Sulphur monoxide exposes a potential molecular disk wind from the planet-hosting disk around HD100546
Alice Booth, Catherine Walsh, Mihkel Kama, Ryan A. Loomis, Luke T. Maud, Attila Juhász
AAstronomy & Astrophysics manuscript no. SO_HD100546_final © ESO 2017December 20, 2017
Sulphur monoxide exposes a potential molecular disk wind fromthe planet-hosting disk around HD100546
Alice Booth , Catherine Walsh , Mihkel Kama , Ryan A. Loomis , Luke T. Maud & Attila Juhász School of Physics and Astronomy, University of Leeds, Leeds LS2 9JT, UKe-mail: [email protected], [email protected] Institute of Astronomy, Madingley Rd, Cambridge, CB3 0HA, UK Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA Leiden Observatory, Leiden University, PO Box 9513, 2300 RA Leiden, The NetherlandsReceived 12 / / / / ABSTRACT
Sulphur-bearing volatiles are observed to be significantly depleted in interstellar and circumstellar regions. This missing sulphur ispostulated to be mostly locked up in refractory form. With ALMA we have detected sulphur monoxide (SO), a known shock tracer,in the HD 100546 protoplanetary disk. Two rotational transitions: J = – 6 (301.286 GHz) and J = – 6 (304.078 GHz) aredetected in their respective integrated intensity maps. The stacking of these transitions results in a clear 5 σ detection in the stackedline profile. The emission is compact but is spectrally resolved and the line profile has two components. One component peaks at thesource velocity and the other is blue-shifted by 5 km s –1 . The kinematics and spatial distribution of the SO emission are not consistentwith that expected from a purely Keplerian disk. We detect additional blue-shifted emission that we attribute to a disk wind. The diskcomponent was simulated using LIME and a physical disk structure. The disk emission is asymmetric and best fit by a wedge ofemission in the north east region of the disk coincident with a ‘hot-spot’ observed in the CO J = Key words. protoplanetary disks - astrochemistry - stars: individual (HD 100546, HD 97048) - submillimeter: planetary systems -stars: pre-main sequence
1. Introduction
Investigating the chemical structure and evolution of protoplan-etary disks is important when studying planet formation as thecomposition of a planet is determined by the content of its parentdisk. The physical and chemical conditions in disks can be tracedwith observations of molecular line emission (see Henning & Se-menov 2013, Dutrey et al. 2014, and references therein). Theseobservations are key to understanding disk chemistry resultingfrom changes in disk structure due to planet-disk interactions.Transition disks were the first disks identified to possesssignatures of clearing by unseen planets (Strom et al. 1989).Protoplanet candidates have been identified in the cavities ofsome transition disks e.g. LkCa 15 (Sallum et al. 2015) andHD 169142 (Reggiani et al. 2014). These disks are expected tohave a rich observable chemistry due to the disk midplane be-ing directly exposed to far-UV radiation from the central star.Molecules that otherwise would be frozen out onto grains in thecold, shielded midplane of the disk may be detectable (Cleeveset al. 2011). High spatial resolution observations with ALMAallow investigation into the physical and chemical conditions as-sociated with planet-forming regions of nearby disks.The dominant volatiles (gas and ice) in protoplanetary disksare composed of hydrogen, oxygen, carbon, nitrogen and sul-phur. Sulphur is observed to be significantly depleted in circum- stellar regions with detections of gas-phase S-bearing moleculesonly accounting for ∼ ffl e et al. 1999).The missing sulphur is thought to reside in or on the solid dustgrains in the disk, the most abundant molecule being H S incometary ices (Bockelée-Morvan et al. 2000) and also in ironsulphides, another main component of primitive comets and me-teorites (Keller et al. 2002). The lack of agreement betweenobserved abundances and chemical models points towards asyet unaccounted for grain-surface processes preventing desorp-tion of H S and the formation of gas phase S-bearing species(Dutrey et al. 2011). This is supported by non-detections ofanticipated molecules despite deep targeted searches in disks(Martín-Doménech et al. 2016).Despite the depletion issue, sulphur-bearing species are use-ful tracers of physical processes in interstellar and circumstel-lar material. For example, SO is frequently detected as a tracerof shocked gas associated with the bipolar outflows from Class0 and I protostars e.g., Tafalla et al. 2010; Podio et al. 2015and Sakai et al. 2016. Searches for SO have found that detec-tions in outflows are ubiquitous but in protoplanetary disks in-frequent (Guilloteau et al. 2013). The first detection of SO ina circumstellar disk was reported by Fuente et al. (2010). Theyobserved the J = – 2 (138.178 GHz) transition in the transi-tional disk AB Aur. This detection was confirmed by Pacheco- Article number, page 1 of 16 a r X i v : . [ a s t r o - ph . E P ] D ec & A proofs: manuscript no. SO_HD100546_final
Vázquez et al. (2015) with detections of the higher frequencyJ = – 4 (206.176 GHz) and J = – 4 (219.949 GHz)transitions. Further observations and modelling suggest that theSO is distributed in a ring (145 to 384 au) and is depleted inthe region of the disk’s horse-shoe shaped dust trap (Pacheco-Vázquez et al. 2016). Most recently, Guilloteau et al. (2016) re-port the detection of the SO J = – 5 (251.826 GHz) andJ = – 5 (261.978 GHz) transitions in four disks out of 30 thatwere observed with the IRAM 30-m telescope. The aforemen-tioned studies show that SO has been detected in both T Tauriand Herbig disks. ALMA observations have revealed ring com-ponents of SO emission in Class 0 protostars (Ohashi et al. 2014;Podio et al. 2015; Sakai et al. 2016) that have been interpretedas accretion shocks at the disk-envelope interface.It is evident from existing observations that SO is a tracerof shocks in Class 0 and I protostars. However, SO is an elusivedisk molecule requiring high sensitivity observations for its de-tection: this is now possible with ALMA. We present the first de-tection of SO in the planet-hosting disk around HD 100546 fromALMA Cycle 0 observations gaining insight into the molecularcontent and structure of transition disks. We also report a non-detection of SO in the protoplanetary disk around HD 97048.The rest of this paper is structured as follows. Section 2 gives anoverview of previous observations of HD 100546 and Section 3describes our observations. The detected line emission and anal-ysis are detailed in Sections 4 and 5. In Section 6 the results arediscussed and we list our conclusions and prospects for furtherwork.
