out-flowing gas hitting the ambient medium of the envelope. Suchan increase in the CH OH abundance has been observed inboth low- (Bachiller et al. 1995) and high-mass star forming re-gions (Liechti & Walmsley 1997; Palau et al. 2017). We there-fore compare in more detail the kinematics of the CH OH emis-sion in Fig. 12 with that of outflow tracers (see Sect. 3.4.2) toexclude its association with outflow shocks. While the CO (3–2)emission is not useful for velocities below v lsr ± − due tostrong self absorption, the HC N = ± − peaksof the CH OH t = ff set from the HC N = OH emission to followthe high-velocity emission from the outflowing gas.Interestingly, a comparison with the HC N = OH. The velocity gradient across the lineis visible, its peak is, however, o ff set from the CH OH peaks.This supports a picture where the CH OH and HC N trace dif-ferent components, the former one tracing more the inner enve-
Fig. 12.
Comparison of the position velocity map along the ∆ α axisand averaged over a region of ∼ ′′ . The color scale shows the CO(3–2) emission which is heavily a ff ected by missing spacings and selfabsorption at the source v lsr . The dark gray contours show the CH OH t =
1, white contours the HC N (37–37) (left), and the HC N = lope, while the HC N = OH into the gas phase. The launching mechanismof the outflowing gas is highly unexplored territory in high-massstar formation, the location of the outflow launching site is there-fore not constrained. Towards low-mass protostars, methanol hasbeen observed to be associated with the launching of the jet / diskwind in the close vicinity, within <
135 au distance from theprotostar (Leurini et al. 2016), which is a considerably smallerscale than probed by our observations. There is some indicationthat towards low-mass protostars outflow can be launched at theouter regions of the Keplerian accretion disk, a recent study ofa low-mass Class I type protostar presents an example where theoutflow is launched at a distance beyond the disk edge, fromthe inner envelope (Alves et al. 2017). With our current angularresolution we do not probe such small scales, which would beunresolved, and peaking on the protostar. However, we cannotexclude that the observed CH OH spots may have contributionfrom the surface of the flattened envelope.Given the extent of the torsionally excited CH OH emission,the best explanation for our observations is that a significantamount of CH OH is liberated into the gas phase by shocks asso-ciated with the inner envelope itself. Around the low-mass Class0 protostar, L1157, CH OH has been detected tracing shockswithin the infalling gas (Goldsmith et al. 1999; Velusamy et al.2002). In both of the latter two scenarios explaining the CH OHemission, the observed maximum velocity o ff set corresponds tothe line-of-sight rotational velocity of the gas at the innermostregions of the envelope. The brightest spots of the torsionally excited methanol emis-sion can be interpreted as tracing shocks emerging in the in-nermost regions of the envelope, hence associated with thecentrifugal barrier. This happens when the inflowing materialfrom the envelope hits material with a smaller radial veloc-ity component that corresponds to an accretion disk surround-ing the central protostar. Such a transition between the enve-lope and the disk material at the centrifugal barrier has been di-rectly observed in nearby low-mass protostars (Sakai et al. 2014;Oya et al. 2016; Alves et al. 2017). Towards these objects boththe gas kinematics and the gas chemistry change at the inner en-velope; some studies interpret the extent of this region as a sharp
Article number, page 12 of 18. Csengeri et al.: The search for high-mass protostars with ALMA revealed up to kilo-parsec scales (SPARKS) (c.f. Alves et al. 2017), while others as a more gradual transitionregion (Oya et al. 2016). Here we observe it as a more extendedemission, which also appears to be asymmetric.Direct evidence for the presence of the disk is hinderedby the angular and spectral resolution of our dataset; however,putting all pieces of evidence together, we find several indica-tions supporting the scenario of accretion shocks at the centrifu-gal barrier. In Sect. 3.1 we already noted a marginally resolved,compact dust component that could correspond to the disk with aresolved major axis of 250 au seen in projection. The elongationof this residual is resolved, and is perpendicular to the outflowaxis within ∼ ◦ . Both its orientation and extent are consistentwith the location of the two CH OH peaks at a projected distanceof 300 to 800 au considering the uncertainty resulting from ourangular resolution of ∼
