Water deuterium fractionation in the high-mass star-forming region G34.26+0.15 based on Herschel/HIFI data
Audrey Coutens, Charlotte Vastel, Ugo Hincelin, Eric Herbst, Dariusz C. Lis, Luis Chavarría, Maryvonne Gérin, Floris F. S. van der Tak, Carina M. Persson, Paul F. Goldsmith, Emmanuel Caux
MMon. Not. R. Astron. Soc. , 000–000 (0000) Printed 22 August 2018 (MN L A TEX style file v2.2)
Water deuterium fractionation in the high-mass star-forming regionG34.26 + Herschel / HIFI data
A. Coutens , (cid:63) , C. Vastel , , U. Hincelin , E. Herbst , D. C. Lis , , L. Chavarr´ıa ,M. G´erin , F. F. S. van der Tak , , C. M. Persson , P. F. Goldsmith , E. Caux , Niels Bohr Institute, University of Copenhagen, Juliane Maries Vej 30, DK-2100 Copenhagen Ø, Denmark Centre for Star and Planet Formation, Natural History Museum of Denmark, University of Copenhagen, Øster Voldgade 5-7,DK-1350 Copenhagen K, Denmark Universit´e de Toulouse, UPS-OMP, IRAP, Toulouse, France CNRS, Institut de Recherche en Astrophysique et Plan´etologie, 9 Av. Colonel Roche, BP 44346, 31028 Toulouse Cedex 4, France Department of Chemistry, University of Virginia, McCormick Road, Charlottesville, VA 22904, USA California Institute of Technology, Cahill Center for Astronomy and Astrophysics 301-17, Pasadena, CA 91125, USA Sorbonne Universit´es, Universit´e Pierre et Marie Curie, Paris 6, CNRS, Observatoire de Paris, UMR 8112, LERMA, Paris, France Universidad de Chile - CONICYT, Camino del Observatorio 1515, Las Condes, Santiago LERMA-LRA, UMR 8112 du CNRS, Observatoire de Paris, Ecole Normale Sup´erieure, UPMC & UCP, 24 rue Lhomond, 75231 Paris Cedex 05, France SRON Netherlands Institute for Space Research, Landleven 12, 9747 AD Groningen, The Netherlands Kapteyn Astronomical Institute, University of Groningen, 9700 AV Groningen, The Netherlands Chalmers University of Technology, Department of Earth and Space Sciences, Onsala Space Observatory, 43992 Onsala, Sweden Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91125, USA
Accepted xxx. Received xxx; in original form xxx
ABSTRACT
Understanding water deuterium fractionation is important for constraining the mechanismsof water formation in interstellar clouds. Observations of HDO and H
O transitions werecarried out towards the high-mass star-forming region G34.26 + Herschel
Space Observatory, as well as with ground-based single-dish telescopes.Ten HDO lines and three H
O lines covering a broad range of upper energy levels (22–204 K) were detected. We used a non-LTE 1D analysis to determine the HDO / H O ratio as afunction of radius in the envelope. Models with di ff erent water abundance distributions wereconsidered in order to reproduce the observed line profiles. The HDO / H O ratio is found tobe lower in the hot core ( ∼ × − –7.5 × − ) than in the colder envelope ( ∼ × − –2.2 × − ). This is the first time that a radial variation of the HDO / H O ratio has been found tooccur in a high-mass source. The chemical evolution of this source was modeled as a functionof its radius and the observations are relatively well reproduced. The comparison between thechemical model and the observations leads to an age of ∼ years after the infrared darkcloud stage. Key words: astrochemistry – ISM: individual object: G34.26 + Water, being necessary for the emergence of life, is one of the mostimportant molecules found in space. As a dominant form of oxygen(the most abundant element in the Universe after hydrogen and he-lium), water controls the chemistry of many other species, whetherin the gas phase or in the solid phase (see for example the reviewby van Dishoeck et al. 2013). Water is a unique diagnostic of thewarmer gas and the energetic processes taking place close to star-forming regions. Water is also a contributor to maintaining the low (cid:63)
E-mail: [email protected] temperature of the gas by spectral line radiative cooling. Low tem-peratures are a requisite for cloud collapse and star formation. Wa-ter is mainly in its solid form (as ice on the surface of dust grains)in the cold regions of the interstellar medium as well as in asteroidsand comets that likely delivered water to the Earth’s oceans (e.g.,Hartogh et al. 2011; Alexander et al. 2012). Therefore, constrain-ing the distribution of water vapor and ice during the entire star andplanet formation phase is mandatory to understand our own origins.Because of its high abundance in our own atmosphere, obser-vations of interstellar water have been primarily carried out fromspace observatories including ISO,
Spitzer , ODIN, SWAS, and re-cently
Herschel . Indeed, water has been detected toward the coldprestellar core L1544 (Caselli et al. 2012), many low-mass proto- c (cid:13) a r X i v : . [ a s t r o - ph . S R ] S e p A. Coutens, C. Vastel, U. Hincelin et al. stars (e.g., Coutens et al. 2012; Kristensen et al. 2010, 2012), high-mass protostars (e.g., van der Tak et al. 2013; Emprechtinger et al.2013), in the disk of a young star TW Hydrae (Hogerheijde et al.2011), as well as in many comets (e.g., 103P / Hartley 2: Hartoghet al. 2011, C / / Honda-Mrkos-Pajduˇs´akov´a: Lis et al. 2013) and in asteroids(24 Themis: Campins et al. 2010, Ceres: K¨uppers et al. 2014). Thewater abundance shows a very large variation from one source cate-gory to another, as well as within each type of sources. The questionthen arises: how is water produced and why are its abundance vari-ations so large? Although production in the gas phase followed bydirect condensation onto dust grains is possible (Bergin et al. 1999),observations favor formation through chemical reactions on the sur-face of cold dust grains. Indeed, in comparison to gas-phase waterabundance, the observed water ice abundance is too high to be en-tirely explained by direct accretion from the gas-phase (Roberts &Herbst 2002). Consequently surface reactions on cold dust grainsto form water molecules have been investigated with modern sur-face science techniques (e.g. Watanabe & Kouchi 2008). Consider-ing the large reservoir of oxygen and hydrogen atoms in molecularclouds, large amounts of water ice might be produced (Dulieu et al.2010) following the successive hydrogenation of oxygen on grainsurfaces: O H −→ OH H −→ H O . (1)Tielens & Hagen (1982) proposed that water ice might also be pro-duced through the successive hydrogenation of molecular oxygen:O −→ HO −→ H O −→ H O + OH , (2)demonstrated by Miyauchi et al. (2008), Ioppolo et al. (2008) andOba et al. (2009), or by hydrogenation of ozone:O −→ O + OH , OH H −−→ H O + H , (3)demonstrated by Mokrane et al. (2009).Deuterated water is likely to be formed through the same pro-cesses. Many rotational transitions have been detected from theground, as well as with Herschel / HIFI, for example in low-massprotostars (Parise et al. 2005; Liu et al. 2011; Coutens et al. 2012,2013; Persson et al. 2013, 2014), high-mass star forming regions(e.g., Jacq et al. 1990; Gensheimer et al. 1996), and comets (e.g.,Bockel´ee-Morvan et al. 1998; Hartogh et al. 2011; Lis et al. 2013).The HDO / H O ratio is an interesting diagnostic tool to help un-derstand the origin of water in the interstellar medium, with adirect comparison with the D / H ratio observed in comets and inthe Earth’s oceans. It is also helpful to constrain the water for-mation conditions. In star-forming regions, observations of bothhigh- and low-excitation water lines with a high spectral reso-lution are needed to disentangle the contributions from the hotcores (or hot corinos in the case of low-mass protostars) and thecolder external envelope, that can be linked to the parental cloud,in which stars form. Near protostars, the grain temperature risesabove ∼
100 K, leading to rapid water ice desorption that increasesthe gas-phase H O (and its deuterated counterparts) abundance inthe inner parts of the envelope. In order to interpret the observedspectra in terms of local physical conditions and relative abun-dances, radiative transfer modeling is necessary. This is illustratedwith the modeling performed by Coutens et al. (2012, 2013) towardthe low-mass protostar IRAS 16293-2422, where numerous HDO,H
O, and D O transitions have been used simultaneously to con-strain the abundances in the hot corino, in the cold envelope, and ina water-rich absorbing layer surrounding the envelope. This paper reports full statistical equilibrium and radia-tive transfer calculations towards the ultra compact HII regionG34.26 + Herschel / HIFI observations of HDO and the less abun-dant H
O water isotopologue. The paper is organized as follows.In Sections 2 and 3, we describe the source and the observationsrespectively. In Section 4, we present results obtained both with asimple local thermal equilibrium modeling (LTE) and with the 1Dnon-LTE modeling. In Section 5, we compare them with a chemicalmodel. Finally, we present our conclusions in Section 6.
Located at a distance of ∼ (cid:48) ) HII region (componentD) in the south-east. Chemical surveys were carried out towards theA, B and C components using single-dish telescopes (MacDonaldet al. 1996; Hatchell et al. 1998) and interferometric observations(Mookerjea et al. 2007). Many complex species, characteristic ofhot cores, have been detected. From molecular line observations,the emission peak does not coincide with the HII components (Watt& Mundy 1999; De Buizer et al. 2003), but is shifted to the East ofthe component C by ∼ (cid:48)(cid:48) (Mookerjea et al. 2007: Figure 3). Thisdi ff erence may arise due to the external influence of the nearby HIIregions, or may reveal separate regions of chemical enrichment.The hot core is likely externally heated by stellar photons ratherthan by shocks, as SiO was not detected at the position of the hotcore (Hatchell et al. 2001). This source is also characterized byinfall motions as suggested by observations of absorption compo-nents of NH , CN, HCN and HCO + (Wyrowski et al. 2012; Liuet al. 2013, Hajigholi et al. in prep.).The hot core of G34 has been the target of many Her-schel / HIFI observations for the past four years, including waterline emission (Flagey et al. 2013) and its deuterated couterparts.We present in Section 3 the H
O and HDO transitions observedfrom the ground as well as the
Herschel / HIFI observations. Notethat the Half-Power Beam Width (HPBW) of those telescopes en-compasses the components A and B and the molecular peak fromcomponent C for all observations.
This source is part of the PRISMAS Key Program (PRobing Inter-Stellar Molecules with Absorption line Studies; Gerin et al. 2010)which was followed by an Open Time Program led by C. Vastel.The targeted coordinates are α (J2000) = h m s , δ (J2000) = ◦ (cid:48) (cid:48)(cid:48) . The observations were performed in the pointed dualbeam switch (DBS) mode using the double sideband (DSB) HIFIinstrument (de Graauw et al. 2010; Roelfsema et al. 2012) onboardthe Herschel
Space Observatory (Pilbratt et al. 2010). The DBSreference positions were situated approximately 3 (cid:48) east and westof the source. The HIFI Wide Band Spectrometer (WBS) was used c (cid:13) , 000–000 ater deuterium fractionation in the high-mass star-forming region G34.26 + Table 1.
HDO and H
O transitions observed towards the ultra-compact HII region G34 (1) .Species Frequency J Ka , Kc E up / k A ij Telescope HPBW F e ff B e ff d (cid:51) rms (2) (cid:82) T mb d (cid:51) FWHM (GHz) (K) (s − ) ( (cid:48)(cid:48) ) (km s − ) (mK) (K km s − ) (km s − )HDO 80.5783 1 , –1 ,
47 1.32 × − IRAM-30m 31.2 0.95 0.81 0.182 56 2.36 5.9225.8967 3 , –2 ,
168 1.32 × − IRAM-30m 11.1 0.92 0.61 0.064 101 10.45 6.7241.5616 2 , –2 ,
95 1.19 × − IRAM-30m 10.4 0.90 0.56 0.061 84 12.27 6.6464.9245 1 , –0 ,
22 1.69 × − CSO 16.5 - 0.35 (3) , -1 ,
66 5.25 × − HIFI 1a 43.9 0.96 0.76 0.305 10 2.13 7.6509.2924 1 , –1 ,
47 2.32 × − HIFI 1a 42.3 0.96 0.76 0.294 44 1.76 9.0599.9267 2 , -2 ,
95 3.45 × − HIFI 1b 35.9 0.96 0.75 0.250 12 2.63 7.7848.9618 2 , -1 ,
84 9.27 × − HIFI 3a 25.4 0.96 0.75 0.176 10 3.92 10.3893.6387 1 , –0 ,
43 8.35 × − HIFI 3b 24.1 0.96 0.74 0.167 63 − (4) (5) , -1 ,
66 1.56 × − HIFI 3b 23.4 0.96 0.74 0.163 20 2.01 6.1p–H
O 203.4075 3 , –2 ,
204 4.81 × − IRAM-30m 12.1 0.93 0.62 0.074 121 8.38 (6) (6) p–H O (7) , –0 ,
53 1.79 × − HIFI 4b 19.2 0.96 0.74 0.136 110 − (8) (9) o–H O (7) , –1 ,
60 3.29 × − HIFI 1a 38.7 0.96 0.75 0.274 12 1.77 (10) (11)(1)
The frequencies, upper energy levels ( E up ) and Einstein coe ffi cients ( A ij ) come from the spectroscopic catalog JPL (Pickett et al. 1998). (2) The rms is calculated at the spectral resolution of the observations, which is indicated in the column d (cid:51) . (3) This value corresponds to the ratio between the main beam e ffi ciency B e ff and the forward e ffi ciency F e ff . (4) The integrated flux of the emission component is ∼ − , whereas it is ∼ − − for the absorbing component. (5) The Full Width at Half Maximum (
FWHM ) of the fundamental line at 894 GHz is estimated to be 5.9 km s − for the emission component (v LSR = − ) and 3.9 km s − for the absorption component (v LSR = − ). (6) After subtraction of the CH OCH line contaminating the para–H O line profile. (7)
Observations from Flagey et al. (2013). (8)
The integrated flux of the emission component is ∼ − , whereas it is ∼ − − for the absorbing component. (9) The Full Width at Half Maximum (
FWHM ) of the fundamental H
O line at 1101 GHz is estimated to be 6.4 km s − for the emission component (v LSR = − ) and 3.5 km s − for the absorption component (v LSR = − ). (10) The integrated flux of the emission component is ∼ − , whereas it is ∼ − − for the absorbing component. (11) The Full Width at Half Maximum (
FWHM ) of the fundamental H
O line at 547 GHz is estimated to be 7.5 km s − for the emission component (v LSR = − ) and 3.3 km s − for the absorption component (v LSR = − ). Figure 1.