2. HD 100546
HD 100546 is a 2.4 M (cid:12)
Herbig Be star (van den Ancker et al.1998) at a distance of 109 +
4– 3 pc (Gaia Collaboration et al.2016a,b) and host to a bright disk with a position angle of 146 ◦ and an inclination of 44 ◦ (e.g. Walsh et al. 2014). This transi-tion disk is well observed because of its interesting structure andproximity.Modelling of the SED and interferometric observations haveshown that there is a cavity in the disk within 10 au of the centralstar and an inner dust disk ( ≤ J planet at 10 auand 15 M J planet at 68 au. Further modelling of the evolution ofthis system suggests that the inner planet formed first, ≥ ff y (Sturmet al. 2010; Pani´c et al. 2010; Carmona et al. 2011; Thi et al.2011; Goto et al. 2012; Liskowsky et al. 2012; Fedele et al. 2013;Hein Bertelsen et al. 2014). The disk hosts a large molecular gasdisk with the CO gas extending out to ≈
400 au (Pineda et al.2014; Walsh et al. 2014). In addition to this, there is evidence forthermal decoupling of gas and dust in the disk atmosphere andradial drift of millimeter-sized dust grains (see Bruderer et al.2012; Meeus et al. 2013; Walsh et al. 2014).
3. Observations
The observations of HD 100546 and the four SO transitionsanalysed in this work are detailed in Table 1 (ALMA program2011.0.00863.S). The continuum and CO line emission from thisdata have already been analysed (see Walsh et al. 2014, for fulldetails) and this work uses the self-calibrated, phase-correctedand continuum subtracted measurements sets. Imaging of thedata was done using Common Astronomy Software Application(CASA) version 4.6.0. The individual lines were each imagedat the spectral resolution of the observations (0.24 km s –1 , ap-plying Hanning smoothing) and then at a coarser resolution of1 km s –1 . Only the J = – 6 and J = – 6 transitions weredetected. These two SO lines were then stacked in the uv planeto increase S / N in the resulting channel maps (see Table 1). Thiswas done as follows. The central frequency of each spectral win-dow was transformed to the central frequency of the SO line itcontained using the CASA tool regridspw . These measurementsets were then concatenated using the CASA task concat . Theline emission was imaged at a velocity resolution of 1 km s –1 us-ing the CLEAN algorithm with natural weighting and a channelby channel mask guided by the spatial extent of the CO J = σ ). The J = – 1 transition has a low excitationenergy (21 K) and we expect the inner-most region of this diskto be warm ( >
20 K). Hence, the detection of the higher en-ergy transitions only is consistent with the expected temperatureof emitting molecular gas. Further, the lower energy transitionis significantly weaker (see Table 1). Including the J = – 7 transition in the stacking increased the noise level in the chan-nel maps thus degrading the S / N in the resulting images. Stack-ing wholly in the image plane gave the same results but with aslightly lower S / N than using the regrid plus concat method(also seen in the analysis of Walsh et al. 2016b).
4. Results
Figure 1 shows the integrated intensity maps of the two SO linesdetected in the imaging. They encompass the significant ( > σ )on source emission detected in the 1 km s –1 channel maps. Thisis emission across 11 channels from –7 km s –1 to 5 km s –1 withrespect to the source velocity. The source velocity of the emis-sion, as inferred from the CO J = –1 . The J = – 6 and J = – 6 transitions are detected with a peak S / N of 7 and 11 respectivelyin the integrated intensity. The r.m.s. noise was extracted fromthe region beyond the 3 σ contour of the integrated intensity. Article number, page 2 of 16ooth et al. 2017: Detection of Sulphur Monoxide in the HD100546 Disk
Table 1: ALMA band 7 observational parameters and sulphur monoxide transitions for HD 100546Date observed 18th November 2012Baselines 21 - 375 mWeighting naturalSO rotational transitions 7 – 6 – 6 – 7 – 1 – 6 + – 6 Rest frequency (GHz) 301.286 304.078 344.311 345.704 -Synthesised beam 1 (cid:48)(cid:48) . 1 × (cid:48)(cid:48) . 6 1 (cid:48)(cid:48) . 1 × (cid:48)(cid:48) . 6 1 (cid:48)(cid:48) . 0 × (cid:48)(cid:48) . 5 1 (cid:48)(cid:48) . 0 × (cid:48)(cid:48) . 5 1 (cid:48)(cid:48) . 1 × (cid:48)(cid:48) . 6Beam P.A. 24° 23° 40° 40° 24°Spectral resolution ( km s –1 ) 0.24 0.24 0.21 0.21 1.00r.m.s noise (channel –1 mJy beam –1 ) 10.9 9.9 17.4 16.2 4.2Peak emission (mJy beam –1 ) - - - - 24.7E u (K) 71.0 62.1 87.5 21.1 -Einstein A coe ffi cient (s –1 ) 3.429e-04 3.609e-04 5.186e-04 1.390e-07 - The values for the line frequencies and Einstein A coe ffi cients are from the Cologne Database for Molecular Spectroscopy(CDMS; Müller et al. 2001) and the Leiden Atomic and Molecular Database (LAMDA; Schöier et al. 2005). Stacking these two transitions results in a convincing detec-tion of SO in the channel maps (6 σ ) and line profile (5 σ ). Theline profiles of the individual transitions, at a channel width of1 km s –1 , are shown in Appendix A. Figure 2 shows the chan-nel maps of the stacked SO emission at a velocity resolution of1 km s –1 and with respect to the source velocity. The emissionreaches a peak S / N of 6 with an r.m.s. noise of 4.2 mJy beam –1 channel –1 . The noise in the channel maps was determined bytaking the r.m.s. of the line-free channels either side of the sig-nificant emission. The CO emission ( > σ ) is also plotted to al-low a comparison between these two molecules. When compar-ing the SO and CO emission, the SO emission is significantlymore compact than the CO emission and there is an excess blue-shifted component of SO emission that is not spatially consistentwith the blue-shifted disk emission traced in CO. This is furtherhighlighted by comparing the SO and CO line profiles shown inAppendix B.Figure 3 shows the line profile extracted from within the 3 σ contour of the stacked integrated intensity and covers a velocityrange of – 50 km s –1 to +