400 au.In addition, we find that the highest excitation molecularemission observed by us in the vibrationally excited HC N = µ m in order to populateits vibrationally excited states. Therefore, it more likely tracesa region considerably closer to the protostar than the torsionallyexcited methanol line. Although this molecule is also present inthe outflow, as suggested by the vibrationally ground state tran-sition, we propose that the high excitation vibrationally excitedHC N = pv -diagrams with a sim-ple toy model adapted from Ohashi et al. (1997) to describe anaxisymmetric rotating thin disk for the HC N = OH t = OH lines trace therotating envelope with v rot ( r ) ∼ r − , and the HC N = × − resolution, a power-law density distribution with n ∝ r − . , constant molecular abundance with respect to H , anda central protostellar mass of 15 M ⊙ (see Sect 4.1). For the ge-ometry of the CH OH emission we use a ring between 300 and900 au corresponding to the extent of the marginally resolvedcompact component, and v rot = − at the inner radius of300 au. For the HC N = OHcorresponding to the highest column densities are reasonablywell reproduced with these models, the observed asymmetriesin the distribution of the emission and projection e ff ects are,however, not included in our models. We break the degeneracybetween the location of the ring of CH OH emitting gas andthe mass of the central object by measuring the position of themethanol shocks. Since we do not include a correction for theinclination angle between the source and our line of sight, thedetermined parameters are uncertain within a factor of a few.These models demonstrate that the CH OH emission can bewell explained by a ring of emitting gas from the infalling en-velope that is more extended than the observed HC N = OH emission at a velocity that is consistent with the Kep-lerian velocity of the estimated source mass of 40 M ⊙ . Due tothe observed asymmetry of the HC N = Fig. 13.
Left column: pv -diagram of models with a rotating ring with v r ∼ r (top), and with a disk in Keplerian rotation (bottom). Right col-umn: The color scale shows the 0th moment maps for the 334.42 GHzCH OH t = N = Fig. 14.
Keplerian velocity as a function of radius, r , from the centralobject for 10, 15, and 20 M ⊙ . The projection corrected velocity deter-mined from the 334.42 GHz CH OH t = OH t = velocity resolution, our models only show that the extent of theobserved velocity range of the HC N = M ⊙ ,and a range of centrifugal barriers around 300-800 au, which cor-responds to the parameter range that could still be consistentwith the observations. To obtain the rotational velocity at thisradius, we correct the observed 4.5 km s − for an inclination an-gle, i , of 56 ◦ based on the axis ratio of the measured envelopesize in Sect. 3.1. We see that the observed velocity o ff set of theCH OH peaks fits well the Keplerian velocity for the plausiblemass range and the range of centrifugal barrier within a resolu-tion element.
Article number, page 13 of 18 & Aproofs: manuscript no. G328p25_14032018_1strevision
Although massive rotating toroids towards precursors of OB-type stars have been frequently detected (Beltrán et al. 2004;Cesaroni et al. 2014; Beltrán et al. 2014; Sánchez-Monge et al.2013b; Cesaroni et al. 2017), clear signatures of accretion disksaround high-mass protostars are still challenging to identify, inparticular towards the precursors of the most massive, O-typestars (Beltrán & de Wit 2016). Our results suggest the presenceof an accretion disk around a still deeply embedded young high-mass protostar that is likely to be a precursor of an O4-O5 typestar. Our findings suggest that a disk may have formed alreadyat this early stage, providing observational support to numeri-cal simulations which predict that despite the strong radiationpressure exerted by high-mass protostars, accretion through flat-tened structures, and disks enable the formation of the highestmass stars (Krumholz et al. 2009; Kuiper et al. 2010, 2011).Together with other examples of envelope-outflow-disk sys-tems (e.g. Johnston et al. 2015; Beltrán & de Wit 2016), thissuggests a physical picture of high-mass star formation on thecore scale that is qualitatively very similar to that of low-massobjects. As observed towards L1527 (Sakai et al. 2014), TMC-1(Aso et al. 2015), and VLA1623A (Murillo et al. 2013), we alsosee evidence for shocks induced by the infall from the envelopeto the disk imposing a change in the chemical composition of theinfalling gas at the centrifugal barrier (see also Oya et al. 2017).The accretion disk around the protostar of L1527 has been con-firmed since then with direct imaging (Sakai et al. 2017).
