arXiv:1211.0275v1 [astro-ph.SR] 1 Nov 2012 h uoenSuhr bevtr,adteOsl pc Obs Space Onsala the and Observatory, Southern f¨ur Radi European Max-Planck-Institut the the APE between 083.C-0996. collaboration program a ESO in (APEX) Experiment Pathfinder otn atcpto rmNASA. from participation portant consorti Investigator Principal European-led by provided tutr ttecnr fteVl oeua lu.Teeo These cloud. molecular C Vela the of centre the at structure epeetAPEX present We etto ftecnrlsa-omn ig.Cmie with Combined ridge. star-forming central the of mentation rjc ebidahg eouinclm est a fter the of map a density with column which resolution ridge high a build we project ietcmaio ihSresSuh fteGudBl Aqui Belt Gould the of South, di Serpens with comparison Direct e words. Key widths. central rjcswl ept xmn h di the formation. examine star high-mass to two help these – Combining will pc respectively. (700 2010), projects HOBYS al. and et Motte 2010) surv kpc; al. 3 Belt Andr´e et pc; Gould re 500 – in the 130 formation (HGBS, complexes: star star-forming high-mass two nearby and particular, low- atively In on Galaxy. focus projects our key in formation and low- star both mass into insights observational important viding 200 Ostriker 2007). Yorke & a & (McKee as Zinnecker formation clear, their less in is involved stars mass processes high establ for well scenario relatively formation been the has form stars the low-mass Although for systems. paradigm planetary t of are hosts and likely galaxy selves host their formati astrophy populate key galactic stars drive a low-mass ecology, stars is high-mass star while a process: of ical formation the mass, of Independent Introduction 1. tostars igs(ile l 01.Azuaine l 21)fudth dominati found (2011) single clustere al. et into be Arzoumanian 2011). or al. can et denses (nests) which (Hill ridges the networks filaments, in disorganised these proceeds into 2010 of formation Galaxy al. ) (Andr´e et Star our structures star (“supercritical” 2010). in filamentary of al. et of birthplaces complexes Molinari comprised the star-forming be studying most to for revealed regime have crucial the tre, ⋆⋆ .K¨onyvesV. .Giannini T. eray7 2018 7, February ⋆ Astrophysics & Astronomy ff rn omto ehnssadhsois h eaCrdea ridge C Vela the histories, and mechanisms formation erent hspbiaini ae ndt curdwt h Atacama the with acquired data on based is publication This Herschel Herschel The .Hill T. eovn h eaCrdewt P-ArT with ridge C Vela the Resolving Herschel 1 S:idvda bet Vl ,RW3)–IM lmns–sub – filaments ISM: – 36) RCW C, (Vela objects individual ISM: h Andr´e Ph. , sa S pc bevtr ihsineinstruments science with observatory space ESA an is 1 6 bevtosi h a-nrrdadsubmillime- and far-infrared the in observations , 2 n .Rev´eret V. and , .Nguy Q. , / P-ArT´eMiS 450 ∼ pc bevtr Plrt ta.21)i pro- is 2010) al. et (Pilbratt Observatory Space . ccnrlwdhi ossetwt htmaue o low for measured that with consistent is width central pc 0.1 1 .Arzoumanian D. , ˆen-Lu˜ aucitn.ms no. manuscript µ 1 .Lortholary M. ’ o otnu bevtoso C 6adteajcn ig,ah a ridge, adjacent the and 36 RCW of observations continuum m .L Pennec Le J. , ’ ng ff 3 rnebtenlw and low- between erence .Palmeirim P. , (A 1 .Motte F. , ffi 1 itoscnb on fe h references) the after found be can liations 1 .Martignac J. , .Rodriguez L. , n ihim- with and a oastronomie, ASfrifae n PR u-ilmteosrain f observations sub-millimetre SPIRE and far-infrared PACS 1 .Peretto N. , Herschel ervatory. srain,a ihrrslto than resolution higher at bservations, go apdwt -r´MS eetatterda density radial the extract P-ArT´eMiS. We with mapped egion 1 nand on dSresSuhfiaetsaecmo hrceitc,inc characteristics, common share filament South Serpens nd .Minier V. , ethe re ished high- hem- acmlx eel aysmlrte ewe h w regio two the between similarities many reveals complex, la ation coe 2012 October ABSTRACT tin at is X key ng ey 7; s- s, l- d ; t 1 .Talvard M. , 1 .Boulade O. , stogtt ea nerysaei t vlto ( evolution its in stage early a an 2011) formation known al. at et be star Hill is 2009; to high-mass C al. thought et Vela and Netterfield is 2003; 2012). intermediate- al. et low-, al. (Massi et Minier house 2012; to al. et Giannini Herschel thickness, or width, central standard a ∼ by characterised filament be interstellar these regions star-forming low-mass n h P-ArT´eMiS camera the ing spa comparable at which project, HGBS sample, resolution. the regio tial by HOBYS targeted star-forming the those low-mass p as 700 in with such at complex comparison Located closest direct gas. the allows io molecular cluster. the is of from star sheet C results Vela initial ionising ridge an the 36 th RCW of that of showed isation the centre (2012) the is al al. et roughly et cloud (Hill Minier at molecular cloud and the C ridge, in Vela this cores to dense Adjacent high-mass 2011). the of majority unn hog h eteo eaCi ig,ahigh-column a ridge, a is C Vela ( of density centre the through Running sdhr ocmlmn the complement to here used ont eisaldo PX(avr ta.2010). al. et (Talvard APEX on installed be to soon ilmte(5,30 500 350, (250, millimetre eg 11.5 (e.g. h C 6rgo fVl a bevduigteP- the using al. observed et Minier was 2008; C al. (Andr´e Vela et camera of bolometer region ArT´eMiS 36 RCW The reduction data and Observations 2. 1 .Schneider N. , 1 2 . pc. 0.1 1 h eaCmlclrcodwsrcnl bevdwith observed recently was cloud molecular C Vela The eew rsn td fteVl ig n C 6us- 36 RCW and ridge C Vela the of study a present we Here hr N Where -r´MSi rttp o h agrfra r´MScam ArT´eMiS format larger the for prototype a P-ArT´eMiS is ilmte–IM ut xicin–Sas al-ye–Sta – early-type Stars: – extinction dust, ISM: – millimetre .Men’shchikov A. , ms trfrigfiaet nthe in filaments star-forming -mass ∼ ′′ spr fteHBSpoet(ile l 2011; al. et (Hill project HOBYS the of part as 0 mag 100 oprdwt 25 with compared 1 H 2 .D Breuck De C. , 1 eMiS ´ = .Doumayrou E. , .4A 0.94 1 , 4 1 , efgaiaigfiaet hc ossthe houses which filament, self-gravitating ) 5 .Bontemps S. , V ⋆ Herschel × 1 g-ashg-oundniyfilamentary density high-column igh-mass .Didelon P. , µ 10 2 and ) tafco – ihrresolution higher 2–3 factor a at m), 21 t450 at 7 ′′ Herschel cm t350 at sSIEcmr,rva la frag- clear reveal camera, SPIRE ’s 1 − .Dubreuil D. , 2 µ mag Herschel nAE.P-ArT´eMiS is APEX. on m 4 , µ 5 Herschel 1 PR ad ntesub- the in bands SPIRE .Louvet F. , m). − .Hennemann M. , o the rom 1 Bhi ta.1978) al. et (Bohlin rfieo h eaC Vela the of profile uigterfilament their luding ol etsurvey. Belt Gould Herschel s ept likely Despite ns. 1 .Gallais P. ,