50 km s –1 about the source veloc-ity. This large velocity range was chosen to highlight the sig-nificance of the emission with respect to the underlying noise.The r.m.s. of the line profile was determined from the line-freechannels either side of the channels with significant emission.The line profile is double peaked which could indicate that theemission is originating from an inclined disk in Keplerian rota-tion. However, the trough of the profile is 2 km s –1 blueshiftedfrom the CO-determined source velocity. Because the emissionis clearly peaking on source, numerous checks were done to seeif the blue-shifted emission shift is real. The CASA velocity ref-erence frame is LSRK as is needed, the line frequencies are cor-rect, and there are no other emission lines within the consideredvelocity range that could attribute to the blue-shifted emission.In addition to this, we checked the removal of 5% to 10% of thelonger and shorter baselines and the double-peaked line profilepersists. We also checked that the frequency axis of the spec-tral cube was not incorrectly indexed. The line profiles of theindividual transitions (Appendix A) both have the same line pro-file shape. Therefore, we did not induce any additional signalthrough stacking.To investigate the spatial distribution of both components ofemission (on source emission and blue shifted emission, respec-tively), a moment zero map was created for both. Figure 4 showsthe integrated intensity maps from -7.5 km s –1 to -2.5 km s –1 andfrom -2.5 km s –1 to 4.5 km s –1 . These velocity ranges are high- Table 2: Moment map S / N and r.m.s.Moment map S / N r.m.s.(mJy beam –1 km s –1 )J = – 6 disk component 4.8 17J = – 6 wind component 5.0 15J = – 6 disk component 8.4 15J = – 6 wind component 6.8 13lighted in the stacked line profile (Figure 3) and the individuallines profiles (Appendix A). The peak S / N and r.m.s. for each ofthese integrated intensity maps are listed in Table 2. The peakemission in the two maps is spatially o ff set but the exact separa-tion of these two components is unclear because the emission isof the order of the same size as the beam. The ratio of the peakemission in the moment zero maps of the J = – 6 transitionto the J = – 6 transition is higher in the blue-shifted compo-nent of emission compared with the emission at source velocity.This likely indicates that the blue-shifted component is tracingwarmer gas. The stacked moment maps over the two velocityranges are shown in Appendix C and the spatial o ff set betweenthe two components is more significant.From this analysis we propose that we are observing twocomponents of emission and not just Keplerian disk emissionthat is blue-shifted relative to the source velocity. The emissionin the line profile that peaks on source and aligns kinematicallywith the CO in the channel maps (Figure 2) can be attributed todisk emission. The morphology of the on-source singly-peakeddisk component might be explained by asymmetric SO emissionwhich is arising from the north-east side of the disk only. Thisis investigated further in Section 5. The blue-shifted emissionpeaks o ff source spectrally and spatially and it is attributed to adisk wind where material is being driven from the surface of thedisk resulting in a blue-shift along the line of sight. We use a matched filter code developed for interferometric datasets to confirm the detection of the two SO lines and to search foremission from the lines undetected in the image plane (Loomiset al. 2017). The detection of weak spectral lines direct from the vis_sample is publicly available at: https://github.com/AstroChem/vis\_sample Article number, page 3 of 16 & A proofs: manuscript no. SO_HD100546_final
Fig. 1: Integrated intensity maps of the two SO transitions taken over a 11 km s –1 velocity range. Left: the J = – 6 transition withan r.m.s of 22 mJy beam –1 km s –1 and a peak emission of 151 mJy beam –1 km s –1 resulting in a S / N of 6.9. Right: the J = – 6 transition with an r.m.s of 19 mJy beam –1 km s –1 and a peak emission of 206 mJy beam –1 km s –1 resulting in a S / N of 10.8. Theblack contours are at intervals of σ going from 3 σ to peak. uv data is optimal as this saves the computational time neededto generate images and eliminates imaging biases. Since theposition-velocity pattern of a disk in Keplerian rotation is wellcharacterised, matched filtering can be used to detect weak spec-tral lines in disks (Loomis et al. 2017). This technique has beenshown to increase S / N of the methanol detection in TW Hya by53% (Walsh et al. 2016b). A filter can either be a strong line de-tected in the same disk or a model of the anticipated emission.The image cube used as a filter is sampled in the uv plane andthese visibilities are cross-correlated with the low S / N visibil-ities. This is done by sliding the filter though the data channel-by-channel along the velocity axis. If there is a detectable signal,i.e. emission with a similar position and velocity distribution asthe filter, the filter response will peak at the source velocity ofthe emission.We use the CLEAN image of the CO line (J = σ with the r.m.s. noise normalised to 1.Figure 4 shows the response of the three detected SO lines tothe compact CO and best fit LIME wedge model filters. For theCO filter three of the four lines were detected, the J = – 6 ,J = –6 and J = –7 transitions. They have a peak responseof 6.8, 4.2 and 4.0 respectively. There is no detection of the lowerenergy J = – 1 transition. We tested di ff erent CO filters withdi ff erent compression factors of 1 /
2, 1 / /
4. We found thatthe J = – 6 transition is picked up in all the filters but thehigher excitation lines have an improved response with the morecompact filters. The best fit LIME model also detects the same three SO lines. The J = – 7 filter response is approximatelythe same as with the compact CO filter but the other two linesresponses are quite di ff erent.The peak of the responses are not all at the expected sourcevelocity supporting the theory that we have multiple velocitycomponents of emission including a possible disk wind. Thematched filter has confirmed the detection of the two lines de-tected in the image plane and they are observed at a substantiallyhigher S / N than in the channel maps. It has also facilitated thedetection of a line that we do not detect in the imaging. Fur-ther, the matched filter line response scales with intensity, so wealso now have rudimentary excitation information to motivateour models.