Fig. 15.
Local specific angular momentum ( j / m = R × v rot ) as afunction of radius for a sample of low- and high-mass protostars andYSOs. The high-mass sample (Guzmán et al. 2014; Beltrán & de Wit2016) corresponds to high-mass YSOs with disk candidates, the low-mass objects (Ohashi et al. 1997; Sakai et al. 2014; Oya et al. 2016)correspond to envelopes and disks identified around low-mass Class0 protostars. Where no explicit information was available, we adopta 50% uncertainty for the rotation radius, and a linearly propagated50% uncertainty on the estimated specific angular momentum. Forour target, we estimate a 52% uncertainty on these values based onthe uncertainty in the disk radius estimate. On this plot we madeno attempt to correct for the source’s inclination angle. Dashed lineshows the j = − line (Belloche 2013), and an angular ve-locity, Ω= − pc − . For the high-mass sample we show thej = − pc − line, and Ω=
50 km s − pc − . The grey shaded areashows our range of disk size estimate. Our findings suggest, however, considerably di ff erent phys-ical conditions and chemical environment for high-mass pro-tostars. The observed prominent bright peak of CH OH, that we explain by shocks at the centrifugal barrier, blends with themore di ff use emission from the envelope. At our spatial reso-lution, these shocks do not outline a sharp boundary, and leaveus with a relatively large range of disk radii that is still con-sistent with the observations. Based on the size of the compactdust continuum source, we measure a minimum projected radiusof ∼
250 au for the disk major axis, while a maximum outer ra-dius is constrained by the peak of the CH OH emission locatedsomewhere between 300 and 800 au. We correct these values foran orientation angle of φ ∼ ◦ based on the fitted position an-gle of the dust residual emission, giving a disk radius between255 and 817 au, that is a factor of few larger than recent ALMAobservations suggest for disks around some low-mass Class 0protostars (e.g. Oya et al. 2016; Sakai et al. 2017). For the rota-tional velocity at this outer radius we take the velocity o ff set ofthe CH OH peaks corrected for the inclination angle (Sect. 4.2).Taking an average disk size between the minimum and maxi-mum expected values implies that the local specific angular mo-mentum, j / m = R × v rot = . × cm s − , where R is the diskradius, and v rot , is the rotational velocity of the disk at the givenradius.We compare our measurement in Fig. 15 to values fromthe literature following Ohashi et al. (1997), and Belloche et al.(2002), and complement it with high-mass disks and toroidsfrom Beltrán & de Wit (2016). We recognise that the local spe-cific angular momentum is considerably higher towards the high-mass case compared to the low-mass case, although the so farobserved disk candidates are still at larger physical scales, andtypically towards objects that are likely in a more evolved stagethan the protostar of G328.2551-0.5321. This suggests that thekinetic energy may be larger at the onset of the collapse in thecase of high-mass star formation. The larger kinetic energy couldbe explained if the collapse sets in at the clump, thus at > T dust =
150 K, we obtainsub-solar mass estimations around M disk < . M ⊙ . Such an el-evated temperature is expected for the disk in the close vicinityof the protostellar embryo, it is, however, still consistent with theSED because of the large optical depth of the cooler dust. Sinceour disk mass estimate is sub-solar, it is likely that the disk massis below 10% of the mass of the central object and thus gravi-tationally stable. The stability of the disk itself is an importantquestion, unstable massive disks could either lead to episodicaccretion bursts, or undergo fragmentation (Vorobyov & Basu2010). Both phenomena are observed towards low-mass proto-stars; in particular multiplicity within low-mass cores has beenrecently explained by disk fragmentation (Tobin et al. 2016).On the other hand, the high-mass disk candidate AFGL 4176(Johnston et al. 2015) is more extended, but also more massive.Fragmentation of massive or unstable disks and toroids aroundhigh-mass stars could explain why short period binaries have thehighest frequency among O-type stars (Sana 2017). The fact that we observe at the same time a massive core which isnot fragmented down to our resolution limit of ∼
