colloquium given at nasa/jpl

Shocks and star formation

Lars KristensenSubMillimeter Array Fellow,

Harvard-Smithsonian Center for AstrophysicsSMA

SMIT

HSO

NIA

N ASTROPHYSICAL OBSERV

ATO

RY

ACADEMIA SINICA IA

A

Colliding galaxiesGalactic outflows

Supernova remnants

Planetary nebulae

Young stars

Solar system / ISM

Earth / SunSupersonic flight

Colliding galaxiesGalactic outflows

Supernova remnants

Planetary nebulae

Young stars

Solar system / ISM

Earth / SunSupersonic flight

Shocks are everywhere

What is a shock?

What is a shock?

• “... any pressure-driven disturbance which is time-independent (in a co-moving reference frame) and which effects an irreversible change in the state of the medium.” (Draine 1980)

What is a shock?

• “... any pressure-driven disturbance which is time-independent (in a co-moving reference frame) and which effects an irreversible change in the state of the medium.” (Draine 1980)

• “A hydrodynamical surprise” (Chernoff 1987)

Why should you care?1000 K

100 K

50 K

10 K2.7 K

103 104 105 106 107

Tem

pera

ture

Density (cm-3)Slide credit: E. Bergin

Why should you care?

Diff

use

1000 K

100 K

50 K

10 K2.7 K

103 104 105 106 107

Tem

pera

ture



Diff

use

Molecular clouds

1000 K

100 K

50 K

10 K2.7 K

103 104 105 106 107

Tem

pera

ture



Diff

use

Molecular clouds

1000 K

100 K

50 K

10 K2.7 K

103 104 105 106 107

Tem

pera

ture

Density (cm-3)

Star formation

Slide credit: E. Bergin


Shoc

ksD

iffus

e

Molecular clouds

1000 K

100 K

50 K

10 K2.7 K

103 104 105 106 107

Tem

pera

ture

Density (cm-3)

Star formation

Slide credit: E. Bergin

CO, H2O: dominant coolants and probes

G. J. Herczeg: Warm water in a Class 0 outflow

Fig. 4. The combined Herschel/PACS and Spitzer/IRS continuum-subtracted spectra of IRAS 4B, with bright emission in many H2O (blue marks),CO (red marks), OH (green marks), and atomic or ionized lines (purple). The Spitzer spectrum is multiplied by a factor of 10 so that the lines arestrong enough to be seen on the plot. The inset shows the combined spectrum including the continuum.

Fig. 5. Top: spectra extracted separately from a spaxel centered on thesub-mm continuum (red) and a spaxel offset by 9.′′4 to the S and cen-tered on the outflow position (blue). The outflow position is dominatedby line emission while the central object is dominated by continuumemission, indicating that the lines and continuum are spatially offset.Bottom: the same two spectra after continuum subtraction show that theon-source line emission gets much weaker to short wavelengths.

point sources, one at the outflow position and one at the sub-mm continuum peak, and are subsequently fit with 2D Gaussianprofiles. More complicated spatial distributions would be unre-solved in our maps.

The right panel of Fig. 6 demonstrates that the location ofthe warm H2O coincides with the Spitzer/IRAC 4.5 µm imag-ing. The two H2O lines near 63.4 µm, o-H2O 818−707 and p-H2O 808−717, (E′ = 1070 K) are centered at 5.2 ± 0.2′′ fromthe 63 µm continuum and are spatially extended relative to thecontinuum emission (assumed to be unresolved for simplicity)by FWHM = 6.7±1.0′′. About 70% of the emission is producedat the southern outflow position (Fig. 8). The flux ratios for thetwo H2O 63.4 µm lines are similar at both the on-source andoff-source positions (Fig. 9). In the 108.5 µm spectral map, boththe o-H2O 221−110 (E′ = 194 K) and CO 24–23 (E′ = 1524 K)emission are centered 2.5 ± 0.4′′ south of the 108.5 µm contin-uum emission and are spatially-extended in the outflow directionby 13.6 ± 0.7′′, relative to the extent of the continuum emis-sion. The larger spatial extent and smaller offset in the 108 µmlines both indicate that the outflow component contributes∼40%of the measured line flux. The spatial differences may be inter-preted as differential extinction across the emission region, dis-cussed in the next subsection.

The 54 µm maps are noisy because PACS has poor sensitiv-ity at <60 µm. The o-H2O 532−505 54.507 µm (E′ = 732 K)emission is offset by 5.9 ± 0.4′′ south from the the peak of thesub-mm continuum emission. The CO 49-48 53.9 µm emission(E′ = 6457 K) is offset 2.9 ± 0.4′′ south, between the peak of thesub-mm emission and the bright outflow location. The highly-excited CO emission is produced in a different location than thehighly-excited H2O emission.

The [O I] emission is offset by 3.7 ± 0.3′′ at PA= 168, justwest of the outflow, and is spatially extended by ∼7.′′1 ± 1.0.

3.3. Extinction estimates to the central source and outflowposition

The extinction to different physical structures within theIRAS 4B system depends on how much envelope material is

A84, page 5 of 23

Herczeg et al. (2012)

Part I Water tracing shocks in embedded protostars

Part II Hot water in disks: a shocked wind origin

Part III Low-mass protostars in high-mass clusters

Part IWater tracing shocks in embedded protostars

Low-mass YSO evolution

the ‘‘average’’ HCO+ abundance in envelopes around Class 0sources found by Jørgensen et al. (2004). It is possible that inthese two sources theHCO+ abundancemay be slightly enhanced,but our data are not sufficient to firmly make this conclusion. Inthe other sources in our sample, the HCO+ is similar to, or below,the standard abundance.

The HCO+ emission seen in Figure 3 arises mainly along theoutflow cavity walls or the outflow lobes, and the observed over-abundance is consistent with chemical models that predict HCO+

enhancement in outflow walls due to the shock-triggered re-lease of molecules from dust mantles and subsequent chemicalreactions in the warm gas (Rawlings et al. 2000, 2004; Viti et al.2002). Hence, it is clear that some of the sources in our samplepower outflows that are affecting the chemical composition of theirsurroundings. On the other hand, HCO+ outflows with abundancesat or below the standard value must arise in dense (nk103) en-velope gas (see Evans 1999) that has been entrained by the out-flow, where shocks have not seriously affected the chemicalproperties of the environment. In some sources the HCO+ outflowemission does not trace the entire extent of the outflow wall (e.g.,HH 300 and L1228) or does not trace both outflow lobes (e.g., HH114mms and RNO 43). This could result from differences in theenvironmental conditions, such as density, shock-induced radia-tion, or chemical composition, or in the effects of optical depth.

6. SUMMARY

Our systematic high angular resolution (300Y700), multimolecularline survey of the circumstellar environment within!104 AU ofnine protostars at different evolutionary stages has enabled detailedstudies of the kinematics, morphology, density distribution, andchemistry of the circumstellar gas. These in turn have permittedinvestigations of how the circumstellar envelopes and proto-stellar outflows change with time.

