Herschel observations of the Herbig-Haro objects HH52-54
P. Bjerkeli, R. Liseau, B. Nisini, M. Tafalla, M. Benedettini, P. Bergman, O. Dionatos, T. Giannini, G. Herczeg, K. Justtanont, B. Larsson, C. McCoey, M. Olberg, A.O.H Olofsson
aa r X i v : . [ a s t r o - ph . GA ] J un Astronomy&Astrophysicsmanuscript no. hh54 c (cid:13)
ESO 2018August 27, 2018
Herschel observations of the Herbig-Haro objects HH 52-54 ⋆ P. Bjerkeli , R. Liseau , B. Nisini , M. Tafalla , M. Benedettini , P. Bergman , O. Dionatos , T. Giannini ,G. Herczeg , K. Justtanont , B. Larsson , C. M c Coey , M. Olberg and A.O.H Olofsson Department of Earth and Space Sciences, Chalmers University of Technology, Onsala Space Observatory, 439 92 Onsala, Sweden INAF - Osservatorio Astronomico di Roma, Via di Frascati 33, 00040 Monte Porzio Catone, Italy Observatorio Astron´omico Nacional (IGN), Calle Alfonso XII,3. 28014, Madrid, Spain INAF - Osservatorio Astrofisico di Arcetri, Largo E. Fermi 5, 50125 Firenze, Italy Centre for Star and Planet Formation, Natural History Museum of Denmark, University of Copenhagen, Øster Voldgade 5-7, 1350Copenhagen, Denmark Max Planck Institut for Extraterestrische Physik, Garching, Germany Department of Astronomy, Stockholm University, AlbaNova, 106 91 Stockholm, Sweden University of Waterloo, Department of Physics and Astronomy, Waterloo, Ontario, CanadaAccepted June 14, 2011
ABSTRACT
Context.
The emission from Herbig-Haro objects and supersonic molecular outflows is understood as cooling radiation behindshocks, initiated by a (proto-)stellar wind or jet. Within a given object, one often observes the occurrence of both dissociative (J-type)and non-dissociative (C-type) shocks, owing to the collective e ff ects of internally varying shock velocities. Aims.
We are aiming at the observational estimation of the relative contribution to the cooling by CO and H O, as this providesdecisive information for the understanding of the oxygen chemistry behind interstellar shock waves.
Methods.
The high sensitivity of HIFI, in combination with its high spectral resolution capability, allows us to trace the H O outflowwings at unprecedented signal-to-noise. From the observation of spectrally resolved H O and CO lines in the HH52-54 system, bothfrom space and from ground, we arrive at the spatial and velocity distribution of the molecular outflow gas. Solving the statisticalequilibrium and non-LTE radiative transfer equations provides us with estimates of the physical parameters of this gas, including thecooling rate ratios of the species. The radiative transfer is based on an Accelerated Lambda Iteration code, where we use the fact thatvariable shock strengths, distributed along the front, are naturally implied by a curved surface.
Results.
Based on observations of CO and H O spectral lines, we conclude that the emission is confined to the HH54 region. Thequantitative analysis of our observations favours a ratio of the CO-to-H O-cooling-rate ≫
1. Formally, we derive the ratio Λ (CO) / Λ ( o -H O) =
10, which is in good agreement with earlier determination of 7 based on ISO-LWS observations. From the best-fit modelto the CO emission, we arrive at an H O abundance close to 1 × − . The line profiles exhibit two components, one of which istriangular and another, which is a superposed, additional feature. This additional feature likely originates from a region smaller thanthe beam where the ortho -water abundance is smaller than in the quiescent gas. Conclusions.
Comparison with recent shock models indicate that a planar shock can not easily explain the observed line strengthsand triangular line profiles. We conclude that the geometry can play an important role. Although abundances support a scenario whereJ-type shocks are present, higher cooling rate ratios than predicted by these type of shocks are derived.
Key words.
Stars: formation - Stars: winds, outflows - ISM: Herbig-Haro objects - ISM: jets and outflows - ISM: molecules
1. Introduction
Outflows have many times been discovered through observationsof Herbig-Haro objects (see e.g. Herbig 1950, 1951; Haro 1952, ⋆ Herschel is an ESA space observatory with science instruments pro-vided by European-led Principal Investigator consortia and with impor-tant participation from NASA.Complementary observations were made with:Odin is a Swedish-led satellite project funded jointly by theSwedish National Space Board (SNSB), the Canadian Space Agency(CSA), the National Technology Agency of Finland (Tekes) and CentreNational d’Etude Spatiale (CNES).The Swedish ESO Submillimetre Telescope (SEST) located at LaSilla, Chile was funded by the Swedish Research Council (VR) and theEuropean Southern Observatory. It was decommissioned in 2003.The Atacama Pathfinder EXperiment (APEX) is a collaborationbetween the Max-Planck-Institut f¨ur Radioastronomie, the EuropeanSouthern Observatory and the Onsala Space Observatory. ff erenttype of shock chemistry (e.g. Bergin et al. 1998). Depending onthe ionisation fraction, magnetic field strength and velocity ofthe shock, water abundances can be elevated to di ff erent levels(Hollenbach et al. 1989). In J-type shocks (Jump shocks), wherethe magnetosonic speed is lower than the propagation of thepressure increase, the involved energies generally dissociate H and all molecules with lower binding energies. As such the wa-ter abundance is generally low in J-type shocks, although it mayreform in the post-shock cooling region once pre-shock densi-ties are su ffi ciently high. In C-type shocks however, the pre-shock gas is partially heated due to traversing magnetic wavesfrom the post-shock gas, and molecules can survive the passageof the shock (Draine 1980) where both the magnetic field andthe gas density is compressed. In this type of shock, the activa-
1. Bjerkeli et al.: Herschel observations of the Herbig-Haro objects HH 52-54 tion barrier for neutral-neutral reactions between molecular hy-drogen and oxygen is reached, and the water abundance is ex-pected to become enhanced. This can be both due to the e ff ectof sputtering from dust grains (Kaufman & Neufeld 1996) anddue to high temperature chemistry occurring in the shocked re-gion (Bergin et al. 1998). After the shock passage, the enhancedwater abundance persists for a long time in the post-shock gas.The atmosphere of the Earth is opaque at the wavelengthof the lowest rotational transitions of water as well as mostother higher excited transitions. Thus, it is not until recentyears, with the use of space based observatories, these transi-tions have been observed successfully. Previous missions suchas SWAS (Melnick et al. 2000) and Odin (Nordh et al. 2003)have put constraints on the water abundance and the dynam-ics of molecular outflows (Franklin et al. 2008; Bjerkeli et al.2009). None of these missions, however, provided the spatialand spectral resolution that is available with the Herschel SpaceObservatory (Pilbratt et al. 2010).HH 54 is a Herbig-Haro object located in the Chamaeleon IIcloud at a distance of approximately 180 pc (Whittet et al. 1997).The visible objects in HH 54 are moving at high Doppler speedout of the cloud toward the observer, with velocities of the or-der 10 - 100 km s − (Caratti o Garatti et al. 2009). The objectsshow a clumpy appearance due to either Rayleigh-Taylor insta-bilities in the flow or due to variability in the jet itself. Anotherpossibility is patchy overlying dust extinction. The source ofthe jet is not well constrained. This is further discussed in ap-pendix A.2, where we also present quantitative arguments foridentifying IRAS 12553-7651 (ISO-Cha II 28) as the HH54 jet-driving source and its associated blueshifted CO outflow. No red-shifted emission is observed as what would be expected from abipolar jet (see Section 3). On the other hand, the extinction isrelatively low, something that might allow the redshifted gas toflow out essentially unhindered into the low density material atthe rear side of the cloud. It can however not be ruled out that theoutflow itself is one-sided and asymmetric. Recent simulationsshow that rapidly rotating stars with complex magnetic fields canbe responsible for such type of flows (Lovelace et al. 2010).HH 54 has previously been observed in various lines ofCO, SiO and H O using Odin and SEST (Bjerkeli et al.2009, R. Liseau, unpublished). Several H O lines and high-J CO lines were also observed with ISO-LWS and publishedin Liseau et al. (1996) and Nisini et al. (1996). During thePerformance Verification Phase of the Heterodyne Instrumentfor the Far-Infrared (HIFI) instrument, aboard the HerschelSpace Observatory, the CO (10 −
9) transition was observed. Aspart of the WISH keyprogram (van Dishoeck et al. 2011), alsothe H O (1 − ) and H O (2 − ) transitions were ob-served using HIFI and the Photodetector Array Camera andSpectrometer (PACS) respectively. We note that the wavelengthregion, covering the H O (2 − ) transition, also can be ob-served with HIFI.In this paper, we present observations, both from space andground, of HH 52-54 in spectral lines of CO and H O. We chooseto observe this region based on the fact that HH54 is free fromcontamination from other objects, spatially confined and re-solved in the infrared regime with the instruments used. Usingresults from observations carried out with SEST, APEX, Odinand Herschel, we aim at improving our understanding of inter-stellar shock waves. The observing modes and the instrumentsthat have been used are described in Section 2 while the detailsof the HIFI data reduction can be found in Appendix B. The ba-sic observational results are summarised in Section 3 whereasthe interpretations are discussed in Section 4. −150−100−50050−150−100−50050 RA offset ["] D e c o ff s e t [ " ] HH54 B HH52HH5319" I n t eg r a t ed i n t en s i t y [ K k m s − ] Fig. 1.