400 au, and find
Article number, page 14 of 18. Csengeri et al.: The search for high-mass protostars with ALMA revealed up to kilo-parsec scales (SPARKS) strong evidence for a centrifugal barrier at a large radius of 300-800 au, may appear contradictory and needs to be discussed. Theobserved indication for the Keplerian disk together with the largeangular momentum suggest that magnetic braking has not beene ffi cient to evacuate and redistribute the angular momentum.Numerical simulations predict that, in particular, at the earlyphase of the collapse, the angular momentum from the accretiondisk can be e ffi ciently removed due to magnetic braking, andthereby suppress the formation of large disks (e.g. Seifried et al.2011; Myers et al. 2013; Hennebelle et al. 2016a). Our obser-vations therefore point to a relatively weak magnetic field. Onthe other hand, despite its large mass which is two orders ofmagnitude larger than the thermal Jeans mass, the core did notfragment and appears to be collapsing monolithically, which isconsistent with the Turbulent Core model (McKee & Tan 2003).This would require additional support to complement the ther-mal pressure which can either be magnetic or turbulent. Neitherthe line widths, nor the large angular momentum are consistentwith strong enough turbulence or magnetic fields, therefore theproperties of the core embedded in G328.2551-0.5321 appeardi ffi cult to explain.However, if turbulence or strong ordered motions arepresent, the misalignment between the magnetic field lines andthe angular momentum vectors can limit the e ff ect of magneticbraking leading to a less e ffi cient removal of the angular momen-tum (Myers et al. 2013). Alternatively, the present day propertiesof the collapsing core and the physical state of the pre-stellarcore prior to collapse may have been significantly di ff erent atonset of the collapse. The small scale properties of high-massprotostars, and the physical properties of their accretion disk,such as in G328.2551-0.5321 may thus challenge high-mass starformation theories. Clearly, more observational examples of ac-cretion disks around high-mass protostars are needed to put fur-ther constrains on the formation and properties of disks, and thecollapse scenario.
5. Summary and conclusions
We presented a case study of one of the targets from the SPARKSproject, which uses high angular-resolution observations fromALMA to study the sample of the most massive mid-infraredquiet massive clumps selected from the ATLASGAL survey. Ourobservations reveal a single massive protostellar envelope as-sociated with the massive clump, G328.2551-0.5321. Based onprotostellar evolutionary tracks, we estimate the current proto-stellar mass to be between 11 and 16 M ⊙ , surrounded by a mas-sive core of ∼ M ⊙ . The estimated envelope mass is an or-der of magnitude larger than the currently estimated protostellarmass, making this object an excellent example of a high-massprotostar in its main accretion phase, similar to the low-massClass 0 phase.We discovered torsionally excited CH OH spots o ff set fromthe protostar with a velocity o ff set of ± − compared tothe source v lsr . These peaks are best explained by shocks fromthe infalling envelope onto the centrifugal barrier. Based on theobserved unblended methanol transitions, we estimate the phys-ical conditions on these spots, and find T kin = −
170 K, and N (CH OH) = . − × cm − , suggesting large CH OH col-umn densities.Our analysis of the dust emission reveals azimuthal elonga-tions associated with the dust continuum peak, and a compactcomponent with a marginally resolved beam deconvolved R radius of ∼
250 au measured along is major axis. This componentis consistent with an accretion disk within the centrifugal barrier outlined by the CH OH shock spots at a distance between ∼ ff set from the protostar. Furthermore, we proposethe vibrationally excited HC N = e J = Acknowledgements.