5. Modelling the SO disk emission using LIME
The abundance of SO was estimated by matching the ob-served emission with simulated emission generated using a HD100456 physical disk structure (Figure 6) and LIME version 1.5(LIne Modelling Engine; Brinch & Hogerheijde 2010). Ray-tracing calculations were done assuming LTE and the appro-priate distance, inclination and position angle for the source.The molecular data files for sulphur monoxide were takenfrom the Leiden Atomic and Molecular Database (LAMDA;http: // home.strw.leidenuniv.nl / moldata / SO.html). To check thatLTE calculations were a good approximation we calculated thecritical density of the transitions. When the number density ofgas is greater than the critical density collisional processes dom-inate. In this regime the level populations are determined by theBoltzmann distribution and LTE is an accurate assumption. The
Article number, page 4 of 16ooth et al. 2017: Detection of Sulphur Monoxide in the HD100546 Disk 12.00 km s 1 11.00 km s 1 10.00 km s 1 9.00 km s 1 8.00 km s 1 7.00 km s 1 6.00 km s 1 5.00 km s 1 4.00 km s 1 3.00 km s 1 2.00 km s 1 1.00 km s 1 0.00 km s 1 1 1 1 1 1 1 1 R e l a t i v e D e c li na t i on ( a r cs e c ) COSO8.00 km s 1 1 1 1 1 Fig. 2: Channel maps of the stacked SO emission (red contours) and the CO J = σ clip (grey colour map).The stacked SO emission has with an r.m.s. noise of 4.2 mJy beam –1 channel –1 and peak emission of 24.7 mJy beam –1 resulting ina peak S / N of 5.9. The contours are from 3 σ to peak in intervals of σ . The velocities stated are with respect to the source velocity ofthe emission.critical density is determined by:n cr = A ul (cid:80) l (cid:48) < u γ ul (cid:48) where A ul and γ ul are the Einstein A and collisional rate coe ffi -cients of the transition. The critical density for the SO J = –6 transition was determined to be 6 × cm –3 at 100 K usingthe LAMDA molecular data with the collisional rates from Liqueet al. (2006). The other transitions are of a similar order of mag-nitude. As the matched filter is a linear process the relative responsesfor a pair of lines can be used as a proxy for their relative inten-sities after correcting for the di ff erence in noise levels in eachline. These relative intensities were converted to line ratios andthis information was used to confine the location of the SO inthe disk with respect to the temperature and density conditions.Model line ratios were calculated from line intensities deter-mined using the RADEX radiative transfer code assuming an SOcolumn density of 10 cm –2 motivated by full chemical mod-els. RADEX is a non-LTE 1D radiative transfer code that canbe used with the intensity of an observed particular molecular Article number, page 5 of 16 & A proofs: manuscript no. SO_HD100546_final
50 40 30 20 10 0 10 20 30 40 50V - V
LSRK (km s -1 )201510505101520253035 F l u x D e n s i t y ( m J y ) σ Stacked
Fig. 3: Line profile extracted from within the 3 σ extent of the SOstacked integrated intensity with an r.m.s. noise of 5.7 mJy anda peak flux of 30.4 mJy resulting in a S / N of 5.3. Highlightedin red and blue are the velocity ranges of emission used in themoment maps of the individual lines in Figure 4.line to estimate the excitation temperature and column densityof the gas, assuming an isothermal, homogeneous medium withno significant velocity gradient (van der Tak et al. 2007). Theline ratios of the three lines were calculated over a grid of tem-peratures and densities and the results are shown in Figure 7.These model line ratios were compared to the observed line ra-tios taken from the matched filter responses. Within the velocityrange we defined as disk emission, these are 1.5, 1.6 and 1.1for the J = – 6 / J = – 6 , J = – 6 / J = – 7 andJ = – 6 / J = – 7 line ratios respectively. A selection ofthe RADEX results are shown in a table in Appendix D alongwith the observed ratios and their associated errors. The regimethat best fits our observations is a H density between 10 to10 cm –3 and a gas temperature between 50 and 100 K. Theseconditions result in the SO being distributed in a ring from 20 to100 au in a layer above the mid-plane (see Figure 6). This is inagreement with previous modelling of sulphur volatiles in disks(e.g. Dutrey et al. 2011) and is in agreement with the compactnature of the SO emission we observe. Modelling the SO in aregion of lower density and higher temperature resulted in sig-nificantly more extended emission than in our observations. Weopt to only model the near surface of the disk as we assume thatin this region of the disk ( <
100 au) the optically thick dust emis-sion will block the emission from the molecular gas in the farsurface of the disk.A set of models were run varying the fractional abundanceof SO with respect to H . This was done in order to match theobserved peak in the integrated intensity of the disk componentfor each of the two transitions: 80 mJy beam –1 km s –1 for theJ = – 6 transition and 124 mJy beam –1 km s –1 for theJ = – 6 . A model for a full disk was calculated with a frac-tional abundance of 3.5 × –7 with respect to H resulting in apeak intensity for each of the lines of 92 mJy beam –1 km s –1 and 109 mJy beam –1 km s –1 respectively. These values matchthe peak emission of the observations within the 1 σ range. Theresidual maps of the observed integrated intensity minus themodel integrated intensity for the two lines are shown in Figure8. The residuals show that the observed emission is asymmet-ric peaking in the north east region of the disk and a full disk isnot an accurate representation of the data. A second model was run restricting the SO to a specific angular region of the disk. A45° wedge of emission was calculated with the optimal positionpicked by eye from the residual maps to be from 0° to 45° fromthe disk’s major axis. A fractional abundance of 5.0 × –6 withrespect to H resulted in a peak intensity for each of the linesof 93 mJy beam –1 km s –1 and 96 mJy beam –1 km s –1 respec-tively. These values match the peak emission of the observationswithin the 2 σ range. This fractional abundance is greater than the‘depleted’ sulphur fractional abundance observed dark clouds(SO ≈ –8 ; Ru ffl e et al. 1999). This suggests that there are en-ergetic processes occurring in HD100546 releasing a source ofrefractory sulphur into the gas phase. The fractional abundanceof SO derived from the LIME modelling is model dependent asit depends on the gas density of the region of the disk wherethe SO is located. This model well reproduces the integrated in-tensity however, the kinematics trace red-shifted disk emission.The peak in both of the line profiles for the wedge models is1.7 × the observed line profile peaks and the model emission isover a narrower velocity range. The model line profiles for boththe disk and the wedge models are compared with the observedline profiles in Appendix E. Further refinement of the disk emis-sion component requires better data as the emission is the samesize scale as the beam.