Energy level diagram of the HDO lines. Green solid arrows: theIRAM-30m observations; blue short dashed arrows: the PRISMAS / HIFIobservations; red long dashed arrows: the Open Time HIFI observations;magenta dotted arrow: the CSO observation. The frequencies are given inGHz. (A color version of this figure is available in the online journal.) with optimization of the continuum, providing a spectral resolu-tion of 1.1 MHz over an instantaneous bandwidth of 4 × ff erent Local Oscillator (LO) settings.This method is necessary in such chemically rich regions in order toensure genuine detection of spectral lines. The HDO data were pro-cessed using the standard HIFI pipeline up to level 2 with the ESA-supported package HIPE 8.0 (Ott 2010) and were then exported asFITS files into CLASS / GILDAS format for subsequent data re-duction. The two linear polarizations were averaged to lower thenoise in the final spectrum. The baselines are well-fitted by straightlines over the frequency range of the whole band and were sub-tracted from all observations. The single sideband continuum tem-perature (that was obtained by dividing by 2 the DSB continuumderived from the linear fit obtained from line free regions in thespectrum, i.e. assuming a sideband gain ratio of unity) was addedto the spectrum of the 1 , –0 , fundamental line. To constrain theHDO / H O ratio, we also used two H
O transitions observed in theframework of the PRISMAS program and previously published byFlagey et al. (2013). We refer to this paper for the data reduction ofthese two lines. A list of all the
Herschel / HIFI observations used inthis paper is provided in Table A1.The ground state 1 , –0 , HDO transition was observed at theCaltech Submillimeter Observatory (CSO) in September 2011 us-ing the Fast Fourier Transform Spectrometer (FFTS) with 500 MHzbandwidth. The data were taken under good weather conditions,with 1.5 mm of precipitable water vapor. The beam switching modehas been used with a chop throw of 240 (cid:48)(cid:48) . The main beam e ffi ciency c (cid:13)000
Herschel / HIFI observations used inthis paper is provided in Table A1.The ground state 1 , –0 , HDO transition was observed at theCaltech Submillimeter Observatory (CSO) in September 2011 us-ing the Fast Fourier Transform Spectrometer (FFTS) with 500 MHzbandwidth. The data were taken under good weather conditions,with 1.5 mm of precipitable water vapor. The beam switching modehas been used with a chop throw of 240 (cid:48)(cid:48) . The main beam e ffi ciency c (cid:13)000 , 000–000 A. Coutens, C. Vastel, U. Hincelin et al. was determined from total power observations of Mars. The systemtemperature was about 3500 K during the run. The single sidebandcontinuum temperature ( ∼ , –1 , ), 226 (3 , –2 , ) and 242 GHz (2 , –2 , ) as well as the ortho–H O transition at203 GHz (3 , –2 , ) were observed with the IRAM-30m telescope.The observations were carried out in December 2011 using the FastFourier Transform Spectrometer (FTS) at a 50 kHz resolution. Thespectral resolution was 0.19, 0.07 and 0.06 km s − for the 81, 226and 242 GHz transitions, respectively. All the observations wereperformed using the Wobbler Switching mode. The 30m beam sizesat the observing frequencies are given in Table 1. During this run,weather conditions were good for winter, with 2 mm of precipitablewater vapor. System temperatures were always less than 200 K.Figure 1 presents the energy level diagram of the HDO transi-tions used for the modeling. Table 1 summarizes the observations. Most of the observed HDO lines show a Gaussian-like profile (seefor example Fig. 4). Only the HDO 1 , –0 , fundamental transi-tion observed with Herschel / HIFI shows an inverse P-Cygni pro-file, i.e a profile showing a red-shifted absorption component anda blue-shifted emission component. A similar profile has alreadybeen observed for this transition in low-mass protostars (Coutenset al. 2012, 2013). The Gaussian FWHM (Full Width at Half-Maximum) was derived for each line with the CASSIS software(see Table 1). Using the available spectroscopic databases CDMS(Cologne Database Molecular Spectroscopy; M¨uller et al. 2011,2005) and JPL (Jet Propulsion Laboratory; Pickett et al. 1998), wealso checked that the di ff erent lines are not contaminated by otherspecies. The HDO 2 , –2 , transition at 242 GHz could be slightlyblended with the CH COCH , –12 , line. However a simpleLTE (Local Thermal Equilibrium) modeling of CH COCH linesobserved in the spectra, shows that the contribution of CH COCH is negligible. With a column density of 4 × cm − , an excitationtemperature of 100 K, a FWHM of 6 km s − and a source size of1.7 (cid:48)(cid:48) , the predicted intensity of the CH COCH line at 241.6 GHzis 0.06 K, to be compared with the observed line intensity 1.75 K.The HDO 2 , –1 , line at 849 GHz is very probably blended withthree CH OH lines (18 , –17 , A-, 18 , –17 , A + , and 18 , –17 , A-) lying in the red-shifted portion of the line profile. Thiscould explain why this line is broader than the others (see Table 1).As the CH OH contribution could be non-negligible, we do notuse this HDO line to constrain the abundances. We present howeverthe modeling of this line for completeness. The other HDO lines donot show any potential blending. The portion of the 600 GHz lineobserved at (cid:51) >
72 km s − is produced by the CH OH 7 , –6 , (cid:51) = + line from the image band ( ν = O 3 , –2 , line is blended with the CH OCH , , –2 , , and 3 , , –2 , , transitions at 203.4101 and203.4114 GHz ( E up =
18 K). We can reproduce, with a LTEmodeling approach, the CH OCH lines observed nearby in thespectra (see Figure 2) as well as in the other bands. The CH OCH lines are well-fitted with a column density of 7 × cm − , anexcitation temperature of 100 K, a FWHM of 6 km s − and asource size of 1.7 (cid:48)(cid:48) . The predicted line profiles of the CH OCH transitions blended with H O are then subtracted from the http://cassis.irap.omp.eu Figure 2.
IRAM-30m observations (in black) of the para–H
O 3 , –2 , transition at 203.4 GHz. The frequency of the H O line is indicated by ablue dotted line ( (cid:51) =
58 km s − ). The other lines observed in this spectra arethe SO (cid:51) = , –11 , line at 81.5 km s − (green long dashed line), theC H CN (cid:51) = , –22 , line at 74.1 km s − (yellow short dashed line)and several CH OCH lines (magenta solid lines). A LTE modeling of theCH OCH lines (in magenta) was carried out to estimate the contaminationof the para–H O line by the CH OCH , , –2 , , transition. (A colorversion of this figure is available in the online journal.) observed line profile to extract the proper H O spectrum. Due tothe high number of CH OCH lines considered in the analysis andthe presence of CH OCH lines with similar upper energy levels(18 K) around the H O feature (see Figure 2), the uncertaintyproduced by this subtraction is negligible with respect to thecalibration uncertainty ( ≤ A simple LTE modeling was first employed to estimate theHDO / H O ratio in the hot core. We plot in Figure 3 the rotationdiagram (Goldsmith & Langer 1999) of the HDO lines shown inTable 1. We exclude the fundamental transition at 894 GHz, whichshows absorption and probably probes colder regions outside of thehot core. We take into account beam dilution and consider di ff erentsource sizes between 1 (cid:48)(cid:48) and 5 (cid:48)(cid:48) , as the exact size of the hot coreis unknown. Indeed, the structure determined by van der Tak et al.(2013) predicts a size of 4.5 (cid:48)(cid:48) for T >
100 K, whereas the interfer-ometric observations of two HDO lines by Liu et al. (2013) favora smaller source size which, however, is not well constrained. Nolinear curve is in reasonable agreement with the complete dataset.Plausible explanations are that the lines are optically thick or thatthey do not all have the same excitation temperature. We estimatethe critical densities of these species using the HDO collisionalcoe ffi cients with ortho– and para–H of Faure et al. (2011). At ∼
100 K, the critical densities are about 5 × – 5 × cm − forall lines, except for the lines at 80, 226, and 242 GHz that have crit-ical densities between 3 × and 3 × cm − . These latter lines c (cid:13) , 000–000 ater deuterium fractionation in the high-mass star-forming region G34.26 +
25 50 75 100 125 150 175
Eup/k[K] l n ( N u / gu ) HDO 19002
Tex = 79 (-8,+10) K N = 1.6 (-0.2,+0.3) E16 /cm ²
25 50 75 100 125 150 175
Eup/k[K] l n ( N u / gu ) HDO 19002
Tex = 79 (-8,+10) K N = 1.6 (-0.2,+0.3) E16 /cm ² Texte Texte l n ( N u / g u ) E up /k (K) Figure 3.
Rotational diagram of the HDO lines. The fundamental line at894 GHz is excluded from the figure. A source size of 4 (cid:48)(cid:48) is assumed. Theerror bars correspond to uncertainties of 20%. A linear fit was made usingonly the lines with low critical densities (81, 226, and 242 GHz, indicatedby red triangles) to estimate the HDO column density ( ∼ × cm − )and the excitation temperature ( ∼
79 K) in the hot core (see text). (A colorversion of this figure is available in the online journal.) are at low frequencies, so that their radiative decay is slower. Theyare consequently probably in LTE, as the density in the hot core isexpected to be (cid:38) cm − (van der Tak et al. 2013). In addition,these three lines are also those with the expected lowest opacities.We consequently fit a straight line to these three points only. Thecolumn density and the excitation temperature of HDO were thenestimated for di ff erent values of the source size (1 (cid:48)(cid:48) –5 (cid:48)(cid:48) ). To derivethe HDO / H O ratio, we used the H
O 3 , –2 , line observed at 203GHz with IRAM. Indeed this line is quite excited and its criticaldensity is relatively low, about 10 cm − . Using the same excitationtemperature as HDO, we calculated the column density of H Oin the hot core and derived an estimate of the HDO / H O ratio be-tween 5.2 × − and 5.7 × − . The H O / H O ratio is assumedto be 400 following the relation determined by Wilson (1999) be-tween the O / O isotopic ratio and the distance of the source fromthe galactic center. The derived HDO / H O ratio is consistent withthe previous estimates by Jacq et al. (1990, 4 × − ) and Liu et al.(2013, 3.0 × − ), who assumed an H O / H O ratio equal to 500.It is also slightly greater than the estimate by Gensheimer et al.(1996, 1.1 × − ). Only three HDO lines (among 10) were used to estimate theHDO / H O ratio in the hot core with the rotational diagram analysis.We consequently decided to employ non-LTE spherical modeling(that also considers the line opacities) to include the informationprovided by all the HDO lines (except the 848 GHz line that isprobably blended with CH OH) and to determine the water deu-terium fractionation in both the hot core and the colder part of theenvelope. We used the RATRAN code (Hogerheijde & van der Tak2000) that assumes spherical symmetry and takes into account con-tinuum emission and absorption by dust. To derive the HDO andH
O abundances, we used the temperature and H density profilesderived by van der Tak et al. (2013, Section 4.1). This structure was determined taking into account JCMT / SCUBA and PACS data.The radial velocity profile ( (cid:51) r ) and the turbulence width (Doppler b-parameter, db ) have also to be provided in RATRAN. We describein Appendix B2 the method employed to constrain them and showthe final (cid:51) r and db profiles used in the analysis. We find that inwardmotions ( (cid:51) r ∼ -3 km s − ) are present in the cold envelope, while out-ward motions ( (cid:51) r ∼ − ) take place in the inner regions. Thesame type of velocity profile was found in SgrB2(M) by Rol ff s et al.(2010). The Doppler b-parameter appears lower in the inner regions( db ∼ − ) than in the outer regions ( db ∼ − ), sim-ilarly to what was found by Caselli & Myers (1995) and Herpinet al. (2012) in other high mass sources. To reproduce the contin-uum levels seen in the observations as best as possible, we used thedust opacities from Ossenkopf & Henning (1994), with thick icemantles and a gas density of 10 cm − . The dust opacities used byvan der Tak et al. (2013, thin ices mantles with gas density of 10 cm − ) to derive the structure would not however di ff er too much,as the predicted continuum is consistent with the observations towithin 10–20% uncertainties. The most recent HDO and H O col-lisional coe ffi cients calculated with ortho– and para–H by Faureet al. (2011) and Daniel et al. (2011), respectively, were used. Theortho / para ratio of H is assumed to be at LTE in each cell of theenvelope. It consequently varies from ∼ − in the coldest regionsup to the equilibrium value of 3 in the warm regions. Most of the studies of water and deuterated water in star-formingregions (e.g., Ceccarelli et al. 2000; Parise et al. 2005; van der Taket al. 2006; Coutens et al. 2012; Herpin et al. 2012) assume anabundance jump at T j =
100 K, corresponding to the temperature atwhich the water molecules are supposed to be released in the gasphase by thermal desorption. In a first step, we consequently as-sumed such an abundance jump for the modeling of the HDO lines.According to the physical structure used here (van der Tak et al.2013), the source size corresponding to T >
100 K is ∼ (cid:48)(cid:48) (diam-eter). We ran a grid of models with various inner ( T >
100 K) andouter ( T <
100 K) abundances and realized that, regardless of thevelocity profiles, the intensities of the di ff erent lines cannot be re-produced simultaneously (see Figure 4). Indeed, when the excitedtransitions observed at 225 and 241 GHz with IRAM are repro-duced, the fluxes of the CSO and HIFI lines are overproduced, inparticular at 491, 600, 849, and 919 GHz (red dashed model in Fig-ure 4). On the contrary, if these latter lines are reproduced, the fluxis underpredicted for the IRAM lines (green dotted model in Figure4). Although the choice of the velocity profiles can a ff ect the lineprofiles, it is not possible to appreciably modify the intensities anddecrease this disagreement.To obtain an agreement for all the transitions, an increase ofthe jump temperature is necessary. Indeed, Figure B4 shows that,with a jump temperature of 120 K, the model that reproduces thefluxes of the most excited HDO lines (226 and 242 GHz) is in bet-ter agreement with the fluxes of the lines at 491, 600, 849, and 919GHz than the model with a jump at T j =
100 K. The fluxes of thesefour lines are, however, still overproduced. Consequently, we rangrid of models for higher jump temperatures (150, 180, 200, and220 K) and compared the influence of the jump temperature on theline intensities. The best-fit predictions obtained for T j =
150 K,180 K, 200 K, and 220 K are shown in Figures B5, B6, 5, and B7,respectively. These four models reproduce relatively well the ob-servations. The 491, 600, 849, and 919 GHz lines are quite sen-sitive to the jump temperature. Their intensities decrease with the c (cid:13)000
150 K,180 K, 200 K, and 220 K are shown in Figures B5, B6, 5, and B7,respectively. These four models reproduce relatively well the ob-servations. The 491, 600, 849, and 919 GHz lines are quite sen-sitive to the jump temperature. Their intensities decrease with the c (cid:13)000 , 000–000 A. Coutens, C. Vastel, U. Hincelin et al.