Upper panel : False colour map of the CO (10 −
9) in-tegrated intensity in the blue line wing (from + − to −
30 km s − ). The positions of HH 54 B, HH 53 and HH 52 areindicated with yellow squares. The readout positions for the on-the-fly and raster maps are indicated with white dots. Middle panel : CO (10 −
9) map of the integrated intensity ob-tained with HIFI in the blue line wing overlaid on an H α im-age from Caratti o Garatti et al. (2009). Contours are from 5.5 to33.6 K km s − in steps of 3.5 K km s − . Lower panel : A zoom of the CO (10 −
9) integrated intensityoverlaid on the H O (2 − ) emission obtained with PACS.The positions of HH 54 B, C and K as indicated in Giannini et al.(2006) are indicated with yellow squares.
2. Bjerkeli et al.: Herschel observations of the Herbig-Haro objects HH 52-54
Table 1.
Molecular line observations carried out with Odin, SEST, APEX and Herschel.
Telescope Molecule Frequency E u / k B HPBW η mb Date t int (GHz) (K) ( ′′ ) (YYMMDD) (hr)SEST CO (2 −
1) 230.538 16.6 23 0.50 970811 - 980806 3SEST CO (3 −
2) 345.796 33.2 15 0.25 970811 - 980806 3APEX CO (4 −
3) 461.041 55.3 14 0.60 070918 0.5Odin CO (5 −
4) 576.268 83.0 118 0.90 050502 - 050620 4APEX CO (7 −
6) 806.652 155.9 8 0.43 070918 0.8Herschel-HIFI CO (10 −
9) 1151.985 304.2 19 0.66 090726 - 100221 9Odin H O (1 − ) 556.936 42.4 126 0.90 090609 - 100420 12Herschel-HIFI H O (1 − ) 556.936 42.4 39 0.76 100729 0.05Herschel-PACS H O (2 − ) 1669.905 79.5 13 N / A 090226 0.1
2. Observations
The observations described in this paper were obtained between1997 and 2010 with several di ff erent facilities. A summary ofthe observations is presented in Table 1 while the line intensitiesare listed in Table 2. During the Performance Verification Phase of HIFI(de Graauw et al. 2010), the CO (10 −
9) data were obtainedon 26-27 July 2009 and 21 February 2010. The H O data wereobtained 29 July 2010. The 3.5 m Cassegrain telescope hasa Full Width Half Maximum (FWHM) of 38 ′′ at 557 GHz,19 ′′ at 1152 GHz and 13 ′′ at 1670 GHz, respectively. TheHIFI H O (1 − ) spectrum presented in this paper wasobtained in point mode with position switch using band 1 (490 -630 GHz). The OFF spectrum is obtained by a single observa-tion of a reference point 10 ′ away. The CO (10 −
9) HIFI mapswere obtained in two di ff erent observing modes using band 5(1120 - 1250 GHz). In the dual-beam-switch raster mode, aninternal chopper mirror is used to obtain an OFF spectrum 3 ′ away from the observed position. In the on-the-fly with positionswitch mode, the telescope is scanning the map area back andforth. The data were calibrated using the Herschel InteractiveProcessing Environment (HIPE) version 4.2 and 5.0 for theCO (10 −
9) and H O (1 − ) observations, respectively (Ott2010). The data reduction of the HIFI maps is described in detailin Appendix B. For the CO (10 −
9) observation, data from oneof the spectrometers on-board have been used. The Wide BandSpectrometer (WBS) is an acousto-optical spectrometer witha 4 GHz frequency coverage. The channel spacing is 500 kHz(0.1 km s − at 1152 GHz and 0.3 km s − at 557 GHz). For theH O (1 − ) observation, data from the High ResolutionSpectrometer (HRS) have also been used. The HRS is anAuto-Correlator System (ACS) where the resolution can bevaried from 0.125 - 1.00 MHz. For this observation it was setto 0.24 MHz. Observations from the horizontal (H) and thevertical (V) polarisations were combined for both observations.The spectra were converted to a T mb scale using main beame ffi ciencies, η mb (1152GHz) = η mb (557GHz) = The PACS spectrograph (Poglitsch et al. 2010) is a 5 × . ′′ µ m region, centeredon the H O (2 − ) line at 179.5 µ m. The blue channel simul-taneously covered the 89.4–90.2 µ m region, which is featurelessand not discussed further. Two di ff erent nod positions, located6 ′ from the target in opposite directions, were used to correctfor the telescopic background. Data were reduced with HIPEversion 4.0. The fluxes were normalised to the telescopic back-ground and subsequently converted to an absolute flux based onPACS observations of Neptune (Lellouch et al. 2010), with anapproximate uncertainty of ∼
20 % at 180 µ m. The spatial res-olution at 180 µ m is nearly di ff raction-limited (see Table 1). Inwell-centered observations of point sources, only about 40 %of the light in the system falls within the central spaxel. TheH O (2 − ) line is spectrally unresolved in the R = ∆ υ ∼
175 km s − ). The Odin space observatory carries a 1.1 m Gregorian tele-scope and was launched into space in 2001 (Nordh et al. 2003;Hjalmarson et al. 2003). It is located in a polar orbit at 600 kmaltitude. At 557 and 576 GHz, the FWHM is 126 ′′ and 118 ′′ respectively. The spectra were converted to a T mb scale using amain beam e ffi ciency, η mb = −
4) observations, su ff ered from frequency driftand have, for that reason, been calibrated, using atmosphericspectral lines, acquired during the time intervals when Odinobserved through the Earth’s atmosphere (Olberg et al. 2003).Since the first publication of the CO (5 −
4) data in Bjerkeli et al.(2009), the frequency calibration scheme has improved. Despitethis, the velocity scale for this particular observation has someuncertainties due to in-orbit variations of the local oscillator unitfrequency. These variations are most likely caused by slight tem-perature changes in the spacecraft during each orbit, due to thefact that the Earth is located very nearby. On a velocity scalethese fluctuations correspond to a ∼ − uncertainty.A spectrum, showing a tentative detection of H O (1 − )at 557 GHz was published in Bjerkeli et al. (2009). Since then,however, additional observations toward HH 54 have been car-ried out in June 2009 and April 2010 for a total on-source timeof 12 hours.
3. Bjerkeli et al.: Herschel observations of the Herbig-Haro objects HH 52-54
Table 2.
Integrated intensities over the line wings and 1 σ uncertainties in parentheses. Line Source ∆ υ LSR R T mb d υ T mb , rms a (km s − ) (K km s − ) (mK)CO (2 −
1) HH 54 0.9 to − . −
2) HH 54 0.9 to − . −
3) HH 54 2.4 to − . −
4) HH 54 2.4 to − . −
6) HH 54 2.4 to − . −
9) HH 52 - - 284HH 53 - - 211HH 54 2.4 to − . O (1 − ) b HH 54 2.4 to − . Notes to the Table: a The velocity bin size when calculating the rms isthe same as the channel spacing. b This refers to the spectra obtainedwith HIFI.
The observing mode for both observations was positionswitching, where the entire telescope is re-orientated to obtain areference spectrum (10 ′ away in June 2009 and 15 ′ away in April2010). The spectrometer used is an acousto-optical spectrometer(AOS) where the channel spacing is 620 kHz (0.33 km s − and0.32 km s − at 557 GHz and 576 GHz respectively). The dataprocessing and calibration are described by Olberg et al. (2003). The observations and calibration of the CO (2 − − − −
2) and SiO (5 −
4) data obtained with SESTwere already described in detail by Bjerkeli et al. (2009), towhich we refer the interested reader.