6. Discussion
We detect SO in the protoplanetary disk around HD 100546 forthe first time. In the image plane we have a clear detection oftwo lines in the integrated intensity maps and we improve theS / N in the channel maps and line profile by stacking. From themorphology of the line profile and the asymmetric distributionof the emission it is likely we are observing two components ofemission: a wedge of disk emission and a blue-shifted compo-nent (-5 km s –1 ). We use a matched filter to better determine therelative intensities of the SO lines in the data set and confirmthe detection of three transitions and a non detection of the low-est energy transition. The relative intensities of the three detectedlines are used to motivate the location of SO in LIME modelling.The residuals from the observed and modelled integrated inten-sity maps reveal that the integrated emission is indeed asymmet-ric peaking north-east of the source position. This is coincidentwith a ‘hot-spot’ observed in CO emission relating to a possibledisk warp (Walsh et al. 2017). The CO J = –1 with re-spect to the source velocity) component is spatially inconsistentwith the expected location of blue-shifted Keplerian disk emis-sion. We attribute this emission to a disk wind. This hypothesisis summarised in a cartoon in Figure 9.We did not detect any of the four SO lines in the comple-mentary HD 97048 Cycle 0 data (see Walsh et al. 2016a) usingthe imaging methods detailed in Sections 3 and 4. The stackedemission in the channel maps at a velocity resolution of 1 km s –1 reaches an r.m.s. noise of 6 mJy beam –1 , and there was no sig-nificant response using the matched filter analysis. This supportsour hypothesis that SO is tracing a physical mechanism uniqueto HD 100546. The disks around HD 100546 and HD 97048 Article number, page 6 of 16ooth et al. 2017: Detection of Sulphur Monoxide in the HD100546 Disk
Fig. 4: Integrated intensity maps of the two SO transitions over two velocity ranges. Top: the J = –6 transition from -7.5 km s –1 to-2.5 km s –1 (left) and -2.5 km s –1 to 4.5 km s –1 (right). Bottom: the J = – 6 transition from -7.5 km s –1 to -2.5 km s –1 (left) and-2.5 km s –1 to 4.5 km s –1 (right). The peak S / N and r.m.s. for each of these maps are listed in Table 2. The black contours are atintervals of σ going from 3 σ to peak.have significantly di ff erent structures: the gap(s) in the sub-mmdust are further from the star in HD 97048 and there are no pro-toplanet candidates yet detected in this disk (Quanz et al. 2012;Walsh et al. 2016a; van der Plas et al. 2017). Further detailedmodelling is required to determine the chemical origin of theSO emission in the HD 100546 disk and how the physical andchemical conditions di ff er from those in the HD 97048 disk.The only other disk from which SO emission has been im-aged is AB Aur (Pacheco-Vázquez et al. 2016). In this transition disk the SO is located further from the star in a ring from ap-proximately 145 to 384 au with a maximum modelled abundanceof 2 × –10 with respect to H . The SO, like in HD 100456, isthought to reside in a layer between the surface and the midplaneof the disk. The relative abundance of SO observed in AB Auris a few orders of magnitude less than in HD 100546 and theemission is not necessarily tracing the same process. In AB Aur,SO is proposed as a chemical tracer of the early stages of planetformation as the abundance of SO appears to decrease towards Article number, page 7 of 16 & A proofs: manuscript no. SO_HD100546_final
50 40 30 20 10 0 10 20 30 40 50V - V
LSRK (km s -1 )32101234567 F il t e r R e s p o n s e ( σ ) − E u = 71K Compact CO J=3-2 Filter
50 40 30 20 10 0 10 20 30 40 50V - V
LSRK (km s -1 )32101234567 F il t e r R e s p o n s e ( σ ) − E u = 62K Compact CO J=3-2 Filter
50 40 30 20 10 0 10 20 30 40 50V - V
LSRK (km s -1 )32101234567 F il t e r R e s p o n s e ( σ ) − E u = 87K Compact CO J=3-2 Filter
Fig. 5: Matched filter responses for the three detected SO transitions. Left: the results using a spatially compact (1 /
4) version ofthe CO channel maps as a filter. Right: the results from using the best fit LIME wedge model as a filter (see Section 5 for details).Highlighted in red and blue are the velocity ranges of emission we attribute to a disk and a wind component respectively.the disk’s dust trap, a local pressure maximum thought to be thesite of future plant formation (e.g. van der Marel et al. 2013).
