A high resolution study of complex organic molecules in hot cores
Hannah Calcutt, Serena Viti, Claudio Codella, Maria T. Beltrán, Francesco Fontani, Paul M. Woods
aa r X i v : . [ a s t r o - ph . S R ] J u l Mon. Not. R. Astron. Soc. , 1–20 (2013) Printed 26 August 2018 (MN L A TEX style file v2.2)
A high resolution study of complex organic molecules inhot cores
Hannah Calcutt ∗ , Serena Viti , Claudio Codella , Maria T. Beltr´an , Francesco Fontani ,Paul M. Woods , Department of Physics and Astronomy, University College London, WC1E 6BT, London, UK INAF, Osservatorio Astrofisico di Arcetri, Largo Enrico Fermi 5, I-50125 Firenze, Italy Astrophysics Research Centre, School of Mathematics & Physics, Queen’s University Belfast, Belfast BT7 1NN, UK
ABSTRACT
We present the results of a line identification analysis using data from theIRAM Plateau de Bure Inferferometer, focusing on six massive star-forming hotcores: G31.41+0.31, G29.96-0.02, G19.61-0.23, G10.62-0.38, G24.78+0.08A1 andG24.78+0.08A2. We identify several transitions of vibrationally excited methyl for-mate (HCOOCH ) for the first time in these objects as well as transitions of other com-plex molecules, including ethyl cyanide (C H CN), and isocyanic acid (HNCO). Wealso postulate a detection of one transition of glycolaldehyde (CH (OH)CHO) in twonew hot cores. We find G29.96-0.02, G19.61-0.23, G24.78+0.08A1 and G24.78+0.08A2to be chemically very similar. G31.41+0.31, however, is chemically different: it man-ifests a larger chemical inventory and has significantly larger column densities. Wesuggest that it may represent a different evolutionary stage to the other hot cores inthe sample, or it may surround a star with a higher mass. We derive column densitiesfor methyl formate in G31.41+0.31, using the rotation diagram method, of 4 × cm − and a T rot of ∼
170 K. For G29.96-0.02, G24.78+0.08A1 and G24.78+0.08A2,glycolaldehyde, methyl formate and methyl cyanide all seem to trace the same ma-terial and peak at roughly the same position towards the dust emission peak. ForG31.41+0.31, however, glycolaldehyde shows a different distribution to methyl for-mate and methyl cyanide and seems to trace the densest, most compact inner part ofhot cores.
Key words:
ISM: molecules — ISM: abundances — Astrochemistry — stars: forma-tion — stars: protostars — line: identification
The process that leads to the formation of a high-mass star,whether it is by accretion (McKee & Tan 2002, 2003), orcoalescence and accretion (Bonnell et al. 1998) is extremelyfast, ∼ –10 yr, and will, according to some evolution-ary models, depend on the final mass of the star (e.g.,Bernasconi & Maeder 1996). The contraction of a core froma low to a very high density ( ∼ cm − ) occurs very quicklyand the star will reach the Zero Age Main Sequence whilestill embedded (Palla & Stahler 1993). This, together withthe fact that massive stars tend to form in association, makesthe determination of the early stages of the evolution of amassive star via observations of Spectral Energy Distribu-tions (SEDs) a rather difficult task. ∗ E-mail: [email protected]
A common way of observing massive young stellar ob-jects (YSOs) has been via the detection of hypercompactand ultracompact HII regions (UCHII): regions of densegas ionized by the newly formed star. However, ionised re-gions trace relatively late evolutionary stages. On the otherhand, the earliest phases are well traced by hot cores, whichare small ( ∼ − –10 − pc), dense ( > cm − ), relativelywarm ( > K), optically thick (A v > mag), and transient( yr) condensations. Spectral surveys have revealed avery rich chemistry (see review by Herbst & van Dishoeck(2009); Garay & Lizano (1999)) which include high abun-dances of small saturated molecules (e.g. H O, NH ,H S, CH OH) as well as large organic species (CH CN,CH CHCN, CH CH CN, C H OH). Such high abundancesare believed to arise from the sublimation of ice mantlesfrozen out onto the dust grains during the collapse of theparent cloud, which is induced by the nearby (proto)star. c (cid:13) Calcutt et al.
Once the protostar is born, the dust is heated andmolecules formed on the dust grain sublimate. This warm-up phase is most likely not instantaneous (Viti & Williams1999); (Viti et al. 2004); (Garrod et al. 2008), and there isexperimental evidence that desorption occurs in as many asfour stages in four distinct narrow temperature ranges (tem-perature bands). The proportion of each species that evap-orates in each band depends on the total amount presenton the grains as well as on the bonding properties of themolecule (Collings et al. 2004). Chemical models incorpo-rating such experimental results show that distinct chem-ical events occur at specific grain temperatures and thesediffer depending on the mass of the star (Viti et al. 2004).The chemistry of the hot cores will, therefore, reflect thistemperature-driven evolution and the evolutionary chem-istry can provide, in principle, a record of the collapse pro-cess as well as the ignition history of the star.Of course, since there is a chemical differentiation overtime due to a time-dependent increase of the dust tempera-ture, one also expects a space-dependent chemical differenti-ation due to the difference in dust temperature progressivelyfurther away from the star. The degeneracy between timeand spatial effects can only be solved by interferometric ob-servations and chemical modelling. A class of molecules thatcan be particularly useful to solve this degeneracy is that ofcomplex organic molecules (COMs) because (i) their detec-tion is a confirmation of the high density warm cores, as mostCOMs are not easily produced in the gas phase chemistryof dark clouds and, ii) most importantly, their emission hasbeen observed to be compact in extent in star forming re-gions outside the Galactic Center (Herbst & van Dishoeck2009; Garay & Lizano 1999) and therefore may trace themost central region of the hot core, close to the YSOs.In fact, many organic molecules have been detected in hotcores and have been used to trace different structures asso-ciated with star formation such as discs and maser activ-ity (e.g. Olmi et al. 1996; Cesaroni et al. 1998; Olmi et al.2003; Beltr´an et al. 2005, 2011). Multi-transition and multi-species observations of complex molecules are essential inorder to derive the temperature and excitation conditions ofhot cores, especially if different transitions and species tracedifferent extensions of the core.Beltr´an et al. (2005) mapped the CH CN emission intwo hot cores and spectrally identified many other organicmolecules in both objects. This work led to the first de-tection outside the Galactic Centre of a transition of gly-colaldehyde, CH (OH)CHO, in G31.41+0.31 (Beltr´an et al.2009). The richness of these spectra gives a great insight intothe chemistry of this type of object, which we investigate indepth here. In addition to the previously-published data forG31.41+0.31 and G24.78+0.08 A2, we also analyse the un-published spectra of four more hot cores to give a sample ofsix (see Sect. 2). Seven emission lines evaded definite identi-fication in the spectrum of G31.41+0.31 at 1.4 mm. In thispaper we identify some of these unidentified lines, and findthat they are seen in other members of our sample. Since allthe new identifications are of complex species, we use thenew data in conjunction with previously-published data toderive the excitation conditions of each core. We then makeuse of a chemical model of a hot core to interpret the molec-ular inventory of the six cores and qualitatively characteriseeach core and its evolutionary stage. Previous work such as Isokoski et al. (2013) have taken a molecular inventory ofhot cores using single dish observations to look for chemi-cal differences in objects with disk-like structures comparedto those without disks. This work will explore the chemicaldifferences between hot cores by making use of high spatialresolution observations. This will allow us to look at the spa-tial chemical variation in hot cores as well as the variabilitybetween objects. This work is based on observations taken with the Plateau deBure Interferometer (PdBI) and reported by Beltr´an et al.(2005), for the hot cores G31.41+0.31 (G31) and A1 and A2in G24.78+0.08 (G24A1 and G24A2), and by Beltr´an et al.(2011) for the hot cores G29.96-0.02 (G29), G19.61-0.23(G19), G10.62-0.38 (G10). Information about the obser-vations can be found in their Section 2. The synthesizedCLEANed beams for maps made using natural weightingcan be found in Table 1. The V
LSR for each hot core listedin Table 1 has been determined from high spatial resolutionobservations of the 12-13 transitions of CH
CN.These observations range from 220 209.95 MHz to220 759.69 MHz for all the hot cores and were analysed usingthe GILDAS software package. We have selected luminous ( L bol > L ⊙ ) objects withtypical signposts of massive star formation such as watermasers and ultracompact (UC) HII regions.Six hot cores were observed towards five star-formingregions on the basis of being bright in the sub-mm, andhaving previously given indications of being chemically rich.We discuss each source in detail below. G31.41+0.31:
This is a well-studied hot core located at7.9 kpc (Cesaroni et al. 1994, 1998), with evidence of a rotat-ing massive molecular toroid, suggested by OH maser emis-sion and confirmed using CH CN emission (Gaume & Mutel1987; Beltr´an et al. 2004, 2005; Cesaroni et al. 2011). It hasbeen mapped in several different molecules including SiO,HCO + and NH , as well as several complex molecules likeCH CN, C H CN and CH (OH)CHO (Cesaroni et al. 1994;Maxia et al. 2001; Beltr´an et al. 2005). Interferometric ob-servations of molecular lines with high excitation energieshave revealed the presence of deeply-embedded YSOs, whichin all likelihood explains the temperature increase towardthe core centre (Beltr´an et al. 2004, 2005; Cesaroni et al.2010). The Spitzer/GLIMPSE images by Benjamin (2003)show that the G31 hot core lies in a complex parsec-scaleregion where both extended emission and multiple stellarsources are detected. G29.96-0.02:
G29 is located at a distance of 3.5 kpc(Beltr´an et al. 2011) and associated with the infrared sourceIRAS 18434-0242. It contains a cometary UCHII regionwith a hot core located in front of the cometary arc(Wood & Churchwell 1989; Cesaroni et al. 1994). It hasbeen mapped in several molecules including NH , HCO + , (cid:13) , 1–20 high resolution study of COMs in hot cores Table 1.