c 1 ⋆⋆ .Elia D. , S 2018 ESO < HOBYS 10 could s s pro- rs: 1 6 , yr). 1 ns, era 6 nd of n- , c, , 1 e - . T. Hill et al.: Resolving the Vela C ridge with P-ArT´eMiS and Herschel

applied (see Hill et al. 2011). The quality of the SED fit was as- sessed using χ2 minimisation. The corresponding column den- sity map of the Vela C ridge and RCW 36 is given in Fig. 2 (left). As our P-ArT´eMiS data, at 10′′ resolution have better reso- lution than that of all of the Herschel SPIRE bands, we devised a method to derive a higher resolution column density map of the Vela C ridge and RCW36 region mapped by P-ArT´eMiS. By combining the PACS 160 µm, SPIRE 250 µm and P-ArT´eMiS 450 µm data we were able to derive a column density map at 11.5′′ resolution, i.e. the resolution of the 160 µm Herschel map. This method is similar to the one used by Palmeirim et al. (2012, submitted) for their higher resolution Herschel column density map of the Taurus B211 region, but adapted for P- ArT´eMiS data as outlined in Appendix A. Due to the area cov- ered by our P-ArT´eMiS observations, only 1/3 of the full Vela C ridge detected by Herschel is measured∼ in the higher res- olution column density map, which shows clear fragmentation into a number of cores/clumps. These cores are clearly visible with P-ArT´eMiS (see Fig. 1) but could not be previously iden- Fig. 1. P-ArT´eMiS 450 µm map of the RCW36/Vela C ridge at tified from the lower resolution Herschel column density map 10′′ resolution. The contours are 3, 7, 11 Jy/10′′ -beam and the (Fig. 2, left). rmsnoise levelin the centralpartofthe map is 1 Jy/10′′ -beam. ∼ 3.2. Filamentary structure