Our 12CO images trace high-velocity outflowing molecularmaterial around all sources. We detect a clear trend in the mor-phology of the protostellar molecular outflows. The youngestsources (Class 0) in our sample power molecular outflows with

jetlike morphologies or cone-shaped lobes with opening anglesless than 55". The sightly more evolved (Class I) sources in oursample have molecular outflows with lobe opening angles ofmore than 75". Outflows from the most evolved young stars inour sample, the Class II sources, have even wider lobes or nodefinite shape or structure. Combining our data with that from anumber of sources in the literature shows that the outflow open-ing angle close to the source widens with time.The denser, lower velocity gas probed by our 13CO and HCO+

observations appears to arise from the outflow cavitywalls or veryclose to the protostar. We suggest that this is dense gas from theouter regions of the circumstellar envelope that has been en-trained by the high-velocity flow, thereby eroding the envelopeand helping to widen the outflow cavities. We also suggest thatevolutionary changes in the morphology and velocity field ofthe dense circumstellar envelopes, detected in our C18O images,are mainly caused by outflow-envelope interactions. Class 0 sourceenvelopes, for example, are elongated with velocity gradientsalong the outflow axis, indicating that the young and powerfulprotostellar outflows can entrain dense envelope gas. Class Ienvelopes, by contrast, are elongated more or less perpendicularto the outflow axis and concentrated outside the outflow lobes,as if most of the dense gas along the outflow axis has alreadybeen displaced. By the Class II stage the sparse C18O emissionobserved is constrained to the outflow lobe edges, consistentwith most, if not all, of the dense gas being cleared.Not unexpectedly, our results show a decrease in envelope

mass as the age of the protostar increases. Comparisons of theestimated envelope mass-loss rate with the dense gas outflowrate imply that the protostellar outflow plays an important rolein envelope mass loss during and after the Class I stage. Foryounger protostars, envelope mass loss is more likely to be domi-nated by large infall rates. Finally, enhanced abundance of HCO+ inthe outflow in some of the sources in our sample indicates thatthe chemical composition of the environment around these pro-tostars is affected by shock-induced chemical processes in theoutflow-envelope interface.

Fig. 8.—Schematic picture of outflow-envelope interaction evolution. Dark gray regions denote high-density (envelope) gas, mostly traced by C18O in ourobservations. Light gray regions show the molecular outflow traced by the 12CO and 13CO. Arrows indicate the gas motion. See x 5.4 for details.

ARCE & SARGENT1082 Vol. 646Class 0 Class IIClass I

Jet / wind present at all evolutionary stages

Arce & Sargent (2006)

Physical processes

101

103

102

104

T (K)

R. Visser

Outflow (entrained)

Passively heated envelope

DiskUV-heating of cavity walls

Shell shocks along cavity walls

Jet w. internal working surfaces

Shocks dominate heating processes

Water traces shocks

“Bullet” “Bullet”

Broad outflow

Envelope emission + absorption = inverse P-Cygni

Offset comp.

BHR7115 L⊙2.7 M⊙• One profile: lots of

H2O moving!

• Bulk of emission in three components

• Velocity resolution allows for decomposition

Herschel-HIFIKristensen et al. (2012)

H2O always trace shocksClass 0 Class I

Kristensen et al. (2012)

H2O always trace shocks

• Profile richness in Class 0/I sources: disentangling dynamics

• When, how, where is the gas moving along the line of sight?

Class 0 Class I


H2O always trace shocks

• Profile richness in Class 0/I sources: disentangling dynamics

• When, how, where is the gas moving along the line of sight?

• The devil is in the detail profile

Class 0 Class I


~1-10 ~102-103 ~104-105 L⊙

Time

Mass

D =

1 k

pc

D = 200 pc

100 km s-1

x275 x25

x250

x140

H2O profile evolutionH2O 110-101

557 GHz

Time more important than mass in evolution

Caselli et al., Hogerheijde et al., Kristensen et al., McCoey et al., van der Tak et al.

Evolutionary scheme

• Class 0: H2O tightly linked to outflow, infall, molecular jet

• Class I: envelope opens, outflow force decreases, expansion

(Kristensen et al. 2012)

Part IIHot water in disks:

a wind origin?

Hot H2O detected with Spitzer

• A true surprise

• Detected toward Class II disk sources

• Interpretation: inner hot disk, r < 1 AU

Organic Molecules and Waterin the Planet Formation Regionof Young Circumstellar DisksJohn S. Carr1 and Joan R. Najita2

The chemical composition of protoplanetary disks is expected to hold clues to the physical andchemical processes that influence the formation of planetary systems. However, characterizingthe gas composition in the planet formation region of disks has been a challenge to date. Wereport here that the protoplanetary disk within 3 astronomical units of AA Tauri possesses a richmolecular emission spectrum in the mid-infrared, indicating a high abundance of simple organicmolecules (HCN, C2H2, and CO2), water vapor, and OH. These results suggest that water isabundant throughout the inner disk and that the disk supports an active organic chemistry.

Disks of gas and dust orbit very youngstars and represent the early stages ofplanetary system formation. Studies of

these disks can establish the conditions underwhich planets form, complementing studies ofsolar system bodies in efforts to reconstruct theorigin of our Solar System. In the same way thatthe chemical record preserved in primitive bodies(comets and meteorites) is probed for clues tothe physical and chemical processes that oper-ated in the early solar nebula (1, 2), the chemicalcomposition of protoplanetary disks is expectedto hold clues to the processes that are active indisks. Although significant advances have beenmade from the study of the dust properties ofdisks (3–5), much less progress has been madein the study of the gaseous component, partic-ularly over the range of radii where planet for-mation is thought to occur (6–8). Measurementsof disk molecular abundances can provide in-sights on issues such as the origin and evolutionof water in the Solar System, the nature of or-ganic chemistry in disks, and the ability ofdisks to synthesize pre-biotic species. Here wereport that a protoplanetary disk around a youngstar displays a rich molecular emission spectrumin the mid-infrared that provides detailedthermal and chemical information on gas in theplanet formation region of the disk.

We observed AA Tauri, a fairly typical clas-sical T Tauri star (CTTS, which are approximatelysolar-mass stars with accretion disks and ages ofless than a few 106 years), with the Infrared Spec-trograph (IRS) on the Spitzer Space Telescope(9, 10), covering wavelengths of 9.9 to 37.2 mmat a spectral resolution (l/Dl) of 600. The highsignal-to-noise ratio (~200 to 300) produced byour observational and data reduction techniques(11) reveals a rich spectrum of molecular emis-sion lines that is dominated by rotational tran-

sitions of H2O (Fig. 1). Rotational transitions ofOH were also observed between 20 and 31 mm.From 13 to 15 mm, ro-vibrational emission bandsof the fundamental bendingmodes of C2H2, HCN,and CO2 are prominent (Fig. 2).

This mid-infrared molecular emission almostcertainly has its origin in a disk. AATauri lacksa surrounding molecular envelope or molecularoutflow as potential alternate sites for the emittinggas. A disk interpretation is also consistent withhigh-spectral-resolution studies of CTTSs, whichdemonstrate a disk origin for their near-infraredmolecular emission features. These include emis-sion from the 5-mm fundamental ro-vibrationalbands of CO, which is very common in CTTSs(12), and emission from the 2.3-mm CO overtonebands and hot bands of H2O and OH (6, 13, 14),which is less frequently observed. All these near-infrared emission bands are thought to arise in atemperature inversion in the atmosphere of theinner [<1 astronomical unit (AU)] disk (6, 12–14).

To determine the properties of the emittinggas, we modeled the emission spectrum and an-alyzed the measured line fluxes using standardtechniques (11). The emission from each molec-

ular species was modeled independently as ther-mally excited emission from a slab with a giventemperature, column density, and emitting area.Although these quantities are expected to varywith radius in a real disk, we cannot constrain theirradial variation without the benefit of velocity-resolved line profiles. Our strategy allows us todetermine the flux-weighted mean gas temper-ature (T ), column density (N), and emitting area(p × R2) for each molecule (Table 1). The com-bined synthetic spectrum reproduces well thedetails of the observed spectrum (Fig. 2). Mo-lecular abundances with respect to CO (Table 1)were then determined by modeling the 5-mmfundamental CO emission spectrum of AATauri,obtained with the near-infrared echelle spectro-graph NIRSPEC (15) at the Keck Observatory.