Parameters of the IRAM PdBI observations
Source α (J2000) † δ (J2000) V LSR
Synthesized Beam P.A. Velocity resolution (kms − )(h m s) ( ◦′′′ ) (kms − ) †† ( ′′ ) ( ◦ )G31.41+0.31 18 47 34.330 -01 12 46.50 96.8 a × b × b × b × a × † Coordinates of the phase centre of the observations. †† The synthesized CLEANed beams for maps made using natural weighting. a Beltr´an et al. (2005) b Beltr´an et al. (2011)
CS, CH CN, HNCO and HCOOCH (Cesaroni et al. 1998;Pratap et al. 1999; Maxia et al. 2001; Olmi et al. 2003;Beuther et al. 2007). A velocity gradient along the east-west direction has been measured in NH , CH CN,and HN C emission which is interpreted as rotation(Cesaroni et al. 1998; Olmi et al. 2003; Beuther et al. 2007).On the other hand, an outflow directed along the southeast-northwest direction has been mapped in H S and SiO byGibb et al. (2004) and Beuther et al. (2007). Beltr´an et al.(2001) confirmed the existence of a rotating molecu-lar toroid around the outflow axis. Masers of H O,CH OH and H CO have also been detected around thishot core (Hofner & Churchwell 1996; Walsh et al. 1998;Hoffman et al. 2003).
G19.61-0.23:
G19 is located at a distance of 12.6 kpc(Kolpak et al. 2003), and associated with the infrared sourceIRAS 18248-1158. It contains several embedded UCHIIregions, detected by Garay et al. (1985), and more re-cently mapped by Furuya et al. (2005). Several moleculartracers, such as CS, NH , CH CH CN, HCOOCH , andCH CN have been mapped in this hot core (Plume et al.1992; Garay et al. 1998; Remijan et al. 2004; Furuya et al.2005). CO and C O emission show inverse P Cygni pro-files indicating infalling gas towards the core (Wu et al.2009; Furuya et al. 2011). L´opez-Sepulcre et al. (2009)mapped a CO outflow without a well-defined morphol-ogy. H O, OH, and CH OH masers have been detected byForster & Caswell (1989), Hofner & Churchwell (1996) andWalsh et al. (1998). Beltr´an et al. (2011) reported velocitygradients, observed in CH CN, oriented perpendicular to thedirection of a molecular outflow.
G10.62-0.38:
G10 is located at a distance of 3.4 kpc(Blum et al. 2001) and contains a well-studied UCHII re-gion (e.g. Wood & Churchwell 1989) associated with the in-frared source IRAS 18075-1956. The hot core in this starforming region has been mapped in NH (Ho & Haschick1986; Keto et al. 1987, 1988; Sollins & Ho 2005) and inSO and OCS (Klaassen et al. 2009). Infall and bulk ro-tation in the molecular gas surrounding the UCHII re-gion have been detected. H66 α emission shows the occur-rence of inward motions in the ionised gas (Keto 2002).CH OH and H O masers have been mapped towards thecore and are distributed linearly in the plane of the rotation(Hofner & Churchwell 1996; Walsh et al. 1998), while OHmasers appear to lie along the axis of rotation (Argon et al.2000). Outflow activity has been detected by Keto & Wood(2006), L´opez-Sepulcre et al. (2009) and by Beltr´an et al. (2011), who reported a CH CN toroid rotating around themain axis of the outflow.
G24.78+0.08A1 and G24.78+0.08A2:
G24 is a high-mass star-forming region located at 7.7 kpc from the Sun,and associated with several YSOs in different evolution-ary phases embedded in their parental cores (Furuya et al.2002). The G24 region has been extensively studied invarious molecules like CO, NH and CS, as well as incomplex molecules and in the continuum (Codella et al.1997; Cesaroni et al. 2003; Beltr´an et al. 2005, 2011). Twomain groups of YSOs, called A1 and A2, are separatedby ∼ . ′′ G24.78+0.08A1:
G24A1 is one of the three massive coreswith a rotating toroid detected in G24 (Beltr´an et al. 2011).At the centre of G24A1, an unresolved hypercompact (HC)HII region is created by a YSO with a spectral type of atleast O9.5 (Codella et al. 1997; Beltr´an et al. 2007). NH (2,2) observations have revealed that the gas in the toroid isundergoing infall towards the HC HII region (Beltr´an et al.2006), suggesting that accretion onto the star might stillbe occurring, even through the ionised region (for a po-tential mechanism, see Keto & Wood 2006). On the otherhand, Very Long Baseline Array proper motion measure-ments of H O masers associated with the HC HII region(Moscadelli et al. 2007) indicate that the ionised regionmight be expanding, thus questioning the possibility of ac-cretion onto the star.
G24.78+0.08A2:
G24A2 is also asso-ciated with a massive CH CN toroid, rotating around themain axis of a bipolar outflow observed in the CO iso-topologues (Furuya et al. 2002; Beltr´an et al. 2005, 2011;Codella et al. 2013). The mid-infrared and radio contin-uum measurements show a compact (1000–2000 AU) sourcewhich could be due to an ionised jet (Vig et al. 2008). Emis-sion from several complex molecules clearly indicate thepresence of a molecular hot core (Beltr´an et al. 2005, 2011;Codella et al. 2013).
Figure 1 shows the reduced spectrum for each hot core; welabel the identified lines (Beltr´an et al. 2005) as well as theseven unidentified lines in G31, A to G in ascending fre-quency in the top panel. As can be seen from the figure, notall lines are present in all cores. Lines C, D, F and G weretoo blended with other lines to be clearly identified. Table 4lists the previously unidentified lines which could be identi- c (cid:13) , 1–20 Calcutt et al. fied (A, B and E), with their frequencies and observationalparameters, and which cores they are detected in. These willbe discussed in Sect. 4.2.We also find that several of the molecular lines identi-fied in G31, G24A1 and G24A2 by Beltr´an et al. (2005) arealso seen in other members of our sample of hot cores. Webriefly discuss these new detections in Sect. 4.1.We determine the identity of all of the spectral linesin our sample using a rigorous method to avoid misidenti-fications. Here we follow the line identification process aspresented in Snyder et al. (2005). While we refer the readerto the Snyder et al. (2005) paper for a detailed explanationof the recipe one has to follow in order to identify a spectralline, here we very briefly summarise the criteria used:(i) We only considered transitions with a frequency error of <
50 MHz.(ii) All the expected transitions of a molecule must have fre-quency agreement. In the case of line blending two linesshould be at least resolved by the Rayleigh criterion.Snyder et al. (2005) also remark that a more stringent cri-terion may be used whereby two overlapping lines can beconsidered to be resolved if they are least separated at thehalf-maximum intensity of the weakest line.(iii) If other transitions from the same molecule are present inthe observed frequency range, have line strengths at de-tectable levels, and there are no mitigating circumstancessuch as maser activity or line self-absorption as to why theyshould not be observed, they should be present in the spec-tra. Their relative intensity must also correspond to the pre-dicted one under LTE.(iv) Transitions from molecules already observed in similar ob-jects and, if not, in the interstellar medium (ISM), shouldbe favoured over new molecules not yet detected in space.(v) Excluding transitions with upper energy levels exceedingthe highest upper energy level of a transition previously de-tected in the hot core sample (931 K).We have used the JPL, CDMS, and Lovas/NIST molec-ular spectroscopy databases for line identification . We findthat there are uncertainties in both the frequencies of linesin the molecular databases and the observed frequency ofthe unidentified astronomical lines. Any line with a mea-sured laboratory frequency uncertainty larger than 50 MHzhas been excluded from consideration. Since the spectralresolution of the observations is 2.5 MHz, we searched forlines within the linewidths of the unidentified lines. The listof potential lines has been reduced by excluding transitionswith an upper energy level over 931 K, as typically excita-tion temperatures of the present hot core sample do notexceed ∼
300 K (e.g. Beltr´an et al. 2005; Beltr`an et al. 2010,and references therein) and to date, the highest upper en-ergy level of a transition previously detected in the hot coresample is 931 K.
In this subsection we identify lines in the G10, G19 and G29hot cores, which have already been detected in G31, and in For references see acknowledgements. most cases already detected in G24A1 and G24A2 as well.Table 2 shows the derived frequency, velocity, integrated in-tensity, FWHM, peak temperature and rms of the baselinefor these newly detected lines in each hot core and the tran-sitional information can be found in Table 3.The 10 , –9 , transition of HNCO is detected in a fur-ther two hot cores, G19 and G29 (see Fig. 2), although it isblended with one transition of methyl cyanide in all cores.The 25 , –24 , transition of C H CN was also de-tected in G31, G24A1 and G24A2 in Beltr´an et al. (2005).We have found it in a further two hot cores: G19 and G29(Fig. 3). This line is very bright in G31 at 19.7 K but is asweak as 4.0 K in G29. The FWHM is similar in all the hotcores, ranging from 6.9 kms − in G29 to 10.8 kms − in G19.It is not detected in G10. Both of these lines are brightestin G31. (OH)CHO) The complex organic molecule glycolaldehyde(CH OHCHO), which is an isomer of both methyl formate(HCOOCH ) and acetic acid (CH COOH), is the simplestof the monosaccharide sugars. This important organicmolecule was first tentatively detected at 1.4 mm with thePdBI outside the Galactic centre, where it was observed inSgr B2(N) (Hollis et al. 2000) towards three massive hot-cores (G31, G24A1, and G24A2) by Beltr´an et al. (2005).Later on, Beltr´an et al. (2009) confirmed the detectionby observing two additional Glycolaldehyde transitionstowards G31 at 2.1 and 2.9 mm with the PdBI. Recently,Jørgensen et al. (2012) have detected 13 transitions towardsthe Class 0 IRAS 16293–2422 object at 0.4 and 1.4 mm,using the ALMA array towards the hot-corino surroundingthe Solar-type protostar.In this present work, we report a new detection of a lineat ∼
220 466 MHz in two hot cores (Fig. 4), G29 (3 σ detec-tion) and G19 (5 σ detection), which we postulate to be the20 , –19 , transition of glycolaldehyde, as was detectedby Beltr´an et al. (2005) in G31, G24A1 and G24A2. We ob-serve a shift in the velocity of 3-4 km/s in this transitioncompared to the V LSR of each hot core. This is consistentwith the glycolaldehyde observations of both Beltr´an et al.(2009), Halfen et al. (2006), and Hollis et al. (2000). Whilstfurther transitions of this molecule are needed to confirm itspresence within G29 and G19, if confirmed, it would sug-gest that glycolaldehyde is a common hot-core tracer. Inboth hot-cores this line is blended with two methyl cyanidelines, as it is in G31, G24A1 and G24A2. We also acknowl-edge that the 46 , –46 , EE (E u = 816 K, S µ = 2843D ) and 11 , , – 10 , , AE (E u = 63 K, S µ = 519 D )transitions of acetone may be contributing to the emissionseen at this frequency. This is consistent with observationsby Fuente et al. (2014), where they detect the 20 , –19 , transition of glycolaldehyde at 220 466 MHz, blended withacetone. In Section 5.3 we model both glycolaldehyde andacetone emission to explore this possibility. For the purposesof a chemical comparison between the hot cores in our sam-ple, we have assumed that the line at 220 466 MHz is gly-colaldehyde in the rest of the paper.Figure 5 shows a map of the 18 , –17 , E methyl for-mate emission (blue contours; line A, see Section 4.2), the20 , –19 , glycolaldehyde emission (white contours) and c (cid:13) , 1–20 high resolution study of COMs in hot cores Figure 1.