2009) on the Atacama Pathfinder Telescope (APEX) telescope At the 36′′ resolution of the Herschel column density map, the on 23 May 2009. An area of 4′ by 4′ was mapped in 0.5 h, RCW36 region is characterised by a single dominating high- ∼ ∼ using a total-power, on-the-fly scanning mode. The atmospheric column density filament containing the Vela C ridge (see Fig. opacity at zenith was measured with a skydip and found to be 2). The filamentary structure seen in our higher resolution col- 0.8 at λ = 450 µm, which corresponds to a precipitable wa- umn density map is consistent with that seen in the lower res- ∼ ter vapour 0.7mm. The pointing and focus of the telescope olution map, with slight deviations as it traces the topological were checked∼ using similar observing procedures to those used structure of the fragmented cores (see Fig. 2, right). The mean with APEX/LABOCA (e.g spiral scans). Flux calibration was column density of the Vela C ridge, as measured from the higher achieved by taking spiral scans of Mars and Saturn. The point- resolution column density map, is 9 1022 cm 2. ∼ × − ing accuracy and absolute calibration uncertainty were estimated The mean radial density profile (ρp) of the Vela C ridge to be 2 and 30%, respectively. The main beam had a full ∼ ′′ ∼ (Fig. 3) was derived by measuring cuts perpendicular to the crest width at half maximum (FWHM) 10′′, known to 10% ac- of the filament at each pixel, and then averaging along the length curacy, and contained 60% of the∼ power, the rest∼ being dis- ∼ of the filament (as detailed in Arzoumanian et al. 2011). In or- tributed in an “error beam” extending up to an angular radius of der to derive the characteristic parameters of the profile (e.g. 80′′. The data were reduced using in-house IDL routines, fol- ∼ central density, power law exponent), we assume a cylindrical lowing the same method as Andr´eet al. (2008) and Minier et al. filament model given by a Plummer-like function, which is a (2009). This procedure includes baseline subtraction, removal density profile that can be expressed in terms of column density. of correlated sky noise and 1/f noise, and subtraction of uncor- Accordingly, related 1/f noise using a method which exploits the high level ρc of redundancy in the data. The final 450 µm continuum map is ρp(r) = (1) [1 + (r/R )2]p/2 presented in Figure 1. f lat 2 The entire Vela C molecular cloud ( 3deg ), including the where ρ is the radial density at the centre of the filament, p is RCW36 region, was mapped with Herschel∼ at 70, 160, 250, 350 c the exponentof the model function,and R f lat is the characteristic and 500 µm as part of the HOBYS key program. The observa- radius for the flat inner portion of the profile. tions and data reduction are as described by Hill et al. (2011). The Vela C ridge can be characterised by an an inner radius R f lat 0.05 0.02pc, and a deconvolved FWHM of 3. Structural Analysis 0.12 0.02 pc∼ which± is consistent with that seen in low-mass star-forming± filaments (Arzoumanian et al. 2011). The Vela C 3.1. Column density and dust temperature maps (2.7 0.2) radial density profile decreases at large radii as r− ± . The filament outer radius 0.4 0.1pc is defined from the deviation Multiple wave-band observations covering the far-infrared and ∼ ± sub-millimetre regime allow construction of column density of the observed (western side) profile from the Plummer fit (cf. Fig. 3). The eastern side of the profile, containing the OB star (NH2 ) and dust temperature maps. The maps of these quanti- / cluster, decreases up to a larger radius 1–1.5pc (see Fig. 4). ties for the Vela C ridge RCW36 region, derived from Herschel ∼ data, were drawn using pixel-by-pixel spectral energy distribu- The radial profiles of the column density can also be used to tion (SED) fitting according to a modified blackbody with a sin- derive a mass per unit length (Mline = R Σobs(r)dr) value for a gle dust temperature (cf. Hill et al. 2012). Only the longest four filament, by integrating the column density over the radius (cf. Herschel wavebands were used. In order to do this, the Herschel Arzoumanian et al. 2011). The Vela C ridge is well defined and observations were first convolved to the resolution of the 500 µm constrained allowing us to estimate, without confusion, its mass band (36′′), and the zero offsets derived from Planck data were per unit length as Mline 320( 75) or 400( 85) M /pc as mea- ∼ ± ± ⊙

2 T. Hill et al.: Resolving the Vela C ridge with P-ArT´eMiS and Herschel

2 Fig. 2. The Vela C ridge/RCW36 column density maps. The units are in H2 cm− . Left: Map derived from the longest four Herschel wavebands with a resolution of 36′′, equivalent to that of the SPIRE 500 µm band. Right: Map constructed using P-ArT´eMiS and Herschel data, with resolution 11.5′′. Appendix A details how this map was created. The filament from which the radial density profile (Fig. 3) was determined (see Section 3.2) is as indicated by the black line, measured on the right map but, overlaid on the left figure for clarity.