CO (4 −
3) and CO (7 −
6) data were obtained with theAPEX / FLASH receiver in service mode in July 2006. Inthis project , both CO lines were observed simultaneously ina grid map around HH54 B centered on α = h m s · δ = − ◦ ′ ′′ . The selected reference position was rela-tively nearby at (120 ′′ , − ′′ ), which resulted in contaminatedline spectra close to the cloud LSR velocity (see below). Thetelescope pointing was checked by observing the nearby novaX Tra (IRAS 15094-6953). The map was spaced by half theinstrument beam for the CO (4 −
3) transition which correspondsto 7 ′′ at 450 GHz. The data reduction was performed in CLASSand the spectra were converted to a T mb scale using the mainbeam e ffi ciencies, 0.60 and 0.43 for the CO (4 −
3) and CO (7 − − at 461 GHz and 0.02 km s − at807 GHz).
3. Results
The results from our observations are summarised in Table 2,where the integrated intensity over the blue line wing is pre-sented. ESO project code: 077.C-4005(A)
Fig. 2. H O (1 − ) spectra obtained with HIFI. Theblue spectral line is the HRS and the black spectral lineis the WBS. Both observations were centered on HH 54 B: α = h m s · δ = − ◦ ′ ′′ . The dashed line in-dicates the position of the velocity of the cloud. CO (10 −
9) is only detected in the HH 54 region. No emission isdetected toward the region of HH 52-53 down to an rms levelof ∼ −
9) emission, we esti-mate the size of the source to ∼ ′′ . The H O (1 − ) line isclearly detected toward HH 54 (see Fig. 2). Simultaneously, theNH (1 − ) line at 572 GHz was covered in the upper side band.No emission is detected down to an rms level of ∼
20 mK. For theH O (2 − ) line observed with PACS, emission is detectedin most of the spaxels and the angular extent of the source is nolarger than ∼ ′′ . The peak flux in the central spaxel is 9 Jy. Inthe lower panel of Figure 1, the CO (10 −
9) integrated intensitycontours are shown overlaid on the H O (2 − ) normalisedflux in each spaxel. In this figure, each spaxel is presented on asquare grid. In reality however, there is a small misalignment be-tween each spaxel (see Poglitsch et al. 2010, Their Fig 10). Ano ff set of ∼ ′′ between the CO (10 −
9) and the H O (2 − )emission peak is also observed, where the peak of the CO emis-sion is located in between the B and C clumps as identified bySandell et al. (1987). This o ff set might be real given a pointingaccuracy of a few arcseconds for Herschel. Noteworthy is thatthe di ff erent clumps in the region show detectable proper mo-tion over a time scale of a few years (e.g. Schwartz et al. 1984;Caratti o Garatti et al. 2006, 2009). However, the CO and H Oobservations with Herschel were obtained over a time span ofonly one year and it is therefore unlikely that proper motion isthe cause of the observed o ff set. The H O (1 − ) spectra ob-tained with HIFI are presented in Fig. 2. The line is self absorbedby the foreground cloud at υ LSR = + − . The Odin H O (1 − ) observations carried out on 9 June2009 and 20 April 2010 confirmed the previously published de-tection with an improved signal to noise. It is this dataset that isused for the comparison with HIFI data in the present paper.
4. Bjerkeli et al.: Herschel observations of the Herbig-Haro objects HH 52-54
Fig. 3.
CO (10 −
9) map obtained with HIFI. The map shows the spectra toward the region close to HH 54 B and the map has beenregridded with map spacing equal to 9 . ′′
3. O ff sets are with respect to HH 54 B: α = h m s · δ = − ◦ ′ ′′ . The velocityscale ( υ LSR ) and intensity scale ( T mb ) is indicated in the upper right corner of the map, υ LSR = + − with a dashed line. The CO (2 −
1) and CO (3 −
2) maps were centered with a slighto ff set with respect to HH 54 B, viz 4 . ′′
8. The spacing in CO (2 − ′′ while the spacing between the observations in CO (3 − ′′ , i.e one full beam width. The number ofpositions observed in the CO (2 −
1) map and the quality of thebaselines in the CO (3 −
2) does not allow us to put constraintson the source size. SiO was not detected, when averaging allspectra together, down to an rms level of 10 mK for SiO (2 − −
2) and 7 mK for SiO (5 − The quality of the FLASH data is not fully satisfactory, asthey su ff er from o ff -beam contamination near the line centre.However, as we here are mainly focussing on the line wings,this should a ff ect our conclusions only little, if at all. In theseCO line maps, a bump feature (See Sec. 4.1) in the line profilewas detected in some positions (at υ LSR = − − , see below).A reliable source size from these maps can however not be deter-mined. In both maps, the feature is detected in a velocity rangespanning over ∆ υ ≃ − (see Fig. 4) . Table 3.
Parameters used in the CO model
Parameters held constant
Distance to source D source =
180 pcCO abundance X (CO) = × − LSR velocity υ LSR = − Velocity profile υ ( r ) = r / R max km s − Shell thickness ∆ r / R max = θ source = ′′ Microturbulence v turb = . − Gas to dust mass ratio M gas / M dust = κ µ m =
25 cm g − Dust frequency dependence β = Free parameters H density n (H ) = × - 1 × cm − Kinetic temperature T kin =
10 - 330 K 5. Bjerkeli et al.: Herschel observations of the Herbig-Haro objects HH 52-54
Fig. 4.
Same as Figure 3, but for the CO (2 − − −
3) and CO (7 −
6) maps obtained with SEST and APEX. The APEXmaps have been regridded with a map spacing equal to 8 ′′ for CO (4 −
3) and 9 ′′ for CO (7 − −
3) spectra maybe contaminated with OFF beam contribution.
4. Discussion
Common to the observed transitions in CO and H O is thatonly blue-shifted emission is detected. For all transitions, themaximum detected velocity in the line wing is of the order of −
20 km s − . A bump-like feature at υ LSR h − − is alsoclearly visible in the observed CO (10 −
9) and CO (2 −
1) spectraand possibly also in the CO (3 − −
4) and H O (1 − )data. In the CO (10 −
9) map, this feature is more prominent insome positions than others, and most likely it is spatially unre-solved to Herschel (see Fig 3). The υ LSR ≃ − − compo-nent is also clearly visible in the CO (4 −
3) and CO (7 −
6) spec-tra observed with APEX (see Figure 4) where the beam sizes are13 ′′ and 8 ′′ , respectively. Also in these maps, this componentseems unresolved, hence it likely originates from a region withan angular extent that is smaller than the telescope beams. In theCO (5 −
4) data, where the beam size is 118 ′′ , the bump-like fea-ture is barely visible. This is expected if the beam filling factor issmall. A position-velocity diagram of the CO (10 −
9) transitionshows a trend of higher velocities being detected at lower decli-nation and closer to the position of HH 54 B (see Figs. 3 and 5),i.e where the H O (2 − ) emission peaks (see Fig. 1). Thisis also clearly visible in the right panel of Fig. 5 where the inte- grated intensity of the bump is presented together with the inte-grated intensity for the underlying triangular profile. The inten-sity maximum of the bump seems o ff set by ∼ ′′ to the southfrom the peak of the bulk outflow emission. The uncertainty at-tributed to this o ff set due to the baseline subtraction is of theorder ∼ ′′ . Assuming that the apex of the shock is located closeto the HH54 B position, one would also expect the highest ve-locities in this region. To compute the line profiles for the observed emission, we usean Accelerated Lambda Iteration (ALI) code (See App. C). TheALI code we use is a non-LTE, one dimensional code, assum-ing spherical geometry where several subshells are used. Thenumber of cells and angles used in the ray tracing can also bearbitrarily chosen. In this work, a curved geometry is comparedwith a plane parallel slab to interpret the observed spectral lines.The formation of the observed spectral lines occurs mostlikely in shocked gas. Using results from detailed models ofC-shocks, Neufeld et al. (2006) presented estimates of the ex-citation conditions for HH 54. From the analysis of H -rotationdiagrams, Neufeld et al. (2006) determined the presence of two
6. Bjerkeli et al.: Herschel observations of the Herbig-Haro objects HH 52-54 −25 −20 −15 −10 −5 0 5−20−1001020304050 Velocity [km s −1 ] D e c o ff s e t [ " ] I n t en s i t y [ K ] D e c o ff s e t [ " ] BCK I n t eg r a t ed i n t en s i t y [ K k m s − ] Fig. 5.