We thank the referee for the careful reading of themanuscript. This paper makes use of the ALMA data: ADS / JAO.ALMA2013.1.00960.S. ALMA is a partnership of ESO (representing its memberstates), NSF (USA), and NINS (Japan), together with NRC (Canada), NSC andASIAA (Taiwan), and KASI (Republic of Korea), in cooperation with the Re-public of Chile. The Joint ALMA Observatory is operated by ESO, AUI / NRAO,and NAOJ. T.Cs. acknowledges support from the
Deutsche Forschungsgemein-schaft, DFG via the SPP (priority programme) 1573 ’Physics of the ISM’. HBacknowledges support from the European Research Council under the Horizon2020 Framework Program via the ERC Consolidator Grant CSF-648505. LB ac-knowledges support by CONICYT Project PFB06. A.P. acknowledges financialsupport from UNAM and CONACyT, México.
References
Alves, F. O., Girart, J. M., Caselli, P., et al. 2017, A&A, 603, L3André, P., Ward-Thompson, D., & Barsony, M. 2000, Protostars and Planets IV,59Aso, Y., Ohashi, N., Saigo, K., et al. 2015, ApJ, 812, 27Bachiller, R., Liechti, S., Walmsley, C. M., & Colomer, F. 1995, A&A, 295, L51Belloche, A. 2013, in EAS Publications Series, Vol. 62, EAS Publications Series,ed. P. Hennebelle & C. Charbonnel, 25–66Belloche, A., André, P., Despois, D., & Blinder, S. 2002, A&A, 393, 927Beltrán, M. T., Cesaroni, R., Neri, R., & Codella, C. 2011, A&A, 525, A151Beltrán, M. T., Cesaroni, R., Neri, R., et al. 2004, ApJ, 601, L187Beltrán, M. T., Cesaroni, R., Neri, R., et al. 2005, A&A, 435, 901Beltrán, M. T. & de Wit, W. J. 2016, A&A Rev., 24, 6Beltrán, M. T., Sánchez-Monge, Á., Cesaroni, R., et al. 2014, A&A, 571, A52Benjamin, R. A., Churchwell, E., Babler, B. L., et al. 2003, PASP, 115, 953Beuther, H., Schilke, P., Menten, K. M., et al. 2002, ApJ, 566, 945Bonfand, M., Belloche, A., Menten, K. M., Garrod, R. T., & Müller, H. S. P.2017, A&A, 604, A60Bontemps, S., Andre, P., Terebey, S., & Cabrit, S. 1996, A&A, 311, 858Bontemps, S., Motte, F., Csengeri, T., & Schneider, N. 2010, A&A, 524, A18Carey, S. J., Noriega-Crespo, A., Mizuno, D. R., et al. 2009, PASP, 121, 76Cesaroni, R., Galli, D., Neri, R., & Walmsley, C. M. 2014, A&A, 566, A73Cesaroni, R., Sánchez-Monge, Á., Beltrán, M. T., et al. 2017, A&A, 602, A59Csengeri, T., Bontemps, S., Wyrowski, F., et al. 2017a, A&A, 601, A60Csengeri, T., Bontemps, S., Wyrowski, F., et al. 2017b, A&A, 600, L10Csengeri, T., Leurini, S., Wyrowski, F., et al. 2016, A&A, 586, A149Csengeri, T., Urquhart, J. S., Schuller, F., et al. 2014, A&A, 565, A75Cutri, R. M., Wright, E. L., Conrow, T., et al. 2012, Explanatory Supplement tothe WISE All-Sky Data Release Products, Tech. rep.Duarte-Cabral, A., Bontemps, S., Motte, F., et al. 2013, A&A, 558, A125Dunham, M. M., Crapsi, A., Evans, II, N. J., et al. 2008, ApJS, 179, 249Flower, D. R. & Pineau des Forêts, G. 2012, MNRAS, 421, 2786Flower, D. R., Pineau Des Forêts, G., & Rabli, D. 2010, MNRAS, 409, 29Garay, G., Mardones, D., Bronfman, L., et al. 2010, ApJ, 710, 567Goldsmith, P. F., Langer, W. D., & Velusamy, T. 1999, ApJ, 519, L173Gómez, L., Wyrowski, F., Pillai, T., Leurini, S., & Menten, K. M. 2011, A&A,529, A161Guzmán, A. E., Garay, G., & Brooks, K. J. 2010, ApJ, 725, 734Guzmán, A. E., Garay, G., Rodríguez, L. F., et al. 2014, ApJ, 796, 117Hennebelle, P. & Ciardi, A. 2009, A&A, 506, L29Hennebelle, P., Commerçon, B., Chabrier, G., & Marchand, P. 2016a, ApJ, 830,L8Hennebelle, P., Lesur, G., & Fromang, S. 2016b, A&A, 590, A22Hosokawa, T. & Omukai, K. 2009, ApJ, 691, 823Hosokawa, T., Yorke, H. W., & Omukai, K. 2010, ApJ, 721, 478