The equivalent radii for the emitting areas(Table 1) indicate that the observed gas resides atdisk radii that correspond to the region of ter-restrial planets in the Solar System; that is, within2 to 3 AU of the star. The derived temperatures forthe emission are also roughly consistent withpredicted gas temperatures at radii ~1 AU formodels that calculate the gas thermal structure inthe temperature inversion region of disk atmo-spheres (16, 17). The significantly lower temperaturefor CO2 suggests that its emission is concentratedat larger radii, which is also consistent with pre-dictions of chemical models (18) that CO2 is de-stroyed at temperatures much higher than 300 K.The higher mean temperature for CO can be un-derstood by the fact that CO fundamental emis-sion is insensitive to gas with T < 400 K, becauseof the higher energy of the upper levels of thetransitions and the shorter wavelength of emission.Vertical gradients in the gas temperature andmolecular abundances could also contribute tosome of the differences; for example, H2Omay beabundant in a cooler part of the atmosphere ascompared to CO (16).

The abundances of simple organics and H2Ofor AATauri (Fig. 3) are about one order of mag-nitude higher than both observations and models(19) of hot molecular cores, the dense and warm

REPORTS

1Remote Sensing Division, Naval Research Laboratory, Code7210, Washington, DC 20375, USA. 2National Optical Astron-omy Observatory, Tucson, AZ 85719, USA.

Fig. 1. Spitzer IRS spec-trum of AA Tauri from 9.8to 38.0 mm. The spectrumcombines all orders ofthe short-high and long-high IRS modules (11).Rotational transitions ofOH are marked with a dia-mond, and the Q branchesof C2H2, HCN, and CO2are labeled along withatomic [Ne II]. The ma-jority of emission featuresare due to rotational tran-sitions of H2O (unmarkedfeatures).

14 MARCH 2008 VOL 319 SCIENCE www.sciencemag.org1504Carr & Najita 2008

0

500

1,000

1,500

2,000

2,500

3,000

3,500

-12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12J

E/k

(K)

ortho -H2O para -H2O

102,9-93,6, 321 GHz

81,8-70,7 63 μm

Hot H2O

• My definition: Anything with Eup > 1000 K

• Typically observed with Spitzer, Herschel-PACS, ground-based NIR/MIR

Watson et al. 2007

HIFI landPACS land

Spitzerland

• Observed the 63 mu line, spectral resolution ~ 100 km/s

• Detected in 24% of gas-rich disks

• Emission, when detected, correlates with [OI] also at 63 mu

• No spatial extent seen: disk origin?

PACS hot water in disksP. Riviere-Marichalar et al.: Detection of warm water vapour in Taurus protoplanetary discs by Herschel

Table 2. Probabilities for correlations between o-H2O line intensity andstellar/disc parameters.

Observable Points n Spearman’s prob. Kendall’s prob.L[OI] 68 0.0009 (0.4090) 0.0000 (0.196)63 µm flux 64 0.0147 (0.2635) 0.0002 (0.4567)850 µm flux(2) 57 0.0145 (0.2596) 0.0131 (0.8496)Lstar

(3) 65 0.1789 (0.1130) 0.1980 (0.5535)LX

(3) 65 0.0225 (0.9912) 0.0087 (0.6012)α(2−8µm) (4) 62 0.026 (0.1547) 0.0023 (0.3850)L[OI]6300 Å 27 0.1291 (0.3255) 0.1008 (0.3450)

Notes. The values obtained for random populations are shown in brack-ets. Accurate only if N > 30. (2) 850 µm continuum fluxes fromAndrews & Williams (2005). (3) Values from Güdel et al. (2007).(4) SED slope from Luhman et al. (2010).

Fig. 2. Plot of the 63.32 µm o-H2O line luminosity versus[OI] 63.18 µm. Filled dots are detections, arrows are upper limits. Solid,dark grey arrows represent objects with non-detections spanning thesame spectral range as the objects with detections. Light grey, emptyarrows represent non-detections with other spectral types.

other abundant species emit strongly at or close to the observedwavelength of the feature. This water feature was observed byHerczeg et al. (2012) in the outflow of NGC 1333 IRAS 4B.The o-H2O emission is only present in spectra with [O i] detec-tions. The targets FS Tau, HL Tau, and T Tau display extendedemission in the [O i] 63.18 µm line, but only T Tau show hintsof extended emission in the 63.32 µm o-H2O line. T Tau is anexceptional object. It is a triple star system that drives at leasttwo jets. It was the most line rich PMS star observed by ISO(Lorenzetti 2005) and resembles more a hot core than a PMSstar, since the continuum emission is quite extended (∼3.5′′ at70 µm). It is impossible to tell how much of the line emissioncomes from the discs, from the outflows, or even from the sur-rounding envelope.

To help us understand the origin of the o-H2O emission, wecompared the line intensity with several star and disc (jet) pa-rameters. We computed survival analysis ranked statistics usingthe ASURV code (Feigelson & Nelson 1985; Isobe et al. 1986).The result of this analysis is summarised in Table 2. We alsocreated random populations to test the validity of the results.

The survival analysis shows a correlation between the o-H2Oline fluxes and the [O i] line fluxes at a significance level of 0.99(see Fig. 2). This relationship suggests that both lines have a sim-ilar origin. The H2O emission is correlated with the continuumemission at 63 µm, but at a significance level of 0.95 in theSpearman statistics (Fig. 3). The 850 µm continuum flux canbe used as a proxy for the amount of dust present in the disc. Asurvival analysis shows a possible correlation with the 850 µmcontinuum flux, at a significance level of 0.95 in both Spearmanand Kendall statistics, although a large scatter is present. A cor-relation with neither the stellar luminosity nor the spectral type

Fig. 3. Plot of the 63.32 µm ortho-H2O line flux versus 63 µm contin-uum flux. Symbols as in Fig. 2.

Fig. 4. H2O luminosity versus X-ray luminosity. Symbols as in Fig. 2.

is found. There seems to be a weak correlation with the slopeof the SED measured between 2 µm and 8 µm, used as a proxyfor the presence of hot dust. The significance of this correla-tion is dominated by T Tau, which has the highest o-H2O fluxin the sample. There is likely no link between mass loss rateand LH2O. The [O i] luminosity at 6300 A is proportional to themass loss rate (Hartigan et al. 1995). Although the sample is toosmall to test this relationship conclusively, we note that whileL[OI]6300 A spans four orders of magnitude, the H2O luminosityspans only one order of magnitude. Finally, the survival anal-ysis statistics points to a possible relationship with the X-rayluminosity Lx (Fig. 4). While the Spearman probability for thereal sample is only one order of magnitude smaller than for therandom sample, the Kendall probability is two orders of magni-tude smaller. Inspection of Fig. 4 also shows that the o-H2O lineflux is detected only for sources with X-ray luminosities higherthan 1030 erg s−1, which is consistent with photochemical discmodels that show that far-IR line fluxes increase significantlyabove this X-ray luminosity threshold (Aresu et al. 2011). Wenote that log Lx = 1030 erg s−1 is above the median (mean) X-rayluminosity in Taurus (log Lx = 29.8 (29.75) erg s−1 respectively,Güdel et al. 2007). Interestingly, more than half of the sourceswith Lx > 1030 erg s−1 do not display the o-H2O line. This be-haviour may stem from either (1) the different shape of the X-rayspectrum (hardness ratio), (2) the duty cycle of the flares respon-sible for the high levels of X-ray fluxes, and/or (3) that X-rays arenot the only driver of H2O chemistry and excitation, let alone anyinner disc geometry and radiative transfer considerations. Wecaution that the correlation with X-rays is significantly weakerwhen TTau is removed from the analysis.