Left panel : Position (in DEC) velocity diagram of the HIFI CO (10 −
9) data showing the observed intensity. The bump-likefeature can be seen at υ LSR = − − . The o ff set in RA is 0 ′′ . υ LSR =+ − is indicated with a dashed line. Right panel : Blue contours are the integrated intensity when the bump-like feature is subtracted from the spectra. Contours are from3.6 to 28.7 K km s − in steps of 3.1 K km s − . Red contours show the integrated intensity for the bump. Contours are from 2.6 to7.5 K km s − in steps of 0.6 K km s − . Underlying colors show the total integrated intensity over the observed line. The positions ofHH 54 B, C and K are indicated with yellow squares.di ff erent temperature regimes, viz. at 400 K and at 10 K, re-spectively.We used their analytical expressions for the column density, N (H ) = . × " n (H )10 . T gas ! − . cm − , (1)and for the velocity gradient, d υ dz = . × " n (H )10 . T gas ! . km s − pc − . (2)to compute the emission in the CO and H O lines. Here n (H )is the pre-shock density and the other quantities have their usualmeaning. Following Neufeld et al. (2006) we take the compres-sion factor of 1.5, which was used by them for H , but we assumethat holds for CO as well. In addition, they assume that the frac-tional beam filling of the two-temperature components is givenby the ratio of the column density derived from the rotation dia-gram and that given by Eq. 1.For these two temperatures, and for pre-shock densitites of10 cm − and 10 cm − , we compute from, Eqs. (1) and (2) inslab geometry, the emission in CO and H O lines, including theshapes of the lines. Other parameters (such as CO abundance,source size, microturbulence etc.) are given in Table 3. For theH O abundance we assume X (H O) = × − , consistentwith the upper limit of < × − determined by Neufeld et al.(2006).The results are presented in Figs. 7 and 9. As seen in thelatter figure, the computed line strength of the ground state lineof o -H O for a temperature of 400 K and pre-shock density of10 cm − is not far from what HIFI has observed. On the otherhand, such parameters are not in agreement with the observedCO (10 −
9) line, the intensity of which is severely over-predicted.As expected, the low-J lines CO (2 −
1) and CO (3 −
2) are not very These were the pre-shock densities used by Neufeld et al. (2006) tocompute the H abundance. Fig. 6.
A cut through the shells of the spherical model describedin the text. Essentially all radiation originating from the rearside of the sphere (grey) is blocked out by the central blackbodysource.sensitive to the changes in temperature from 400 to 1000 K. In allcases the computed rectangular line shape is di ff erent from theobserved profiles, which have a pronounced triangular shape. The spatial distribution of the gas does also influence the ob-served line profiles (see e.g. Hartigan et al. 1987). For that rea-son we also investigate a scenario where the emission originatesfrom a curved geometry. To implement this we use a simplemodel that mimics an expanding shell with a diameter of 30 ′′ lo-cated at a distance of 180 pc (see Fig 6). The interior of the spher-
7. Bjerkeli et al.: Herschel observations of the Herbig-Haro objects HH 52-54 −30 −20 −10 0 10 20 30−1012345 I n t en s i t y [ K ] Velocity [km s −1 ]SEST CO(2−1) −30 −20 −10 0 10 20 30−50510 I n t en s i t y [ K ] Velocity [km s −1 ]SEST CO(3−2)−30 −20 −10 0 10 20 30−5051015 I n t en s i t y [ K ] Velocity [km s −1 ]APEX CO(4−3) −30 −20 −10 0 10 20 30−15−10−50510152025 I n t en s i t y [ K ] Velocity [km s −1 ]APEX CO(7−6)−30 −20 −10 0 10 20 30−0.200.20.40.60.811.2 I n t en s i t y [ K ] Velocity [km s −1 ]Odin CO(5−4) −30 −20 −10 0 10 20 30−101234 I n t en s i t y [ K ] Velocity [km s −1 ]Herschel/HIFI CO(10−9) Fig. 7.
The red (1000 K) and green (400 K) lines represent the model described in Neufeld et al. (2006), using pre-shock densitiesof 10 cm − (dashed) and 10 cm − (solid). The observed spectra from SEST, APEX, Odin and HIFI are plotted with black solidlines. Note that the modelled spectra, not have been baselines subtracted.ical shell is empty, i.e. at the temperature of cosmic backgroundradiation field. This cold sphere occupies 88% of the radius ofthe sphere, based on a shell thickness of 5 × cm. This valueis in between the slab thickness estimated by Liseau et al. (1996)and the analytical expression for the slab thickness at 180 K.The shell thickness is also consistent with the cooling length es-timated by Kaufman & Neufeld (1996). Using this method, weblock out essentially all the radiation originating from the oppo-site side of the sphere.To compare with the observed line profiles, spectra are com-puted viewing the curved surface from the front. This is sup-ported by the absence of detectable SiO emission which is of-ten observed in molecular outflows with relatively high veloci-ties (see e.g. Nisini et al. 2007). Shock modelling carried out byGusdorf et al. (2008) suggests that sputtering is not very e ffi cientin the velocity regime below 25 km s − . Therefore, assuming thatthe outflow in HH 54 is observed from the front, and that it hasa small inclination angle with respect to the line of sight, themaximum velocity of the molecular gas is likely lower than this. Inspection of the CO and H O spectra show a maximum ra-dial velocity of ∼
20 km s − . This is also consistent with the mod-elling carried out by Giannini et al. (2006) who suggested a C + Jtype shock with a maximum velocity of 18 km s − . The velocityin the shell increases linearly from 0 to 20 km s − (see Table 3).The true velocity profile is most likely more complicated. In abow shock, the velocity component perpendicular to the jet di-rection is expected to be smaller than the component parallel tothe jet direction. This makes the model somewhat simpler thanreality. The velocity field within the shocked region probablyalso has a more complicated profile. Furthermore, we assumethat the emission in all lines stem from the same region of size30 ′′ . The bump-like feature discussed in Section 4.1, indicatinga deviation from a linear velocity profile, has not been consid-ered in the modelling presented here. The parameters used in themodel are summarised in Table 3.We set up a grid where we vary the H density from 10 to 10 cm − and the kinetic temperature of the gas from 30to 330 K. Thus, our model is steady state and in equilibrium.We choose this approach in order to keep the number of free
8. Bjerkeli et al.: Herschel observations of the Herbig-Haro objects HH 52-54 −30 −20 −10 0 10 20 30−1012345 I n t en s i t y [ K ] Velocity [km s −1 ]SEST CO(2−1) −30 −20 −10 0 10 20 30−50510 I n t en s i t y [ K ] Velocity [km s −1 ]SEST CO(3−2)−30 −20 −10 0 10 20 30051015 I n t en s i t y [ K ] Velocity [km s −1 ]APEX CO(4−3) −30 −20 −10 0 10 20 30−5051015 I n t en s i t y [ K ] Velocity [km s −1 ]APEX CO(7−6)−30 −20 −10 0 10 20 30−0.200.20.40.60.81 I n t en s i t y [ K ] Velocity [km s −1 ]Odin CO(5−4) −30 −20 −10 0 10 20 30−10123 I n t en s i t y [ K ] Velocity [km s −1 ]Herschel/HIFI CO(10−9) Fig. 8.
The six spectra obtained with SEST, APEX, Odin and Herschel compared to the best-fit model (see Sec. 4.3.1).parameters as small as possible. Also, it is worth noting, thatwe have not achieved a better fit to the bump-like feature (seeSec. 4.1) when varying the density and temperature profiles overthe shells. To find the best fit density and kinetic temperature,the reduced χ is minimised, where the di ff erence between theobserved and the modelled intensity is evaluated in each veloc-ity bin (see Fig. 10). The CO (2 − − −
4) andCO (10 −
9) lines are included in the χ - minimisation whereasthe CO (4 −
3) and CO (7 −
6) lines are exluded due to the con-tamination from the o ff position and the high noise level of theCO (7 −
6) observation. For all the CO spectra, we only take theline wings into consideration, due to the fact that emission fromthe surrounding cloud is clearly visible in the CO (2 −
1) spectra.