HDO 1 -0
894 GHz
40 50 60 702.83.03.23.43.63.84.04.2 T m b ( K ) HDO 1 -0
465 GHz
40 50 60 701.52.02.53.03.54.04.5
HDO 1 -1
81 GHz
40 50 60 70-0.20.00.20.40.6
HDO 3 -2
226 GHz
40 50 60 70-0.50.00.51.01.52.0
HDO 2 -2
242 GHz
40 50 60 70-0.50.00.51.01.52.0
HDO 2 -1
491 GHz
40 50 60 70v
LSR (km s -1 )-0.20.00.20.40.60.81.0 T m b ( K ) HDO 1 -1
509 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.3 HDO 2 -2
600 GHz
40 50 60 70v
LSR (km s -1 )-0.20.00.20.40.60.8 HDO 2 -1
849 GHz
40 50 60 70v
LSR (km s -1 )-0.20.00.20.40.60.81.0 HDO 2 -1
919 GHz
40 50 60 70v
LSR (km s -1 )-0.20.00.20.40.60.8 Figure 4.
Black solid line : HDO lines observed with HIFI, IRAM, and CSO.
Red dashed line : Modeling for a jump temperature T j =
100 K, an inner abundance X in = × − and an outer abundance X out = × − . Green dotted line : Modeling for a jump temperature T j =
100 K, an inner abundance X in = × − and an outer abundance X out = × − . (A color version of this figure is available in the online journal.) increase of T j . In particular, for T j =
150 K and 180 K, their inten-sities are slightly overpredicted, while, for T j ≥
220 K, they startto be under-predicted. In view of these results, the best-fit modelis obtained for T j ∼
200 K. Table 2 summarizes the HDO best-fitabundances found for each jump temperature and the correspond-ing size of the jump abundance. The best-fit was determined witha χ minimization of the line profiles similar to what was done inCoutens et al. (2012), assuming a calibration uncertainty of 20%for each line. As the HDO line at 849 GHz is probably blendedwith CH OH, we did not include it in the calculation. The re-duced χ obtained for the model with an abundance jump at 200 Kis 1.3. The HDO inner abundance is strongly constrained by thehigh number of emission lines used in the analysis. If we just con-sider the grid with T j =
200 K, its value is between 1.7 × − and2.1 × − . Consequently, the main uncertainty on the HDO innerabundance comes from the value assumed for the jump tempera-ture ( ∼ × − –3 × − for T j ∼ × − and 9 × − . TheHDO 1 , –1 , line at 81 GHz is not reproduced by any of the mod-els within the 20% calibration uncertainty and could maybe su ff erof calibration problems at this low frequency with the IRAM-30mtelescope. Models with a two-jump abundance profile such as inComito et al. (2010) were also attempted but do not improve the fit(see Appendix B3).It clearly appears that, to reproduce the HDO line profiles, anincrease of the jump temperature in the model is necessary. We can-not conclude, however, that the sublimation temperature for waterice is significantly higher than 100 K. Although some experimentsactually favor an evaporation temperature of 110–120 K (Fraseret al. 2001), it is not su ffi cient to perfectly reproduce all the HDOtransitions. The main reason for the modification of the jump tem-perature would be related to the size of the hot core, rather than tothe water sublimation temperature itself. In this case, the size of thehot core in which the abundance of water increases after the evap-oration of the icy mantles should be smaller ( ∼ (cid:48)(cid:48) instead of 4.5 (cid:48)(cid:48) ),in order to lead to a better agreement between the model and theobservations. This is also in agreement with the interferometric ob- servations of the HDO lines at 225 and 241 GHz by Liu et al. (2013)that are not spatially resolved with a beam size of 3.7 (cid:48)(cid:48) × (cid:48)(cid:48) . Twoexplanations can be provided to explain the smaller size of the hotcore. One would be that the physical structure derived by van derTak et al. (2013) is unreliable at small scales. Indeed the structuredetermined here is only based on large-scale maps and the densityprofile is assumed to follow a power-law. The density and temper-ature profiles could therefore be uncertain at small scales ( θ (cid:46) (cid:48)(cid:48) ).In this case, the temperature actually would reach 100 K at a ra-dius which is smaller than what the physical structure predicts (vander Tak et al. 2013). The second possible explanation is providedby the chemical models coupled with a dynamical approach, wherethe dynamical timescales can be in competition with the chemi-cal and adsorption / desorption timescales. Indeed, as it can be seenfor example in Aikawa et al. (2012) and Wakelam et al. (submit-ted), the abundance increases gradually for a certain temperaturerange before a constant inner abundance is reached. The temper-ature where the inner abundance is constant is higher than 100 Kbut its exact value is dependent on the model parameters. It seemstherefore possible that the constant inner abundance can be reachedonly at ∼
200 K. Some tests assuming a gradual abundance increasewere attempted in Section 4.2.5 and this explanation seems to holdhere.
O lines with an abundance jump andestimate of the HDO / H O ratios
A similar model with an abundance jump was carried out with theH
O lines detected in this source to determine the HDO / H O ra-tio throughout the envelope. The H
O transitions detected with theHIFI instrument by Flagey et al. (2013) are not well suited to mea-sure abundances, because of their large opacities. With its excita-tion level, the para–H
O 3 , –2 , transition observed at 203 GHzwith the IRAM-30m telescope is suitable to probe the hot core andderive the HDO / H O ratio in the warm inner region. The H
O fun-damental lines previously detected with
Herschel / HIFI by Flageyet al. (2013) are also used to constrain the HDO / H O ratio in the en-velope, as these lines combine both emission and absorption. Note c (cid:13) , 000–000 ater deuterium fractionation in the high-mass star-forming region G34.26 + HDO 1 -0
894 GHz
40 50 60 702.53.03.54.04.5 T m b ( K ) HDO 1 -0
465 GHz
40 50 60 701.52.02.53.03.54.0
HDO 1 -1
81 GHz
40 50 60 70-0.20.00.20.40.6
HDO 3 -2
226 GHz
40 50 60 70-0.50.00.51.01.52.0
HDO 2 -2
242 GHz
40 50 60 70-0.50.00.51.01.52.0
HDO 2 -1
491 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.3 T m b ( K ) HDO 1 -1
509 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.3 HDO 2 -2
600 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.4 HDO 2 -1
849 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.4 HDO 2 -1
919 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.4 Figure 5.
Black solid line : HDO lines observed with HIFI, IRAM, and CSO.
Red dashed line : Modeling for a jump temperature T j =
200 K, an inner abundance X in = × − and an outer abundance X out = × − . (A color version of this figure is available in the online journal.) p-H O 3 -2
203 GHz
40 50 60 70v
LSR (km s -1 )-0.50.00.51.01.52.0 T m b ( K ) p-H O 1 -0
40 50 60 70v
LSR (km s -1 )4.55.05.56.06.5 o-H O 1 -1
548 GHz
40 50 60 70v
LSR (km s -1 )0.80.91.01.11.21.31.4 Figure 6.
Black solid line : H
O lines observed with HIFI and IRAM.
Red dashed line : Modeling for a jump temperature T j =
200 K, an inner abundance X in = × − and an outer abundance X out = × − . (A color version of this figure is available in the online journal.) that we only use here the ortho 1 , –1 , transition at 548 GHz andthe para 1 , –0 , transition at 1102 GHz. The fundamental ortho2 , –1 , transition, which is observed at 1656 GHz in absorption,was not taken into account because of pointing problems a ff ectingthe observations. The source being fairly peaked on the continuum,an o ff set could lead to a significant loss of flux.All the physical parameters are kept similar to those of thestudy of HDO. Figures B8, B9, 6, and B10 show the best-fit modelsobtained for these three lines for the jump temperatures previouslyassumed for deuterated water, T j = >
50 K). This value is also consistent with the ratio determined inmost of the foreground clouds on the line of sight towards brightcontinuum sources (Lis et al. 2010; Flagey et al. 2013). The best-fitinner and outer H
O abundances are summarized in Table 2. Thereduced χ is about 1.5 for the case T j =
200 K. Assuming an obser-vational uncertainty of 20% for the excited para–H
O line at 203GHz, the inner abundance cannot be higher than 1.2 × − or lowerthan 7 × − for the model with T j =
200 K. The outer abundanceis estimated to be between 1.0 × − and 1.5 × − , based onan observational uncertainty of 20% for the absorbing componentof the para–H O transition at 1102 GHz. The H
O abundancesin Table 2 are estimated using an H O / H O ratio of 400 (Wilson1999). The best-fit HDO / H O ratios are then equal to ∼ (5–6) × − in the hot core and ∼ × − in the outer envelope. Even whenconsidering the HDO and H O results with a 20% calibration un-certainty, the outer HDO / H O ratio (1.0 × − –2.2 × − ) is stillhigher than the inner HDO / H O ratio (3.5 × − –7.5 × − for T j =
200 K). We ran models with a constant ortho / para H ratio equalto 3 to check that the ortho / para H ratio assumed in the model doesnot a ff ect the results. The HDO and H O line profiles are exactlythe same as with an LTE ortho / para ratio, confirming the variationof the HDO / H O ratio from the cold to the warm regions.
Although we used a constant abundance of HDO and H
O in thecold envelope of G34, it is very probable that the water abun-dance shows variations in this region due to non-thermal desorp-tion mechanisms. In particular, Mottram et al. (2013) showed thatthe desorption by the cosmic ray-induced UV field leads to an outerabundance of water decreasing gradually from the cold to the warmregions of low-mass protostars. To confirm that the presence of agradual abundance decrease in the cold envelope does not a ff ect thederived value of the HDO / H O ratio in this region, we ran a mod-eling considering an equilibrium state between the desorption bythe cosmic ray-induced UV field and the re-depletion on the grains. c (cid:13) , 000–000 A. Coutens, C. Vastel, U. Hincelin et al.
Table 2.
HDO and H O abundances obtained for di ff erent jump temperatures T j T j (K) ( a ) θ ( (cid:48)(cid:48) ) ( a ) X in (HDO) X out (HDO) X in (H O) X out (H O) X in (H O) ( b ) X out (H O) ( b ) (HDO / H O) in( b ) (HDO / H O) out( b ) ( c ) ( c ) × − × − × − × − × − × − × − × −
180 1.9 1.5 × − × − × − × − × − × − × − × − ( d ) × − × − × − × − × − × − × − × −
220 1.5 3 × − × − × − × − × − × − × − × − Notes: ( a ) Size of the region where the temperature is higher than T j (diameter). It is derived from the structure determined by van der Tak et al. (2013). ( b ) Assuming H O / H O = ( c ) Fit is not good enough to determine the HDO abundances. ( d ) Best-fit. radius (AU)10 -13 -12 -11 -10 -9 -8 -7 -6 HD O a bund a n ce Figure 7.