Sulphur chemistry, particularly the evolution of S-bearingmolecules on grain surfaces, is not fully understood as current models fail to reproduce observed abundances (e.g. Guilloteauet al. 2016). If we are observing a wedge or partial ring of SO,the inner edge of the emission coincides with the inner edge ofthe sub-millimetre dust ring at approximately 20 au. Since HD100546 is a transition disk the midplane material is exposed tofar-UV photons from the central star. This will cause the desorp-tion of molecules from icy grain mantles. H O ice has been ob-
Article number, page 8 of 16ooth et al. 2017: Detection of Sulphur Monoxide in the HD100546 Disk
Fig. 6: The HD 100546 disk physical structure from Kama et al. (2016). Top left and moving clockwise: the dust temperature (K),gas temperature (K), UV flux (in units of the interstellar radiation field) and number density (cm –3 ). The two white contours in eachof the temperature plots correspond to temperatures 30 K and 50 K. The shaded region highlights the location of the SO motivatedby RADEX calculations and used in the LIME modelling.Fig. 7: The RADEX modelling results of the selected SO line ratios. From left to right are the three ratios: J = – 6 / J = – 6 ,J = – 6 / J = – 7 and J = – 6 / J = – 7 . The solid black line is the observed line ratio and the dotted black lines arethe error bars. The shaded region highlights the temperature and density conditions chosen for the location of the SO in the LIMEmodelling.served in this disk (Honda et al. 2016) and H S ice is a primarycomponent of cometary ices (Bockelée-Morvan et al. 2000). Thephotodissociation of these molecules originating from cosmicrays or UV photons, depending on the height of the gas in the disk, would create the reactants possible to form SO;H O + h ν −→ OH + H −→ O + H + HH S + h ν −→ HS + H −→ S + H + HHS + O −→ SO + HS + OH −→ SO + H.The activation energy, E A , of the two SO formation reac-tions is zero (KIDA: KInetic Database for Astrochemistry Article number, page 9 of 16 & A proofs: manuscript no. SO_HD100546_final
Fig. 8: Residual maps from the disk emission integrated intensity and the LIME models for each of the transitions. Top: diskemission minus disk model. Bottom: disk emission minus wedge model. Overlaid are dashed -5,-4 and -3 σ contours and solid 3, 4and 5 σ contours.http: // kida.obs.u-bordeaux1.fr / ). There are a few possible reac-tions for the destruction of SO to form SO (Millar & Herbst1990);SO + O −→ SO + h ν SO + OH −→ SO + H.In AB Aur the abundance of SO decreases with increasingdensity towards the disk’s dust trap. This is attributed to theincrease in conversion of SO to SO via radiative associationwith atomic oxygen and then freeze out of SO onto dust grains (Pacheco-Vázquez et al. 2016). In HD 100546, the density andtemperature of the disk may have been perturbed creating theconditions for the localised formation of the observed asymmet-ric SO. However, chemical modeling of warped disks and asso-ciated temperature perturbations is required to confirm this hy-pothesis.From observations of cometary volatiles in the particular, forthe case of 67P, the total abundance of sulphur-bearing speciesdetected is consistent with the solar abundance of sulphur (Cal-monte et al. 2016). This means that if our solar system is typical Article number, page 10 of 16ooth et al. 2017: Detection of Sulphur Monoxide in the HD100546 Disk
Fig. 9: Cartoon of HD 100546 wedge disk emission and SO diskwind.then the observed depletion of sulphur in circumstellar regionsmay be an observational e ff ect as we have not been able to detectthe various forms of sulphur. The form of the sulphur, whetherit resides in refractory or volatile form in planet-forming disksis still an open question. For the SO we detect in HD 100546 itis unclear as to its origin, e.g., if it is a result of the volatile re-actions described above or whether it has been released from re-fractory materials due to a shock as suggested by the abundancewe determine. Further observations may make this clearer. The influence of the massive companion at approximately 10 auin the disk may cause the disk velocity structure to depart fromsimple Keplerian rotation in the inner region. The protoplanetembedded in the disk at 50 au may also have an e ff ect (Quillen2006). If the disk is warped, the line of sight inclination willvary radially changing the velocity structure. Previous observa-tions with APEX show the CO emission from the HD 100546protoplanetary disk is asymmetric (Pani´c et al. 2010) suggestingthat one side of the outer disk is colder by 10-20 K than the otheror that there is a shadow on the outer disk caused by a warped ge-ometry of the inner disk. Shadows resulting from disk warps andtheir e ff ect on gas kinematics have been observed in a few othersources e.g. HD 142527 (Casassus et al. 2015). There is evidencefor a possible warp in the inner 100 au of the HD 100546 diskfrom a detailed study of the CO J = ff ect on the chemical content of the disk.In planet-forming Class II disks, jets / outflows are not the maindriver of disk dispersal. Instead the removal of angular momen-tum from the disk material can be achieved by slower disk winds( <
30 km s –1 ). Photoevaporative disk winds are thought to be theprimary disk dispersal mechanism (Alexander et al. 2014). MHDdisk winds also drive disk dispersal but are less well understood(Ercolano & Pascucci 2017). Evidence of photoevaporative diskwinds has been detected from a number of sources in the formof blue-shifted (up to 10 km s –1 ) line profiles of forbidden lineemission in the optical (e.g. Pascucci et al. 2011; Ercolano &Owen 2016). In addition to this, ALMA observations show aspatially resolved molecular disk wind originating from the HD163296 disk (Klaassen et al. 2013), and a molecular protostel-lar outflow from TMC1A launched by a disk wind originatingfrom a Keplerian disk (Bjerkeli et al. 2016). We checked for anylarge scale or high velocity ( >
10 km s –1 ) CO or SO emissionfrom HD100546 but none was detected. Gas launched from theHD 100546 disk surface with a velocity of a few km s –1 wouldhave a blue shift along the line of sight to the observer and couldaccount for the blue-shifted emission we observe. The mecha-nism for launching this material and why it is traced in the SOemission is unclear. The lack of observed excess blue-shiftedCO emission is due to the SO originating from a layer in thedisk that is higher than the emitting layer of the CO J = u = ff ect in CO will require ob-servations of higher J lines which will be tracing the warmergas in the atmosphere. We surmise that the red-shifted counter-part of the disk wind, launching from the far side of the disk, isobscured by the optically thick dust disk. Determination of thechemical origin of the SO will help to shed light on whether thewind is MHD driven (ion-molecule chemistry) or photoevapora-tive (photon-dominated chemistry) in nature.An alternative explanation for the SO emission is that it is theresult of an accretion shock due to a circumplanetary disk. Theposition angle of the SO emission coincides with the observedinfrared point source in the disk at approximately 50 au that hasbeen attributed to a protoplanet (Quanz et al. 2013; Currie et al.2015). We have shown that SO is detectable in protoplanetary disks withALMA uncovering a sulphur reservoir in the HD 100546 proto-planetary disk. In addition, we have shown that SO may be atracer of a molecular disk wind. New data with better spatial andspectral resolution are required to truly disentangle the disk andwind components of the emission. This will allow for an accuratedetermination of the spatial distribution of the disk emission andthe launch region of the wind. Observations of sulphur-bearingspecies in this disk and others will help to constrain the fractionof the cosmic abundance of sulphur partitioned in the refractoryand the volatile materials. Transition disks are important targetsas their cavities may expose the hidden sulphur.
Acknowledgements.