From the curved geometry model, the best fit gas density and ki-netic gas temperature are n (H ) = × cm − and T kin =
180 K(formally 177 K) respectively. This implies a total H mass of ∼ × − M ⊙ . The value of the reduced χ is 2.3. The CO spec-tra obtained with Odin, SEST and HIFI are plotted in Figure 8together with the modelled spectra. The model fits the observa- tions well, and we conclude that the geometry of the region canbe a crucial parameter determining the shape of the line profiles.The CO (10 −
9) line observed with Herschel, however, shows aslightly more complicated profile than predicted (see discussionin Section 4.1). Thus, the triangular form is distorted by the pres-ence of the υ LSR ≃ − − feature. A minor change in thekinetic temperature (i.e to 170 K) provides a better fit to the un-derlying triangular shape of the CO (10 −
9) line without a ff ectingthe lower-J CO lines by much. The computed maximum opti-cal depths are ≤ × − for all the modelled CO lines. Usingthe gas density and kinetic temperature obtained from the COmodelling, the observed o -H O ground state transition is bestfit with an ortho -water abundance with respect to H of X ( o -H O) = × − (see Figure 11). The observed total flux in themap obtained with PACS corresponds to an integrated line in-tensity of 20 K km s − . This is in agreement with the integratedintensity from the predicted line profile to within a factor of 2. O Line profile predictions
The line profiles for the six lowest rotational transitions of or-tho -water have been computed. The abundance is set to 1 × −
9. Bjerkeli et al.: Herschel observations of the Herbig-Haro objects HH 52-54
Table 4.
The predicted integrated intensities, maximum intensities, continuum levels and optical depths for the six lowest rota-tional transitions of ortho -water. The maximum and integrated intensities are for the baselines subtracted spectra. The foregroundabsorption, visible in the observed H O (1 − ) spectrum, is not considered in this table. Line ν FWHM Receiver R T mb d υ R F λ d λ T mb , max T cont τ max (GHz) ( ′′ ) (K km s − ) (erg cm − s − ) (K) (mK)1 -1 -1 / PACS 13.0 28 0.80 80 442 -1 -2 / PACS 1.8 3.8 0.13 80 0.33 -2 / PACS 2.3 5.4 0.15 80 0.53 -2 < -3 < -2 −30 −20 −10 0 10 20 3000.20.40.6 Velocity [km s −1 ] I n t en s i t y [ K ] Fig. 9.
The red (1000 K) and green (400 K) lines represent themodel described in Neufeld et al. (2006), using pre-shock densi-ties of 10 cm − (dashed) and 10 cm − (solid). The observedspectrum for the H O (1 − ) transition with HIFI is plot-ted with a black solid line. The o -H O abundance is set to X ( o -H O) = × − .and the lines, that are predicted to be strong enough to be read-ily detected with HIFI, are displayed in Figure 12. Also in thisfigure, red-shifted emission originating from the opposite sideof the sphere is present for the model with a curved geometry.Significant changes in the excitation temperature in the inner andouter shells show up in these spectra as weak absorption features.For this reason, the spectra have been computed using more than100 shells to avoid any drastic changes in excitation temperature,due to optical depth e ff ects between each shell. Simultaneouslywith the CO (10 −
9) observation, H O (3 − ) at 1153 GHzwas also observed. The observed noise level of ∼ ∼
10 mK. A summary of these predictions is presented inTable 4. In this table, the observed integrated intensity in theH O (1 − ) line is 30 % lower than the predicted value dueto the absorption from foreground gas. If this gas is at a low tem-perature, however, the higher transitions should not be a ff ectedby much.
140 160 180 200 2204.94.9555.055.1 l og ( n ) [ c m − ] Temperature [K]
Fig. 10.
Contour plot showing the sum of the chi-squares, com-paring the modelled spectra and the observations. The 5%, 10%,20% and 30% deviations from the minimum value are indicatedwith solid lines. The best-fit density and temperature are indi-cated with a cross.
For the curved model described in Sec. 4.3, we derive the cool-ing ratio Λ (CO) / Λ ( o -H O) =
10 where the 557 GHz line is thedominant contributor to the water cooling. This value agrees wellwith the earlier determination of 7 based on ISO-LWS data andpresented by Liseau et al. (1996).
As already discussed in Sec. 4.1, the bump-like feature seemsunresolved to Herschel, i.e the size of the source is uncer-tain. A source size comparable to the beam size of Herschel,at 1152 GHz (19 ′′ ), would have to have a high temperature( > ∼ cm − ) to fit the observations.This implies that the flux in the high-J CO lines would be higherthan what was observed with ISO-LWS (Giannini et al. 2006).However, a source size as small as ∼ ′′ , where the temperatureand density would have to be ∼
400 K and ∼ cm − respec-tively, is a plausible scenario. In that case however, the ortho -water abundance has to be less than 10 − to fit with the ISO-LWS observations. In addition, the source is likely not smaller
10. Bjerkeli et al.: Herschel observations of the Herbig-Haro objects HH 52-54 −30 −20 −10 0 10 20 3000.20.40.60.811.21.4 I n t en s i t y [ K ] Velocity [km s −1 ] Fig. 11.
The HIFI H O (1 − ) spectrum compared with themodel spectrum for X ( o -H O) = × − . Spectra with the o -H O abundance increased by a factor of two (dashed line) anddecreased by a factor of two (dotted line) are also included in-cluded. Note that the foreground gas is visible in absorption inthe observed spectrum. −30 −20 −10 0 10 20 3000.20.40.60.8
Velocity [km s −1 ] I n t en s i t y [ K ] H O (2 −1 ) H O (2 −2 ) H O (3 −2 ) H O (3 −3 ) Fig. 12.
Line profile predictions for the H O (2 − ),H O (2 − ), H O (3 − ) and H O (3 − ) spectra asobserved with HIFI.than 1 ′′ . Also in this case, the high-J CO lines would be strongerthan what is actually observed.Using a source size of 10 ′′ (this size is estimated fromH maps presented in Neufeld et al. (2006)), the observed CO”bullet”-emission is well fit with a temperature, T kin =
600 K,and a density, n (H ) = × cm − . This yields a column den-sity, N (H ) = × cm − . For an ortho -water abundance of X ( o -H O) = × − , the H O (3 − ) emission is entirelyaccounted for by the ISO-LWS observations. In this case theH O (2 − ) emission would be merely about 25% of whatwas observed with PACS and no significant contribution wouldbe observed in the H O (1 − ) line. Therefore, we concludethat the o -H O abundance in the ”bullet” is likely lower than10 − . −16 −15 −14 −13 −12 −11 Rot. quant. number [J] F l u x [ e r g c m − s − ] SESTOdinHIFIISOAPEX
Fig. 13.
Observed CO line fluxes (squares) as a function of therotational quantum number J. The errorbars are the estimatedcalibration uncertainties (10% for SEST, Odin and HIFI and30% for APEX and ISO). The best fit model, described in thepresent paper (see Sec. 4.3.1), is indicated with a solid line. Thefit to the high-J CO lines is indicated with black circles (seeSec. 4.3.5).
The derived H density implies a column density of5 × cm − which is approximately one order of magnitudehigher than what is derived for the warm gas (Neufeld et al.2006). The best fit temperature, T kin =
180 K, is significantlylower than the temperature in the gas responsible for the high-J CO emission [i.e CO (14 −
13) – CO (20 − ff erent tempera-ture regimes are present in HH 54 and a one-temperature, one-density model cannot explain all the infrared observations. Thehigh-J CO emission observed with ISO can be explained usingthe same geometry and source size but having a shell thickness4 × cm, i.e one order of magnitude thinner. In that case aslightly higher density ( n (H ) = × cm − ) and tempera-ture (T =
500 K) fits the high-J CO lines well. On the other hand,this secondary component also makes a significant contributionto the CO (10 −
9) line and this emission may originate from dif-ferent components. Recently, Takami et al. (2010) report mor-phological di ff erences between Spitzer observations in the 3.6,4.5, 5.8 and 8.0 µ m bands. The emission is observed to be lesspatchy in the long wavelength bands (i.e. 5.8 and 8.0 µ m) andthey interpret this as thermal H emission being more enhancedin regions of lower density and temperature. Given the fact thathot gas obviously is present in this region, the secondary com-ponent may in reality be in smaller regions of high temperaturesimilar to those observed in H . In Figure 13, the CO line flux isplotted as a function of the rotational quantum number, J.The predicted integrated intensities can be compared withthe modelling presented in Giannini et al. (2006). These au-thors present a multi-species analysis where they conclude thatthe observed H , CO and H O lines can only be explained
11. Bjerkeli et al.: Herschel observations of the Herbig-Haro objects HH 52-54 by a J-shock with magnetic precursor. The C + J shock model,presented in their paper, explains the observed H O (2 − )and H O (3 − ) line fluxes of 7 × − erg cm − s − and2 × − erg cm − s − well, and they predict the line flux for theH O (1 − ) line to be 2.0 × − erg cm − s − . Convertedto a K km s − scale these values correspond to 6.8, 1.4 and53 K km s − respectively. Taking the beam size into account,the predicted integrated intensity for the H O (1 − ) line inthis paper is at least a factor of four lower. This could be dueto the fact that this line is very sensitive to the type of shockpresent. This is also discussed in the paper by Giannini et al.(2006), where they note that a J type shock would change theflux by more than a factor of two downwards.Recently, Flower & Pineau des Forˆets (2010) presented the-oretical predictions of CO and H O line intensities, based ondetailed C- and J-type shock model calculations. Expectedly,the assumed magnetic fields are di ff erent for their models of C-and J-shocks, i.e. for b = b = .