Spectra of the hot cores G31, G29, G19, G10, G24A1 and G24A2 observed with the PdBI (Beltr´an et al. 2005, 2011), integratedover the 3 σ contour level area. The species labelled are those that have already been identified in Beltr´an et al. (2005) and Beltr´an et al.(2011) as well as the 7 lines in G31 that were found but not identified by Beltr´an et al. (2005). The numbers indicate the position of theCH CN (12 K –11 K ) K-components in the upper part (in pink) of each spectra and of the CH CN (12 K –11 K ) K-components in thelower part (in red). the averaged emission of the K=0, 1, 2 (12–11) transitionsof methyl cyanide emission (colour scale) in G31, G24A1,G24A2 and G29 (i.e. the four hot-cores for which the threemolecular species have been detected). Methyl cyanide is atypical hot core tracer. For G24A1, G24A2, and G29 thethree species seem to be tracing the same material andpeak at roughly the same position towards the dust emissionpeak. On the other hand, in G31 glycolaldehyde peaks to-wards the centre of the core where the continuum source(s)is(are) embedded, whilst methyl formate and methyl cyanideshow a different morphology. In particular, as reported byBeltr´an et al. (2005) the methyl cyanide traces a toroidalstructure with the strongest emission eastwards of the mil-limetre continuum emission peak, which is located towardsthe central dip. On the other hand, methyl formate does notshow a toroidal morphology but peaks towards the easternside of the core, at a position barely coincident with thatof the methyl cyanide. (Beltr´an et al. 2005) explained theseapparent toroidal morphology as caused by the high opticaldepth and the existence of a temperature gradient in the core. In this scenario, glycolaldehyde appears to be less af-fected by excitation conditions, and a better tracer of theinner conditions of the hot-core closer to the embedded pro-tostar(s). Methyl formate emission extends to 0.13 pc (3 . ′′ ′′ ).In G24, both methyl formate emission and glycolaldehydeemission extend to 0.11 pc (3 ′′ ). In G29, both methyl for-mate emission and glycolaldehyde emission extend to 0.05pc (3 ′′ ). We now discuss some of the lines that were previouslyunidentified (see Fig. 1), namely A, B and E , for each source,and we compare the frequency, area of the Gaussian fitted,FWHM and peak brightness temperature in each hot corefor each of the lines (see Table 4). It is clear from Fig. 1that the spectra do vary among hot cores and that G31, thebrightest hot core with L bol > L ⊙ (Beltr´an et al. 2005) c (cid:13) , 1–20 Calcutt et al.
Table 2.
The observed frequency, velocity, area of Gaussian fit, FWHM, peak brightness temperature and the rms of the baseline formolecules previously detected in Beltr´an et al. (2005) and confirmed in new cores in this work. Spectra are integrated over the 3 σ contourlevel. Molecule Frequency † V LSR V peak † R T d v (Error) FWHM T B rms of baseline(MHz) (kms − ) (kms − ) (Kkms − ) (kms − ) (K) (K)G31.41+0.31CH OHCHO 220 466.35 96.8 93.6 (0.7) 40.5 (1.2) 6.8 (0.1) 5.6 (0.2) 0.3HNCO 220 585.58 96.8 96.5 (1.0) 174.9 (22.5) 14.9 (2.4) 11.1 (0.9) 0.9C H CN 220 661.45 96.8 96.3 (0.3) 203.5 (37.6) 9.7 (2.0) 19.7 (1.2) 0.3G29.96-0.02CH OHCHO 220 467.53 98.9 93.9 (3.4) 6.5 (1.1) 10.9 (2.1) 0.6 (0.1) 0.2HNCO 220 585.88 98.9 98.0 (0.2) 85.6 (18.7) 10.5 (2.9) 7.7 (0.2) 0.3C H CN 220 661.96 98.9 97.5 (0.4) 29.3 (9.2) 6.9 (2.5) 4.0 (0.1) 0.8G19.61-0.23CH OHCHO 220 467.67 41.6 36.4 (0.2) 5.8 (1.4) 6.9 (1.9) 0.8 (0.1) 0.2HNCO 220 586.00 41.6 40.5 (0.2) 87.5 (41.4) 10.2 (5.5) 8.1 (0.3) 0.6C H CN 220 662.77 41.6 39.1 (0.4) 61.0 (4.2) 10.8 (0.9) 5.3 (0.5) 0.9G24.78+0.08 A1CH OHCHO 220 467.22 110.8 106.5 (0.2) 8.0 (0.3) 4.3 (0.5) 1.8 (0.1) 0.2HNCO 220 584.53 110.8 111.9 (0.2) 67.9 (16.3) 9.0 (2.5) 7.1 (0.3) 0.8C H CN 220 661.67 110.8 110.0 (0.2) 68.5 (9.1) 7.8 (1.2) 8.2 (0.5) 1.1G24.78+0.08 A2CH OHCHO 220 466.94 110.8 106.8 (0.5) 10.3 (1.8) 5.8 (1.2) 1.7 (0.1) 0.3HNCO 220 584.65 110.8 111.8 (0.1) 59.2 (35.6) 6.4 (4.1) 8.7 (1.5) 0.8C H CN 220 661.67 110.8 110.0 (0.3) 83.0 (13.9) 8.4 (1.6) 9.3 (0.6) 1.0 † The small spread in frequency (0.2 – 2.4 MHz) between the rest frequency of the line and the observed frequency for each molecule ineach hot core, is due to the difference between the V
LSR and the V peak in each hot core. This difference is within the spectralresolution and V peak error of the observations.
Table 3.
Transitional information for the molecules previously detected in Beltr´an et al. (2005) and confirmed in new cores in this work.All this transitional information was taken from the JPL spectral line catalog.Molecule Transition Frequency E u S µ (MHz) (K) (D )CH OHCHO 20(2,18)–19(3,17) 220 463.88 120.05 89.74HNCO 10(1,9)–9(1,8) 220 585.20 101.50 27.83C H CN 25(2,24)–24(2,23) 220 660.92 143.02 367.60 is the most chemically-rich hot core in our sample. Lines C,D, F and G labelled in this figure could not be identified.Line A is seen in four of the hot cores in our sample:G31, G29, G24A1 and G24A2 (Fig. 7). We have fitted aGaussian profile to this line in each of the hot cores to de-termine their rest frequency, integrated intensity, FWHM,and brightness temperature (see Table 4). There is onlya small spread in the observed rest frequency of this linefrom 220 258.240 MHz to 220 260.657 MHz, suggesting it isthe same line in each hot core. Line A is brightest in G31,as expected. The FWHM range from 7.6 kms − in G29 to11.1 kms − in G24A1.The potential identities of line A consist of three lines ofmethyl formate (HCOOCH ): 18 , -17 , E, 24 , -241 , , -24 , E; one line of methylene amidogen (H CN):3 , -2 , ,F=11/2-9/2, (n=0-0)*; and one line of vinyl alco-hol (C H OH): 11 , -11 , . Methylene amidogen has previ-ously been found in the cold core TMC-1 (Ohishi et al. 1994)and methyl formate and vinyl alcohol have previously been detected in hot cores (Fontani et al. 2007; Turner & Apponi2001). Using criterion (iii) from Section 4 we can rule outmethylene amidogen as there are several expected transi-tions in the frequency range of our observations that arenot seen. Vinyl alcohol can also be ruled out because the11 , –11 , transition has a low line intensity (S µ = 0.03),which leads to a column density of 2.7 × cm − , six or-ders of magnitude higher than the vinyl alcohol column den-sity in Sagittarius B2(N) (Turner & Apponi 2001) and muchlarger than those of more commonly-observed molecules.We therefore identify line A with the 18 , –17 , E transi-tion of methyl formate, but the 24 , –24 , E, and 24 , –24 , E transitions also contribute to the line A emission.The 18 , –17 , E transition of methyl formate is a high en-ergy transition therefore it was not identified previously.One notes however that, while all the expected transi-tions of methyl formate in our frequency range are detectedin G31, this is not the case for the other hot cores (seeFig. 6). For example the 220 551.31 MHz 18 , –17 , , vt=1 c (cid:13) , 1–20 high resolution study of COMs in hot cores Table 4.