This paper makes use of the following ALMA data:ADS / JAO.ALMA / NRAO and NAOJ. A.B. acknowledges the studentship funded bythe Science and Technology Facilities Council of the United Kingdom (STFC).C.W. acknowledges financial support from the Netherlands Organisation for Sci-entific Research (NWO, grant 639.041.335) and start-up funds from the Uni-versity of Leeds, UK. This work is partly funded by STFC consolidated grant
Article number, page 11 of 16 & A proofs: manuscript no. SO_HD100546_final number
References
Acke, B. & van den Ancker, M. E. 2006, A&A, 449, 267Alexander, R., Pascucci, I., Andrews, S., Armitage, P., & Cieza, L. 2014, Proto-stars and Planets VI, 475, ed. H. Beuther et al. (Tucson, AZ: Univ. ArizonaPress)Avenhaus, H., Quanz, S. P., Meyer, M. R., et al. 2014, ApJ, 790, 56Benisty, M., Stolker, T., Pohl, A., et al. 2017, A&A, 597, A42Benisty, M., Tatulli, E., Ménard, F., & Swain, M. R. 2010, A&A, 511, A75Bjerkeli, P., van der Wiel, M. H. D., Harsono, D., Ramsey, J. P., & Jørgensen,J. K. 2016, Nature, 540, 406Bockelée-Morvan, D., Lis, D. C., Wink, J. E., et al. 2000, A&A, 353, 1101Bouwman, J., de Koter, A., Dominik, C., & Waters, L. B. F. M. 2003, A&A, 401,577Brinch, C. & Hogerheijde, M. R. 2010, A&A, 523, A25Brittain, S. D., Carr, J. S., Najita, J. R., Quanz, S. P., & Meyer, M. R. 2014, ApJ,791, 136Brittain, S. D., Najita, J. R., & Carr, J. S. 2009, ApJ, 702, 85Bruderer, S., van Dishoeck, E. F., Doty, S. D., & Herczeg, G. J. 2012, A&A, 541,A91Calmonte, U., Altwegg, K., Balsiger, H., et al. 2016, MNRAS, 462, S253Carmona, A., van der Plas, G., van den Ancker, M. E., et al. 2011, A&A, 533,A39Casassus, S., Marino, S., Perez, S., et al. 2015, ApJ, 811, 92Cleeves, L. I., Bergin, E. A., Bethell, T. J., et al. 2011, ApJ, 743, L2Currie, T., Cloutier, R., Brittain, S., et al. 2015, ApJ, 814, L27Currie, T., Muto, T., Kudo, T., et al. 2014, ApJ, 796, L30Dutrey, A., Guilloteau, S., & Guelin, M. 1997, A&A, 317, L55Dutrey, A., Semenov, D., Chapillon, E., et al. 2014, Protostars and Planets VI,317, ed. H. Beuther et al. (Tucson, AZ: Univ. Arizona Press)Dutrey, A., Wakelam, V., Boehler, Y., et al. 2011, A&A, 535, A104Ercolano, B. & Owen, J. E. 2016, MNRAS, 460, 3472Ercolano, B. & Pascucci, I. 2017, Royal Society Open Science, 4, 170114Fedele, D., Bruderer, S., van Dishoeck, E. F., et al. 2013, A&A, 559, A77Fuente, A., Cernicharo, J., Agúndez, M., et al. 2010, A&A, 524, A19Gaia Collaboration, Brown, A. G. A., Vallenari, A., et al. 2016a, A&A, 595, A2Gaia Collaboration, Prusti, T., de Bruijne, J. H. J., et al. 2016b, A&A, 595, A1Garufi, A., Quanz, S. P., Schmid, H. M., et al. 2016, A&A, 588, A8Goto, M., van der Plas, G., van den Ancker, M., et al. 2012, A&A, 539, A81Guilloteau, S., Di Folco, E., Dutrey, A., et al. 2013, A&A, 549, A92Guilloteau, S., Reboussin, L., Dutrey, A., et al. 2016, A&A, 592, A124Hein Bertelsen, R. P., Kamp, I., Goto, M., et al. 2014, A&A, 561, A102Henning, T. & Semenov, D. 2013, Chemical Reviews, 113, 9016Honda, M., Kudo, T., Takatsuki, S., et al. 2016, ApJ, 821, 2Kama, M., Bruderer, S., van Dishoeck, E. F., et al. 2016, A&A, 592, A83Keller, L. P., Hony, S., Bradley, J. P., et al. 2002, Nature, 417, 148Klaassen, P. D., Juhasz, A., Mathews, G. S., et al. 2013, A&A, 555, A73Lique, F., Dubernet, M.-L., Spielfiedel, A., & Feautrier, N. 2006, A&A, 450, 399Liskowsky, J. P., Brittain, S. D., Najita, J. R., et al. 2012, ApJ, 760, 153Loomis, R. A., Öberg, K. I., Andrews, S. M., et al. 2017, A&A, submittedMartín-Doménech, R., Jiménez-Serra, I., Muñoz Caro, G. M., et al. 2016, A&A,585, A112Meeus, G., Salyk, C., Bruderer, S., et al. 2013, A&A, 559, A84Millar, T. J. & Herbst, E. 1990, A&A, 231, 466Mulders, G. D., Paardekooper, S.-J., Pani´c, O., et al. 2013, A&A, 557, A68Müller, H. S. P., Thorwirth, S., Roth, D. A., & Winnewisser, G. 2001, A&A, 370,L49Ohashi, N., Saigo, K., Aso, Y., et al. 2014, ApJ, 796, 131Pacheco-Vázquez, S., Fuente, A., Agúndez, M., et al. 2015, A&A, 578, A81Pacheco-Vázquez, S., Fuente, A., Baruteau, C., et al. 2016, A&A, 589, A60Pani´c, O., van Dishoeck, E. F., Hogerheijde, M. R., et al. 2010, A&A, 519, A110Pascucci, I., Sterzik, M., Alexander, R. D., et al. 2011, ApJ, 736, 13Pineda, J. E., Quanz, S. P., Meru, F., et al. 2014, ApJ, 788, L34Pinilla, P., Birnstiel, T., & Walsh, C. 2015, A&A, 580, A105Podio, L., Codella, C., Gueth, F., et al. 2015, A&A, 581, A85Quanz, S. P., Amara, A., Meyer, M. R., et al. 2015, ApJ, 807, 64Quanz, S. P., Amara, A., Meyer, M. R., et al. 2013, ApJ, 766, L1Quanz, S. P., Birkmann, S. M., Apai, D., Wolf, S., & Henning, T. 