tend to have inner widths of 0.1pc. With higher resolution P- ArT´eMiS data at hand it is possible∼ to check the application of such a characteristic filament width to more distant, higher mass regions, such as the Vela C complex at d 700 pc. Here we have shown that, based on its radial column∼ density profile, the Vela C ridge has a filament inner width consistent with the char- acteristic inner width suggested by Arzoumanian et al. (2011). We have extended our study of filamentary structures to the Serpens South filament, part of the Aquila star-forming complex at d 260 pc (see Fig. 5). The Serpens South filament is particu- larly∼ interesting as it, as one of the most extreme column density filaments of the HGBS, is comprisedof high-columndensity ma- terial, similarly to Vela C. Bontemps et al. (2010) detected seven Class 0 protostars, confirming that Serpens South is undergo- ing low- to intermediate-mass star formation, while Maury et al. (2011) used evolutionary tracks to estimate the lifetime of these Fig. 3. The mean radial density profile (western component) per- Class-0 protostars and showed that Serpens South is at a very pendicular to the Vela C ridge (black line on Fig. 2, left), shown early phase of forming stars. here in log-log format. The complementaryeastern profile (in the The column density map of Serpens South has been de- direction of the star cluster) is given in Fig. 4. The area in yel- rived in the same manner as that of Vela C (see section 3.1 and low shows the dispersion of the radial profile along the filament. K¨onyves et al. 2010), with only the longest four Herschel bands. The solid blue line corresponds to the effective 11.5′′ HPBW The 36′′ resolution of this map, at the distance of Serpens South, resolution of the column density map (0.04pc at 700pc) used to corresponds to a spatial resolution of 0.05pc on the sky, compa- construct the profile. The dotted blue line indicates the Gaussian rable to that of Vela C when using our higher-resolution column profile, while the dotted red line shows the best fit to the model, density maps with P-ArT´eMiS data (0.04pc at 700pc). These often called the Plummer profile. two regions - Serpens South, forming low- to intermediate-mass stars, and Vela C forming intermediate and potentially high-mass stars - provide a good comparative opportunity. sured from the integrated radial profile of 0.4 and 1.5pc, respec- The crest shown in Fig. 5 (left) for the Serpens South fil- 21 2 tively, after subtracting a background of 3.6 10 cm− . ament is 1.2pc in length and has an average column den- × ∼ 22 2 sity value of 6.4 10 cm− (after subtracting a background of 21 2 × 3.7 10 cm− ). Thecrestshownin Fig.2 forthe Vela Cridgeis 4. Comparison with the Serpens South filament × 22 2 0.8pc in length and is slightly more dense at 8.6 10 cm− . Recent Herschel observations have revealed the prolificity of in- The∼ Vela C ridge and Serpens South filament have× a similar terstellar filaments in star-forming complexes, though only those mass per unit length (Mline = 320 and 290 M /pc, respectively) ⊙ filaments above an AV of 8mag are supercritical and capable and a similar outer radius ( 0.4pc; though the outer radius for of forming stars (Andr´eet∼ al. 2010, 2011). Arzoumanian et al. both regions may be as large∼ as 1–1.5pc, see Figs. 4 and 5). (2011) showed that the filaments in low-mass HGBS regions The radial density profile of the Serpens∼ South filament (Fig. 5)