Best-fit abundance profiles obtained for HDO when the outerabundance (region at T <
200 K) is constant (black solid line, Section 4.2.2),when it decreases from the cold regions to the region at T =
200 K (reddashed line, Section 4.2.4), and when it decreases from the cold regions tothe region at T =
100 K and increases from T =
100 K to T =
200 K (greendotted line, see Section 4.2.5). The temperature reaches 200 K at a radius of2700 AU ( ∼ (cid:48)(cid:48) , see Figure B1). (A color version of this figure is availablein the online journal.) Using similar equations to those in Hollenbach et al. (2009) andMottram et al. (2013), we get by equating desorption to depletion: G cr F Y x f s , x n gr σ gr = n (x) n gr σ gr (cid:51) th , x (4)with F the local interstellar flux of 6–13.6 eV photons assumedto be equal to 10 photons cm − s − , G cr the scaling factor of theUV flux, Y x the photodesorption yield for the molecule x ( ∼ − for H O, ¨Oberg et al. 2009), f s , x the fraction of the molecule x ongrains, n gr the grain density, σ gr the cross sectional area of the grainand (cid:51) th , x the thermal velocity. The thermal velocity is calculatedaccording to the following formalism: (cid:51) th , x = (cid:114) k b T k π m x , (5)where k b is the Boltzmann constant, T k the gas temperature and m x the mass of the molecule x. The outer abundance of H O withrespect to H is then equal to: X out (H O) = G cr F Y H O f s , H O (cid:51) th , H O n H , (6) with n H the H density. Similarly we obtain for HDO: X out (HDO) = G cr F Y HDO f s , H O (cid:51) th , HDO n H (cid:32) HDOH O (cid:33) , (7)and for H O: X out (H O) = G cr F Y H O f s , H O (cid:51) th , H O n H (cid:32) H OH O (cid:33) . (8)The photodesorption yields for HDO and H O are assumed simi-lar to those for H O ( ¨Oberg et al. 2009). The thermal velocity isapproximatively the same due to their relatively similar masses.All the other parameters are independent of the molecules exceptthe fraction f s , x of these molecules contained in the grain mantleswhich reflects the isotopic ratios, HDO / H O and H O / H O, onthe grains. The external UV field should also a ff ect the externalpart of the outer envelope. But, due to the very small constraints onthese di ff erent mechanisms, we only considered the desorption bythe cosmic ray-induced UV field.We ran a grid of models for the case T j =
200 K, keep-ing the inner abundances determined previously ( X in (HDO) = × − and X in (H O) = × − ). Di ff erent values were then as-sumed for the factors W HDO = G cr f s , H O HDO / H O and W H O = G cr f s , H O H O / H O. Assuming H O / H O =
400 (Wilson 1999)and f s , H O = O), the best-fit model of the H
O lines gives a scaling factor G cr of about 1.6 × − . If water represents only 50% of the grainmantles, G cr is then equal to 3.2 × − leading to a cosmic ray-induced UV field of ∼ × photons cm − s − . These values rep-resent, however, only upper limits, since the desorption by the ex-ternal UV field is not taken into account in the analysis. The typicalvalue of the cosmic-ray induced UV flux ( G cr ∼ − ; e.g., Prasad& Tarafdar 1983; Shen et al. 2004) is then consistent with the upperlimit derived here ( G cr (cid:46) × − ).The best-fit abundance profile determined for HDO when theouter abundance decreases from the cold to the warm regions ispresented in Figure 7 (red dashed line). The HDO / H O ratio in theouter envelope is equal to 1.3 × − . It is then, once again, higherthan in the hot core ( ∼ (5–6) × − ). The HDO and H O line pro-files predicted with the RATRAN code (see red dashed lines in Fig-ures 8 and 9) are relatively similar to those in Figures 5 and 6 thatassume a two-step abundance profile with a jump at 200 K. Thefit is even better for the H
O and HDO fundamental transitions(HDO: 894 GHz; H
O: 548 and 1102 GHz), as the predicted in-tensity of their emission is now in agreement with the observations.Some of the HDO lines (509, 600, and 919 GHz) show small self-absorptions on their blue-shifted side. However, these defects couldprobably disappear with slightly di ff erent velocity profiles. Indeed, c (cid:13) , 000–000 ater deuterium fractionation in the high-mass star-forming region G34.26 + HDO 1 -0
894 GHz
40 50 60 702.83.03.23.43.63.84.04.2 T m b ( K ) HDO 1 -0
465 GHz
40 50 60 701.52.02.53.03.54.0
HDO 1 -1
81 GHz
40 50 60 70-0.20.00.20.40.6
HDO 3 -2
226 GHz
40 50 60 70-0.50.00.51.01.52.0
HDO 2 -2
242 GHz
40 50 60 70-0.50.00.51.01.52.0
HDO 2 -1
491 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.3 T m b ( K ) HDO 1 -1
509 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.3 HDO 2 -2
600 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.4 HDO 2 -1
849 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.4 HDO 2 -1
919 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.4 Figure 8.
Black solid line : HDO lines observed with HIFI, IRAM, and CSO.
Red dashed line : Modeling for a constant inner abundance X in = × − (T ≥
200 K) and an outer abundance (T <
200 K) decreasing from the cold to the warm regions (see Section 4.2.4).
Green dotted line : Modeling for a constant innerabundance X in = × − (T ≥
200 K), an abundance gradually increasing from 100 to 200 K, and an outer abundance (T <
100 K) decreasing from the coldto the warm regions (see Section 4.2.5). (A color version of this figure is available in the online journal.) p-H
O 3 -2
203 GHz
40 50 60 70v
LSR (km s -1 )-0.50.00.51.01.52.0 T m b ( K ) p-H O 1 -0
40 50 60 70v
LSR (km s -1 )4.55.05.56.06.5 o-H O 1 -1
548 GHz
40 50 60 70v
LSR (km s -1 )0.80.91.01.11.21.3 Figure 9.
Black solid line : H
O lines observed with HIFI and IRAM.
Red dashed line : Modeling for a constant inner abundance X in = × − (T ≥
200 K)and an outer abundance (T <
200 K) decreasing from the cold to the warm regions (see Section 4.2.4).
Green dotted line : Modeling for a constant innerabundance X in = × − (T ≥
200 K), an abundance gradually increasing from 100 to 200 K, and an outer abundance (T <
100 K) decreasing from the coldto the warm regions (see Section 4.2.5). (A color version of this figure is available in the online journal.) the velocity profiles used here were only adapted for the modelswith the abundance jumps (see Appendix B2). / hot core transition In Section 4.2.2, we mentioned that a model with a gradual in-crease of the HDO abundance at the cold envelope / hot core tran-sition could potentially explain why we need a higher jump tem-perature than 100 K to reproduce the HDO line profiles. Here weshow the results obtained with both a decrease of the outer abun-dance from the outermost regions to the regions at 100 K and agradual increase from 100 to 200 K. This type of profile is then rel-atively similar to the predictions of chemical models coupled witha dynamical approach (Aikawa et al. 2012, Wakelam et al. sub-mitted). The HDO inner abundance is equal to 2 × − and theouter abundance follows the trend described in Section 4.2.4. Theabundance profile used and the result of the model for HDO areshown in Figures 7 and 8 (green dotted line), respectively. Thismodel also appears very similar to the model with an abundancejump at 200 K (see Figure 5). A model with both an abundance de-crease (with temperature) in the colder envelope and an increase of the abundance towards the hot core is probably more realistic thanthe jump abundance assumption and could explain why the hot coreis smaller than expected. It is, however, important to note that thetemperature range of the gradual abundance increase is not known.We assume here the range 100–200 K but it could be slightly dif-ferent and a specific range is probably dependent on the dynamics.The result of this modeling should thus be considered only qualita-tively. We can however conclude that this type of abundance profileallows to reproduce the HDO line profiles as well as the abundancejump models at ∼ O lines. The model (pre-sented in Section 4.2.5) with a gradual increase of the abundance atthe cold envelope / hot core transition is presented in Figure 9 (greendashed lines). The lines are here again reproduced as well as by thejump abundance models. The HDO / H O ratio shows consequentlythe same variation between the inner and outer regions as foundbefore, i.e. 5.6 × − at T >
200 K and 1.3 × − at T <
100 K.
The singly deuterated form of water has been studied toward manyhigh-mass hot cores with ground-based telescopes (Jacq et al. 1990; c (cid:13)000
The singly deuterated form of water has been studied toward manyhigh-mass hot cores with ground-based telescopes (Jacq et al. 1990; c (cid:13)000 , 000–000 A. Coutens, C. Vastel, U. Hincelin et al.
Gensheimer et al. 1996; Pardo et al. 2001; van der Tak et al. 2006).These studies are relevant for the hot core study but do not directlyaddress for the cold external envelope, since the observations ofthe ground HDO transition at 894 GHz with a very good signal-to-noise ratio are necessary in order to disentangle the contribu-tion from the hot core to the contribution of the cold envelope. Thelaunch of the
Herschel
Space Observatory dramatically changedthe situation, with the access to the high frequency range with manyHDO transitions available in addition to the ground-state transi-tion. The D / H ratio in water remained for a long time very poorlyknown since the study of water was based on observations su ff eringfrom dilution in the large beams of the Infrared Space Observatory(ISO), the Submillimeter Wave Astronomy Satellite (SWAS) andthe ODIN satellite as well as from large opacities. The only way tostudy water from the ground was to use the H O transition avail-able with some telescopes at 203 GHz (Jacq et al. 1988; van derTak et al. 2006; Jørgensen & van Dishoeck 2010; Persson et al.2012, 2013). With the help of this line, the water deuterium frac-tionation was previously estimated in the high-mass star-formingregion G34 by Jacq et al. (1990), Gensheimer et al. (1996), and Liuet al. (2013). They found, in its hot core, HDO / H O ratios rangingbetween 1 × − and 4 × − . Since our modeling in the hot coreregion is mostly dominated by the 81, 226 and 241 GHz transitionsaccessible from the ground, these values are relatively consistentwith our estimate of (5–6) × − both with the rotational diagramapproach and the non-LTE 1D analysis. Note that we assumed anH O / H O ratio of 400, whereas the previous studies assumed500. In addition, the HDO / H O ratio found in this hot core is con-sistent with the average HDO / H O ratio (a few 10 − ) found in otherhigh-mass sources (Jacq et al. 1990; Gensheimer et al. 1996; Pardoet al. 2001; van der Tak et al. 2006; Emprechtinger et al. 2013).In the hot core, we also determined the water abundance (rel-ative to H ) to be a few × − . Similar values were estimated inother high-mass hot cores (Chavarr´ıa et al. 2010; Herpin et al. 2012;Neill et al. 2013), although lower values were also found, for ex-ample, in NGC 6334 I ( ∼ − , Emprechtinger et al. 2013). Thevalue of 10 − is comparable to the observed abundance of solid wa-ter and together with the derived HDO / H O abundance ratios of10 − − − suggests that the origin of the observed water is evap-oration of grain mantles.Recently, Liu et al. (2013) also attempted to constrain the D / Hratio for water in the outer envelope of G34 using the 894 GHztransition observed from the ground with APEX. From a RATRANmodeling using an abundance jump profile at 100 K, they failedto reproduce the profile of this ground state transition leading toa very uncertain value for the D / H ratio in the outer region of theenvelope of (1 . . × − . With the sensitivity of Herschel / HIFIobservations of the 894 GHz transition, it became possible to mea-sure accurately the D / H ratio of water in low-mass (Coutens et al.2012, 2013) and high-mass protostars, from the hot core region tothe cold external envelope. We showed here that, with a value of(1.0–2.2) × − in the colder envelope, the HDO / H O ratio is in-deed higher than the estimate by Liu et al. (2013). It is also higherthan in the hot core. A similar behavior was discovered in the low-mass sources IRAS16293 and NGC1333 IRAS4A (Coutens et al.2013a, 2013b). But this is the first time that a radial variation ofthe D / H ratio has been observed towards a high-mass star-formingregion. The HDO / H O ratio derived in the colder envelope of G34is among the highest values found in high-mass sources. It is closeto the high value of (2–4) × − found in Orion KL (Persson et al.2007; Neill et al. 2013) but lower by more than a factor 10 than in the absorbing layer of low-mass protostars (Coutens et al. 2012,2013). In order to study the chemical pathways that could lead to the ob-served HDO and H O abundances and their corresponding ratio,we modeled the chemical evolution of the source as a function ofits radius, using the full gas-grain chemical model Nautilus (Her-sant et al. 2009).