2012, A&A,538, A92Quillen, A. C. 2006, ApJ, 640, 1078Reggiani, M., Quanz, S. P., Meyer, M. R., et al. 2014, ApJ, 792, L23Ru ffl e, D. P., Hartquist, T. W., Caselli, P., & Williams, D. A. 1999, MNRAS, 306,691 Sakai, N., Oya, Y., López-Sepulcre, A., et al. 2016, ApJ, 820, L34Sallum, S., Follette, K. B., Eisner, J. A., et al. 2015, Nature, 527, 342Schöier, F. L., van der Tak, F. F. S., van Dishoeck, E. F., & Black, J. H. 2005,A&A, 432, 369Strom, K. M., Strom, S. E., Edwards, S., Cabrit, S., & Skrutskie, M. F. 1989, AJ,97, 1451Sturm, B., Bouwman, J., Henning, T., et al. 2010, A&A, 518, L129Tafalla, M., Santiago-García, J., Hacar, A., & Bachiller, R. 2010, A&A, 522,A91Tatulli, E., Benisty, M., Ménard, F., et al. 2011, A&A, 531, A1Thi, W.-F., Ménard, F., Meeus, G., et al. 2011, A&A, 530, L2Tieftrunk, A., Pineau des Forets, G., Schilke, P., & Walmsley, C. M. 1994, A&A,289, 579van den Ancker, M. E., de Winter, D., & Tjin A Djie, H. R. E. 1998, A&A, 330,145van der Marel, N., van Dishoeck, E. F., Bruderer, S., et al. 2013, Science, 340,1199van der Plas, G., van den Ancker, M. E., Acke, B., et al. 2009, A&A, 500, 1137van der Plas, G., Wright, C. M., Ménard, F., et al. 2017, A&A, 597, A32van der Tak, F. F. S., Black, J. H., Schöier, F. L., Jansen, D. J., & van Dishoeck,E. F. 2007, A&A, 468, 627Walsh, C., Daley, C., Facchini, S., & Juhász, A. 2017, A&A, 607, A114Walsh, C., Juhász, A., Meeus, G., et al. 2016a, ApJ, 831, 200Walsh, C., Juhász, A., Pinilla, P., et al. 2014, ApJ, 791, L6Walsh, C., Loomis, R. A., Öberg, K. I., et al. 2016b, ApJ, 823, L10 Article number, page 12 of 16ooth et al. 2017: Detection of Sulphur Monoxide in the HD100546 Disk
Appendix A: Individual line profiles
50 40 30 20 10 0 10 20 30 40 50V - V
LSRK (km s -1 )201510505101520253035 F l u x D e n s i t y ( m J y ) σ − E u = 71K
50 40 30 20 10 0 10 20 30 40 50V - V
LSRK (km s -1 )201510505101520253035 F l u x D e n s i t y ( m J y ) σ − E u = 62K Fig. A.1: Line profiles of the individual J = – 6 (left) and J = – 6 (right) transitions extracted from within the 3 σ extent of theirrespective intensity maps. Both line profiles reach a S / N of 3 with r.m.s. noise of 10.8 mJy and 8.7 mJy respectively. Highlighted inred and blue are the velocity ranges of emission used in the moment maps in Figure 4.
Appendix B: SO and CO J=3-2 line profiles
Fig. B.1: SO stacked (red) line profile and CO J = σ extent of their respectivestacked integrated intensity maps. The SO stacked reaches a S / N of 5 and the CO J = / N of 375. Note that the r.m.s.for the CO J = Article number, page 13 of 16 & A proofs: manuscript no. SO_HD100546_final
Appendix C: Stacked moment maps
Fig. C.1: Integrated intensity maps of the stacked SO emission over two velocity ranges: -7.5 km s –1 to -2.5 km s –1 (left) with a S / Nof 7 and -2.5 km s –1 to 4.5 km s –1 (right) with a S / N of 9. The r.m.s. noise these maps is 11 and 14 mJy beam –1 km s –1 and the peakemission is 80 and 123 mJy beam –1 km s –1 respectively. The black contours are at intervals of σ going from 3 σ to peak. Article number, page 14 of 16ooth et al. 2017: Detection of Sulphur Monoxide in the HD100546 Disk
Appendix D: RADEX Line Ratios
Table D.1: Model and observed line ratios for detected SO transitions.Observed line ratios- - 1.5 ± ± ± H (cm –3 ) T k (K) J = – 6 / J = – 6 J = – 6 / J = – 7 J = – 6 / J = – 7
25 1.909 7.000 3.66710
25 1.726 3.982 2.29610
25 1.607 2.481 1.54410
25 1.587 2.302 1.45110
25 1.583 2.288 1.44610
25 1.583 2.288 1.44610
50 1.659 4.056 2.44410
50 1.489 2.285 1.53510
50 1.386 1.530 1.10410
50 1.367 1.449 1.06010
50 1.364 1.442 1.05810
50 1.364 1.442 1.05810
100 1.546 2.930 1.89510
100 1.400 1.683 1.20310
100 1.296 1.197 0.92410
100 1.273 1.150 0.90310
100 1.269 1.144 0.90210
100 1.270 1.145 0.90210
250 1.555 2.348 1.51010
250 1.371 1.318 0.96110
250 1.239 1.015 0.81910
250 1.212 0.990 0.81610
250 1.211 0.984 0.81210
250 1.203 0.978 0.81310
500 1.551 2.182 1.40710
500 1.358 1.203 0.88610
500 1.218 0.962 0.79010
500 1.192 0.939 0.78810
500 1.191 0.939 0.78910
500 1.193 0.939 0.787
Article number, page 15 of 16 & A proofs: manuscript no. SO_HD100546_final
Appendix E: Individual line profiles with LIME disk model line profiles
Fig. E.1: Line profiles of the individual J = – 6 (left) and J = – 6 (right) transitions with the LIME model line profiles plottedon the top: disk model (red) and wedge model (blue).(right) transitions with the LIME model line profiles plottedon the top: disk model (red) and wedge model (blue).