The observed frequency, velocity, area of Gaussian fit, FWHM, peak brightness temperature and the rms of the baseline forlines A, B and E in each of the hot cores in our sample. Spectra are averaged over the 3 σ contour level.Object Frequency † V LSR V peak R T d v (Error) FWHM T B , peak rms of baseline(MHz) (kms − ) (kms − ) (K.kms − ) (kms − ) (K) (K)Line AG31 220 258.26 96.8 96.8 72.0 (1.8) 8.2 (0.3) 7.8 (0.2) 0.3G29 220 260.05 98.9 96.2 8.3 (0.3) 7.6 (0.3) 1.02 (0.03) 0.2G19 No detection rms = 0.2G10 No detection rms = 0.1G24A1 220 260.65 110.8 107.5 17.4 (2.9) 11.1 (2.2) 1.5 (0.2) 0.2G24A2 220 259.28 110.8 109.2 19.4 (2.1) 10.3 (1.2) 1.8 (0.2) 0.3Line BG31 220 307.69 96.8 96.6 37.4 (2.0) 7.1 (0.5) 5.0 (0.3) 0.3G29 No detection rms = 0.2G19 No detection rms = 0.2G24A1 No detection rms = 0.2G24A2 No detection rms = 0.3G10 No detection rms = 0.1Line EG31 220 552.70 96.8 95.1 16.0 (4.5) 5.0 (2.8) 2.56 (0.9) 0.9G29 No detection rms = 0.3G19 No detection rms = 0.6G10 No detection rms = 0.2G24A1 No detection rms = 0.8G2A2 No detection rms = 0.8 † The small spread in frequency (0.2 – 2 MHz) between the rest frequency of the line and the observed frequency for each molecule ineach hot core, is due to the difference between the V
LSR and the V peak in each hot core. This difference is within the spectralresolution and V peak error of the observations.
Table 5.
The observed lines transitions, E u , S µ , and line list used for lines A, B and E.Line Molecule Transition Frequency E u S µ Line list(MHz) (K) D Line A HCOOCH v =1 18(8,10) – 17(8,9)E 220 258.09 331 38.5 JPLLine B HCOOCH v =1 18(10,9) – 17(10,8)E 220 307.38 354 33.2 JPLLine E H COOCH v =1-1 220 551.30 313 43.9 TopModel transition of methyl formate falls in a region of no detectableemission in the hot cores G29, G24A1 and G24A2. Thismay be an indication of temperature differences amonghot cores as the intensity of methyl formate transitions issignificantly temperature dependent.Line B is only detected in G31 (see Fig. 8). Its potentialline identities are three transitions of methylene amidogen(H CN) and one line of methyl formate (18 , –17 , E).Methylene amidogen can, again, be ruled out accordingto the (iii) criterion of Section 4, so the methyl formate18 , –17 , E transition is the best candidate for line B.Line E is only detected in G31 (see Fig. 9). There aretwo potential line identities for line E, one line of a carbon-13isotopologue of methyl formate (H COOCH ), 18 , –17 , , vt=1 and one line of propanal (CH CH CHO),9 , –8 , . There is only one bright line of propanal in the frequency range of these observations. This line falls inan area of little emission in the spectrum of G31 therebyruling out propanal as a candidate. The 18 , –17 , , vt=1transition of methyl formate is therefore the best candidatefor line E. The transitional information for these newlyidentified lines can be found in Table 5.Clearly G31 is the most chemically rich object and infact, of the three unidentified lines, B and E are only presentin this source. In the next sections we derive column densi-ties assuming Local Thermodynamic Equilibrium (LTE) andanalyse our results through the use of a chemical model. c (cid:13) , 1–20 Calcutt et al.
Table 6.
Estimated rotational temperatures and column density of methyl formate calculated using the rotational temperature andusing a source size measured in this work.Object Source size ( ′′ ) T rot (K) Column density (cm − )G31 3.5 169 ±
39 4 × Figure 2.
HNCO spectra with Gaussian fits, integrated over the3 σ contour level area toward G31, G29, G19, G24A1 and G24A2as seen with the PdBI. This transition was previously detected inG31, G24A1 and G24A2 (Beltr´an et al. 2005) and is detected inG19 and G29 in this work. Methyl formate in G31 has been extensively studied and,aside from three newly-identified transitions, 26 transitionshave been previously detected (Fontani et al. 2007) usingthe IRAM-30m. We therefore use the 29 lines of methyl for-mate in G31 to derive column densities and temperatureestimates using a rotational diagram. We have accountedfor the beam dilution of the IRAM-30m data using a source
Figure 3. C H CN spectra with Gaussian fits, integrated overthe 3 σ contour level area toward G31, G29, G19, G24A1 andG24A2 as seen with the PdBI. This transition was previouslydetected in G31, G24A1 and G24A2 in Beltr´an et al. (2005) andis detected in G19 and G29 in this work. size of 3 . ′′ × cm − )and Isokoski et al. (2013, 1.7 × cm − ) when assuminga source size of 3 . ′′ c (cid:13) , 1–20 high resolution study of COMs in hot cores Figure 4. CH (OH)CHO spectra with Gaussian fits, integratedover the 3 σ contour level area toward G31, G29, G19, G24A1 andG24A2 as seen with the PdBI. This transition was previously de-tected in G31, G24A1 and G24A2 in Beltr´an et al. (2005) and isdetected in G19 and G29 in this work. In G29 it is only detectedat a 3 σ level. More transitions of glycolaldehyde are needed toconfirm its presence in G29 and G19, and this line may be con-taminated by the 46 , -46 , EE and 11 , , – 10 , , AEtransitions of acetone. a single gas component with a temperature of ∼
169 K canfit all 29 transitions. Note that line A and line B are vibra-tionally excited transitions of methyl formate and we wouldhave expected them to trace a hotter region of gas.
Table 7 lists calculated column densities for all identifiedmolecules across our sample, derived using Equation 1 be-low, by assuming LTE at a constant temperature of 300 K,225 K and 150 K, and optically thin emission (see Section5.1 for a detailed discussion of the temperature of our hotcores). N u /g u = N tot Q ( T rot ) e − E u /T rot = (cid:18) πν k R T dvhc Ag u (cid:19) (1)where g u is the statistical weight of the level u , N tot is the total column density of the molecule, Q(T rot ) is therotational partition function, E u is the energy of the upperenergy level, k is the Boltzmann constant, ν is the frequencyof the line transition, A is the Einstein coefficient of thetransition, R T dv is the integrated line intensity. All columndensities have been corrected for beam dilution by divid-ing the integrated line intensity by the beam dilution factor(Equation 2). η BD = θ S θ S + θ B (2)where θ S is the source size and θ B is the beam size.For the IRAM-30m observations of G31 we use a beamsize ranging from 10 ′′ – 24 . ′′
4. For the PdBI observations weassume the source fills the beam.Table 7 shows that all our species peak in their densityin G31. However, it is surprising that there is relatively littlevariation across the sample for HNCO and C H CN, bothof which differ by a factor of 5 or less from source to source.HNCO densities are remarkably consistent throughout thesample. CH CN and CH (OH)CHO show more variation,but if we exclude G31, then calculated column densities forthe remaining objects in the sample again agree very well,to within a factor of ∼ We have also analysed the observations using the spec-tral modelling software CASSIS and using the JPL Cat-alog. CASSIS has been developed by IRAP-UPS/CNRS(http://cassis.irap.omp.eu). We use the LTE (local thermo-dynamic equilibrium) analysis tool to determine the columndensities and excitation temperatures, T ex , required to re-produce the emission. The brightness temperature, T b , of agiven species is calculated by CASSIS according to: T b = T C e − τ + (1 − e − τ )( J ν ( T ex ) − J ν ( CMB )) (3)where T C is the temperature of the continuum, τ isthe opacity, CMB is the cosmic microwave background at2.7 K, and J ν (T) = hν/k ) / (1 − e hν/kT −
1) is the radiationtemperature.The input parameters for CASSIS are the columndensity, excitation temperature, source size, FWHM,and V
LSR for each species we observe. Values for thesource size, FWHM, and V
LSR are taken from ourobservations. We vary the column densities and exci-tation temperatures for each species until a best fit isachieved. Further details of the CASSIS software and LTEanalysis tool can found in the CASSIS documentation(http://cassis.irap.omp.eu/docs/RadiativeTransfer.pdf).In particular, the spectral modelling focuses on theemission of CH CN, HCOOCH and a possible contamina-tion of the CH (OH)CHO emission with (CH ) CO, to testthe validity of our line assignments. In this model we assumethe gas is in LTE conditions at a temperature T ex . A com-parison of the column densities we derive from observation c (cid:13) , 1–20 Calcutt et al.
Figure 5.
Spectral line map of the 18 , -17 , E transition of methyl formate (blue contours), the 20 , –19 , transition of glyco-laldehyde (white contours) and the averaged emission of the K=0, 1, 2 (12-11) transitions of methyl cyanide emission (colour scale) inG31, G29, G24A1 and G31. For the methyl formate in G31, the channels averaged were 91.7 – 101.9 kms − , with contour levels of 0.04– 0.28 Jybeam − , in steps of 0.04 Jybeam − . In G24 the channels averaged for methyl formate were 102.4 – 112.6 kms − , with contourlevels of 0.02 – 0.12 Jybeam − , in steps of 0.02 Jybeam − . In G29 the channels averaged for methyl formate were 91.1 – 101.3 kms − ,with contour levels of 0.015 – 0.06 Jybeam − , in steps of 0.015 Jybeam − . For glycolaldehyde in G31, the channels averaged were 91.9– 95.3 kms − with contour levels of 0.10 – 0.46 Jybeam − , in steps of 0.12 Jybeam − . In G24 the channels averaged for glycolaldehydewere 104.8 – 108.2 kms − , with contour levels of 0.04 – 0.12 Jybeam − , in steps of 0.04 Jybeam − . In G29 the channels averaged forglycolaldehyde were 92.2 – 95.6 kms − , with contour levels of 0.016 – 0.08 Jybeam − , in steps of 0.016 Jybeam − . For methyl cyanidethe contour levels in G31 are 0.1–0.94 Jybeam − in steps of 0.12, in G24 they are 0.1 – 1.0 Jybeam − in steps of 0.18 Jybeam − , inG29 they are 0.09 – 1.14 Jybeam − in steps of 0.15 Jybeam − . Figure 6.