Nautilus is a gas grain chemical code adapted from the originalcode developed by the Herbst group (Hasegawa & Herbst 1993).It solves the kinetic equations of gas-phase chemistry, takes intoaccount grain surface chemistry, and interactions between bothphases (adsorption, thermal and non-thermal desorption). The rateequations follow Hasegawa et al. (1992) and Caselli et al. (1998).More details on the processes included in the code are presentedby Semenov et al. (2010). The chemical network is adapted fromAikawa et al. (2012) and Furuya et al. (2012). As pointed out byPagani et al. (1992), Flower et al. (2004, 2006a,b), Walmsley et al.(2004), and Pagani et al. (2009), considering ortho and para spinmodifications of various H and D bearing species is important dueto some reactions which are much faster with ortho–H than para–H , and can change the entire chemistry of deuterium fractionation.Thus, we extended the network including the ortho, para, and metastates of H , D , H + , H D + , D H + , and D + . For the reactions in-volving these species, we have applied spin selection rules to knowwhich reactions are allowed, and have determined branching ra-tios assuming a total scrambling and a pure nuclear spin statisticalweight. Some of the rate coe ffi cients of these reactions have beentheoretically or experimentally determined (Marquette et al. 1988;Jensen et al. 2000; McCall et al. 2004; Dos Santos et al. 2007; Hugoet al. 2009; Honvault et al. 2011a,b; Dislaire et al. 2012) and forthese we used the calculated or measured values. We have bench-marked our model against some previous work that includes spin-state chemistry, using the same conditions as described in Figure 8of Pagani et al. (2009) and Figure 4 of Sipil¨a et al. (2013): a tem-perature of ∼
10 K and a density from ∼ to ∼ cm − . Minordi ff erences in abundances do exist, since the networks, the models,and the input parameters can be slightly di ff erent, but the result isglobally similar. A notable di ff erence is however seen for HD af-ter 10 yrs as compared with Sipil¨a et al. (2013). They predict adecrease of its gas phase abundance by one order of magnitude at10 yr. Under the same conditions, we predict a decrease in the gasphase HD abundance of only a factor ≈
2, similar to the modelof Albertsson et al. (2013, 2014, priv. com.). The inclusion in ourmodel of photodesorption and reactive desorption may have somee ff ect on HD depletion. Photodesorption due to direct interstellarUV photons and secondary photons generated by cosmic rays, aswell as the exothermic association between the surface species Hand D, may both release enough HD molecules to the gas phaseto lower the HD depletion. The network and a benchmark will bepresented in more detail in a forthcoming paper (U. Hincelin et al.,in preparation).In our model, elemental and initial abundances followHincelin et al. (2011). Initially, the ortho-to-para H ratio is set toits statistical value of 3, and deuterium is assumed to be entirely c (cid:13) , 000–000 ater deuterium fractionation in the high-mass star-forming region G34.26 + in HD form with an abundance of 1 . × − relative to total hy-drogen, following Kong et al. (2013). Note that the timescale forconversion to a thermal ortho-to-para H ratio is a few times 10 toa few times 10 yr at 10 K depending on the density, as in Paganiet al. (2009). In the evolutionary sequence of high-mass star for-mation proposed by Beuther et al. (2007) and Zinnecker & Yorke(2007), infrared dark clouds (IRDCs) are expected to be the firststage. Comparing observations of high-mass star-forming regionswith advanced gas-grain chemical modeling, Gerner et al. (2014)derived a chemical age for this stage of around 10 yrs. The meandensity and temperature of IRDCs are respectively 10 cm − and16 K (Sridharan et al. 2005). From the initial elemental and chem-ical abundances, we have computed the chemical evolution overa period of 10 yrs, corresponding to t IRDC in Figure 10, with atemperature of 16 K, a proton density of 2 × cm − , and a visualextinction of 30. In our standard model, we use a cosmic ray ioniza-tion rate of 1 . × − s − , but also use a value ten times higher, asdiscussed in Section 5.2. Following this first phase, we switched toa time-independent one-dimensional physical structure of G34 de-rived by van der Tak et al. (2013) as seen in Figure B1, and allowedthe time-dependent chemistry to continue to evolve independentlyat each value of the radius of the source. Figure 10 shows the computed fractional abundances for gaseousHDO and H O relative to the total proton density and their ratio as afunction of the radius of the source, at di ff erent times following theIRDC stage. The computed values can be compared with the valuesthat best fit the observations, as listed in Table 2. The observationalvalues are given for two points in the table, the inner hot core andthe colder envelope, but these values are represented as areas in thefigures with their height referring to uncertainty and their length tothe length of the inner and outer regions. Note that observationalresults may not be constant as a function of radius, as shown forthe abundances in Sections 4.2.4 and 4.2.5.During the IRDC phase, water and HDO are present mainly onthe grain surfaces, with the water abundance ≈ − . Once we applythe physical profile of the source, the temperature in the inner re-gion, greater than ∼
100 K, is high enough to allow the rapid desorp-tion of H O and HDO, and a transition region is observed around6 × AU, which corresponds to ∼
100 K. Beyond 6 × AU,the reverse e ff ect is observed: molecules are slowly adsorbed ontograin surfaces depending on the radius, because the density of thesource is now higher than during the IRDC phase. The rate of ad-sorption is directly proportional to the density, and since the densityis higher for small radii, the gaseous molecules are adsorbed morequickly closer to the transition region. This e ff ect is clearly seen attimes of 10 yrs and longer. While the gas-phase water fractionalabundance predicted by the chemical model in the inner core (ra-dius ≤ − to 10 − relative to thetotal proton density, in the colder envelope, the water abundanceslie between a few × − and 10 − depending on the radius and thetime. This dependence also holds for HDO, which possesses aninner-core abundance between 10 − and 10 − , and an outer abun-dance between a few × − and 10 − .In addition to these gas-grain interactions, chemical reactionsare also occurring. In the inner core, gaseous water is mainly de-stroyed by reactions with atomic hydrogen: H + H O −→ OH + ortho- H , and H + H O −→ OH + para- H . However, water is e ffi -ciently reformed by the reverse reactions, so its abundance does notchange significantly. In the same region, HDO is also mainly de- t IRDC + 1 yrt
IRDC + 10 yrt IRDC + 10 yrt IRDC + 10 yrt IRDC + 1.8x10 yrt IRDC + 5.6x10 yr H OHDO
Figure 10.
Top and center: calculated gas-phase abundances of H O andHDO relative to the total density of protons. Gray areas show observationalvalues and uncertainties of H O and HDO, observed in the hot inner coreand the colder envelope. Bottom: HDO / H O gas-phase abundance ratio.Gray areas show observational values and uncertainties of HDO / H O gas-phase abundance ratio observed in the hot inner core and the colder enve-lope (see Sections 4.2.2 and 4.2.3 for information about uncertainties). Bothabundances are plotted as a function of the radius of the source. The resultsare time dependent, and the colors and types of lines correspond to di ff erentvalues after the first initial phase: black solid lines ( t = t IRDC + t = t IRDC + yr), green short dashed lines ( t = t IRDC + yr),blue dashed dotted (1 dot) lines ( t = t IRDC + yr), orange dashed dot-ted (3 dots) lines ( t = t IRDC + . × yr), and purple long dashed lines( t = t IRDC + . × yr). (A color version of this figure is available in theonline journal.)c (cid:13)000
Top and center: calculated gas-phase abundances of H O andHDO relative to the total density of protons. Gray areas show observationalvalues and uncertainties of H O and HDO, observed in the hot inner coreand the colder envelope. Bottom: HDO / H O gas-phase abundance ratio.Gray areas show observational values and uncertainties of HDO / H O gas-phase abundance ratio observed in the hot inner core and the colder enve-lope (see Sections 4.2.2 and 4.2.3 for information about uncertainties). Bothabundances are plotted as a function of the radius of the source. The resultsare time dependent, and the colors and types of lines correspond to di ff erentvalues after the first initial phase: black solid lines ( t = t IRDC + t = t IRDC + yr), green short dashed lines ( t = t IRDC + yr),blue dashed dotted (1 dot) lines ( t = t IRDC + yr), orange dashed dot-ted (3 dots) lines ( t = t IRDC + . × yr), and purple long dashed lines( t = t IRDC + . × yr). (A color version of this figure is available in theonline journal.)c (cid:13)000 , 000–000 A. Coutens, C. Vastel, U. Hincelin et al. stroyed by reactions with atomic hydrogen: H + HDO −→ OH + HD,H + HDO −→ OD + ortho- H , and H + HDO −→ OD + para- H .Although HDO is also reformed by the reverse reactions, these pro-cesses are su ffi ciently slower than the destruction reactions that theHDO abundance decreases, with an e ffi ciency depending on the lo-cal temperature. This is indicated by the dashed lines in the upperpanel of Figure 10, particularly within a radius of 1000 AU. Thus,we observe a general decrease of the HDO / H O ratio in the hotinner core as a function of time.In the colder envelope, at larger radii, the H O gas phase abun-dance is reduced due to adsorption, as discussed above, and ion-molecule reactions, particularly the reaction with HCO + , whichforms H O + and CO. Before t = t IRDC + yrs, HCO + mainlyreacts with carbon atoms, and after this time, the carbon atom abun-dance is low enough to allow an increase of the HCO + abundancethrough ion-molecule reactions involving CO. Although HDO alsoreacts with HCO + , it is partially reformed by ion-molecule reac-tions involving H DO + , and dissociative recombination of H DO + with an electron. H O is also reformed by reactions involvingH O + , but not as e ffi ciently as HDO. At later times, the abundancesof HDO + and H DO + are increased, while the ones of H O + andH O + are decreased, so that the HDO / H O abundance ratio in-creases.At 10 AU, next to the transition region, the temperature anddensity are respectively equal to 80 K and 10 cm − . Here, there isa complex competition between the formation of HDO and H O inthe gas phase and the adsorption and desorption of these molecules.For this reason, we get temporarily a peak in the HDO / H O ra-tio around 10 and 10 yrs (respectively the green and blue peak).The main gas phase reactions involved are the following: H O + andH DO + react with DCN, DNC, HCN, and HNC, which form HDOand H O. Besides, after 10 yrs, H CO plays also a role: it is slowlyreleased from the grain surface, and reacts e ffi ciently with OH andOD to form respectively H O and HDO. However, at this tempera-ture and density, adsorption of HDO and H O is still quite e ffi cient,and removes a part of these molecules from the gas phase.If we compare the computed abundances of water and HDOwith the observational values, seen as gray areas in Figure 10, theH O abundances are in good agreement in both the hot inner coreand the colder outer envelope. This also holds true for HDO inthe colder envelope; however, our model does not produce enoughHDO in the hot inner core at all times. Specifically, our values arefive to fifty times less than those indicated by the observations, de-pending on the time and the radius. Given the low abundance ofHDO in the hot inner core, our calculated gaseous HDO / H O ra-tio is lower than the observed one throughout this region, whilein the colder outer envelope, our ratio lies within the range of theobservational values at selected times. Note that the observationalabundances and ratio may not be constant as a function of radius inthe two regions, so more constraints are necessary to compare withthe model results.The HDO abundance profiles in the cold outer region fromFigure 10 favor the best-fit abundance profile for water from Figure7, which increases with radius in the cold envelope. A comparisonbetween the HDO profiles of these two figures leads to the bestagreement around t = t IRDC + yrs. This time corresponds tothe best-fit chemical age of Gerner et al. (2014): their high massprotostellar object stage, the stage just after the IRDC stage, lasts ∼ × yrs, and the following stage, the hot molecular core stage,lasts ∼ × yrs, which give a total similar to ours.We have tested the sensitivity of H O and HDO to the cosmic-ray ionization rate, using a value of 1 . × − s − , which is ten times higher than the standard rate. This value is close to the up-per limit derived in this source (see Section 4.2.4). Cosmic rays arethe main source of ions in clouds, and formation and destruction ofneutral species involve mainly reactions with charged species. As aconsequence, most of the molecules are sensitive to the cosmic-rayionization rate (Wakelam et al. 2010). Compared with our standardmodel, in the cold envelope, the gas phase H O abundance is de-creased by one order of magnitude at early times after the IRDCphase. Then, H O is reformed quite e ffi ciently so that the finalabundance is one order of magnitude higher than with our stan-dard model. In the same region, the HDO abundance is increasedby a factor 10 to 100 depending on the time. The HDO / H O ratio isthen enhanced, and higher than the observational value by a factorof ∼
100 and < O abundance is slightly increased to a value ≥ − atall times. The HDO abundance is more sensitive at early times tothe cosmic ionization rate: it is firstly increased by a factor of 100,but then the value tends to decrease to the same one as in our stan-dard model. The HDO / H O ratio is also enhanced, up to a factor of100 at early times, but tends to decrease to the same value as in ourstandard model. In the IRDC phase and the cold envelope, gaseousH O is mainly formed by reactions involving H O + and destroyedby reactions with HCO + and C + , while H DO + is the main reac-tant involved in the formation of HDO. In the inner hot core, theabundances of H O and HDO are mainly changed due to OH andOD which are sensitive to the cosmic ray ionization rate (Wakelamet al. 2010).We have also tested the sensitivity of our modeling to the in-clusion of spin-state chemistry, and provide in Appendix C the re-sults of a simulation using our chemical network without consider-ing the spin states. Our main conclusion is that the gas phase HDOabundance is not only sensitive to the inclusion of spin-state chem-istry at low temperature, but also at high temperature, although thedi ff erence is less strong. In addition, the H O abundance is slightlysensitive at longer times to the spin-state chemistry in the cold en-velope region, but not in the hot inner core region. The overall ratioHDO / H O decreases if we take into account spin state chemistry,as it can be predicted based simply on the thermodynamics of pro-tonated ion-HD exchange reactions.