All the stars detected in o-H2O are outflow/jet sources, al-though three of them AA Tau, DL Tau, and RY Tau do notshow any excess in [OI], i.e. all the emission is consistent withcoming from the disc (Howard et al., in prep.). Two of these

L3, page 3 of 5

A&A 538, L3 (2012)

Fig. 1. Spectra for the objects with a 63.32 µm feature detection (>3σ). The red lines indicate the rest wavelength of the [O i] and o-H2O emission.

Photodetector Array Camera and Spectrometer (Poglitsch et al.2010).

In this Letter, we report the first detection of the o-H2O lineat 63.32 µm in a subsample of protoplanetary discs around TTauri stars in the 1–3 Myr old Taurus star forming region.

2. Observations and data reduction

This study is based on a sample of 68 classical and weak-lineT Tauri stars from the Taurus star forming region with spectralmeasurements from Herschel-PACS centred at the wavelengthof the [O i] 3P1 →3P2 63.184 µm line. The Taurus star-formingregion is one of the main targets for the study of protoplane-tary systems, because it is among the nearest star-forming re-gions (d = 140 pc) with a well-known population of more than300 young stars and brown dwarfs according to Kenyon et al.(2008), Luhman et al. (2010), and Rebull et al. (2010).

The observations described in this letter are part of theHerschel open time key programme GASPS [P.I. W. Dent],(see Mathews et al. 2010), a flux-limited survey devoted to thestudy of the gas and dust in circumstellar systems around youngstars. The survey focuses on the detection of the [O i] emissionat the 63.18 µm feature. For this study, we analyzed 68 starswith spectral types ranging from late F-early G to mid M. ThePACS spectral observations were made in chop/nod pointed linemode. The observing times ranged from 1215 to 6628 s, depend-ing on the number of nod cycles. The data were reduced usingHIPE 7.0.1751. A modified version of the PACS pipeline wasused, which included: saturated and bad pixel removal, chopsubtraction, relative spectral response-function correction, andflat fielding. Many observations suffer from systematic pointingerrors, in some cases as large as 8′′, and are always shifted to theEast. This is due to a plate scale error in the star tracker, whichis normally negligible except in areas where the tracked starsare asymmetrically distributed within the field, as in Taurus. Themis-pointing translates into systematic small shifts in the line

Table 1. Line positions and fluxes from PACS spectra.

Name Sp. type λH2O − λ[OI] [O i] flux o−H2O flux– – µm 10−17 W/ m2 10−17 W/m2

AA Tau K7 0.140 2.2 ± 0.13 0.80 ± 0.13DL Tau K7 0.141 2.4 ± 0.15 0.65 ± 0.14FS Tau M0 0.139 37 ± 0.26 2.00 ± 0.33RY Tau K1 0.139 10 ± 0.42 1.95 ± 0.38T Tau K0 0.138 830 ± 0.75 27.8 ± 0.7XZ Tau M2 0.139 32.2 ± 0.48 2.11 ± 0.48HL Tau K7 0.142 54.3 ± 0.71 8.14 ± 0.80UY Aur M0 0.139 33.6 ± 0.37 1.90 ± 0.32

Notes. All spectral types from the compilation of Luhman et al. (2010).

centre position. When the star was well centred within a singlespaxel, we extracted the flux from that spaxel and applied theproper aperture correction. When the flux was spread over morethan one spaxel, we co-added the spaxels.

3. Results and discussion

Among the sample of 68 Taurus targets studied in this let-ter, 33 have discs that are rich in gas. These 33 all show the[O i] 3P1 →3 P2 line in emission at 63.18 µm (signal-to-noiseratio >3, with values ranging from 3 to 375). In 8 of these 33 tar-gets (∼24%), an additional fainter emission-line at 63.32 µmis detected (Fig. 1). We computed 63.32 µm line fluxes by fit-ting a Gaussian plus continuum curve to the spectrum usingDIPSO1. To improve the line fitting, the noisier edges of thespectral range were removed (i.e., λ < 63.0 and λ > 63.4).The results are listed in Table 1, where we report the peak po-sition of the feature with respect to the observed wavelengthof the [O i] 63.18 µm line. According to these fits, the peak ofthe feature is at λ0 = 63.32 µm. The FWHM is 0.020 µm, i.e.,the instrumental FWHM for an unresolved line. We identify thefeature as the ortho-H2O 818 → 707 transition at 63.324 µm(EUpper Level = 1070.7 K, Einstein A = 1.751 s−1) since no

1 http://star-www.rl.ac.uk/docs/sun50.htx/sun50.html

L3, page 2 of 5

Riviere-Marichalar et al. 2012

Origin of hot PACS H2O

• Following Spitzer hot H2O interpretation: inner warm disk, T ~ 600 K

• No need for shocks/winds?

P. Riviere-Marichalar et al.: Detection of warm water vapour in Taurus protoplanetary discs by Herschel

Appendix A: Non detections

GASPS is a flux-limited survey. All sources have detection lim-its for the o-H2O line at 63.32 µm around 10−17 W/m2, with anuncertainty of a factor of two both ways. The list of non detec-tions is given in Table A.1.

Table A.1. Source list, spectral types, and 63.32 µm o-H2O line fluxes.

Name SP. Type o − H2O flux– – 10−17 W/m2

Anon 1 M0 <1.17BP Tau K7 <0.94CIDA2 M5.5 <0.89CI Tau K7 <0.92CoKu Tau/4 K3 <1.37CW Tau K3 <1.26CX Tau M2.5 <0.94CY Tau M1.5 <1.41DE Tau M1 <1.23DF Tau M2 <1.28DG Tau K6? <1.32DG Tau B <K6 <1.25DH Tau M1 <1.07DK Tau K6 <0.53DL Tau K7 <1.14DM Tau M1 <0.89DN Tau M0 <0.77DO Tau M0 <1.21DP Tau M0.5 <0.92DQ Tau M0 <0.98DS Tau K5 <0.58FF Tau K7 <1.11FM Tau M0 <1.24FO Tau M3.5 <0.96FQ Tau M3 <0.91FT Tau – <1.25FW Tau M5.5 <1.08FX Tau M1 <1.82GG Tau M5.5 <1.24GH Tau M2 <0.84GIK Tau K7 <0.74GM Aur K7 <1.21GO Tau M0 <1.23Haro 6-13 M0 <1.43Haro6-37 M1 <1.84HBC 347 K11 <1.69HBC 356 K21 <0.94HBC 358 M3.5 <1.29HK Tau M0.5 <0.90HN Tau K5 <0.64HO Tau M0.5 <1.46HV Tau M1 <0.72IP Tau M0 <0.83IQ Tau M0.5 <1.3804158+2805 M5.25 <1.06LkCa 1 M4 <1.06LkCa 3 M1 <0.94LkCa 4 K7 <0.98LkCa 5 M2 <1.16LkCa 7 M0 <0.88LkCa 15 K5 <0.52SAO 76428 F8? <1.58SU Aur G2 <0.72UX Tau K5 <0.99UZ Tau M2 <0.60V710 Tau M0.5 <1.17V773 Tau K3 <0.77V819 Tau K7 <1.21V927 Tau M4.75 <0.88VY Tau M0 <1.30

Notes. Spectral types from the compilation by Luhman et al. (2010).(1) from Herbig & Bell (1988).