1, respectively, where b is defined through B = b √ n H µ G. For their high density C-shocks (see below), this means that pre-shock fields of order450 µ G should be present. Based on OH Zeeman observations,Troland & Crutcher (2008) detected 9 dark clouds in a sample of34. Corresponding line-of-sight magnetic field strengths were inthe range B los = − µ G, with typical values around 15 µ G. Asdiscussed by Troland & Crutcher (2008), for randomly orientedfields, B los = . × | B | . Hence, magnetic fields in dark cloudsdo not likely exceed levels of 20 − µ G. This refers to the ob-served scales of 3 ′ . However, these authors also showed that ina given cloud, B los did not change appreciably from one posi-tion (active molecular outflow) to another (quiescent surround-ing cloud). It seems, therefore, that fields as strong as 450 µ Gmay not be that common. On the other hand, order of magnitudelower field strengths (45 µ G for b = .
1) are more consistentwith the observational evidence and may promote the occurrenceof J-shocks.For the radiative transfer Flower & Pineau des Forˆets (2010)used an LVG approximation in slab geometry. In their Fig. 8,line profiles for CO (5-4) and H O (1 − ) are shown for C-shocks with two densities and four shock velocities. These linesare close in frequency and observations of HH 54 with Odin aremade with essentially the same telescope beam, rendering reso-lution issues to be of only minor importance, i.e. any scaling ofthe intensities due to the source size should a ff ect both observedlines in the same way. It should be feasible, therefore, to directlycompare the line profiles of our Odin observations with thoseof the models by Flower & Pineau des Forˆets (2010). Based onthe observed maximum radial velocities, we consider only mod-els with υ s ≥
20 km s − . Their models with shock velocities of20 km s − could correspond to a head-on view, whereas theirmodels for υ s =
30 and 40 km s − could correspond to inclina-tions of the flow with respect to the line of sight of 48 ◦ and 60 ◦ ,respectively .For the C-shock models, several of the line profiles displayshapes that are qualitatively similar to those observed with Odinand HIFI. These line shapes are a consequence of the computedflow variables, not a geometry e ff ect. However, in particular forthe range of ∆ υ ∼ − , the Odin / HIFI linesexhibit CO-to-H O intensity ratios very much in excess of unity,i.e. T CO ( ∆ υ ) / T H O ( ∆ υ ) ≫ ∆ υ is the velocity relative to therest frame of the flow, i.e. υ LSR . For the emission knots of HH 54 Caratti o Garatti et al. (2006) de-termined an average inclination of 27 ◦ , which would imply that υ s ∼ υ obs . −30 −25 −20 −15 −10 −5 0 5 1000.20.40.60.81 Velocity [km s −1 ] I n t en s i t y [ K ] Fig. 14.
Comparison between the CO (5 −
4) line (black) and theH O (1 − ) line obtained with Odin (red). The H O spec-trum has been multiplied with 6 for clarity. Also the HIFIH O (1 − ) spectrum has been plotted in this figure for com-parison (blue). The intensity of the emission has in this case beenmultiplied by 6 and corrected for beam filling. −1 v/v shock T C O / T H O Fig. 15.
Line ratios between CO (5 −
4) and H O (1 − ). Theratio as measured from Odin is indicated with a black line. Thepredictions presented by Flower & Pineau des Forˆets (2010) for υ shock =
20 km s − are plotted in blue for n (H) = × cm − and in red for n (H) = × cm − .These observed ratios are very much larger than the theoreti-cal values (Fig. 14 and 15). Albeit the ratios are larger than unityfor the low-density ( n H = × cm − ) C-shocks, these fall stillfar below the observed ones. On the other hand, the high den-sity ( n H = × cm − ) cases could directly be dismissed, asthese tend to show inverted ratios, i.e. T CO / T H O ≪
1, contraryto what is observed (Fig. 15).For the J-shock models, Flower & Pineau des Forˆets (2010)list the predicted integrated intensities. Also in this case, the pre-dicted CO-to-H O ratios (for a shock velocity of 20 km s − ) aremuch lower than the observed ratio, viz 0.06 and 0.005 for thepre-shock densities n H = × cm − and n H = × cm − respectively.
12. Bjerkeli et al.: Herschel observations of the Herbig-Haro objects HH 52-54
The modelling of planar C-shocks show that low densities are re-quired for cooling rate ratios, Λ (CO) / Λ ( o -H O) >
1, in contrastto our own findings, where densities at least as high as 10 cm − seemed implied by the observations (Sect. 4.3.1). The cause forthis mismatch is not clear to us, but one of the reasons could bethe di ff erence between modelled planar and curved geometry.Such modelling should therefore be attempted. The relativelyhigh cooling rate ratio, Λ (CO) / Λ ( o -H O) =
10, is also noteasily reconcilable with the presence of a C-type shock where aratio, Λ (CO) / Λ ( o -H O) ≪ ortho -water abundance deter-mined from the modeling indicate that J-type shocks may con-tribute.As already discussed in Sect. 4.3.5, di ff erent temperatureregimes are present in the HH 54 region. It would be adequatetherefore, to repeat the ISO-LWS observations, using the highersensitivity and spatial resolution provided by PACS. The mid-JCO lines fall in the wavelength range covered by the Spectraland Photometric Imaging Receiver (SPIRE) and these should beobserved.
5. Conclusions
Based on spectral mapping with Herschel of the region contain-ing the Herbig-Haro objects HH 52 to HH 54 we conclude thefollowing: • The CO (10 −
9) 1152 GHz line was clearly detected onlytoward the position of HH 54 with a FWHM < ∼ ′′ , compa-rable to the extent of the HH-emission knots in the visibleand infrared. • The H O (2 − ) 1669 GHz line was clearly detected to-ward HH 54 and with a similar extent as the CO (10 −
9) line.An o ff set of 9 ′′ (2 . × cm) between the two species isobserved, but the reality of this can at present not be firmlyassessed. The H O (1 − ) 557 GHz line was also clearlydetected toward HH 54. • The CO (10 −
9) spectra show only blueshifted emission,with maximum relative velocities in excess of −
20 km s − .The line profiles exhibit typically a triangular shape, whichin certain positions is however contaminated by a bump-likefeature at υ LSR = − − . This feature is constant in veloc-ity and width. It is limited in spatial extent and may be iden-tified by what is commonly called a ”bullet”. Comparison ofthe observed spectra with analytical bow shock line profileslimits the viewing angle to greater than zero but < ∼ ◦ . • The bump is clearly seen in position-velocity cuts, reveal-ing two peaks, in addition to a smooth velocity gradient ofabout 10 km s − pc − . The low-velocity peak appears closeto the ambient cloud velocity, whereas the second peak cor-responds to the bump. For this feature we determine fromGaussian fit measurement of the CO (10 −
9) data a relativepeak- T mb = . υ LSR = − − and FWHM = − . • These line features are also observed in the CO (5 −
4) spec-trum observed with Odin and in spectra obtained from theground, i.e. in maps of the (2 − − −
3) and (7 −
6) COtransitions. In addition to the triangular line shape, these datado therefore also confirm the reality of the spectral bump. • We initially use physical parameters for shock models ofHH 54 found in the literature to compute the CO spectra.These models presented two-component fits, given by ana- lytical expressions for temperature and density in the respec-tive slabs. These models were only moderately successful inreproducing the observational results, in particular what re-gards the line shapes. Considerable improvement was found,however, in computed spectra using a curved geometry in-stead. • Using the best fit model parameters (in a χ -sense) for theCO data we computed spectra for H O. This model fits theH O 557 GHz line observed with HIFI for an ortho-waterabundance with respect to molecular hydrogen, X (H O) = × − . • A cooling rate ratio, Λ (CO) / Λ ( o -H O) ≫
1, is not eas-ily reconcilable with recent shock modelling. On the otherhand, a relatively low water abundance ( ∼ − ), supports ascenario where J-shocks contribute significantly to the ob-served emission. This is also consistent with magnetic fieldstrengths measurements toward dark clouds, where B -valueslower than what is needed for C-type shocks, typically areobserved. • Comparison of our Odin data with line profiles from recentdetailed C-shock model computations in the literature sug-gests that some refinements of these models may be required.