Below we compare our results for water and HDO both in the gasand on ice mantles with those of earlier studies. We first considerice mantles. Some of these studies include spin-state chemistrywhile others do not.Several groups theoretically studied deuteration of water instar forming regions (i.e. Cazaux et al. 2011; Aikawa et al. 2012;Sipil¨a et al. 2013; Taquet et al. 2013, 2014). These studies focuson low mass star-formation regions or cold conditions, and as aconsequence generally deal with lower temperatures and densitiesthan ours. However, considering the external region of the cold en-velope of our source, where the conditions are the closest to thesestudies (30 K and 10 cm − ), it is worth making some comparisonswith our ice results. Our HDO / H O ice ratio in the cold envelopevaries between 10 − and 10 − depending on the time. The largerthe time, the larger the ratio. We can compare our values to thosein Figures 11 and 12 in Sipil¨a et al. (2013) and Figure 8 in Taquetet al. (2013). These studies also include spin state chemistry. Ingeneral, we predict a lower HDO / H O ice ratio than these studies.Despite our slightly higher temperature, and multiple di ff erencesbetween our models, the initial ortho-to-para H ratio may be the c (cid:13) , 000–000 ater deuterium fractionation in the high-mass star-forming region G34.26 + main reason, since a higher value tends to decrease the deuteriumfractionation. Cazaux et al. (2011) and Aikawa et al. (2012), whodid not consider the spin state chemistry, predicted an HDO / H Oice ratio of ∼ .
01, which then can be considered as an upper limit.Aikawa et al. (2012) and Taquet et al. (2014) studied thedeuteration of molecules as a function of the radius of a formingprotostellar core. Here we can compare calculated HDO / H O ra-tios in the gas phase. Their temperature and density gradient alongthe radius is quite important, from ∼
10 K and ∼ cm − to sev-eral hundred Kelvin and 10 cm − , close to the range of condi-tions of our source. Note that these studies include a dynamicalphysical structure instead of a static structure. Despite the di ff er-ences between our model and these earlier studies, we obtain thesame qualitative pattern, in which the gas phase water abundance ishigher in the inner and hot region, while it is lower in the outer andcold region. In the outer region, the abundance is governed mainlyby the density, and as a consequence, tends to be lower when thedensity gets higher. Their HDO / H O ratio changes by one to twoorders of magnitude between the cold region and the hot region,and is higher in the colder region.
Ten lines of HDO and three lines of H
O covering a broad rangeof upper energy levels (22–204 K) were detected with the
Her-schel / HIFI instrument, the IRAM-30m telescope, and the CSO to-wards the high-mass star-forming region G34.26 + O throughout the envelope from thehot core to the colder regions. To reproduce the HDO line pro-files, it is necessary to assume an abundance jump at a temperaturehigher than 100 K ( ∼ / H O ratio is estimated to be about (3.5–7.5) × − in thehot core. It is in agreement with the value derived with the rota-tional diagram analysis of the IRAM-30m lines as well as with pre-vious studies (Jacq et al. 1990; Liu et al. 2013). In the colder gas,we determined the HDO / H O ratio to be about (1.0–2.2) × − ,including the uncertainties. Although radial variations of the waterdeuterium fractionation have already been observed in low-massprotostars (Coutens et al. 2012, 2013a, 2013b), this is the first timethat a decrease of the water deuterium fractionation in the warmerregions has been measured in a high-mass star-forming region. Fi-nally, we modeled the chemical evolution of G34 as a function ofits radius and showed that our model reproduces relatively well theobservational results that assumed an increase of the water abun-dance with radius in the cold regions (see Figures 7 and 10). Thecomparison of the chemical model and the observations favors an age of 10 years after the IRDC stage, which is consistent with theage derived for hot molecular cores by Gerner et al. (2014). ACKNOWLEDGMENTS
The authors are grateful to the anonymous referee for his / her use-ful and pertinent comments and suggestions. They thank K. Furuyaand Y. Aikawa for providing the initial chemical network of deuter-ated species and N. Flagey for providing the reduced HIFI dataof H O. They would also like to thank M. Hajigholi for fruitfuldiscussions regarding the source modeling. A. C. and C. V. thankPCMI for support of the Herschel HIFI project on deuterated wa-ter. C. M. P. acknowledges generous support from the Swedish Na-tional Space Board. Support for this work was also provided byNASA through an award issued by JPL / Caltech.This work is based on observations carried out with the HIFIinstrument onboard the
Herschel Space Observatory , the Institutde RadioAstronomie Millim´etrique (IRAM) 30m Telescope andthe Caltech Submillimeter Telescope (CSO).
Herschel is an ESAspace observatory with science instruments provided by European-led principal Investigator consortia and with important participationfrom NASA. HIFI has been designed and built by a consortium ofinstitutes and university departments from across Europe, Canadaand the United States under the leadership of SRON NetherlandsInstitute for Space Research, Groningen, The Netherlands and withmajor contributions from Germany, France and the US. Consor-tium members are: Canada: CSA, U.Waterloo; France: IRAP (for-merly CESR), LAB, LERMA, IRAM; Germany: KOSMA, MPIfR,MPS; Ireland, NUI Maynooth; Italy: ASI, IFSI-INAF, OsservatorioAstrofisico di Arcetri-INAF; Netherlands: SRON, TUD; Poland:CAMK, CBK; Spain: Observatorio Astron´omico Nacional (IGN),Centro de Astrobiolog´ıa (CSIC-INTA). Sweden: Chalmers Uni-versity of Technology - MC2, RSS & GARD; Onsala Space Ob-servatory; Swedish National Space Board, Stockholm University -Stockholm Observatory; Switzerland: ETH Zurich, FHNW; USA:Caltech, JPL, NHSC. IRAM is supported by INSU / CNRS (France),MPG (Germany) and IGN (Spain). The CSO is operated by the Cal-ifornia Institute of Technology under cooperative agreement withthe National Science Foundation (AST-0838261).
REFERENCES
Aikawa Y., Wakelam V., Hersant F., Garrod R. T., Herbst E., 2012,ApJ, 760, 40Albertsson T., Indriolo N., Kreckel H., Semenov D., CrabtreeK. N., Henning T., 2014, ApJ, 787, 44Albertsson T., Semenov D. A., Vasyunin A. I., Henning T., HerbstE., 2013, ApJS, 207, 27Alexander C. M. O. D., Bowden R., Fogel M. L., Howard K. T.,Herd C. D. K., Nittler L. R., 2012, Science, 337, 721Bergin E. A., Neufeld D. A., Melnick G. J., 1999, ApJL, 510,L145Beuther H., Churchwell E. B., McKee C. F., Tan J. C., 2007, Pro-tostars and Planets V, pp 165–180Bockel´ee-Morvan D., Biver N., Swinyard B., de Val-Borro M.,Crovisier J., Hartogh P., Lis D. C., Moreno R., Szutowicz S.,Lellouch E., Emprechtinger M., Blake G. A., Courtin R., et al.2012, A&A, 544, L15 c (cid:13) , 000–000 A. Coutens, C. Vastel, U. Hincelin et al.
Bockel´ee-Morvan D., Gautier D., Lis D. C., Young K., Keene J.,Phillips T., Owen T., Crovisier J., Goldsmith P. F., Bergin E. A.,Despois D., Wootten A., 1998, Icarus, 133, 147Campins H., Hargrove K., Pinilla-Alonso N., Howell E. S., KelleyM. S., Licandro J., Moth´e-Diniz T., Fern´andez Y., Zi ff er J., 2010,Nature, 464, 1320Caselli P., Hasegawa T. I., Herbst E., 1998, ApJ, 495, 309Caselli P., Keto E., Bergin E. A., Tafalla M., Aikawa Y., DouglasT., Pagani L., Y´ıld´ız U. A., van der Tak F. F. S., Walmsley C. M.,Codella C., Nisini B., Kristensen L. E., van Dishoeck E. F., 2012,ApJL, 759, L37Caselli P., Myers P. C., 1995, ApJ, 446, 665Cazaux S., Caselli P., Spaans M., 2011, ApJL, 741, L34Ceccarelli C., Castets A., Caux E., Hollenbach D., Loinard L.,Molinari S., Tielens A. G. G. M., 2000, A&A, 355, 1129Chavarr´ıa L., Herpin F., Jacq T., Braine J., Bontemps S., BaudryA., Marseille M., van der Tak F., et al. 2010, A&A, 521, L37Comito C., Schilke P., Rol ff s R., Lis D. C., Belloche A., BerginE. A., Phillips T. G., Bell T. A., Crockett N. R., et al. 2010, A&A,521, L38 + Coutens A., Vastel C., Cabrit S., Codella C., Kristensen L. E.,Ceccarelli C., van Dishoeck E. F., Boogert A. C. A., BottinelliS., Castets A., et al. 2013, A&A, 560, A39Coutens A., Vastel C., Caux E., Ceccarelli C., Bottinelli S.,Wiesenfeld L., Faure A., Scribano Y., Kahane C., 2012, A&A,539, A132Coutens A., Vastel C., Cazaux S., Bottinelli S., Caux E., Cecca-relli C., Demyk K., Taquet V., Wakelam V., 2013, A&A, 553,A75Daniel F., Dubernet M.-L., Grosjean A., 2011, A&A, 536, A76De Buizer J. M., Radomski J. T., Telesco C. M., Pi˜na R. K., 2003,ApJ, 598, 1127de Graauw T., Helmich F. P., Phillips T. G., Stutzki J., Caux E.,Whyborn N. D., Dieleman P., Roelfsema P. R., Aarts H., As-sendorp R., Bachiller R., Baechtold W., et al. 2010, A&A, 518,L6 + Dislaire V., Hily-Blant P., Faure A., Maret S., Bacmann A., PineauDes Forˆets G., 2012, A&A, 537, A20Dos Santos S. F., Kokoouline V., Greene C. H., 2007,J. Chem. Phys., 127, 124309Dulieu F., Amiaud L., Congiu E., Fillion J.-H., Matar E., MomeniA., Pirronello V., Lemaire J. L., 2010, A&A, 512, A30 + Emprechtinger M., Lis D. C., Rol ff s R., Schilke P., Monje R. R.,Comito C., Ceccarelli C., Neufeld D. A., van der Tak F. F. S.,2013, ApJ, 765, 61Faure A., Wiesenfeld L., Scribano Y., Ceccarelli C., 2011, MN-RAS, p. 2015Flagey N., Goldsmith P. F., Lis D. C., Gerin M., Neufeld D., Son-nentrucker P., De Luca M., Godard B., Goicoechea J. R., MonjeR., Phillips T. G., 2013, ApJ, 762, 11Flower D. R., Pineau des Forˆets G., Walmsley C. M., 2004, A&A,427, 887Flower D. R., Pineau Des Forˆets G., Walmsley C. M., 2006a,A&A, 456, 215Flower D. R., Pineau Des Forˆets G., Walmsley C. M., 2006b,A&A, 449, 621Fraser H. J., Collings M. P., McCoustra M. R. S., Williams D. A.,2001, MNRAS, 327, 1165Furuya K., Aikawa Y., Tomida K., Matsumoto T., Saigo K.,Tomisaka K., Hersant F., Wakelam V., 2012, ApJ, 758, 86Garay G., Reid M. J., Moran J. M., 1985, ApJ, 289, 681Garay G., Rodriguez L. F., van Gorkom J. H., 1986, ApJ, 309, 553 Gaume R. A., Fey A. L., Claussen M. J., 1994, ApJ, 432, 648Gensheimer P. D., Mauersberger R., Wilson T. L., 1996, A&A,314, 281Gerin M., de Luca M., Goicoechea J. R., Herbst E., Falgarone E.,Godard B., Bell T. A., Coutens A., Ka´zmierczak M., Sonnen-trucker P., Black J. H., et al. 2010, A&A, 521, L16Gerner T., Beuther H., Semenov D., Linz H., Vasyunina T., BihrS., Shirley Y. L., Henning T., 2014, A&A, 563, A97Goldsmith P. F., Langer W. D., 1999, ApJ, 517, 209Hartogh P., Lis D. C., Bockel´ee-Morvan D., de Val-Borro M.,Biver N., K¨uppers M., Emprechtinger M., Bergin E. A., Cro-visier J., Rengel M., Moreno R., Szutowicz S., Blake G. A.,2011, Nature, 478, 218Hasegawa T. I., Herbst E., 1993, MNRAS, 261, 83Hasegawa T. I., Herbst E., Leung C. M., 1992, ApJS, 82, 167Hatchell J., Fuller G. A., Millar T. J., 2001, A&A, 372, 281Hatchell J., Thompson M. A., Millar T. J., MacDonald G. H.,1998, A&AS, 133, 29Herpin F., Chavarr´ıa L., van der Tak F., Wyrowski F., vanDishoeck E. F., Jacq T., Braine J., Baudry A., Bontemps S., Kris-tensen L., 2012, A&A, 542, A76Hersant F., Wakelam V., Dutrey A., Guilloteau S., Herbst E., 2009,A&A, 493, L49Hincelin U., Wakelam V., Hersant F., Guilloteau S., Loison J. C.,Honvault P., Troe J., 2011, A&A, 530, A61Hogerheijde M. R., Bergin E. A., Brinch C., Cleeves L. I., Fo-gel J. K. J., Blake G. A., Dominik C., Lis D. C., Melnick G.,Neufeld D., Pani´c O., Pearson J. C., Kristensen L., Yıldız U. A.,van Dishoeck E. F., 2011, Science, 334, 338Hogerheijde M. R., van der Tak F. F. S., 2000, A&A, 362, 697Hollenbach D., Kaufman M. J., Bergin E. A., Melnick G. J., 2009,ApJ, 690, 1497Honvault P., Jorfi M., Gonz´alez-Lezana T., Faure A., Pagani L.,2011a, Physical Review Letters, 107, 023201Honvault P., Jorfi M., Gonz´alez-Lezana T., Faure A., PaganiL., 2011b, Physical Chemistry Chemical Physics (IncorporatingFaraday Transactions), 13, 19089Hugo E., Asvany O., Schlemmer S., 2009, J. Chem. Phys., 130,164302Ioppolo S., Cuppen H. M., Romanzin C., van Dishoeck E. F., Lin-nartz H., 2008, ApJ, 686, 1474Jacq T., Henkel C., Walmsley C. M., Jewell P. R., Baudry A.,1988, A&A, 199, L5Jacq T., Walmsley C. M., Henkel C., Baudry A., Mauersberger R.,Jewell P. R., 1990, A&A, 228, 447Jensen M. J., Bilodeau R. C., Safvan C. P., Seiersen K., AndersenL. H., Pedersen H. B., Heber O., 2000, ApJ, 543, 764Jørgensen J. K., van Dishoeck E. F., 2010, ApJL, 710, L72Kong S., Caselli P., Tan J. C., Wakelam V., 2013, ArXiv1312.0971Kristensen L. E., van Dishoeck E. F., Bergin E. A., Visser R.,Yıldız U. A., San Jose-Garcia I., Jørgensen J. K., Herczeg G. J.,Johnstone D., Wampfler S. F., et al. 2012, A&A, 542, A8Kristensen L. E., Visser R., van Dishoeck E. F., Yıldız U. A., DotyS. D., Herczeg G. J., Liu F.