Article number, page 15 of 18 & Aproofs: manuscript no. G328p25_14032018_1strevision
Jiménez-Serra, I., Zhang, Q., Viti, S., Martín-Pintado, J., & de Wit, W.-J. 2012,ApJ, 753, 34Johnston, K. G., Robitaille, T. P., Beuther, H., et al. 2015, ApJ, 813, L19Kroupa, P. & Weidner, C. 2003, ApJ, 598, 1076Krumholz, M. R., Klein, R. I., & McKee, C. F. 2007, ApJ, 656, 959Krumholz, M. R., Klein, R. I., McKee, C. F., O ff ner, S. S. R., & Cunningham,A. J. 2009, Science, 323, 754Kuiper, R., Klahr, H., Beuther, H., & Henning, T. 2010, ApJ, 722, 1556Kuiper, R., Klahr, H., Beuther, H., & Henning, T. 2011, ApJ, 732, 20Langer, W. D. & Penzias, A. A. 1990, ApJ, 357, 477Leurini, S., Codella, C., Cabrit, S., et al. 2016, A&A, 595, L4Liechti, S. & Walmsley, C. M. 1997, A&A, 321, 625Liu, H. B., Galván-Madrid, R., Jiménez-Serra, I., et al. 2015, ApJ, 804, 37Lumsden, S. L., Hoare, M. G., Urquhart, J. S., et al. 2013, ApJS, 208, 11Maret, S., Hily-Blant, P., Pety, J., Bardeau, S., & Reynier, E. 2011, A&A, 526,A47McKee, C. F. & Tan, J. C. 2003, ApJ, 585, 850Milam, S. N., Savage, C., Brewster, M. A., Ziurys, L. M., & Wycko ff , S. 2005,ApJ, 634, 1126Molinari, S., Brand, J., Cesaroni, R., & Palla, F. 2000, A&A, 355, 617Molinari, S., Schisano, E., Elia, D., et al. 2016, A&A, 591, A149Molinari, S., Swinyard, B., Bally, J., et al. 2010, PASP, 122, 314Moscadelli, L., Sánchez-Monge, Á., Goddi, C., et al. 2016, A&A, 585, A71Motte, F., Bontemps, S., & Louvet, F. 2017, ArXiv e-prints[ arXiv:1706.00118 ]Motte, F., Bontemps, S., Schilke, P., et al. 2007, A&A, 476, 1243Mottram, J. C., Hoare, M. G., Urquhart, J. S., et al. 2011, A&A, 525, A149Murillo, N. M., Lai, S.-P., Bruderer, S., Harsono, D., & van Dishoeck, E. F. 2013,A&A, 560, A103Myers, A. T., McKee, C. F., Cunningham, A. J., Klein, R. I., & Krumholz, M. R.2013, ApJ, 766, 97Ohashi, N., Hayashi, M., Ho, P. T. P., & Momose, M. 1997, ApJ, 475, 211Ossenkopf, V. & Henning, T. 1994, A&A, 291, 943Oya, Y., Sakai, N., López-Sepulcre, A., et al. 2016, ApJ, 824, 88Oya, Y., Sakai, N., Watanabe, Y., et al. 2017, ApJ, 837, 174Palau, A., Fuente, A., Girart, J. M., et al. 2013, ApJ, 762, 120Palau, A., Walsh, C., Sánchez-Monge, Á., et al. 2017, MNRAS, 467, 2723Peretto, N., Fuller, G. A., Duarte-Cabral, A., et al. 2013, A&A, 555, A112Pérez, L. M., Carpenter, J. M., Andrews, S. M., et al. 2016, Science, 353, 1519Phillips, C. J., Norris, R. P., Ellingsen, S. P., & McCulloch, P. M. 1998, MNRAS,300, 1131Purser, S. J. D., Lumsden, S. L., Hoare, M. G., et al. 2016, Monthly Notices ofthe Royal Astronomical Society, 460, 1039Rosero, V., Hofner, P., Claussen, M., et al. 2016, ApJS, 227, 25Sakai, N., Oya, Y., Higuchi, A. E., et al. 2017, MNRAS, 467, L76Sakai, N., Sakai, T., Hirota, T., et al. 2014, Nature, 507, 78Sana, H. 2017, ArXiv