The expected bright emission of methyl formate transitions, that occur in the frequency range of our observations, plottedon the spectra of hot cores G31, G29, G24A1 and G24A2. The numbers on the plot represent the intensity (S µ ) of methyl formatetransitions as determined by laboratory studies, in units of Debye . and those we derive from spectral modelling can be foundin Table 8.Spectral models of CH CN emission, using only asingle column density and excitation temperature, do notreproduce the observations accurately (Figure 11). Wehypothesise that this ‘poor’ fit may be due to a combination of factors: (i) different transitions of CH CN may peak,or may be tracing, different temperatures and densitieswithin our emission region, (ii) the emission we observemay not be in LTE, (iii) contributions due to blending,(iv) CH CN opacities are highly variable from the K=1 tothe K=7 transitions, (v) the poor signal-noise ratio of the c (cid:13) , 1–20 high resolution study of COMs in hot cores Table 7.
Column densities (cm − ) of organic molecules in our sample, assuming LTE at 300 K, 225 K, and 150 K.300 KObject HNCO C H CN CH CN a CH (OH)CHO HCOOCH G31 3 × × × × †† × ††† G29 2 × × × × × G19 1 × × × × . . .G24A1 1 × × × × × G24A2 2 × × × × ×
225 KG31 2 × × × × †† × ††† G29 1 × × × × × G19 1 × × × × . . .G24A1 7 × × × × × G24A2 8 × × × × ×
150 KG31 1 × × × × †† × ††† G29 7 × × × × × G19 5 × × × × . . .G24A1 4 × × × × × G24A2 5 × × × × × . a CH CN has been observed to be optically thick in all these objects (Beltr´an et al. 2005, 2011) so the column densities have beenderived using observations of CH
CN. †† The column density of glycolaldehyde for G31 has been obtained using the rotation diagram method by Beltr´an et al. (2009) †††
The column density for this object was determined using the rotation diagram method using a source size of 3 . ′′
5, see 5.1. high energy K-transitions. The best fit of all of the CH CNtransitions is achieved at T ex ranging from 410 – 450 Kacross our hot core sample. For HCOOCH emission wehave produced a spectral fit for the transition at 220.258GHz (Figure 12) and the transition at 220.307 GHz. Weare unable to model any isotopologues of methyl formatewith CASSIS and have therefore omitted the transition at220.553 GHz from our fit. In G31, where the observationalcolumn density was derived from multiple transitions ofmethyl formate, we find the required modelling columndensity to be the same as the observed value (4 × cm − ). For the other hot cores we require a modelledcolumn density a factor of 2 – 3 larger than the observedvalue. This is not surprising as the observed column densitywas derived from only one transition in these cores andtherefore represents a lower estimate of the methyl formatecolumn density in these objects. As we only have oneconfirmed methyl formate transition in several of ourhot cores, it is difficult to accurately measure the T ex required to reproduce these observations. We are confidentfrom this modelling that we have correctly identified the18 , –17 , E and 18 , –17 , E transitions of HCOOCH in our hot core sample.For CH (OH)CHO emission we have produced a spec-tral fit for the transition at 220.466 GHz (Figure 13). ForG31 the required column density to reproduce the spectrais similar to the observed value. For G29, G19, G24A1, andG24A2 the required modelling column densities are a factorof 1.5 – 3 larger than the observed values. This is likelydue, again, to the observed column densities for these hot core being derived using only one CH (OH)CHO transition,and therefore represents a lower estimate of the columndensities in these hot cores.We have also explored the possibility that this transi-tion of glycolaldehyde is blended with the 46 , –46 , EE(E u = 816 K, S µ = 2843 D ) and the 11 , , – 10 , , AE(E u = 63 K, S µ = 519 D ) transitions of acetone. Wefind for column densities and excitation temperatures ofacetone which produce enough emission to explain the lineat 220 466 MHz, we overproduce emission at 220 368 MHz.It is possible that both acetone and glycolaldehyde couldbe present in the hot cores in our sample, however, thecontribution of acetone to the line seen at 220 466 MHzis not significant ( < CN) as well as sulphur-bearingspecies could serve as indicators of both mass and age(Hatchell et al. 1998; Viti et al. 2001; Buckle & Fuller 2003;Viti et al. 2004). However much care has to be taken dueto the fact that most species do not necessarily trace thevery inner part of the core and their emission region proba-bly spans a range of densities and temperatures that makethe interpretation difficult (Wakelam et al. 2004). The com- c (cid:13) , 1–20 Calcutt et al.
Table 8.
Spectral modelling column densities and excitation temperatures of methyl cyanide, methyl formate and glycolaldehyde.Hot Core T ex (K) Modelled column density (cm − ) Observed at column density (cm − )CH CNG31 450 9 × × (300 K)G29 410 5 × × (300 K)G19 450 4 × × (300 K)G24a1 310 4 × × (300 K)G24a2 450 6 × × (300 K)HCOOCH G31 130 4 × × (150 K)G29 300 5 × × (300 K)G24A1 300 9 × × (300 K)G24A2 300 1 × × (300 K)CH (OH)CHOG31 130 3 × × (150 K)G29 300 8 × × (300 K)G19 300 9 × × (300 K)G24A1 300 1 × × (300 K)G24A2 300 1 × × (300 K) plex species we have been discussing in this paper, on theother hand, may provide us with a better choice of age/massdiscrimination as they all seem to trace a more compact re-gion than the more widely-observed species such as CH OHand CH CN. In the next section we make use of a chemicalmodel, UCL CHEM (Viti et al. 2004) to simulate the for-mation and evolution of hot cores and see if the differencesin ratios of the species listed in Table 7, between G31 andthe rest of the cores, can shed some light on the masses andages of our sources.
UCL CHEM (Viti et al. 2004) is a two-phase time-dependent model which follows the collapse of a prestel-lar core (Phase I), followed by the subsequent warming andevaporation of grain mantles (Phase II). Phase I starts fromthe number density of a diffuse cloud (10 cm − ) and al-lows a free-fall collapse to take place until a final density(which varies between 10 − cm − ) is reached. This occursisothermally at a temperature of 10 K. During the collapse,atoms and molecules collide with, and freeze on to, grainsurfaces. The depletion efficiency is determined by the frac-tion of the gas-phase material that is frozen on to the grains(Rawlings et al. 1992). This fraction is arrived at by adjust-ing the grain surface area per unit volume, and assuminga sticking probability of unity for all species. The fractionof material on grains is then dependent on the product ofthe sticking probability and the amount of cross-section pro-vided per unit volume by the adopted grain size distribution.Grains are considered to be spheres. We assume that hydro-genation occurs rapidly on the grain surfaces, so that, forexample, some percentage of carbon atoms accreting willrapidly become frozen out methane (CH ) etc. In PhaseII we increase the dust and gas temperature up to 300 K,to simulate the presence of a nearby infrared source in the core. As the temperature increases, mantle species desorbin various temperature bands (see Collings et al. (2004)).UCL CHEM treatment of the temperature and of the icesublimation is as in (Viti et al. 2004), where details of howthe temperature increases with time leading to the subse-quent time dependent sublimation of the ice mantles can befound.Initial atomic abundances are taken from Sofia & Meyer(2001), as in Viti et al. (2004). For the gas-phase chem-istry, reaction rate coefficients are taken from the UMISTdatabase (Woodall et al. 2007). Some coefficients havebeen updated with those from the KIDA database(Wakelam et al. 2009). Our database also includes somesimple grain-surface reactions (mainly hydrogenation) as inViti et al. (2004) as well as selected routes for grain-surfaceformation of glycolaldehyde as in Woods et al. (2012) andmethyl formate as in Occhiogrosso et al. (2011). In PhaseI non-thermal desorption is considered as in Roberts et al.(2007).We investigate how varying key model parameters, suchas the final collapse density, lifetime of the cold phase, thetype of evaporation, and the efficiency of the freeze-out ofspecies (measured as a percentage of the total CO in thesolid phase) affect the abundances of methyl cyanide, methylformate, glycolaldehyde, ethyl cyanide and isocyanic acidduring the hot core evolution. We have modelled hot coreswith final densities of 10 cm − , 10 cm − and 10 cm − .We also vary the efficiency of the freeze-out of species from14%–100%. Since the period in which the grains are warmedfrom very low temperatures to the temperatures observed intypical hot cores is determined by the time taken for a pre-stellar core to evolve towards the Main Sequence, and henceby its mass (Viti & Williams 1999) we have explored theeffect of new-born stars with different masses (15 M ⊙ and25 M ⊙ ), corresponding to contraction times of 118 000 yr and70 000 yr respectively – see Bernasconi & Maeder (1996).In the following section we explore the sensitivity of c (cid:13) , 1–20 high resolution study of COMs in hot cores Figure 7.
Spectra of Line A with Gaussian fits, integrated overthe 3 σ contour level area toward G31, G29, G24A1 and G24A2as seen with the PdBI. This transition was previously detected inG31 but was not identified, in Beltr´an et al. (2005), and is alsodetected in G24A1, G24A2 and G29 in this work. Figure 8.
Spectra of Line B with a Gaussian fit, integrated overthe 3 σ contour level area toward G31 as seen with the PdBI. Thistransition was previously detected in G31 but was not identifiedin Beltr´an et al. (2005). Figure 9.
Spectra of Line E with a Gaussian fit, integrated overthe 3 σ contour level area toward G31 as seen with the PdBI. Thistransition was previously detected in G31 but was not identifiedin Beltr´an et al. (2005). Figure 10.
A rotation diagram of methyl formate transitions inG31, extending the work of Fontani et al. (2007) to higher excita-tion temperatures. A C/ C ratio of 41 was calculated accord-ing to Wilson & Rood (1994) using the Galactic Coordinates ofG31 and a source size of 3 . ′′ v =1 excited transitionsare so small ( < the aforementioned species to changes in the physical andchemical parameters described above, before comparing ourtheoretical models with observations. Plots 14–16 show gas-phase abundances in Phase II of the model, for ease of com-parison to observationally-derived results. Figure 14 shows an example of two models, each with a finaldensity at the end of Phase I of 10 cm − and a freeze-outpercentage of ≈
60% in the solid phase, differing in the finalmass of the star in Phase II. At ≈ . years the suddenjump in abundance of methyl formate and ethyl cyanideis due to the evaporation of these species from the grainmantle surface. At late times varying the final mass of thestar does not seem to affect the fractional abundances ofthe complex species we consider; however, as expected (seeViti et al. 2004), a higher mass implies an earlier icy mantleevaporation. In general a very low abundance of COMs im-plies a young age with the exception of methyl cyanide (butsee Sect. 6.3) whose abundance is high regardless of the age c (cid:13) , 1–20 Calcutt et al.