Appendix B: Location of emission for H2O lines

0.1 1.0 10.0 100.0r [AU]

0.0

0.1

0.2

0.3

0.4

z/r

AV

=1

AV=1

AV=1

AV

=10

AV=10

15.738µm

63.324µm89.988µm

269.27µm

0 2 4 6 8 10log n

H2O [cm-3]

Fig. B.1. Location of the H2O emission for a typical disc model (M∗ =0.8 M⊙, L∗ = 0.7 L⊙, Teff = 4400 K, UV excess fUV = 0.01,X-ray luminosity LX = 1030 erg s−1, Rin = 0.1 AU, Rout = 300 AU,Mdisc = 0.01 M⊙, dust/gas = 0.01, ϵ = −1, scale height H0 = 1 AU atr0 = 10 AU, flaring β = 1.1, astronomical silicate with uniform dustsize distribution amin = 0.05 µm, amax = 1 mm, and index = 3.5). The15.738 µm line is used as a typical H2O line detected by Spitzer. Wehave marked the radii where the cumulative line flux reaches 15% and85% with vertical lines, for the four selected H2O lines. In addition, thehorizontal lines mark the heights above the midplane where 15% and85% of the line flux, from every vertical column, originate in. The en-circled regions are hence responsible for 70%× 70% = 49% of the totalline flux (for a pole-on disc).

L3, page 5 of 5

Riviere-Marichalar et al. 2012, ProDiMo prediction

Temperature

Wavelength

SpitzerHot PACS

Warm PACS

HIFI

SMA observations

• Test interpretation: warm inner disk, or outflow/wind?

• Targets: brightest 63 mu emitters

• Observed on Dec 18, 2013 compact configuration (2” beam ~ 300 AU) @ 321 GHz

P. Riviere-Marichalar et al.: Detection of warm water vapour in Taurus protoplanetary discs by Herschel

Table 2. Probabilities for correlations between o-H2O line intensity andstellar/disc parameters.

Observable Points n Spearman’s prob. Kendall’s prob.L[OI] 68 0.0009 (0.4090) 0.0000 (0.196)63 µm flux 64 0.0147 (0.2635) 0.0002 (0.4567)850 µm flux(2) 57 0.0145 (0.2596) 0.0131 (0.8496)Lstar

(3) 65 0.1789 (0.1130) 0.1980 (0.5535)LX

(3) 65 0.0225 (0.9912) 0.0087 (0.6012)α(2−8µm) (4) 62 0.026 (0.1547) 0.0023 (0.3850)L[OI]6300 Å 27 0.1291 (0.3255) 0.1008 (0.3450)

Notes. The values obtained for random populations are shown in brack-ets. Accurate only if N > 30. (2) 850 µm continuum fluxes fromAndrews & Williams (2005). (3) Values from Güdel et al. (2007).(4) SED slope from Luhman et al. (2010).

Fig. 2. Plot of the 63.32 µm o-H2O line luminosity versus[OI] 63.18 µm. Filled dots are detections, arrows are upper limits. Solid,dark grey arrows represent objects with non-detections spanning thesame spectral range as the objects with detections. Light grey, emptyarrows represent non-detections with other spectral types.

other abundant species emit strongly at or close to the observedwavelength of the feature. This water feature was observed byHerczeg et al. (2012) in the outflow of NGC 1333 IRAS 4B.The o-H2O emission is only present in spectra with [O i] detec-tions. The targets FS Tau, HL Tau, and T Tau display extendedemission in the [O i] 63.18 µm line, but only T Tau show hintsof extended emission in the 63.32 µm o-H2O line. T Tau is anexceptional object. It is a triple star system that drives at leasttwo jets. It was the most line rich PMS star observed by ISO(Lorenzetti 2005) and resembles more a hot core than a PMSstar, since the continuum emission is quite extended (∼3.5′′ at70 µm). It is impossible to tell how much of the line emissioncomes from the discs, from the outflows, or even from the sur-rounding envelope.

To help us understand the origin of the o-H2O emission, wecompared the line intensity with several star and disc (jet) pa-rameters. We computed survival analysis ranked statistics usingthe ASURV code (Feigelson & Nelson 1985; Isobe et al. 1986).The result of this analysis is summarised in Table 2. We alsocreated random populations to test the validity of the results.

The survival analysis shows a correlation between the o-H2Oline fluxes and the [O i] line fluxes at a significance level of 0.99(see Fig. 2). This relationship suggests that both lines have a sim-ilar origin. The H2O emission is correlated with the continuumemission at 63 µm, but at a significance level of 0.95 in theSpearman statistics (Fig. 3). The 850 µm continuum flux canbe used as a proxy for the amount of dust present in the disc. Asurvival analysis shows a possible correlation with the 850 µmcontinuum flux, at a significance level of 0.95 in both Spearmanand Kendall statistics, although a large scatter is present. A cor-relation with neither the stellar luminosity nor the spectral type

Fig. 3. Plot of the 63.32 µm ortho-H2O line flux versus 63 µm contin-uum flux. Symbols as in Fig. 2.

Fig. 4. H2O luminosity versus X-ray luminosity. Symbols as in Fig. 2.

is found. There seems to be a weak correlation with the slopeof the SED measured between 2 µm and 8 µm, used as a proxyfor the presence of hot dust. The significance of this correla-tion is dominated by T Tau, which has the highest o-H2O fluxin the sample. There is likely no link between mass loss rateand LH2O. The [O i] luminosity at 6300 A is proportional to themass loss rate (Hartigan et al. 1995). Although the sample is toosmall to test this relationship conclusively, we note that whileL[OI]6300 A spans four orders of magnitude, the H2O luminosityspans only one order of magnitude. Finally, the survival anal-ysis statistics points to a possible relationship with the X-rayluminosity Lx (Fig. 4). While the Spearman probability for thereal sample is only one order of magnitude smaller than for therandom sample, the Kendall probability is two orders of magni-tude smaller. Inspection of Fig. 4 also shows that the o-H2O lineflux is detected only for sources with X-ray luminosities higherthan 1030 erg s−1, which is consistent with photochemical discmodels that show that far-IR line fluxes increase significantlyabove this X-ray luminosity threshold (Aresu et al. 2011). Wenote that log Lx = 1030 erg s−1 is above the median (mean) X-rayluminosity in Taurus (log Lx = 29.8 (29.75) erg s−1 respectively,Güdel et al. 2007). Interestingly, more than half of the sourceswith Lx > 1030 erg s−1 do not display the o-H2O line. This be-haviour may stem from either (1) the different shape of the X-rayspectrum (hardness ratio), (2) the duty cycle of the flares respon-sible for the high levels of X-ray fluxes, and/or (3) that X-rays arenot the only driver of H2O chemistry and excitation, let alone anyinner disc geometry and radiative transfer considerations. Wecaution that the correlation with X-rays is significantly weakerwhen TTau is removed from the analysis.

All the stars detected in o-H2O are outflow/jet sources, al-though three of them AA Tau, DL Tau, and RY Tau do notshow any excess in [OI], i.e. all the emission is consistent withcoming from the disc (Howard et al., in prep.). Two of these

L3, page 3 of 5

Riviere-Marichalar et al. 2012

HL Tau: recent headliner

• Quiescent disk, main accretion phase done?