3 T. Hill et al.: Resolving the Vela C ridge with P-ArT´eMiS and Herschel has a (deconvolved) Gaussian FWHM width of 0.10 0.05pc, Sect. 4). The lack of known Class-0 protostars in Vela C may in agreement with that found here for Vela C and the± charac- arise from sensitivity limitations, but the same can not be said teristic width of other nearby interstellar filaments measured by for the lack of high-mass dense cores in Serpens South. Arzoumanian et al. (2011) using HGBS data. The inner width While the Vela C ridge and Serpens South filament are likely values measured for the Vela C ridge and Serpens South fila- to have formed in different environments, under different con- ment remain almost unchanged after removing the cores within ditions, and to have experienced different histories (with, e.g., them: the deconvolved FWHM values become 0.11 0.01pc the strong ionizing effect of a massive star cluster in the case of and 0.14 0.01pc, respectively. As such, the common± inner Vela C, and no such effect in Serpens South), these two regions width of ±0.1pc found in both cases may not be attributed to display a similar filament inner width 0.1pc which is also con- core properties.∼ sistent with the characteristic inner width∼ found for low-mass The total length and mass of the Serpens South filament are star-forming filaments by Arzoumanian et al. (2011). This sug- of the order2pc and 500M , respectively, while the total length gests that supercritical filaments and ridges, regardless of their and mass of the Vela C ridge⊙ are 4 pcand 600M . It should formation process, have the same inner width which may be also be noted that Vela C is more distant∼ than∼ Serpens⊙ South, and characteristic across modes of star formation. These results ad- thus core surveys in Vela C are intrinsically biased toward more vocate a high degree of commonality among star-forming fila- massive objects than those accessible in Serpens South. ments, independently of the masses of the stars they form, rather than the aforementioned idea of distinct environmental condi- tions for higher mass stars. 5. Universality of star-forming filament profiles? More work would be needed to establish the universality of Star formation requires a reservoir of material, concentrated star-forming filament profiles in the high-mass regime. Recently, into a small volume, from which the burgeoning young star Hennemannet al. (2012) used HOBYS data to show that the can accumulate mass. The idea of a minimum mass, or den- high-mass DR21 ridge in the Cygnus X complex (d 1.4 kpc) sity requirement, for star formation inside molecular clouds is has an apparent mean central width of 0.3pc when∼ observed then not surprising. Evans (2008) suggested that star forma- at a resolution 0.17pc. The flat inner portion∼ of the DR21 ridge tion is restricted to dense gas within molecular clouds, which was however only marginally resolved with Herschel observa- also has a higher star formation efficiency than lower density tions. It should be stressed that higher mass star-forming ridges gas. Comparing Spitzer inventories of young stellar objects with and filaments occur at greater distances than lower mass ones, dust extinction (AV ) maps of nearby molecular clouds, both and observations of these are thus subject to lower spatial reso- Lada et al. (2010) and Heiderman et al. (2010) suggested that lution. At distances exceeding that of Vela C (> 700pc) a char- star formation requires a minimum gas density (corresponding acteristic inner width of 0.1pc would remain unresolved, or ∼ to AV 8mag). Essentially the same column density threshold at best marginally resolved with Herschel. Higher resolution ob- was recently≈ found by Andr´eet al. (2010, 2011) from an analy- servations of high-mass star-forming filaments and ridges, with sis of the prestellar core population in the Aquila complex based for example the full ArT´eMiS camera to be installed soon on on Herschel data (see also Section 1). APEX or the Atacama Large Millimetre Array (ALMA), over In addition to a minimum mass/density for core and star a greater number of regions are needed now to address the hy- formation, a minimum threshold for high-mass star formation pothesis of a characteristic inner width for interstellar filaments has also been suggested both theoretically (Krumholz & McKee independently of the masses of the stars they form. 2008) and observationally (Kauffmann& Pillai 2010). According to these authors, low- and high-mass stars likely do Acknowledgements. T.H. is supported by a CEA/Marie-Curie Eurotalents not form in the same environments, with the latter requiring a Fellowship. Part of this work was supported by the ANR (Agence Nationale minimum density, and the two modes of star formation (low- pour la Recherche) project ‘PROBeS’, number ANR-08-BLAN-0241. and high-mass) are distinct from each other. In this letter we have compared two nearby (< 700pc) re- gions undergoing clustered star formation. The Vela C molecu- References lar cloud is known as a low- to intermediate-mass star-forming Andr´e, P., Men’shchikov, A., Bontemps, S., et al. 2010, A&A, 518, L102 molecular cloud. The Vela C ridge however is associated with Andr´e, P., Men’shchikov, A., K¨onyves, V., & Arzoumanian, D. 2011, in IAU a high-mass ionising star cluster (Minier et al. 2012), and hosts Symposium, Vol. 270, IAU Symposium, ed. J. Alves, B. G. Elmegreen, seven high-mass dense cores (Hill et al. 2011) each with the po- J. M. Girart, & V. Trimble, 255–262 Andr´e, P., Minier, V., Gallais, P., et al. 2008, A&A, 490, L27 tential to form high-mass stars. Ridges are the extreme den- Arzoumanian, D., Andr´e, P., Didelon, P., et al. 2011, A&A, 529, L6 sity filaments of high-mass star-forming regions whose large Bohlin, R. C., Savage, B. D., & Drake, J. F. 1978, ApJ, 224, 132 areas of influence suggest that they may have been formed Bontemps, S., Andr´e, P., K¨onyves, V., et al. 2010, A&A, 518, L85 through dynamic scenarios such as converging flows, filament Evans, II, N. J. 2008, in Astronomical Society of the Pacific Conference Series, / Vol. 390, Pathways Through an Eclectic Universe, ed. J. H. Knapen, T. J. mergers and or ionisation pressure from nearby star clusters Mahoney, & A. Vazdekis, 52 (Hill et al. 2011; Nguyen Luong et al. 2011; Hennemann et al. Giannini, T., Elia, D., Lorenzetti, D., et al. 2012, A&A, 539, A156 2012; Minier et al. 2012). Comparatively, the Serpens South Gutermuth, R. A., Bourke, T. L., Allen, L. E., et al. 2008, ApJ, 673, L151 filament is populated by low- to intermediate-mass protostars Heiderman, A., Evans, II, N. J., Allen, L. E., Huard, T., & Heyer, M. 2010, ApJ, (Gutermuth et al. 2008; Maury et al. 2011). Both of these fila- 723, 1019 Hennemann, M., Motte, F., Schneider, N., et al. 2012, A&A, 543, L3 mentary star-forming clumps have supercritical masses per unit Hildebrand, R. H. 1983, QJRAS, 24, 267 length and are thus likely to form more stars in the future. Hill, T., Motte, F., Didelon, P., et al. 2011, A&A, 533, 94 Our analysis of the filamentary structure in the Vela C ridge Hill, T., Motte, F., Didelon, P., et al. 2012, A&A, 542, A114 and the Serpens South filament indicates that the two are quite Kauffmann, J. & Pillai, T. 2010, ApJ, 723, L7 K¨onyves, V., Andr´e, P., Men’shchikov, A., et al. 2010, A&A, 518, L106 similar with respect to their column density profile and mass per Krumholz, M. R. & McKee, C. F. 2008, Nature, 451, 1082 unit length, yet the Vela C ridge is a factor of 2 longer and Lada, C. J., Lombardi, M., & Alves, J. F. 2010, ApJ, 724, 687 is slightly more massive than the Serpens South∼ filament (see Massi, F., Lorenzetti, D., & Giannini, T. 2003, A&A, 399, 147