Figure 11.
CASSIS modelling (blue line) of methyl formate (HCOOCH ), glycolaldehyde (CH (OH)CHO), methyl cyanide (CH CN),and ethyl cyanide (C H CN) overlaid on the PdBI observations (black line). We label the molecules detected in this work in the G31panel. The numbers indicate the position of the CH CN (12 K –11 K ) K-components in the upper part (in pink) of each spectra and ofthe CH CN (12 K –11 K ) K-components in the lower part (in red). We have excluded the CO line from these models since there is anissue of missing line flux in several of the hot core observations. of the core; hence, a core where this species is the only abun-dant complex molecule may be a very young core. The the-oretical low fractional abundances before evaporation trans-late into column densities of the order of < cm − , muchlower than the lowest observed values for our sample. Figure 15 is a plot of the fractional abundances of selectedspecies at 10 yr after Phase II starts, as a function of thepercentage of CO left in the solid phase at the end of PhaseI. Note that after few thousand years from the beginning of Phase II CO will have completely evaporated back intothe gas phase so the percentage on the x -axis is not an in-dication of the final gas CO abundance. In the figure, wepresent models which use two different values of the reac-tion rate coefficient for the formation of glycolaldehyde. Thisis due to the uncertainties in the value of this quantity (seeWoods et al. (2012)). Based on the sensitivity analysis of(Woods et al. 2012), we only consider one formation routefor glycolaldehyde for simplicity: g-CH OH + g-HCO → g-CH (OH)CHO, where g- refers to a species which is frozenonto the grain surface. As expected, the abundance of mostcomplex species increases with an increase in freeze-out ef-ficiency (corresponding to a decrease in percentage of CO c (cid:13) , 1–20 high resolution study of COMs in hot cores Figure 12.
Models of the 18 , –17 , E transition of HCOOCH using CASSIS (blue line) overlayed on the PdBI observations(black line). in the gas phase). However it is interesting to note thatethyl cyanide seems to decrease at a very high freeze outefficiency while the abundance of methyl cyanide is more orless constant. The main route of formation for ethyl cyanideis via hydrogenation of HC N on the grains so in principleits abundance should increase with the efficiency of freeze-out. However, as HC N is efficiently formed in the gas phasevia reactions involving H O, HCN and other hydrocarbons,when freeze-out is very efficient the reactants are depletedfrom gas deposited on the grains at a higher efficiency thanthat of the reactions forming HC N. Figure 16 shows the abundance of our selected moleculesas a function of time for three different final densities. Thefractional abundance of most species increases with densitywhen going from 10 to 10 cm − although the increase isless pronounced at high densities. An interesting effect ofincreasing densities is observed with methyl cyanide whoseabundance before thermal evaporation is still high at thelowest density while it drops considerably at high densities:we interpret this as a direct effect of freeze-out, which is Figure 13. CH (OH)CHO spectra with CASSIS models (blueline) overlaid on the PdBI observations (black line). From left toright: the 20 , -19 , transition of glycolaldehyde, the 12(8)-11(8) transition of CH CN, and the 12(6)-11(6) transition ofCH
CN. directly proportional to collisional frequency of the parentspecies forming CH CN during phase I. Finally we note that,at a late stage of the evolution of the core, the ratio of theselected COMs varies as a function of density: for example,at lower densities CH CN is always higher than the other or-ganic molecules while already at 10 cm − CH CN/C H CNis ∼ <
1. Similar considerations canbe made about other species, making the ratios of our se-lected COMs ideal tracers of densities for evolved hot cores.
Table 7 shows that the differences in column densities amongcores, for most species, are seldom larger than one order ofmagnitude; considering the uncertainties in model parame-ters as well as formation and destruction rates for COMs, itis therefore not possible, at present, to use chemical models c (cid:13) , 1–20 Calcutt et al. -14-12-10-8-6-4-2 4 4.5 5 5.5 6 l og [ n ( X ) / n H ] log(time) in yearsEthyl Cyanide (C2H5CN)Isocyanic acid (HNCO)Methyl Cyanide (CH3CN)Methyl Formate (HCOOCH3)Glycoladehyde (CH2OHCHO) -14-12-10-8-6-4-2 4 4.5 5 5.5 6 l og [ n ( X ) / n H ] log(time) in yearsEthyl Cyanide (C2H5CN)Isocyanic acid (HNCO)Methyl Cyanide (CH3CN)Methyl Formate (HCOOCH3)Glycoladehyde (CH2OHCHO) Figure 14.
Gas phase fractional abundances of selected species asa function of time for a 15 M ⊙ (top) and a 25 M ⊙ star (bottom),both with a final gas density after Phase II of 10 cm − . At ≈ . years the sudden jump in abundance of methyl formate andethyl cyanide is due to the evaporation of these species from thegrains mantle surface. to infer the age or the mass of each individual core. Nev-ertheless, we do attempt a qualitative comparison with ob-servations by estimating theoretical column densities fromour models. We derive the theoretical column density N byusing the approximate formula below: N = X × A V × N H , (4)where X is the fractional abundance from our models andN H is the column density of H that provides 1 mag of ex-tinction (Bohlin et al. 1978), which is 1.6 × cm − . For A V we adopt a typical hot core visual extinction of 600 mags(noting that the column density simply scales linearly withextinction and that the fractional abundances calculatedfrom the models are insensitive to visual extinctions abovea critical value of ∼
10 mags, when photons do not penetrateany further).The first general conclusion we can draw by comparingTable 7 with our models (see Figs. 13, 15 and 16) is thatall cores are evolved enough that all the mantles have evap-orated. From our models this means that they are at least -12-10-8-6-4-2 0 5 10 15 20 25 30 35 40 45 l og [ n ( X ) / n H ] Percentage of CO in the solid phaseEthyl cyanide (C2H5CN)Isocyanic acid (HNCO)Methyl cyanide (CH3CN)Methyl formate (HCOOCH3) Glycolaldehyde (CH2OHCHO) with coefficient = 3e-17Glycolaldehyde (CH2OHCHO) with coefficient = 3e-12
Figure 15.
Fractional abundances of selected species at 10 yrafter the ‘switch on’ of the star as a function of the percentage ofCO left in the solid phase at the end of Phase I, for a 25 M ⊙ starwith a final gas density after phase I of 10 cm − . × and 3 × cm − where the uppervalues coincide with models with a high density and freezeout (as expected since HNCO forms mainly on the grains).HNCO seems to be fairly constant among cores and its ob-served column densities are closer to the upper theoreticalvalues; hence the only tentative conclusion we can derivehere is that the density of the observed cores is > cm − ,possibly as high as 10 cm − ; this is consistent with theemission of COMs being so compact that they trace the gascloser to the protostar.C H CN is, on the other hand, always under abundantin our models although, again, it is highest in models wherethe density and freeze out are high. We note that, amongthe cores, the highest value for this species is found in G31,implying a higher density for this core.CH CN shows a larger variation in observed columndensities, from 10 cm − in G19 to 5 × in G31. A similarrange is found in our models as a function, again, of density,although interestingly not of freeze out (see Fig. 14). Againwe conclude that G31 is the densest core, with n H close to10 cm − .The observed CH (OH)CHO column density is similaramong all cores but G31, where it is two orders of magnitudehigher than in the other objects. It is interesting to note thatwe can match its column density with models of gas densitiesof the order of 10 cm − for G31 and 10 –10 cm − for therest of the cores, again supporting the conclusion that G31has a higher gas densities than the rest of the objects in oursample. We note that only models where we use a high ratecoefficient for the formation of glycolaldehyde are able toreproduce the observations, in agreement with the findingsof (Woods et al. 2012).Finally, HCOOCH shows the same behaviour as glyco-laldehyde in that it is higher by at least one order of magni-tude in G31 compared to the rest of the sample where it is c (cid:13) , 1–20 high resolution study of COMs in hot cores -14-13-12-11-10-9-8-7-6-5 4 4.5 5 5.5 6 l og [ n ( X ) / n H ] log(time) in years Ethyl cyanide (C2H5CN)Isocyanic acid (HNCO)Methyl cyanide (CH3CN)Methyl formate (HCOOCH3)Glycoladehyde (CH2OHCHO) -14-13-12-11-10-9-8-7-6-5 4 4.5 5 5.5 6 l og [ n ( X ) / n H ] log(time) in years Ethyl cyanide (C2H5CN)Isocyanic acid (HNCO)Methyl cyanide (CH3CN)Methyl formate (HCOOCH3)Glycoladehyde (CH2OHCHO) -14-13-12-11-10-9-8-7-6-5 4 4.5 5 5.5 6 l og [ n ( X ) / n H ] log(time) in years Ethyl cyanide (C2H5CN)Isocyanic acid (HNCO)Methyl cyanide (CH3CN)Methyl formate (HCOOCH3)Glycoladehyde (CH2OHCHO)
Figure 16.