• Or what???

SMA data• H2O detected at 5σ

(single beam), 8σ integrated, toward HL Tau

• Matches disk position, tentative (3-4σ) extended emission

• Profile: blue-shifted (-15 km/s) and broad (20 km/s)!

Kristensen et al. in prep.40 20 0 20 40

Velocity (km s1)

0.05

0.00

0.05

0.10

0.15

0.20

0.25

Flux

dens

ity(J

y)

13COv = 1–0

H2O102,9–93,6

OutflowDisk

Vibrational CO: wind

• 4.7 mu rovibrational emission from CO supports wind hypothesis

• Identical line profiles, within uncertainty

• CO emission unresolved at 0.2” resolution (30 AU)

Herczeg et al. 201140 20 0 20 40

Velocity (km s1)

0.05

0.00

0.05

0.10

0.15

0.20

0.25

Flux

dens

ity(J

y)

13COv = 1–0

H2O102,9–93,6

Excitation conditions• Use 321 GHz and 63 mu

lines to estimate excitation conditions, n, T, N(H2O)

• Highly degenerate

• N(H2O) ~ 1019 cm-2, n(H2) ~ 109-1012 cm-3, T ~ 500-1500 K, r(H2O) ~ 2 AU

5 6 7 8 9 10 11 12log n (cm3)

16

17

18

19

20

log

N(o

-H2O

)(cm

2 ) Masing

RM12

log(321 GHz / 63µm)

-3

-3

-2

-2

-1

0

1

5 6 7 8 9 10 11 12log n (cm3)

0

2

4

6

8

10

Emitt

ing

radi

us(A

U)

Same emitting region

Different emitting region

Molecular protostellar winds

• Model calculations show molecules survive launching in MHD winds

• Physical conditions in wind close to what is inferred from radiative transfer

A&A 538, A2 (2012)

-50 0 500

50

100

150

cylindrical radius (AU)r0

hei

ght

(AU

)z

Fig. 1. Overall geometry of the “slow” MHD disk wind solution of(Casse & Ferreira 2000) used in this article. Solid white curves showvarious magnetic flow surfaces, with that anchored at 1 AU shown indashed. The density for Macc = 10−6 M⊙/yr and M⋆ = 0.5 M⊙, is codedin the contour plot starting at 8 × 104 cm−3 and increasing by fac-tors of 2. The bottom dotted line traces the slow magnetosonic surface(at 1.7 disk scale heights) where our chemical integration starts.

Eq. (9) of Garcia et al. (2001a). In particular, the midplane fieldin the adopted MHD solution is,

Bz ≃ 0.7 G! r0

1 AU

"−5/4#

Macc

10−6 M⊙ yr−1

$1/2 #M⋆

0.5 M⊙

$1/4

, (3)

which follows the same scaling as the available magnetic fieldmeasurements in protostellar disks (Shu et al. 2007). The re-quired field strengths thus appear plausible. In the following, wewill vary M⋆, Macc, and r0 in order to explore their effect on themolecular content and thermal state of the disk wind.

2.2. Thermo-chemical evolution

The thermo-chemical evolution along wind streamlines wasimplemented by adapting the 2003 version (Flower &Pineau des Forêts 2003) of a code constructed to calculate thesteady-state structure of planar molecular MHD multi-fluidshocks in interstellar clouds (Flower et al. 1985). The ion-neutraldrift, FUV field, and dust-attenuation were calculated followingthe methods developed by Garcia et al. (2001a) in the atomicdisk wind case. Irradiation by stellar X-rays was also added, fol-lowing the approach of Shang et al. (2002).

Given the high densities and low ion-neutral drift speedsin the wind (see Sect. 3) the same temperature and velocity isadopted here for all particles. The latter is prescribed by theunderlying single-fluid MHD wind solution. We thus integratenumerically the following differential equations on mass den-sity ρ, species number density n(a), and temperature T along thestreamline, as a function of altitude z above the disk midplane:dρ f

dz=

S f − ρ f∇ · uvz

, (4)

dn(a)dz=

Ra − n(a)∇ · uvz

, (5)

dTdz=Γ − Λ − ntotkBT∇ · u − 3

2 kBTR32 kBntotvz

· (6)

Here, vz and (∇ · u) are the bulk vertical flow speed and the(3D) flow divergence interpolated from the MHD wind solu-tion, ntot is the total number density of particles, T is the tem-perature, kB is the Boltzmann constant, S and R are the rates ofchange in mass and number of particles, respectively, per unitvolume, and Γ and Λ are the heating and cooling rates per unitvolume. The equations on n(a) apply to each species a as wellas to the individual populations of the first 49 levels of H2 (up toan energy of 20 000 K) which are integrated in parallel with theother variables. The equations on ρ f apply to each “fluid” f (neu-tral, positive, negative) and are used mainly for internal check-ing purposes, as the total mass density ρ(z) is prescribed by theMHD solution.

Cooling and heating mechanisms include

– radiative cooling by H2 lines excited by collisions with H,H2, He, and electrons (Le Bourlot et al. 1999);

– radiative cooling by CO, H2O, and 13CO in the LargeVelocity Gradient approximation (Neufeld & Kaufman1993), and by OH and NH3 in the low-density limit (Floweret al. 1985);

– atomic cooling by fine-structure and metastable lines of C,N, O, S, Si, C+, N+, O+, S+, Si+ (Flower et al. 2003) and Fe+(Giannini et al. 2004);

– inelastic scattering of electrons on H and H2 (Aggarwal et al.1991; Hummer 1963; Rapp & Englander-Golden 1965);

– energy released by collisional ionization and dissociationand exo/endo-thermicity of chemical reactions (Flower et al.1985);

– energy heat/loss through thermalization with grains (Tielens& Hollenbach 1985);

– ambipolar-diffusion heating by elastic scattering between theneutral fluid and charged ions and grains (Garcia et al. 2001a,see Sect. 2.3);

– ohmic heating arising from the drift between electrons andother fluids (Garcia et al. 2001a, see Sect. 2.3);

– photoelectric effect on grains (Bakes & Tielens 1994,Eq. (42)) irradiated by the (attenuated) FUV field of hot ac-cretion spots (see Sect. 2.4);

– heating through cosmic-rays and (attenuated) coronal stellarX-rays (Dalgarno et al. (1999), see Sect. 2.5).

The reader is referred to the corresponding references for a dis-cussion of the physical context and hypotheses involved in mod-elling each of the above processes.

2.2.1. Chemical network

The chemical network consists of 134 species, including atomsand molecules (either neutral or singly ionized) as well as

A2, page 4 of 21

Panoglou et al. 2012

D. Panoglou et al.: Molecules in protostellar MHD disk winds. I.

102

103

104

10-3 10-2 10-1 100 101 102 103

T (

K)

z (AU)

a

class 0class Iclass II

106

108

1010

10-3 10-2 10-1 100 101 102 103n H

(cm

-3)

z (AU)

b class 0class I

class II

10-3

10-2

10-1

100

101

102

10-3 10-2 10-1 100 101 102 103

Av

z (AU)

c class 0class Iclass II

10-2

10-1

100

101

102

103

10-3 10-2 10-1 100 101 102 103

polo

idal

vel

ocity

(km

/s)

z (AU)

bulk

drift

d


10-1510-1410-1310-1210-1110-1010-910-810-7

10-3 10-2 10-1 100 101 102 103

ζ(H

2+ ) (s

-1)

z (AU)

e


10-6

10-4

10-2

100

102

104

106

108

10-3 10-2 10-1 100 101 102 103

χe-3

Av

z (AU)

f

10-7

10-6

10-5

10-4

10-3

10-2

10-3 10-2 10-1 100 101 102 103

n(io

ns)/

n H

z (AU)

g

10-6

10-5

10-4

10-3

10-2

10-1

100

10-3 10-2 10-1 100 101 102 103

n(H

2)/n

H

z (AU)

h


10-1110-1010-910-810-710-610-510-4

10-3 10-2 10-1 100 101 102 103

n(C

O)/

n H

z (AU)

i

Fig. 7. Same as Fig. 6 for streamlines anchored just beyond the sublimation radius: r0 = Rsub = 0.2 AU in the Class 0 and Class I, and 0.11 AUin the Class II. These calculations assume no self-shielding of H2 or CO (i.e. no molecules surviving on streamlines launched inside Rsub). Theresulting abundances set a minimum photodissociation timescale on outer streamlines, to be compared with the flow timescale (see Fig. 10).