4 T. Hill et al.: Resolving the Vela C ridge with P-ArT´eMiS and Herschel

Maury, A. J., Andr´e, P., Men’shchikov, A., K¨onyves, V., & Bontemps, S. 2011, Likewise, the third term of Eq. (A.1) may be written as A&A, 535, A77 Σ250 Σ250 G350 250, where G350 250 is a circular Gaussian with McKee, C. F. & Ostriker, E. C. 2007, ARA&A, 45, 565 − ∗ FWHM √24.92 18.22 17.0 , and may be understood as Minier, V., Andr´e, P., Bergman, P., et al. 2009, A&A, 501, L1 − ≈ ′′ Minier, V., Tremblin, P., Hill, T., et al. 2012, A&A, submitted a term adding information on spatial scales only accessible to Molinari, S., Swinyard, B., Bally, J., et al. 2010, A&A, 518, L100 Herschel observations at wavelengths 250µm. In order to de- Motte, F., Zavagno, A., Bontemps, S., et al. 2010, A&A, 518, L77 ≤ rive an estimate Σ¯ 250 of Σ250 on the right-hand side of Eq. (A.1), Netterfield, C. B., Ade, P. A. R., Bock, J. J., et al. 2009, ApJ, 707, 1824 we first smoothed the PACS 160 µm map to the 18.2 resolu- Nguyen Luong, Q., Motte, F., Hennemann, M., et al. 2011, A&A, 535, A76 ′′ Palmeirim, P., Andr´e, P., & Gould Belt consortium. 2012, submitted, A&A tion of the SPIRE 250 µm map and then derive a color tem- Pilbratt, G. L., Riedinger, J. R., Passvogel, T., et al. 2010, A&A, 518, L1 perature map between 160 µm and 250 µm from the observed Starck, J.-L. & Murtagh, F. 2006, Astronomical Image and Data Analysis I250µm(x, y)/I160µm(x, y) intensity ratio at each pixel (x, y). The Talvard, M., Andr´e, P., Le-Pennec, Y., et al. 2010, in Society of Photo-Optical SPIRE 250 µm map was converted to a gas surface density map Instrumentation Engineers (SPIE) Conference Series, Vol. 7741, Society of ¯ Photo-Optical Instrumentation Engineers (SPIE) Conference Series (Σ250), assuming optically thin dust emission at the temperature Zinnecker, H. & Yorke, H. W. 2007, ARA&A, 45, 481 given by the color temperature map and a dust opacity at 250 µm 2 2 κ250µm = 0.1 (300/250) cm /g. An estimate of the third term of Eq. (A.1) can× then be obtained by subtracting a smoothed ver- sion of Σ¯ (i.e., Σ¯ G ) to Σ¯ itself, i.e., by removing Appendix A: Deriving a high-resolution column 250 250 ∗ 350 250 250 low spatial frequency information from Σ¯ 250. density map with P-ArTeMiS´ and Herschel data The fourth term on the right-hand side of Eq. (A.1) is the component containing information on the smallest scales 10– The P-ArT´eMiS 450 µm map (Fig. 1) provides information on ∼ small-scales in the RCW 36 ridge but is not sensitive to large- 18′′ and which was estimated using the P-ArT´eMiS data. It may Σ Σ scale structures because any emission at low spatial frequen- be written as 160 160 G250 160, where G250 160 is a circular − √∗ 2 2 cies has been completely filtered out during the process of at- Gaussian with FWHM 18.2 11.5 14.1′′. In order to de- ¯ − ≈ mospheric skynoise removal. Conversely, the Herschel column rive an estimate Σ160 of Σ160 on the right-hand side of Eq. (A.1), we used the color temperature map between 160 µm and 250 µm density map producedby Hill et al. (2011) at 36′′ resolution con- (18 resolution) to convert a slightly smoothed version of the P- tains little information on scales < 36′′ but provides an accurate ′′ view of larger-scale features. In order to combine these two com- ArT´eMiS 450 µm data (11.5′′ resolution) and the PACS 160 µm plementary data sets, we used the following procedure (inspired data to a column density map at 11.5′′ resolution, assuming opti- from Palmeirim et al. 2012, submitted). cally thin dust emission and the same dust opacity law as above. ¯ Following the spirit of a multi-resolution data decomposi- Due to skynoise filtering, the latter map (Σ160) does not contain > tion (cf. Starck & Murtagh 2006), the gas surface density dis- information on angular scales 40′′, but