Acknowledgements.
The authors appreciate the support from A. Caratti oGarrati for providing the H α image shown in Figure 1. Aa. Sandqvist is thankedfor scheduling the Odin observations of HH 54. We also thank the WISHinternal referees Tim van Kempen and Claudio Codella for their e ff orts.HIFI has been designed and built by a consortium of institutes and uni-versity departments from across Europe, Canada and the United States underthe leadership of SRON Netherlands Institute for Space Research, Groningen,The Netherlands and with major contributions from Germany, France and theUS. Consortium members are: Canada: CSA, U.Waterloo; France: CESR, LAB,LERMA, IRAM; Germany: KOSMA, MPIfR, MPS; Ireland, NUI Maynooth;Italy: ASI, IFSI-INAF, Osservatorio Astrofisico di Arcetri- INAF; Netherlands:SRON, TUD; Poland: CAMK, CBK; Spain: Observatorio AstronomicoNacional (IGN), Centro de Astrobiologia (CSIC-INTA). Sweden: ChalmersUniversity of Technology - MC2, RSS & GARD; Onsala Space Observatory;Swedish National Space Board, Stockholm University - Stockholm Observatory;Switzerland: ETH Zurich, FHNW; USA: Caltech, JPL, NHSC.
13. Bjerkeli et al.: Herschel observations of the Herbig-Haro objects HH 52-54
References
Alcal´a, J. M., Spezzi, L., Chapman, N., et al. 2008, ApJ, 676, 427Bergin, E. A., Neufeld, D. A., & Melnick, G. J. 1998, ApJ, 499, 777Bjerkeli, P., Liseau, R., Olberg, M., et al. 2009, A&A, 507, 1455Cambr´esy, L. 1999, A&A, 345, 965Caratti o Garatti, A., Eisl¨o ff el, J., Froebrich, D., et al. 2009, A&A, 502, 579Caratti o Garatti, A., Giannini, T., Nisini, B., & Lorenzetti, D. 2006, A&A, 449,1077Contopoulos, I. & Sauty, C. 2001, A&A, 365, 165de Graauw, T., Helmich, F. P., Phillips, T. G., et al. 2010, A&A, 518, L6 + Draine, B. T. 1980, ApJ, 241, 1021Dubernet, M.-L., Daniel, F., Grosjean, A., & Lin, C. Y. 2009, A&A, 497, 911Faure, A., Crimier, N., Ceccarelli, C., et al. 2007, A&A, 472, 1029Flower, D. R. & Pineau des Forˆets, G. 2010, MNRAS, 406, 1745Franklin, J., Snell, R. L., Kaufman, M. J., et al. 2008, ApJ, 674, 1015Giannini, T., McCoey, C., Nisini, B., et al. 2006, A&A, 459, 821Gredel, R. 1994, A&A, 292, 580Gregorio-Hetem, J. C., Sanzovo, G. C., & Lepine, J. R. D. 1989, A&AS, 79, 452Gusdorf, A., Cabrit, S., Flower, D. R., & Pineau Des Forˆets, G. 2008, A&A, 482,809G¨usten, R., Nyman, L. Å., Schilke, P., et al. 2006, A&A, 454, L13Haro, G. 1952, ApJ, 115, 572Haro, G. 1953, ApJ, 117, 73Hartigan, P., Raymond, J., & Hartmann, L. 1987, ApJ, 316, 323Hayakawa, T., Cambr´esy, L., Onishi, T., Mizuno, A., & Fukui, Y. 2001, PASJ,53, 1109Herbig, G. H. 1950, ApJ, 111, 11Herbig, G. H. 1951, ApJ, 113, 697Hildebrand, R. H. 1983, QJRAS, 24, 267Hjalmarson, Å., Frisk, U., Olberg, M., et al. 2003, A&A, 402, L39Hollenbach, D. J., Cherno ff , D. F., & McKee, C. F. 1989, in ESA SpecialPublication, Vol. 290, Infrared Spectroscopy in Astronomy, ed. E. B¨ohm-Vitense, 245–258Justtanont, K., Bergman, P., Larsson, B., et al. 2005, A&A, 439, 627Kaufman, M. J. & Neufeld, D. A. 1996, ApJ, 456, 611Knee, L. B. G. 1992, A&A, 259, 283K¨onigl, A. & Pudritz, R. E. 2000, Protostars and Planets IV, 759Lellouch, E., Hartogh, P., Feuchtgruber, H., et al. 2010, A&A, 518, L152 + Liseau, R., Ceccarelli, C., Larsson, B., et al. 1996, A&A, 315, L181Liseau, R. & Sandell, G. 1986, ApJ, 304, 459Lovelace, R. V. E., Romanova, M. M., Ustyugova, G. V., & Koldoba, A. V. 2010,MNRAS, 408, 2083Maercker, M., Sch¨oier, F. L., Olofsson, H., Bergman, P., & Ramstedt, S. 2008,A&A, 479, 779Melnick, G. J., Stau ff er, J. R., Ashby, M. L. N., et al. 2000, ApJ, 539, L77Neufeld, D. A., Melnick, G. J., Sonnentrucker, P., et al. 2006, ApJ, 649, 816Nisini, B., Codella, C., Giannini, T., et al. 2007, A&A, 462, 163Nisini, B., Lorenzetti, D., Cohen, M., et al. 1996, A&A, 315, L321Nordh, H. L., von Sch´eele, F., Frisk, U., et al. 2003, A&A, 402, L21Olberg, M. 2010, Technical Note: ICC / + Pilbratt, G. L., Riedinger, J. R., Passvogel, T., et al. 2010, A&A, 518, L1 + Poglitsch, A., Waelkens, C., Geis, N., et al. 2010, A&A, 518, L2 + Rybicki, G. B. & Hummer, D. G. 1991, A&A, 245, 171Sandell, G., Zealey, W. J., Williams, P. M., Taylor, K. N. R., & Storey, J. V. 1987,A&A, 182, 237Schwartz, R., Jones, B. F., & Sirk, M. 1984, AJ, 89, 1735Shu, F. H., Najita, J. R., Shang, H., & Li, Z. 2000, Protostars and Planets IV, 789Takami, M., Karr, J. L., Koh, H., Chen, H., & Lee, H. 2010, ApJ, 720, 155Troland, T. H. & Crutcher, R. M. 2008, ApJ, 680, 457van Dishoeck, E. F., Kristensen, L. E., Benz, A. O., et al. 2011, PASP, 123, 138van Kempen, T. 2008, PhD thesis, Leiden Observatory, Leiden University,P.O. Box 9513, 2300 RA Leiden, The NetherlandsWhittet, D. C. B., Prusti, T., Franco, G. A. P., et al. 1997, A&A, 327, 1194Wirstr¨om, E. S., Bergman, P., Black, J. H., et al. 2010, A&A, 522, A19 + Wu, Y., Wei, Y., Zhao, M., et al. 2004, A&A, 426, 503Yang, B., Stancil, P. C., Balakrishnan, N., & Forrey, R. C. 2010, ApJ, 718, 1062Ybarra, J. E. & Lada, E. A. 2009, ApJ, 695, L120Young, K. E., Harvey, P. M., Brooke, T. Y., et al. 2005, ApJ, 628, 283
Appendix A:
A.1. Extinctionby dust and dust parameters
At the position of HH 54, the molecular cloud is relatively ten-uous as the dust extinction through the cloud amounts merelyto A V ∼ -column density of N (H ) = × cm − .From [Fe II] line observations, Gredel (1994) determined thevisual extinction toward the HH object as A V = . However, an R V -value of > ∼
5, which is significantly higher than that used by Gredel (1994)and by others, needs to be invoked to describe the Cha II dust(see the discussion by Alcal´a et al. 2008, and references therein).This reflects a predominantly ’big-grain’ contribution to the ex-tinction (almost 2 magnitudes larger A V per unit E B − V ) and weconsequently chose a shallower dust opacity dependence on thewavelength in the FIR and submm, identified by the parameter β in Table 3, i.e. β = β = λ = µ m, we adopt the grain opacity κ =
25 cm g − (Ossenkopf & Henning 1994). The corresponding mass absorp-tion coe ffi cient κ at long wavelengths is then readily found from κ λ = κ ( λ /λ ) β .Combining our own results with those found in the literatureyields an SED of a putative core at 10 K which provides a strictupper limit to the luminosity. Any associated central point sourcewould have a limiting luminosity of L bol < . L ⊙ . Assumingoptically thin emission at long wavelengths and using standardtechniques, the mass of the 10 K core / envelope is estimated at amere 2 M ⊕ , excluding the possibility of a very young age of thehypothetical central source. A.2. The excitingsource