The AV towards the star is highest at the wind base, wherethe line of sight crosses the densest wind layers (see Fig. 1) anddrops steadily as material climbs along the streamline (Figs. 6,7c). This causes a steep rise in the effective attenuated radia-tion field, until the 1/R2 dilution factor takes over (Figs. 6, 7e,f).Combined with the wind density fall-off (Figs. 6, 7b), this gen-erates a global increase in ionisation fraction along the stream-lines out to 30−100 AU, until recombination sets in (Figs. 6, 7g).In our models, ionization is dominated by stellar X-rays forAV > 3 mag, and by FUV photons further out.

As the wind density drops by a factor 10 from Class 0 toClass I to Class II (Figs. 6, 7b), the ionisation fraction, X(i+) ≡n(ions)/nH, increases by a factor 3 to 10. Indeed, the smallerattenuating column through the inner wind (Figs. 6, 7c) leads tohigher X-ray and FUV ionization rates. At the same time, thelower wind density reduces recombination. Both effects work inthe same direction.

Figure 8 shows the main ionization contributors along the1 AU streamline for the Class 0, Class I, and Class II models.From these results, it can be seen that the main contributors de-pend on the overall ionization degree, X(i+):

when X(i+) ≤ 10−6: the main contributors to the negative chargeare the PAHs. The main positive carriers are molecular ions,with the most abundant being HSO+ and H3O+; and H+3(a direct product of X-ray ionization) also contributing in theClass II.

when X(i+) > 10−6: the main contributors to the negative chargeare the electrons. The main positive carriers are the atomicions S+, C+, and H+, which recombine more slowly withelectrons than molecular ions do.

An interesting result of our calculations is the relatively highabundances of CH+, SH+, and H+ of 10−6−10−4 reached alongthe Class I and Class II streamlines (Fig. 8). CH+ and SH+ areformed by endothermic reaction of C+ and S+ with H2, whenthe wind temperature exceeds about 3000 K. H+ forms by X-rayionisation and charge exchange, but also by photodissociationof CH+, with carbon playing the role of a catalyst through thereaction chain:

C + hν→ C+, (40)C+ + H2 → CH+ + H − 4640 K, (41)CH+ + hν→ C + H+. (42)

Through this repeated formation cycle, H2 is partly convertedinto H+, helping to offset H+ destruction by charge exchange.The flow crossing timescale is also too short for significantH+ radiative recombination. This example clearly illustratesthe out-of-equilibrium and hybrid nature of the chemistry inMHD disk winds, combining elements from both C-shocks(strong heating by ion-neutral drift) and photodissociation re-gions (abundant C+ and S+).

A2, page 11 of 21

Wind solution: a match

• N(H2O) / N(CO) ~ 1

• Model predictions reproduce observations: wind origin

D. Panoglou et al.: Molecules in protostellar MHD disk winds. I.

10-7

10-6

10-5

10-4

10-2 10-1 100 101 102 103

n(A

)/n H

z (AU)

class 0

10-7

10-6

10-5

10-4

10-2 10-1 100 101 102 103

z (AU)

class I

10-7

10-6

10-5

10-4

10-2 10-1 100 101 102 103

z (AU)

class II

X(H2O)X(OH)

X(O)X(SO)X(CO)

X(C)X(S)

Fig. 11. Abundances relative to H nuclei of important O, C, and S-bearing species along a flow line launched at 1 AU, for our Class 0 (left), Class I(middle) and Class II models (right).

H2O: when the wind temperature exceeds ≃400 K, oxygen notlocked in CO is converted to H2O through the endothermic reac-tions of O and OH with H2 in Eqs. (43), (44). Together withH3O+ recombination, this reaction is efficient enough to bal-ance the water destruction processes active in Class I/II jets(FUV photodissociation, photoionization, reaction with H+3 ).The asymptotic abundance of water is thus similar and quite highon the 1 AU streamline for all 3 classes5, at ≃5−10 × 10−5.

C, O, S atoms, OH and SO: OH and atomic C and O have lowabundances along the Class 0 streamline, where CO and H2Oare well shielded. In contrast, high abundances of these species≃0.5−1 × 10−4 are predicted in the outer regions of the Class Iand Class II streamlines, following CO and H2O partial pho-todissociation (see Fig. 11). Concerning sulfur, it is mainly inthe form of atomic S beyond 100 AU along the Class 0 stream-line, and is mainly photoionized into S+ in the Class I/II cases,where the FUV field is more intense (see Fig. 8). The asymptoticabundance of SO is substantial in the Class 0 only, at a level of3 × 10−7. These predicted characteristics for Class 0 jets are inline with the relative abundance of SO to CO ≃ 2 × 10−3 re-ported by Tafalla et al. (2010), and with the mass-flux traced byatomic sulfur lines being possibly as large as that inferred fromCO (Dionatos et al. 2010).

3.2. Effect of increasing launch radius in the Class I model

In Fig. 12 we illustrate in the Class I case the effect of the launchradius on the wind chemistry and temperature. We plot a rangeof r0 of 1−9 AU corresponding to a range in final flow speedof 90 km s−1 to 30 km s−1. The streamline from Rsub = 0.2 AUis also shown for comparison. The main effect is that the flowhas an “onion-like” thermal-chemical structure, with stream-lines launched from larger radii having higher H2 abundanceand lower temperature and ionization. The behavior with r0 isdiscussed in more detail below.

We do not present results for the interval 0.2 < r0 <1 AU in the Class I jet, as the minimum H2 photodissociationtimescale there becomes similar to or less than the flow time (see

5 In the Class 0 streamline, cold dust grains hold an additional H2Oreservoir in the form of ice, of assumed abundance 10−4 (Flower &Pineau des Forêts 2003), that could be later released in the gas phasein shock waves.

discussion in Sect. 3.1.3). Accurate H2 abundances on these in-ner streamlines await a detailed non-local treatment of H2 shield-ing, not taken into account in the present preliminary approach.For the same reason, we do not present results at r0 > 1 AU forthe Class II jet, but we note that the effect of increasing r0 will bequalitatively the same as in the Class I case (i.e. higher H2 abun-dance and lower temperature). We also do not present results atr0 > 1 AU in the Class 0 case, since most H2 and CO moleculessurvive already at a launch radius just beyond the sublimationradius Rsub = 0.2 AU (see Sect. 3.1.3).

3.2.1. Ionization and drift speed at various r0

Wind streamlines are initially less ionized at larger r0(Figs. 12f,g), due to the larger attenuation of X-rays andFUV photons by intervening streamlines (Fig. 12c). At large dis-tances, they recombine to an asymptotic fractional ionization ofa few 10−5, dominated by S+.