this is not a problem ∼ tribution of the RCW 36 region, smoothed to the resolution of since our estimate of the fourth term of Eq. (A.1) was obtained ¯ ¯ the Herschel/PACS 160 µm observations, may be expressed as a by subtracting a smoothed version of Σ160 (i.e., Σ160 G250 160) ¯ ∗ > sum of four terms: to Σ160 itself, i.e., by removing any information on scales 18′′ from Σ¯ 160. This subtraction of scales larger than 18′′ from∼ Σ¯ 160 Σ = Σ + (Σ Σ ) + (Σ Σ ) + (Σ Σ ) . (A.1) also suppresses the effect of the bowls of negative∼ emission seen 160 500 350 − 500 250 − 350 160 − 250 at 30′′ on either side of the central ridge in the P-ArT´eMiS where Σ500, Σ350, Σ250, and Σ160 represent smoothed versions of 450±µm map (Fig. 1). the intrinsic gas surface density distribution Σ after convolution Our final estimate Σ˜ 160 of the gas surface density distribution with the Herschel beam at 500 µm, 350 µm, 250 µm, and 160 µm at 11.5′′ resolution was produced by summing the above esti- respectively, i.e.: Σ = Σ B , Σ = Σ B , Σ = Σ B , 500 ∗ 500 350 ∗ 350 250 ∗ 250 mates of the four terms on the right-hand side of Eq. (A.1): and Σ160 = Σ B160. The first∗ term of Eq. (A.1) is simply the surface density Σ˜ 160 = Σ¯ 500 + Σ¯ 350 Σ¯ 350 G500 350 + Σ¯ 250 Σ¯ 250 G350 250 distribution smoothed to the resolution of the Herschel/SPIRE  − ∗   − ∗  500 µm data. An estimate, Σ¯ , of this term can be obtained 500 + Σ¯ Σ¯ G . (A.2) in a manner similar to Hill et al. (2011) through pixel-by-pixel  160 − 160 ∗ 250 160 SED fitting to the longest four Herschel data points, assum- The resulting 11.5 resolution column density map N˜ for the ing the following dust opacity law, very similar to that advo- ′′ H2 cated by Hildebrand (1983) at submillimetre wavelengths: κ = RCW36 region is displayed in the right panel of Fig. 2 in units of ν mean molecules per cm2, where Σ˜ = µ m N˜ and µ = 2.33 0.1 (ν/1000 GHz)β = 0.1 (300 µm/λ)β cm2/g, with β = 2. 160 H H2 ×The second term of Eq.× (A.1) may be written as Σ is the mean molecular weight. This high-resolution column den- 350 − sity map is subject to larger uncertainties than the standard Σ350 G500 350, where G500 350 is a circular Gaussian with full ∗ 36.3′′-resolution column density map derived from Herschel width at half maximum (FWHM) √36.32 24.92 26.4 . (To − ≈ ′′ data. In particular, the absolute calibration uncertainty of the first order, the SPIRE beam at 500 µm is a smoothed version ground-based P-ArT´eMiS 450 µm data is larger ( 30%) than of the SPIRE beam at 350 µm, i.e., B500 = B350 G500 350.) ∼ ∗ that of the Herschel data (< 10%), and the beam shape of the P- The second term of Eq. (A.1) may thus be viewed as a term ArT´eMiS instrument is also∼ more uncertain3 than the beams of adding information on spatial scales accessible to SPIRE obser- the SPIRE and PACS cameras on Herschel. To test the reliabil- vations at 350 µm, but not at 500 µm. In practice, one can derive ity and robustness of this 11.5′′ resolution column density map, and estimate Σ¯ 350 of Σ350 in a manner similar to Σ¯ 500, through we smoothed it to the 36.3′′ resolution of the standard column pixel-by-pixel SED fitting to three Herschel data points between density map (corresponding to Σ¯ ) and inspected the ratio map 160 µm and 350 µm (i.e., ignoring the lower resolution 500 µm 500 data point). An estimate of the second term of Eq. (A.1) can 3 Note, however, that the < 80 error beam of the P-ArT´eMiS in- ¯ ′′ then be obtained by subtracting a smoothed version of Σ350 (i.e., strument is effectively filtered∼ out during the data reduction and map ¯ ¯ Σ350 G500 350) to Σ350 itself, i.e., by removing low spatial fre- reconstruction process, so that only the 10% uncertainty in the 10′′ ∗ ∼ ∼ quency information from Σ¯ 350. main beam of P-ArT´eMiS matters here for Σ˜ 160.