e-prints [ arXiv:1703.01608 ]Sánchez-Monge, Á., Cesaroni, R., Beltrán, M. T., et al. 2013a, A&A, 552, L10Sánchez-Monge, Á., Cesaroni, R., Beltrán, M. T., et al. 2013b, A&A, 552, L10Sanna, A., Surcis, G., Moscadelli, L., et al. 2015, A&A, 583, L3Schöier, F. L., van der Tak, F. F. S., van Dishoeck, E. F., & Black, J. H. 2005,A&A, 432, 369Schuller, F., Menten, K. M., Contreras, Y., et al. 2009, A&A, 504, 415Seifried, D., Banerjee, R., Klessen, R. S., Du ffi n, D., & Pudritz, R. E. 2011,MNRAS, 417, 1054Sridharan, T. K., Beuther, H., Schilke, P., Menten, K. M., & Wyrowski, F. 2002,ApJ, 566, 931Tanaka, K. E. I., Tan, J. C., & Zhang, Y. 2017, ApJ, 835, 32Tobin, J. J., Kratter, K. M., Persson, M. V., et al. 2016, Nature, 538, 483Urquhart, J. S., Busfield, A. L., Hoare, M. G., et al. 2007, A&A, 461, 11Velusamy, T., Langer, W. D., & Goldsmith, P. F. 2002, ApJ, 565, L43Vorobyov, E. I. & Basu, S. 2010, ApJ, 719, 1896Wyrowski, F., Güsten, R., Menten, K. M., et al. 2016, A&A, 585, A149Wyrowski, F., Güsten, R., Menten, K. M., Wiesemeyer, H., & Klein, B. 2012,A&A, 542, L15Zhang, Q., Hunter, T. R., Brand, J., et al. 2005, ApJ, 625, 864 Max-Planck-Institut für Radioastronomie, Auf dem Hügel 69,53121 Bonn, Germany e-mail: [email protected] OASU / LAB-UMR5804, CNRS, Université Bordeaux, allée Geof-froy Saint-Hilaire, 33615 Pessac, France INAF - Osservatorio Astronomico di Cagliari, Via della Scienza 5,I-09047 Selargius (CA), Italy Max Planck Institute for Astronomy, Königstuhl 17, 69117 Heidel-berg, Germany Departamento de Astronomía, Universidad de Chile, Casilla 36-D,Santiago, Chile Univ. Lyon, ENS de Lyon, Univ Lyon1, CNRS, Centre de RechercheAstrophysique de Lyon UMR5574, F-69007, Lyon, France IRAM, 300 rue de la piscine, 38406, Saint-Martin-d’Héres, France Astrophysics Research Institute, Liverpool John Moores Instituto de Radioastronomía y Astrofísica, Universidad NacionalAutónoma de México, P.O. Box 3-72, 58090, Morelia, Michoacán,México Dept. of Space, Earth & Environment, Chalmers University of Tech-nology, Gothenburg, Sweden Dept. of Astronomy, University of Virginia, Charlottesville, VA,USA School of Physical Sciences, University of Kent, Ingram Building,Canterbury, Kent CT2 7NH, UKArticle number, page 16 of 18. Csengeri et al.: The search for high-mass protostars with ALMA revealed up to kilo-parsec scales (SPARKS)
Appendix A: Dust spectral energy distribution andmodelling
We construct the spectral energy distribution (SED) of the pro-tostar embedded in the clump G328.2551-0.5321, from the mid-infrared wavelengths up to the radio regime (Fig. A.1) in orderto estimate its bolometric luminosity ( L bol ), and constrain a rep-resentative dust temperature ( T d ) for the bulk of the mass usinga model of greybody emission.In Fig. A.1 we show the flux densities from the GLIMPSEcatalog at the shortest indicated wavelengths (Benjamin et al.2003), as well as the WISE band 4 photometry at 22 µ m(Cutri et al. 2012). To illustrate the complexity of the region,we show the emission from the far-infrared and millime-tre wavelength range in Fig. A.2 using Herschel / Hi-GAL data(Molinari et al. 2010), and the ATLASGAL-Planck