Several IRAS point sources are found in the surroundings ofHH 54 and it is plausible to assume that one of these would be theexciting source of the HH object. Most often, the exciting sourceis a young ( < ∼ Myr) stellar object, situated at the geometrical cen-tre of the bipolar outflow and this seems also to be the case forthe CO outflow associable with IRAS 12515 − near HH 52and HH 53 (Knee 1992). For HH 54, the situation is di ff erenthowever: Knee (1992) identified IRAS 12522 − µ m-band, and it wasshown that this IRAS flux is entirely attributable to [O i ] 63 µ mline emission (Liseau et al. 1996). More recent observations by Spitzer are in support of this conclusion, as these were unable toreveal any point source at this position (Alcal´a et al. 2008).On the basis of proper motion measurements and morphol-ogy considerations, Caratti o Garatti et al. (2009) proposed theClass I object IRAS 12500 − ′ south of HH 54, i.e. atthe projected distance of 1 pc. It was not detected at 1200 µ m From H -line emission at various positions of the HH ob-ject, Giannini et al. (2006) derived the essentially constant value of A V = Alcal´a et al. (2008) identify this source as a likely background K-giant.14. Bjerkeli et al.: Herschel observations of the Herbig-Haro objects HH 52-54 by Young et al. (2005, see below). What regards HH 52 andHH 53, no satisfactory candidate was found by these authors.The possibility of di ff erent exciting sources was also consid-ered by Nisini et al. (1996) and Alcal´a et al. (2008), associatingHH 52 and HH 53 with IRAS 12496 − µ m, no emission was de-tected toward HH 54 nor from the Cha II cloud itself ( S µ m <
20 mJy beam − : van Kempen 2008). However, jet-like exten-sions are seen to emanate from both DK Cha and IRAS 12553 − ′ to 20 ′ south / southeast of HH 54. Narrow, extended emission from theseIRAS sources was also reported by Young et al. (2005), whoused SEST-SIMBA to map the region at 1200 µ m. The align-ment with the HH objects is poor, however. Based on the ob-served spectral indices, this emission is dominated by opticallythin radiation from dust. The dust features appear as tori or rings,seen edge-on, but given their sizes (3 × AU), they do not rep-resent what commonly is called a (protostellar) “disc”. Perhaps,these dust features are left-overs from an earlier history, duringwhich the IRAS sources were formed. A “conventional disc” ofsize ∼ AU could very well hide inside and any outflow wouldbe orthogonal to these disc features.Whereas the dust extension from DK Cha is not, whetherdirect or perpendicular, pointing anywhere near HH 52-54, thevector orthogonal to the linear feature of ISO-Cha II 28, at po-sition angle 51 ◦ , is within 5 ◦ from the current direction towardHH 54. Taking the uncertainty of this estimate into account, thiscoincidence appears very compelling. If this source drives / hasdriven a jet, that would very well be aligned with HH 54. Theprojected distance is 16 ′ (0.8 pc), i.e. jet travel times would beof the order of 10 ( υ/
100 km s − ) − yr. With a total luminosityof ∼ L ⊙ and a mass accretion rate of ˙M acc = × − M ⊙ yr − (Alcal´a et al. 2008), ISO-Cha II 28 would make an excellent can-didate for the exciting source of HH 54, the (distance cor-rected) stellar mass loss rate for which has been determined as ˙M loss ≤ × − M ⊙ yr − (Knee 1992), assuming a flow velocityof ≥
100 km s − . A ratio of loss-to-accretion rate of ≤ / Appendix B: HIFI data reduction
The CO (10 −
9) data published in this paper are public to thescientific community and can be downloaded from the HerschelScience Archive (HSA). The two CO (10 −
9) observations werecarried out in dual-beam-switch raster mode (observation id:1342180798) and on-the-fly with position switch mode (obser-vation id: 1342190901). The H O (1 − ) observation wascarried out in point mode with position switch. The data reduc- From the data for CO outflows compiled by Wu et al. (2004), andcomplementing information in the literature, it is found that L bol ∝ ˙M x loss ,where x = . ˙M loss < − M ⊙ yr − (for υ wind =
100 km s − ) andwhich steepens to x = . L bol = L ⊙ if ˙M loss = × − M ⊙ yr − . http: // herschel.esac.esa.int / Science Archive.shtml
Fig. B.1.
CO (10 −
9) spectra toward HH 54. The black spectra arefrom the HIFI observation 1342180798, while the blue spectraare from the observation 1342190901. The horisontal and verti-cal polarisations have been averaged together (see text).tion and production of the HIFI maps followed the followingsteps:1. The raw data were imported to HIPE from the HSA.2. All data were re-calibrated using v4.2 of the HIFI pipelinefor CO (10 −
9) and v5.0 for H O (1 − ). There is a slighto ff set between the pointings for the horisontal and verti-cal polarisations. For that reason, separate readout positionswere calculated for the two polarisations.3. The level 2 data were exported to Class using the HiClasstool in HIPE.4. A baseline was fitted to and subtracted from each individualspectrum. This step and the subsequent data reduction wascarried out in xs .5. From Figure 1, it is clear that the readout positions forthe two CO (10 −
9) maps are not on a regular grid in RAand Dec. Instead of averaging nearby spectra together, aGaussian weighting procedure was used. The FWHM of thetwo dimensional Gaussian used in the weighting was takento be 9.3 ′′ . This size is larger than the average spacing be-tween individual spectra but on the other hand small enoughnot to change the spatial resolution significantly.6. In order to verify the quality of the CO (10 −
9) observations,spectra taken toward the same positions in the two di ff er-ent observing modes were compared, showing no signifi-cant di ff erences. (see Figure B.1). Also the horisontal andvertical polarisation data were reduced individually for bothCO (10 −
9) and H O (1 − ) showing excellent agree-ment. Appendix C: Radiative transfer analysis
An Accelerated Lambda Iteration (ALI) method (e.g.Rybicki & Hummer 1991) is used to compute the line pro-files for the observed transitions, varying only the density,kinetic temperature and the geometry of the shocked region.ALI is a method, where the coupled problem of radiativetransfer and statistical equilibrium is solved exactly. Data reduction software developed by P.Bergman at the Onsala Space Observatory, Sweden;http: // / rss / oso-en / observations /
15. Bjerkeli et al.: Herschel observations of the Herbig-Haro objects HH 52-54
C.1. AcceleratedLambdaIteration
The ALI code has in recent years been used in several publica-tions (see e.g Justtanont et al. 2005; Wirstr¨om et al. 2010) andwas benchmarked with other codes in a paper by Maercker et al.(2008). In that paper, the ALI technique is described in more de-tail. In the present work, typically 30 shells and 16 angles areused. The statistical equilibrium equations and the equation ofradiative transfer are solved iteratively until the relative levelpopulations between each iteration changes by less than 10 − .In the modelling we set the energy limit to 2000 K. 27 rotationalenergy levels, 26 radiative transitions and 351 collisional tran-sitions are included in the calculations for CO. For o -H O, thecorresponding numbers are 45, 164 and 990.
C.1.1. Collision rates
In recent years, substantial e ff orts have been made to deter-mine cross-sections for H O excitation due to collisions with H .Quantal calculations governing collisions between H O and H are to this date not complete and for o -H O, only collision ratesfor interaction with p -H are available (Dubernet et al. 2009).For this reason we make the assumption that all p -H are inthe lowest energy state when solving the statistical equilibriumequations. It is also assumed that the p -H stays in the groundstate after the collision. The ortho-to-para ratio has been esti-mated for warm H by Neufeld et al. (2006). These authors de-rive ratios significantly lower than 3, viz ∼0.4 - 2 in the HH 54region. The computed spectra from the ALI modelling have alsobeen compared with those, using the collision rates presented byFaure et al. (2007), and with similar results. In the case of CO,we use the collision rates presented by Yang et al. (2010).