-C., Parise B., Jørgensen J. K., et al.2010, A&A, 521, L30 + Kuchar T. A., Bania T. M., 1994, ApJ, 436, 117K¨uppers M., O’Rourke L., Bockel´ee-Morvan D., Zakharov V.,Lee S., von Allmen P., Carry B., Teyssier D., Marston A., M¨ullerT., Crovisier J., Barucci M. A., Moreno R., 2014, Nature, 505,525Lis D. C., Biver N., Bockel´ee-Morvan D., Hartogh P., Bergin c (cid:13) , 000–000 ater deuterium fractionation in the high-mass star-forming region G34.26 + E. A., Blake G. A., Crovisier J., de Val-Borro M., Jehin E.,K¨uppers M., Manfroid J., Moreno R., Rengel M., Szutowicz S.,2013, ApJL, 774, L3Lis D. C., Phillips T. G., Goldsmith P. F., Neufeld D. A., HerbstE., Comito C., Schilke P., M¨uller H. S. P., Bergin E. A., GerinM., Bell T. A., Emprechtinger M., et al. 2010, A&A, 521, L26Liu F., Parise B., Kristensen L., Visser R., van Dishoeck E. F.,G¨usten R., 2011, A&A, 527, A19 + Liu F.-C., Parise B., Wyrowski F., Zhang Q., G¨usten R., 2013,A&A, 550, A37Liu T., Wu Y., Zhang H., 2013, ApJ, 776, 29MacDonald G. H., Gibb A. G., Habing R. J., Millar T. J., 1996,A&AS, 119, 333Marquette J. B., Rebrion C., Rowe B. R., 1988, J. Chem. Phys.,89, 2041McCall B. J., Huneycutt A. J., Saykally R. J., Djuric N., DunnG. H., Semaniak J., Novotny O., Al-Khalili A., Ehlerding A.,Hellberg F., Kalhori S., Neau A., Thomas R. D., Paal A.,¨Osterdahl F., Larsson M., 2004, Phys. Rev. A, 70, 052716Miyauchi N., Hidaka H., Chigai T., Nagaoka A., Watanabe N.,Kouchi A., 2008, Chemical Physics Letters, 456, 27Mokrane H., Chaabouni H., Accolla M., Congiu E., Dulieu F.,Chehrouri M., Lemaire J. L., 2009, ApJL, 705, L195Mookerjea B., Casper E., Mundy L. G., Looney L. W., 2007, ApJ,659, 447Mottram J. C., van Dishoeck E. F., Schmalzl M., Kristensen L. E.,Visser R., Hogerheijde M. R., Bruderer S., 2013, A&A, 558,A126M¨uller H. S. P., Coutens A., Walters A., Grabow J.-U., SchlemmerS., 2011, Journal of Molecular Spectroscopy, 267, 100M¨uller H. S. P., Schl¨oder F., Stutzki J., Winnewisser G., 2005,Journal of Molecular Structure, 742, 215Neill J. L., Wang S., Bergin E. A., Crockett N. R., Favre C., PlumeR., Melnick G. J., 2013, ApJ, 770, 142Oba Y., Miyauchi N., Hidaka H., Chigai T., Watanabe N., KouchiA., 2009, ApJ, 701, 464¨Oberg K. I., Linnartz H., Visser R., van Dishoeck E. F., 2009, ApJ,693, 1209Ossenkopf V., Henning T., 1994, A&A, 291, 943Ott S., 2010, in Astronomical Data Analysis Software and Sys-tems XIX Vol. 434 of Astronomical Society of the Pacific Con-ference Series, The Herschel Data Processing System HIPE andPipelines Up and Running Since the Start of the Mission. p. 139Pagani L., Salez M., Wannier P. G., 1992, A&A, 258, 479Pagani L., Vastel C., Hugo E., Kokoouline V., Greene C. H., Bac-mann A., Bayet E., Ceccarelli C., Peng R., Schlemmer S., 2009,A&A, 494, 623Pardo J. R., Cernicharo J., Herpin F., Kawamura J., Kooi J.,Phillips T. G., 2001, ApJ, 562, 799Parise B., Caux E., Castets A., Ceccarelli C., Loinard L., TielensA. G. G. M., Bacmann A., Cazaux S., Comito C., Helmich F.,Kahane C., Schilke P., van Dishoeck E., Wakelam V., WaltersA., 2005, A&A, 431, 547Persson C. M., Olofsson A. O. H., Koning N., Bergman P.,Bernath P., Black J. H., Frisk U., Geppert W., Hasegawa T. I.,Hjalmarson Å., Kwok S., Larsson B., Lecacheux A., NummelinA., Olberg M., Sandqvist A., Wirstr¨om E. S., 2007, A&A, 476,807Persson M. V., Jørgensen J. K., van Dishoeck E. F., 2012, A&A,541, A39Persson M. V., Jørgensen J. K., van Dishoeck E. F., 2013, A&A,549, L3 Persson M. V., Jørgensen J. K., van Dishoeck E. F., Harsono D.,2014, A&A, 563, A74Pickett H. M., Poynter R. L., Cohen E. A., Delitsky M. L., PearsonJ. C., M¨uller H. S. P., 1998, J. Quant. Spec. Radiat. Transf., 60,883Pilbratt G. L., Riedinger J. R., Passvogel T., Crone G., Doyle D.,Gageur U., Heras A. M., Jewell C., Metcalfe L., Ott S., SchmidtM., 2010, A&A, 518, L1 + Prasad S. S., Tarafdar S. P., 1983, ApJ, 267, 603Reid M. J., Ho P. T. P., 1985, ApJL, 288, L17Roberts H., Herbst E., 2002, A&A, 395, 233Roelfsema P. R., Helmich F. P., Teyssier D., Ossenkopf V., Mor-ris P., Olberg M., Shipman R., Risacher C., Akyilmaz M., As-sendorp R., Avruch I. M., Beintema D., Biver N., Boogert A.,Borys C., Braine J., et al. 2012, A&A, 537, A17Rol ff s R., Schilke P., Comito C., Bergin E. A., van der Tak F. F. S.,Lis D. C., Qin S.-L., Menten K. M., G¨usten R., et al. 2010, A&A,521, L46Semenov D., Hersant F., Wakelam V., Dutrey A., Chapillon E.,Guilloteau S., Henning T., Launhardt R., Pi´etu V., Schreyer K.,2010, A&A, 522, A42Sewiło M., Churchwell E., Kurtz S., Goss W. M., Hofner P., 2004,ApJ, 605, 285Sewiło M., Churchwell E., Kurtz S., Goss W. M., Hofner P., 2011,ApJS, 194, 44Shen C. J., Greenberg J. M., Schutte W. A., van Dishoeck E. F.,2004, A&A, 415, 203Sipil¨a O., Caselli P., Harju J., 2013, A&A, 554, A92Sridharan T. K., Beuther H., Saito M., Wyrowski F., Schilke P.,2005, ApJL, 634, L57Taquet V., Charnley S. B., Sipil¨a O., 2014, ApJ, 791, 1Taquet V., Peters P. S., Kahane C., Ceccarelli C., L´opez-SepulcreA., Toubin C., Duflot D., Wiesenfeld L., 2013, A&A, 550, A127Tielens A. G. G. M., Hagen W., 1982, A&A, 114, 245Turner B. E., Balick B., Cudaback D. D., Heiles C., Boyle R. J.,1974, ApJ, 194, 279van der Tak F. F. S., Chavarr´ıa L., Herpin F., Wyrowski F., Walm-sley C. M., van Dishoeck E. F., Benz A. O., Bergin E. A., CaselliP., Hogerheijde M. R., Johnstone D., Kristensen L. E., Liseau R.,Nisini B., Tafalla M., 2013, A&A, 554, A83van der Tak F. F. S., Walmsley C. M., Herpin F., Ceccarelli C.,2006, A&A, 447, 1011van Dishoeck E. F., Herbst E., Neufeld D. A., 2013, ChemicalReviews, 113, 9043Wakelam V., Herbst E., Le Bourlot J., Hersant F., Selsis F., Guil-loteau S., 2010, A&A, 517, A21Walmsley C. M., Flower D. R., Pineau des Forˆets G., 2004, A&A,418, 1035Watanabe N., Kouchi A., 2008, Progress In Surface Science, 83,439Watt S., Mundy L. G., 1999, ApJS, 125, 143Wilson T. L., 1999, Reports on Progress in Physics, 62, 143Wood D. O. S., Churchwell E., 1989, ApJS, 69, 831Wyrowski F., G¨usten R., Menten K. M., Wiesemeyer H., Klein B.,2012, A&A, 542, L15Zinnecker H., Yorke H. W., 2007, ARA&A, 45, 481 c (cid:13) , 000–000 A. Coutens, C. Vastel, U. Hincelin et al.
APPENDIX A:
HERSCHEL / HIFI OBSERVATIONS
The observing IDs of the
Herschel / HIFI data are listed in Table A1.
APPENDIX B: NON-LTE SPHERICAL RADIATIVETRANSFER MODELINGB1 Density and temperature profiles
The density and temperature profiles used for the model of theHDO and H
O lines are shown in Figure B1.
B2 Velocity field profiles
The radial velocity profile ( (cid:51) r ) and the turbulence width (Doppler b-parameter, db ) are required inputs for the RATRAN radiative trans-fer modeling. We do not have direct information on them, exceptthat inward motions are necessary to reproduce the inverse P-Cygniprofile observed on the HDO 1 , –1 , fundamental line observedwith HIFI. Indeed, with a static envelope, the model predicts an ab-sorbing component at the local standard velocity of rest, V LSR = − , whereas the absorption is observed at 61 km s − only. Con-sequently, it is not possible to reproduce the line profile of this HDOfundamental transition with a static envelope and an unrelated ab-sorbing layer situated on the line of sight. The radial velocity wasthen fixed at -3 km s − in the cold outer regions.To estimate the (cid:51) r and db profiles throughout the envelope, weassumed abundance profiles with a jump and proceeded as follows: i) First, we assume an initial profile for the radial velocity and theDoppler b-parameter. Then we run a grid of models with variousinner ( T > T j ) and outer ( T < T j ) abundances. ii) If the predictedline profiles do not fit at all the data, we choose the model thatgives the best agreement with respect to the intensities. Keepingthe same abundances as this model, we modify the velocity profilesto obtain better agreement. iii)
Then we run another grid of modelswith these new velocity profiles and go back to step (ii) if necessaryand so on.Models with infall profiles were attempted but, as shown inFigure B3, some of the HDO lines (509, 600, 849, and 919 GHz)are shifted in velocity with respect to the observations when thistype of profile is taken into account. On the contrary, models withoutward motions in the inner regions give a good agreement for thedi ff erent HDO lines. These outward motions could be produced bystellar winds or outflows.To limit the number of free parameters in the study, we onlyconsidered constant (cid:51) r and db values in the inner and outer regions.Based on the widths of the di ff erent lines, the (cid:51) r parameter wasdetermined to be about 4 km s − in the inner regions in expansionand about -3 km s − in the infalling outer regions. The Doppler b-parameter is estimated to be about 2.5 km s − in the outer region.This constraint is based on the width of the absorption componentof the HDO line at 894 GHz. In the inner regions, the fit of theemission lines is slightly better with db ∼ − . The db pa-rameter seems to decrease from the outer to the inner regions, sim-ilarly to other studies in high mass sources (Caselli & Myers 1995;Herpin et al. 2012). To be able to reproduce the HDO and H Olines with a unique velocity profile for all the models with di ff erentjump temperatures , we delimited the change of the db and (cid:51) r val-ues between the inner and outer region at 100 K. The final velocityfields used in the paper are presented in Figure B2. We cannot how-ever exclude that di ff erent velocity fields (possibly more complex)would reproduce the lines. For the models assuming an increase G34.26+0.15 radius(AU)10 n ( H ) ( c m - ) T e m pe r a t u r e ( K ) Figure B1. H density (solid line) and temperature (dashed line) profilesfrom van der Tak et al. (2013). v r ( k m / s ) db ( k m / s ) Figure B2.