combineddata at 870 µ m (Csengeri et al. 2016). This shows that the emis-sion is largely dominated by extended structures at all thesewavelengths.As a comparison we show the corresponding sourcesidentified in the Herschel / Hi-GAL point source catalog(Molinari et al. 2016), and the ATLASGAL Gaussclumps sourcecatalog (Csengeri et al. 2014). The 870 µ m flux density measure-ment by ALMA reveals the MDC at a size-scale of 0.06 pc (seealso Csengeri et al. 2017b) and therefore puts constraints on theextent of the embedded source. To derive therefore the prop-erties of the gas representative of the embedded protostar, ouraim is to extract and scale the flux densities corresponding to asource at ∼ ′′ (the geometric mean of the PACS 70 µ m beam,Molinari et al. 2016) with a density profile of n ∼ r − , and ne-glecting any temperature gradient. To do this, we perform aper-ture photometry on the PACS 70 µ m, and 160 µ m maps by firstmeasuring the peak intensity at the position of our target withina single beam (taking 5.8 ′′ × ′′ , and 11.4 ′′ × ′′ , respec-tively), and then measure the background emission in 3 di ff er-ent annuli at increasing distance from 15 ′′ to 35 ′′ with respectto the source. For the SPIRE 250, 350, and 500 µ m data, we usethe values from the Herschel / Hi-GAL point source catalog, andscale them adopting the geometric mean of the measured majorand minor axes (Molinari et al. 2016).To obtain the bolometric luminosity of the protostar, we addup all emission from the near / mid-infrared to 870 µ m, and obtain1 . × L bol . As a comparison, we also calculated the proto-star’s internal luminosity using the empirical relation between L bol and the flux density measured at 70 µ m (Dunham et al.2008), and obtain 1 . × L bol . The significant confusion dueto extended emission from the mid-infrared to the submillimeterwavelengths adds, however, some uncertainty to our estimate,the measured values likely correspond to an upper limit to theluminosity.To obtain an estimate of the dust temperature, T d , we per-form a greybody fit to the far-infrared points of the SED between70 µ m and 870 µ m. Since at 70 µ m the emission is mostly dom-inated by the heated dust in the vicinity of the protostar, we usea two component greybody to model the SED, which reveals thetemperature corresponding to the cold gas component dominat-ing the bulk of the emission, and puts strong constrain on thewarm gas temperature, as well as the fraction of the heated gasmass. We use κ
345 GHz = . − and an emissivity indexof β =
2, where κ ν = ν
345 GHz − β . We obtain a cold gas componentat 22 K, and a warm gas component at 48 K that contains <
5% ofthe total gas mass. The result of the SED fit is shown in Fig. A.1.
Fig. A.1.
SED of the embedded protostar within the ATLASGALclump, G328.2551-0.5321. The origin of the shown flux densities arelabeled in the figure legend, and are described in the text. Solid lineshows the result of a two (warm and cold) component grebody fit with48 K and 22 K. The individual components are shown in a dashed grayline. Article number, page 17 of 18 & Aproofs: manuscript no. G328p25_14032018_1strevision
Fig. A.2.