The gas phase abundance of isocyanic acid, ethylcyanide, methyl cyanide, methyl formate, and glycolaldehyde asa function of time for the percentage of CO left in the solid phaseof 100% , a mass of 15 M ⊙ , equivalent to a contraction time of115 000 yr, with step evaporation of the icy mantles. The top panelcorresponds to a final density after phase I of 10 cm − . The mid-dle panel corresponds to a final density after phase I of 10 cm − .The bottom panel corresponds to a final density after phase I of10 cm − . more or less constant at 10 cm − . In our models, we reacha column density of 10 cm − at gas densities of 10 cm − ;interestingly an increase to 10 cm − only yields an increasein methyl formate by a factor of few and only for a relativelyshort period of time, since this species seems to decline inabundance after 5 × yrs. However we point out that thecolumn density for G31 was derived using a temperature de-rived from a rotational diagram, while for all the other coresa simple LTE calculation at 300K was performed. In fact, ingeneral, LTE calculations at lower temperature would yielda lower column densities. We have analysed IRAM PdBI data, in the frequencyrange 220 209.95 MHz to 220 759.69 MHz, towards six hotcores: G31.41+0.31, G29.96-0.02, G19.61-0.23, G10.62-0.38,G24.78+0.08A1 and G24.78+0.08A2. The aim was toidentify seven lines that were unidentified by Beltr´an et al.(2005) in G31 and look for their presence in the other fivehot cores, as well as identify other complex molecules thatwere identified by Beltr´an et al. (2005) in G31 and G24 butnot in G29, G19 and G10.We have identified three new transitions of methylformate (HCOOCH3) in G31, two of which are vibrationallyexcited lines. A rotation diagram analysis of these linescombined with those in (Fontani et al. 2007) yields acolumn density for methyl formate of 4 × cm − . Thisis at least two orders of magnitude larger than the columndensities in the other hot cores in our sample. We have alsofound a single temperature component of ∼
170 K (usinga source size of 3 . ′′
5) that fits these transitions, althoughthe vibrationally-excited transitions we have found andour very large column densities would suggest that methylformate may trace multiple temperature components ofG31. This may also be true of the hot cores in G29 and G24.The spatial distribution of methyl formate does indicatethat it traces the dense and compact parts of hot cores.Comparatively methyl formate traces a region slightly morecompact than that of methyl cyanide but glycolaldehydeemission still remains the most compact to date. At thisstage we can not conclude whether the fact that methylformate is more extended than glycolaldehyde is a questionof excitation or chemistry. We do find, however, that inG29 methyl formate and glycolaldehyde are tracing a anemission region of 0.05 pc which is comparable to thecompact emission of glycolaldehyde in G31 (0.08pc). InG24 methyl formate and glycolaldehyde both trace a regioncomparable in size to the region traced by methyl formatein G31. In our models we do not find a density wherethere is more methyl formate than glycolaldehyde whichwould suggest that glycolaldehyde forms in a smaller denserregion. We note, however our glycolaldehyde network ofreactions is far from complete and it is therefore likely thatwe form too much glycolaldehyde in our model.In this work we postulate that we have found the20 , -19 , transition of glycolaldehyde in two more hotcores bringing the total number of detections in high massstar forming regions, outside the Galactic Centre, to five c (cid:13) , 1–20 Calcutt et al. hot cores. This emission whilst being far more compactin G31, is of comparable compactness to methyl formatein G24A1, G24A2 and G29. We have spectrally modelledour emission to explore the possible contamination ofthis transition with the 46 , –46 , EE and 11 , , – 10 , , AE transitions of acetone. We find that anycontribution of acetone emission to the lie seen at 220466 MHz is not significant. More observations are neededto confirm the presence of glycolaldehyde in these hotcores. Our other complex molecule detections in our samplehighlight chemical homogeneity among G29, G19, G24A1and G24A2, not only in terms of presence or absenceof certain transitions but also when comparing columndensities. G31, however, is the most chemically rich objectand the significantly different column densities we find inthis core and the variety of transitions seen may suggestthat it represents a different evolutionary stage to the otherhot cores in our sample, or it may surround a star with ahigher mass.We have also undertaken a comparison between obser-vations and a chemical model, UCL CHEM, to interpret themolecular inventory of the six cores and qualitatively char-acterise each core and its evolutionary stage. We note thatof the species we are modelling, only methyl formate andmethyl cyanide have been extensively studied in the labo-ratory (Modica & Palumbo 2010; Bennett & Kaiser 2007;Khlifi et al. 1996; Defrees et al. 1985; Huntress & Mitchell1979). The other complex molecules are little-known andit is very likely that our models are missing routes offormation and destruction for these species. Moreover, wenote that the LTE calculations shown in Table 6 werefor a temperature of 300K; a lower temperature would ingeneral yield a lower column density. The uncertaintiesrelated to the size of the emission region and temperatures,together with the incompleteness of the chemical networksfor COMs makes a more quantitative comparison withchemical modelling inappropriate.In conclusion, a qualitative comparison between ourmodelling and observations seem to consistently yield ahigher density for G31 than the other objects in our sample,a result consistent with the fact that most lines are indeedthe brightest in G31. We can also safely conclude that oursample only contains evolved hot cores, with an age of atleast 20,000 years. We are unable to constrain the mass ofeach core; this information would have led to a better con-straint for the age of each core.
ACKNOWLEDGMENTS
This research is supported by an STFC PhD studentship andthe LASSIE Initial Training Network under the EuropeanCommunitys Seventh Framework Programme FP7/2007-2013 under grant agreement No. 238258. Funding for thiswork was also provided by the Leverhulme Trust.Data was reduced using the Gildas software pack-age . Spectral line data were taken from the Spectral Line Atlas of Interstellar Molecules (SLAIM; Available at . F. J. Lovas, private commu-nication, Remijan et al. 2007), the JPL Spectral Line Cat-alog (Pickett et al. 1998), the Cologne Database for Molec-ular Spectroscopy (M¨uller et al. 2005) and the Lovas/NISTdatabase (Lovas & Dragoset 2004).
REFERENCES
Argon A. L., Reid M. J., Menten K. M., 2000, AstrophysicalJournal, Supplement, 129, 159Beltr´an M. T., Cesaroni R., Codella C., Testi L., FuruyaR. S., Olmi L., 2006, Nature, 443, 427Beltr´an M. T., Cesaroni R., Moscadelli L., Codella C.,2007, Astronomy and Astrophysics, 471, L13Beltr´an M. T., Cesaroni R., Neri R., Codella C., 2011, As-tronomy and Astrophysics, 525, A151Beltr´an M. T., Cesaroni R., Neri R., Codella C., FuruyaR. S., Testi L., Olmi L., 2004, Astrophysical Journal, Let-ters, 601, L187Beltr´an M. T., Cesaroni R., Neri R., Codella C., FuruyaR. S., Testi L., Olmi L., 2005, Astronomy and Astro-physics, 435, 901Beltr´an M. T., Codella C., Viti S., Neri R., Cesaroni R.,2009, Astrophysical Journal, Letters, 690, L93Beltr`an M. T., Codella C., Viti S., Neri R., Cesaroni R.,2010, Highlights of Astronomy, 15, 701Beltr´an M. T., Estalella R., Anglada G., Rodr´ıguez L. F.,Torrelles J. M., 2001, Astronomical Journal, 121, 1556Benjamin R. A., 2003, in Bailes M., Nice D. J., ThorsettS. E., eds, Radio Pulsars Vol. 302 of Astronomical Societyof the Pacific Conference Series, The Distribution of theWarm Ionized Medium in the Galaxy. p. 271Bennett C. J., Kaiser R. I., 2007, Astrophysical Journal,661, 899Bernasconi P. A., Maeder A., 1996, Astronomy and Astro-physics, 307, 829Beuther H., Zhang Q., Bergin E. A., Sridharan T. K.,Hunter T. R., Leurini S., 2007, Astronomy and Astro-physics, 468, 1045Blum R. D., Damineli A., Conti P. S., 2001, AstronomicalJournal, 121, 3149Bohlin R. C., Savage B. D., Drake J. F., 1978, Astrophys-ical Journal, 224, 132Bonnell I. A., Bate M. R., Zinnecker H., 1998, MonthlyNotices of the RAS, 298, 93Buckle J. V., Fuller G. A., 2003, Astronomy and Astro-physics, 399, 567Cesaroni R., Beltr´an M. T., Zhang Q., Beuther H.,Fallscheer C., 2011, Astronomy and Astrophysics, 533,A73Cesaroni R., Churchwell E., Hofner P., Walmsley C. M.,Kurtz S., 1994, Astronomy and Astrophysics, 288, 903Cesaroni R., Codella C., Furuya R. S., Testi L., 2003, As-tronomy and Astrophysics, 401, 227Cesaroni R., Hofner P., Araya E., Kurtz S., 2010, Astron-omy and Astrophysics, 509, A50Cesaroni R., Hofner P., Walmsley C. M., Churchwell E.,1998, Astronomy and Astrophysics, 331, 709 c (cid:13)000