Radial velocity ( (cid:51) r , solid line) and Doppler b-parameter ( db ,dashed line) profiles assumed for the di ff erent models presented in the pa-per. of the abundance between 100 and 200 K (see Section 4.2.5), wealso checked that a radial velocity increasing gradually from -3 to4 km s − between 100 and 200 K would give similar results. B3 Modeling of the HDO lines with a two-jump modeling
In their study of Sgr B2(M), Comito et al. (2010) showed that amodel with two abundance jumps, one at 100 K and another at200 K, is necessary to reproduce the di ff erent HDO lines observedtowards this star-forming region. To check the influence on our re-sults, we ran grids with such assumptions for our modeling of G34.Several models give a good agreement with the data. However thefit to the data is not better than with one jump. It is extremely sim-ilar to the model shown in Figure 5. Unsurprisingly, the best-fitgives an inner abundance ( T >
200 K) of 2 × − , consistent witha modeling with a unique jump at a temperature of 200 K. The outerabundance ( T <
100 K) is estimated to be 8 × − , and the abun-dance between 100 and 200 K is only constrained by an upper limitof 1 × − . Indeed, with abundances higher than 1 × − , the pre-dicted intensities for the lines at 491, 600 and 919 GHz becometoo high compared with the observations. These results are consis- c (cid:13) , 000–000 ater deuterium fractionation in the high-mass star-forming region G34.26 + Table A1.
List of the
Herschel / HIFI obsIDsSpecies Frequency (GHz) Transition ObsID-A ObsID-B ObsID-C Observing programHDO 490.5966 2 , –1 , , –1 , , –2 , , –1 , , –0 , , –1 , O 547.6764 1 , –1 , O 1101.6983 1 , –0 , HDO 1 -0
894 GHz
40 50 60 702.53.03.54.04.5 T m b ( K ) HDO 1 -0
465 GHz
40 50 60 701.52.02.53.03.54.0
HDO 1 -1
81 GHz
40 50 60 70-0.20.00.20.40.6
HDO 3 -2
226 GHz
40 50 60 70-0.50.00.51.01.52.0
HDO 2 -2
242 GHz
40 50 60 70-0.50.00.51.01.52.0
HDO 2 -1
491 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.3 T m b ( K ) HDO 1 -1
509 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.3 HDO 2 -2
600 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.4 HDO 2 -1
849 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.4 HDO 2 -1
919 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.4 Figure B3.
Black solid line : HDO lines observed with HIFI, IRAM, and CSO.
Red dashed line : Modeling with an infall profile (free-fall model with M = M (cid:12) ). Green dotted line : Modeling with a velocity profile with inward motions in the outer regions and outward motions in the inner regions. tent with a modeling having a single jump, and the two-jump as-sumption does not improve the fit of the model results to the data.Although we cannot exclude that a double abundance jump occurshere, we estimate that the hypothesis of one jump is probably morereasonable. Indeed, chemical reasons for a three-step abundanceprofile are unclear. All the HDO trapped in the grain mantles shoulddesorb thermally at approximately 100 K. Additional formation ofwater is possible into the gas phase at higher temperatures. But dueto the relatively high temperatures in the hot core, the formation ofdeuterated water should be negligible. In addition, the HDO abun-dances derived between 100 and 200 K are very low ( ≤ × − )comparatively to the abundance above 200 K ( ∼ × − ). Uncer-tainty in the temperature profile in the inner region with a singlejump produced by the desorption from grain mantles or a gradualincrease of the abundance probably provide a better explanationthan a two jump model. B4 Summary of the di ff erent models Table B1 summarizes the di ff erent types of models shown in thepaper. APPENDIX C: SENSITIVITY OF OUR MODELING TOTHE INCLUSION OF SPIN-STATE CHEMISTRY
We studied the sensitivity of our results to the chemistry involv-ing the spin states of H , D , H + , H D + , D H + , and D + . Using thesame physical condition as described in section 5.1, we modeledthe chemical evolution of our source using our basic chemical net-work adapted from Aikawa et al. (2012) and Furuya et al. (2012),without taking into account spin states. Figure C1 presents calcu-lated gaseous HDO and H O abundances, and HDO / H O ratios,with and without spin-state chemistry.The gaseous HDO abundance is higher throughout the sourceby a factor 2 (inner core) to 4 (cold envelope), if we neglect spinstate chemistry. We expected this result since the inclusion of spinstates tends to reduce deuterium fractionation by populating orthospin states of H . The e ff ect is more pronounced in the cold enve-lope because the endothermic reactions involving deuterated ionsand o-H are greatly enhanced in rate. The di ff erences are time de-pendent, and are less pronounced as the system progresses from theIRDC phase to the end of the simulation because o-H is convertedin p-H .In the hot inner region ( ≤ AU) both models give the sameresults for gas-phase water. In the cold envelope, there is more gasphase water if we take into account spin-state chemistry. Contraryto HDO, the di ff erences grow larger with time, and this trend is par- c (cid:13) , 000–000 A. Coutens, C. Vastel, U. Hincelin et al.
HDO 1 -0
894 GHz
40 50 60 702.83.03.23.43.63.84.04.2 T m b ( K ) HDO 1 -0
465 GHz
40 50 60 701.52.02.53.03.54.0
HDO 1 -1
81 GHz
40 50 60 70-0.20.00.20.40.6
HDO 3 -2
226 GHz
40 50 60 70-0.50.00.51.01.52.0
HDO 2 -2
242 GHz
40 50 60 70-0.50.00.51.01.52.0
HDO 2 -1
491 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.40.50.6 T m b ( K ) HDO 1 -1
509 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.3 HDO 2 -2
600 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.40.5 HDO 2 -1
849 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.40.50.6 HDO 2 -1
919 GHz
40 50 60 70v
LSR (km s -1 )-0.20.00.20.40.6 Figure B4.
Black solid line : HDO lines observed with HIFI, IRAM, and CSO.
Red dashed line : Modeling for a jump temperature T j =
120 K, an innerabundance X in = × − and an outer abundance X out = × − . HDO 1 -0
894 GHz
40 50 60 702.53.03.54.04.5 T m b ( K ) HDO 1 -0
465 GHz
40 50 60 701.52.02.53.03.54.0
HDO 1 -1
81 GHz
40 50 60 70-0.20.00.20.40.6
HDO 3 -2
226 GHz
40 50 60 70-0.50.00.51.01.52.0
HDO 2 -2
242 GHz
40 50 60 70-0.50.00.51.01.52.0
HDO 2 -1
491 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.40.5 T m b ( K ) HDO 1 -1
509 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.3 HDO 2 -2
600 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.40.5 HDO 2 -1
849 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.40.5 HDO 2 -1
919 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.40.5 Figure B5.
Black solid line : HDO lines observed with HIFI, IRAM, and CSO.
Red dashed line : Modeling for a jump temperature T j =
150 K, an innerabundance X in = × − and an outer abundance X out = × − . HDO 1 -0
894 GHz
40 50 60 702.53.03.54.04.5 T m b ( K ) HDO 1 -0
465 GHz
40 50 60 701.52.02.53.03.54.0
HDO 1 -1
81 GHz
40 50 60 70-0.20.00.20.40.6
HDO 3 -2
226 GHz
40 50 60 70-0.50.00.51.01.52.0
HDO 2 -2
242 GHz
40 50 60 70-0.50.00.51.01.52.0
HDO 2 -1
491 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.4 T m b ( K ) HDO 1 -1
509 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.3 HDO 2 -2
600 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.4 HDO 2 -1
849 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.4 HDO 2 -1
919 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.4 Figure B6.
Black solid line : HDO lines observed with HIFI, IRAM, and CSO.
Red dashed line : Modeling for a jump temperature T j =
180 K, an innerabundance X in = × − and an outer abundance X out = × − . c (cid:13) , 000–000 ater deuterium fractionation in the high-mass star-forming region G34.26 + Table B1.
List of the models presented in the paper.Model Abundance profile Sections & Figures1 1 jump T j =
100 K HDO : Sect. 4.2.2, Fig. 42 1 jump T j =
120 K HDO : Sect. 4.2.2, Fig. B43 1 jump T j =
150 K HDO : Sect. 4.2.2, Fig. B5H
O : Sect. 4.2.3, Fig. B84 1 jump T j =
180 K HDO : Sect. 4.2.2, Fig. B6H
O : Sect. 4.2.3, Fig. B95 1 jump T j =
200 K HDO : Sect. 4.2.2, Fig. 5H
O : Sect. 4.2.3, Fig. 65 1 jump T j =
220 K HDO : Sect. 4.2.2, Fig. B7H
O : Sect. 4.2.3, Fig. B106 2 jumps T j1 =
100 K, T j2 =
200 K HDO : Appx. B37 Constant inner abundance ( T ≥
200 K) and HDO : Sect. 4.2.4, Fig. 8decrease of the outer abundance from the H
O : Sect. 4.2.4, Fig. 9cold to the warm regions (see Fig. 7)8 Constant inner abundance, gradual HDO : Sect. 4.2.5, Fig. 8increase of the abundance at the transition H
O : Sect. 4.2.5, Fig. 9hot core / cold envelope (100–200 K) and decreaseof the outer abundance from the cold regionsto the regions at 100 K (see Fig. 7) HDO 1 -0
894 GHz
40 50 60 702.53.03.54.04.5 T m b ( K ) HDO 1 -0
465 GHz
40 50 60 701.52.02.53.03.54.0
HDO 1 -1
81 GHz
40 50 60 70-0.20.00.20.40.6
HDO 3 -2
226 GHz
40 50 60 70-0.50.00.51.01.52.0
HDO 2 -2
242 GHz
40 50 60 70-0.50.00.51.01.52.0
HDO 2 -1
491 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.3 T m b ( K ) HDO 1 -1
509 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.3 HDO 2 -2
600 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.4 HDO 2 -1
849 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.4 HDO 2 -1
919 GHz
40 50 60 70v
LSR (km s -1 )-0.10.00.10.20.30.4 Figure B7.
Black solid line : HDO lines observed with HIFI, IRAM, and CSO.
Red dashed line : Modeling for a jump temperature T j =
220 K, an innerabundance X in = × − and an outer abundance X out = × − . p-H O 3 -2
203 GHz
40 50 60 70v
LSR (km s -1 )-0.50.00.51.01.52.0 T m b ( K ) p-H O 1 -0
40 50 60 70v
LSR (km s -1 )4.55.05.56.06.5 o-H O 1 -1
548 GHz
40 50 60 70v
LSR (km s -1 )0.80.91.01.11.21.31.4 Figure B8.
Black solid line : H
O lines observed with HIFI and IRAM.
Red dashed line : Modeling for a jump temperature T j =
150 K, an inner abundance X in = × − and an outer abundance X out = × − .c (cid:13)000
150 K, an inner abundance X in = × − and an outer abundance X out = × − .c (cid:13)000 , 000–000 A. Coutens, C. Vastel, U. Hincelin et al. p-H
O 3 -2
203 GHz
40 50 60 70v
LSR (km s -1 )-0.50.00.51.01.52.0 T m b ( K ) p-H O 1 -0
40 50 60 70v
LSR (km s -1 )4.55.05.56.06.5 o-H O 1 -1
548 GHz
40 50 60 70v
LSR (km s -1 )0.80.91.01.11.21.31.4 Figure B9.
Black solid line : H
O lines observed with HIFI and IRAM.
Red dashed line : Modeling for a jump temperature T j =
180 K, an inner abundance X in = × − and an outer abundance X out = × − . p-H O 3 -2
203 GHz
40 50 60 70v
LSR (km s -1 )-0.50.00.51.01.52.0 T m b ( K ) p-H O 1 -0
40 50 60 70v
LSR (km s -1 )4.55.05.56.06.5 o-H O 1 -1
548 GHz
40 50 60 70v
LSR (km s -1 )0.80.91.01.11.21.31.4 Figure B10.
Black solid line : H
O lines observed with HIFI and IRAM.
Red dashed line : Modeling for a jump temperature T j =
220 K, an inner abundance X in = × − and an outer abundance X out = × − . ticularly strong in the transition area ( (cid:39) AU). In this area, thetemperature is low enough to not allow e ffi cient desorption, and thedensity is high so adsorption is important. In consequence, water isprimarily on the grain surfaces. In this situation, gas phase reac-tions are the principal pathway to form gas phase water, mainlythrough H O + : the spin-state chemistry reduces the formation ofH D + from H + , and the extra amount of H + helps to produce moreH O + through successive reactions involving OH, H O + , and H .Globally, since the HDO abundance is lower and the H Oabundance is higher if we take into account spin-state chemistry,the HDO / H O ratio is lower. The decrease is within a factor of 10,and depends on the radius and the time. The di ff erences are lesscritical at high temperature than at low temperature, but are notnegligible compared with the variation of our measured ratios. c (cid:13) , 000–000 ater deuterium fractionation in the high-mass star-forming region G34.26 + t IRDC + 1 yrt
IRDC + 10 yrt IRDC + 10 yrt IRDC + 10 yrt IRDC + 1.8x10 yrt IRDC + 5.6x10 yr Figure C1.
Calculated gas-phase abundances of HDO (top panel) and H O(middle panel) vs radius with (solid lines) and without (dashed lines) spin-state chemistry. The bottom panel shows the corresponding HDO / H O ra-tios. The results are time dependent, and the colors correspond to di ff erentvalues after the first initial phase as in figure 10.c (cid:13)000