5 T. Hill et al.: Resolving the Vela C ridge with P-ArT´eMiS and Herschel

Fig. A.1. Ratio map of the 11.5′′ resolution column density map (corresponding to Σ˜ 160) smoothed to 36.3′′ resolution divided by the standard column density map (corresponding to Σ¯ 500). The same filament crest as in Fig. 2 is overlaid. The maximum value of the ratio is 1.35 and the minimum value is 0.85. between the two, which has a mean value of 1.00 and a standard deviation of 0.03 (see Fig. A.1). The two column density maps agree to better than 35% everywhere. The morphology and am- plitude of the features seen in the ratio map (Fig. A.1) suggest that the potential artefacts present in the high-resolution column density map translate into an uncertainty of < 40% in the radial column density profiles shown in Fig. 3 and∼ Fig. 4 (left). This uncertainty is comparable to the yellow error bars shown on the profiles (corresponding to the dispersion of the radial profiles observed along the ridge) and therefore does not significantly affect our conclusions regarding the shape and central width of the radial density profile. Finally, we note that our column density map of the Vela C ridge is also more uncertain than that of the Serpens South fil- ament owing to the strong heating effect of the RCW36 cluster and its uncertain location along the line of sight. Simple tests suggest that this effect leads to an additional 50% uncertainty in the absolute calibration of the column density map of the Vela C ridge but has little influence on the shape of the radial column density profile.

1 Laboratoire AIM, CEA/IRFU CNRS/INSU Universit´eParis Diderot, CEA-Saclay, 91191 Gif-sur-Yvette Cedex, France 2 Institut d’Astrophysique Spatiale, CNRS/Universit´e Paris-Sud 11, 91405 Orsay, France 3 Canadian Institute for Theoretical Astrophysics - CITA, University of Toronto, 60 St. George Street, Toronto, Ontario, M5S 3H8, Canada 4 Universit´e de Bordeaux, Bordeaux, LAB, UMR 5804, 33270, Floirac, France 5 CNRS, LAB, UMR 5804, 33270, Floirac, France 6 IAPS- Instituto di Astrofisica e Planetologia Spaziali, via Fosso del Cavaliere 100, 00133 Roma, Italy 7 European Southern Observatory, Karl Schwarzschild Str. 2, 85748 Garching bei Munchen, Germany e-mail: [email protected]

6 T. Hill et al.: Resolving the Vela C ridge with P-ArT´eMiS and Herschel , Online Material p 1

Fig. 4. Left: Mean radial column density profile measured on the eastern side of the Vela C ridge in the 11.5′′ resolution column density map shown in Fig. 2 (right). This figure is complementary to Fig. 3 for the side containing the OB cluster. This profile has a p value of 2.3 0.5 and R f lat=0.05 0.02pc. The error bars in yellow show the dispersion of the radial profile along the filament, while the lines± on the plot are as per± Fig. 3 and as indicated on the key. Right: Comparison of the mean background-subtracted radial profiles measured on the western side of the Vela C ridgein the 11.5′′ resolution column density map (black curve and yellow error bars) and in the 36.3′′ resolution column density map (orange curve). The black and orange dash-dotted curves represent the effective 11.5′′ and 36.3′′ HPBW resolutions of the corresponding data, respectively.

Fig. 5. Left: Column density map of the Serpens South filament (36′′ resolution) derived from HGBS data (see K¨onyves et al. 2010), with the corresponding topological filament identified overlaid in blue. Right: Mean radial density profile (taken from both sides of the filament) measured perpendicular to the supercritical Serpens South filament (left) shown here in log-log format. Lines and colour coding are consistent with Figs. 3 and 4. The Gaussian fit to the inner part of the profile (dotted blue curve) has a deconvolved FWHM width 0.10 0.05 pc. The best Plummer-like model fit (dashed red curve) has an inner radius R f lat 0.03 0.01pc and a power-law index p =±2.02 0.27. ∼ ± ±