Argon A. L., Reid M. J., Menten K. M., 2000, AstrophysicalJournal, Supplement, 129, 159Beltr´an M. T., Cesaroni R., Codella C., Testi L., FuruyaR. S., Olmi L., 2006, Nature, 443, 427Beltr´an M. T., Cesaroni R., Moscadelli L., Codella C.,2007, Astronomy and Astrophysics, 471, L13Beltr´an M. T., Cesaroni R., Neri R., Codella C., 2011, As-tronomy and Astrophysics, 525, A151Beltr´an M. T., Cesaroni R., Neri R., Codella C., FuruyaR. S., Testi L., Olmi L., 2004, Astrophysical Journal, Let-ters, 601, L187Beltr´an M. T., Cesaroni R., Neri R., Codella C., FuruyaR. S., Testi L., Olmi L., 2005, Astronomy and Astro-physics, 435, 901Beltr´an M. T., Codella C., Viti S., Neri R., Cesaroni R.,2009, Astrophysical Journal, Letters, 690, L93Beltr`an M. T., Codella C., Viti S., Neri R., Cesaroni R.,2010, Highlights of Astronomy, 15, 701Beltr´an M. T., Estalella R., Anglada G., Rodr´ıguez L. F.,Torrelles J. M., 2001, Astronomical Journal, 121, 1556Benjamin R. A., 2003, in Bailes M., Nice D. J., ThorsettS. E., eds, Radio Pulsars Vol. 302 of Astronomical Societyof the Pacific Conference Series, The Distribution of theWarm Ionized Medium in the Galaxy. p. 271Bennett C. J., Kaiser R. I., 2007, Astrophysical Journal,661, 899Bernasconi P. A., Maeder A., 1996, Astronomy and Astro-physics, 307, 829Beuther H., Zhang Q., Bergin E. A., Sridharan T. K.,Hunter T. R., Leurini S., 2007, Astronomy and Astro-physics, 468, 1045Blum R. D., Damineli A., Conti P. S., 2001, AstronomicalJournal, 121, 3149Bohlin R. C., Savage B. D., Drake J. F., 1978, Astrophys-ical Journal, 224, 132Bonnell I. A., Bate M. R., Zinnecker H., 1998, MonthlyNotices of the RAS, 298, 93Buckle J. V., Fuller G. A., 2003, Astronomy and Astro-physics, 399, 567Cesaroni R., Beltr´an M. T., Zhang Q., Beuther H.,Fallscheer C., 2011, Astronomy and Astrophysics, 533,A73Cesaroni R., Churchwell E., Hofner P., Walmsley C. M.,Kurtz S., 1994, Astronomy and Astrophysics, 288, 903Cesaroni R., Codella C., Furuya R. S., Testi L., 2003, As-tronomy and Astrophysics, 401, 227Cesaroni R., Hofner P., Araya E., Kurtz S., 2010, Astron-omy and Astrophysics, 509, A50Cesaroni R., Hofner P., Walmsley C. M., Churchwell E.,1998, Astronomy and Astrophysics, 331, 709 c (cid:13)000 , 1–20 high resolution study of COMs in hot cores Codella C., Beltr´an M. T., Cesaroni R., Moscadelli L., NeriR., Vasta M., Zhang Q., 2013, Astronomy and Astro-physics, 550, A81Codella C., Testi L., Cesaroni R., 1997, Astronomy andAstrophysics, 325, 282Collings M. P., Anderson M. A., Chen R., Dever J. W., VitiS., Williams D. A., McCoustra M. R. S., 2004, MonthlyNotices of the RAS, 354, 1133Defrees D. J., McLean A. D., Herbst E., 1985, Astrophysi-cal Journal, 293, 236Fontani F., Pascucci I., Caselli P., Wyrowski F., CesaroniR., Walmsley C. M., 2007, Astronomy and Astrophysics,470, 639Forster J. R., Caswell J. L., 1989, Astronomy and Astro-physics, 213, 339Fuente A., Cernicharo J., Caselli P., McCoey C., JohnstoneD., Fich M., van Kempen T., Palau A., Yildiz U. A., Ter-cero B., L´opez A., 2014, ArXiv e-printsFuruya R. S., Cesaroni R., Codella C., Testi L., BachillerR., Tafalla M., 2002, Astronomy and Astrophysics, 390,L1Furuya R. S., Cesaroni R., Shinnaga H., 2011, Astronomyand Astrophysics, 525, A72Furuya R. S., Kitamura Y., Wootten A., Claussen M. J.,Kawabe R., 2005, Astronomy and Astrophysics, 438, 571Garay G., Lizano S., 1999, Publications of the ASP, 111,1049Garay G., Moran J. M., Rodriguez L. F., Reid M. J., 1998,Astrophysical Journal, 492, 635Garay G., Reid M. J., Moran J. M., 1985, AstrophysicalJournal, 289, 681Garrod R. T., Weaver S. L. W., Herbst E., 2008, Astro-physical Journal, 682, 283Gaume R. A., Mutel R. L., 1987, Astrophysical Journal,Supplement, 65, 193Gibb A. G., Wyrowski F., Mundy L. G., 2004, Astrophys-ical Journal, 616, 301Halfen D. T., Apponi A. J., Woolf N., Polt R., Ziurys L. M.,2006, Astrophysical Journal, 639, 237Hatchell J., Thompson M. A., Millar T. J., MacDonaldG. H., 1998, Astronomy and Astrophysics, 338, 713Herbst E., van Dishoeck E. F., 2009, Annual Review ofAstron and Astrophys, 47, 427Ho P. T. P., Haschick A. D., 1986, Astrophysical Journal,304, 501Hoffman I. M., Goss W. M., Palmer P., Richards A. M. S.,2003, Astrophysical Journal, 598, 1061Hofner P., Churchwell E., 1996, Astronomy and Astro-physics, Supplement, 120, 283Hollis J. M., Lovas F. J., Jewell P. R., 2000, AstrophysicalJournal, Letters, 540, L107Huntress Jr. W. T., Mitchell G. F., 1979, AstrophysicalJournal, 231, 456Isokoski K., Bottinelli S., van Dishoeck E. F., 2013, As-tronomy and Astrophysics, 554, A100Jørgensen J. K., Favre C., Bisschop S. E., Bourke T. L., vanDishoeck E. F., Schmalzl M., 2012, Astrophysical Journal,Letters, 757, L4Keto E., 2002, Astrophysical Journal, 568, 754Keto E., Wood K., 2006, Astrophysical Journal, 637, 850Keto E. R., Ho P. T. P., Haschick A. D., 1988, Astrophys-ical Journal, 324, 920 Keto E. R., Ho P. T. P., Reid M. J., 1987, AstrophysicalJournal, Letters, 323, L117Khlifi M., Paillous P., Bruston P., Raulin F., GuilleminJ. C., 1996, Icarus, 124, 318Klaassen P. D., Wilson C. D., Keto E. R., Zhang Q., 2009,Astrophysical Journal, 703, 1308Kolpak M. A., Jackson J. M., Bania T. M., Clemens D. P.,Dickey J. M., 2003, Astrophysical Journal, 582, 756L´opez-Sepulcre A., Codella C., Cesaroni R., Marcelino N.,Walmsley C. M., 2009, Astronomy and Astrophysics, 499,811Lovas F. J., Dragoset R. A., 2004, Technical report, NISTRecommended Rest Frequencies for Observed InterstellarMolecular Microwave Transitions - 2002 Revision, (ver-sion 2.0.1), http://physics.nist.gov/restfreq . Na-tional Institute of Standards and Technology, Gaithers-burg, MD.Maxia C., Testi L., Cesaroni R., Walmsley C. M., 2001,Astronomy and Astrophysics, 371, 287McKee C. F., Tan J. C., 2002, Nature, 416, 59McKee C. F., Tan J. C., 2003, Astrophysical Journal, 585,850Modica P., Palumbo M. E., 2010, Astronomy and Astro-physics, 519, A22Mookerjea B., Casper E., Mundy L. G., Looney L. W.,2007, Astrophysical Journal, 659, 447Moscadelli L., Goddi C., Cesaroni R., Beltr´an M. T., Fu-ruya R. S., 2007, Astronomy and Astrophysics, 472, 867M¨uller H. S. P., Schl¨oder F., Stutzki J., Winnewisser G.,2005, Journal of Molecular Structure, 742, 215Occhiogrosso A., Viti S., Modica P., Palumbo M. E., 2011,Monthly Notices of the RAS, 418, 1923Ohishi M., McGonagle D., Irvine W. M., Yamamoto S.,Saito S., 1994, Astrophysical Journal, Letters, 427, L51Olmi L., Cesaroni R., Hofner P., Kurtz S., Churchwell E.,Walmsley C. M., 2003, Astronomy and Astrophysics, 407,225Olmi L., Cesaroni R., Neri R., Walmsley C. M., 1996, As-tronomy and Astrophysics, 315, 565Palla F., Stahler S. W., 1993, Astrophysical Journal, 418,414Pickett H. M., Poynter R. L., Cohen E. A., Delitsky M. L.,Pearson J. C., M¨uller H. S. P., 1998, Journal of Quanti-tiative Spectroscopy and Radiative Transfer, 60, 883Plume R., Jaffe D. T., Evans II N. J., 1992, AstrophysicalJournal, Supplement, 78, 505Pratap P., Megeath S. T., Bergin E. A., 1999, AstrophysicalJournal, 517, 799Rawlings J. M. C., Hartquist T. W., Menten K. M.,Williams D. A., 1992, Monthly Notices of the RAS, 255,471Remijan A., Shiao Y.-S., Friedel D. N., Meier D. S., SnyderL. E., 2004, Astrophysical Journal, 617, 384Remijan A. J., Markwick-Kemper A., ALMA WorkingGroup on Spectral Line Frequencies 2007, in AmericanAstronomical Society Meeting Abstracts Vol. 39 of Bul-letin of the American Astronomical Society, Splatalogue:Database for Astronomical Spectroscopy. p. 132.11Roberts J. F., Rawlings J. M. C., Viti S., Williams D. A.,2007, Monthly Notices of the RAS, 382, 733Snyder L. E., Lovas F. J., Hollis J. M., Friedel D. N., JewellP. R., Remijan A., Ilyushin V. V., Alekseev E. A., Dyubko c (cid:13) , 1–20 Calcutt et al.
S. F., 2005, Astrophysical Journal, 619, 914Sofia U. J., Meyer D. M., 2001, Astrophysical Journal, Let-ters, 554, L221Sollins P. K., Ho P. T. P., 2005, Astrophysical Journal, 630,987Turner B. E., Apponi A. J., 2001, Astrophysical Journal,Letters, 561, L207Vig S., Cesaroni R., Testi L., Beltr´an M. T., Codella C.,2008, Astronomy and Astrophysics, 488, 605Viti S., Caselli P., Hartquist T. W., Williams D. A., 2001,Astronomy and Astrophysics, 370, 1017Viti S., Collings M. P., Dever J. W., McCoustra M. R. S.,Williams D. A., 2004, Monthly Notices of the RAS, 354,1141Viti S., Williams D. A., 1999, Monthly Notices of the RAS,305, 755Wakelam V., Caselli P., Ceccarelli C., Herbst E., CastetsA., 2004, Astronomy and Astrophysics, 422, 159Wakelam V., Loison J.-C., Herbst E., Talbi D., Quan D.,Caralp F., 2009, Astronomy and Astrophysics, 495, 513Walsh A. J., Burton M. G., Hyland A. R., Robinson G.,1998, Monthly Notices of the RAS, 301, 640Wilson T. L., Rood R., 1994, Annual Review of Astron andAstrophys, 32, 191Wood D. O. S., Churchwell E., 1989, Astrophysical Journal,Supplement, 69, 831Woodall J., Ag´undez M., Markwick-Kemper A. J., MillarT. J., 2007, Astronomy and Astrophysics, 466, 1197Woods P. M., Kelly G., Viti S., Slater B., Brown W. A.,Puletti F., Burke D. J., Raza Z., 2012, Astrophysical Jour-nal, 750, 19Wu Y., Qin S.-L., Guan X., Xue R., Ren Z., Liu T., HuangM., Chen S., 2009, Astrophysical Journal, Letters, 697,L116 c (cid:13)000