The drift speed does not exceed about 10 km s−1 and re-mains less than 30% of the bulk flow speed out to r0 = 9 AU(Fig. 12d). Therefore, magnetocentrifugal forces appear able tolift molecules out to disk radii of 9 AU for the typical parametersof Class I sources, in the MHD solution investigated here.

3.2.2. Temperature and molecule survival at various r0

The gas temperature at larger launch radii follows a similar be-havior to that found at r0 = 1 AU (i.e. a gradual rise on ascale z ≃ r0, followed by a shallow decrease), but the tem-perature peak is shifted to larger z (i.e. lower nH) and reachesa smaller peak value (see Fig. 12a). Therefore, H2 chemicaloxydation and collisional dissociation are less efficient than forr0 = 1 AU. As a result, the H2 abundance increases with r0 andmore than 90% of the molecules survive for launch radii ≥3 AU(Fig. 12h). Photodissociation of H2, which was already too slowfor r0 = 1 AU, is further reduced by the additional screeningfrom intervening wind streamlines and plays no role.

In contrast, we find that photodissociation remains the ma-jor destruction process for CO out to r0 = 9 AU. The drop inCO abundance towards the inner r0 = Rsub streamline largelyexceeds the factor 2 assumed in our “local” computation of theCO self-screening column (see Fig. 12i). The plotted CO abun-dances for r0 > Rsub should thus be considered as upper limits

A2, page 15 of 21

Part IIILow-mass protostars in

high-mass clusters

Stars form in clusters

But how do low-mass stars form in the presence of high-mass stars? Cygnus X, Herschel

Hennemann & Motte

Outflows scale with Menv

• Idea: Outflows provide a low-contrast tracer of low-mass population in clusters

100 101

Lbol[L⊙]

10−6

10−5

10−4

10−3

FCO(6−5)[M

⊙km

s−1yr

−1 ]

Class 0

Class I

10−1 100 101

Menv[M⊙]10−4 10−3 10−2 10−1

MCO (6−5) [M⊙]

e.g.

Yild

iz, K

riste

nsen

et a

l. 201

5

Recipe

• Step I: Build a cluster-in-a-box

• Step II: Measure outflow emission from nearby low-mass clouds

• Step III: Assign emission to cluster members

• Step IV: Observe 2 1 0 1 2x (pc)

2

1

0

1

2

y(p

c)

0.1 M1 M10 M30 M

21 0 1 2log(M) (M)

0

1

2

3

log(

dN/d

M)

Kristensen & Bergin (to be subm.)ALMA observations: Higuchi et al. 2015, IRAS16547

Recipe




• Step IV: Observe


050100150200250300Da (arcsec)

150

100

50

0

50

100

Dd(a

rcse

c)

IRAS2A-West

IRAS2A-East

IRAS4A-North

IRAS4A-South

IRAS4B-Total

0

2

4

6

8

R T MB

dv(K

kms

1 )

CH3OH emission traces outflows

Recipe




• Step IV: Observe


20 10 0 -10 -20Offset (arcsec)

20

10

0

-10

-20

Off

set(

arcs

ec)

Model

0

0.2

0.4

0.6

0.8

1.0

Jykm

s1

beam

1


20

10

0

-10

-20

Off

set(

arcs

ec)

ALMA

0

0.2

0.4

0.6

0.8

1.0

Jykm

s1

beam

1

0 5 10 15 20Distance (arcsec)

0

0.1

0.2

0.3

0.4

0.5

Inte

nsity

(Jy

kms

1be

am

1 ) ModelObservations

5 6 7 8 9 100.00

0.01

0.02

0.03

0.04

0.05

Benchmarking model

• Observations contain hot core, high-mass outflow, low-mass outflows

• Radial average matches at 50%

• Toy model reproduces observations: fine-tuning required


20

10

0

-10

-20

Off

set(

arcs

ec)

Model

0

0.2

0.4

0.6

0.8

1.0

Jykm

s1

beam

1


20

10

0

-10

-20

Off

set(

arcs

ec)

ALMA

0

0.2

0.4

0.6

0.8

1.0

Jykm

s1

beam

1

0 5 10 15 20Distance (arcsec)

0

0.1

0.2

0.3

0.4

0.5

Inte

nsity

(Jy

kms

1be

am

1 ) ModelObservations

5 6 7 8 9 100.00

0.01

0.02

0.03

0.04

0.05

Next steps• Include contribution from

high-mass outflows

• Explore parameter space: cluster age, IMF, cluster mass, …

• Other species: H2O, high-J CO as unique shock tracers

• Extrapolate to high-z starburst galaxies

The Astrophysical Journal Letters, 741:L38 (5pp), 2011 November 10 Van der Werf et al.

Table 1Parameters of H2O Line Emission from APM 08279+5255

Transition 20,2–11,1 21,1 − 20,2 32,1 − 31,2 42,2 − 41,3 11,0 − 10,1

ν0a 987.926 752.033 1162.911 1207.638 556.936 GHz

νobsb 201.166 153.132 236.797 245.905 GHz

Eu/kc 101 137 305 454 61 KFlux density calibrator 3C273 3C84 3C273 0923+392

and flux densityd 8.0 8.5 7.3 4.3 JySynthesized beame 0′′. 58 × 0′′. 60, 33 2′′. 98 × 1′′. 87, 47 0′′. 62 × 0′′. 56, 48 0′′. 61 × 0′′. 44, 49

σ f 2.5 1.7 7.1 4.3 mJyScont

g 16.5 ± 0.8 5.4 ± 0.3 26.6 ± 1.3 31.4 ± 2.0 mJyIline

h 9.1 ± 0.9 4.3 ± 0.6 8.0 ± 0.9 7.5 ± 2.1 <0.7n Jy km s−1

Imodeli 8.9 4.5 8.4 6.5 0.25 Jy km s−1

v0j 72 ± 35 47 ± 25 93 ± 16 50 ± 43 km s−1

FWHMk 492 ± 84 480 ± 58 312 ± 37 429 ± 96 km s−1

L′line

l,m (7.9 ± 0.8) × 1010 (6.5 ± 1.0) × 1010 (5.1 ± 0.5) × 1010 (4.5 ± 1.3) × 1010 <1.9 × 1010 K km s−1 pc2

Llinel,m (2.4 ± 0.5) × 109 (8.9 ± 1.3) × 108 (2.6 ± 0.3) × 109 (2.6 ± 0.7) × 109 <1.1 × 108 L⊙

Notes.a Rest frequency of the line.b Observing frequency.c Upper level energy of the transition.d Assumed flux density of flux calibrator at observing frequency.e Full width at half-maximum and position angle.f rms noise in the integrated spectrum at ∆ν = 40 MHz.g Observed integrated continuum flux density.h Observed integrated line flux.i Modeled integrated line flux.j Center velocity of fitted Gaussian relative to z = 3.911.k Full width at half-maximum of fitted Gaussian.l Using z = 3.911, H0 = 70 km s−1 Mpc−1, ΩΛ = 0.73, and Ωtot = 1.m Not corrected for lensing.n Upper limit from Wagg et al. (2006).

et al. (2009b). With CO lines detected from rotational levels upto J = 11 (Weiß et al. 2007; Riechers et al. 2009a) and HCN,HNC, and HCO+ lines from levels up to J = 6 (Wagg et al. 2005;Garcıa-Burillo et al. 2006; Riechers et al. 2010), its molecularmedium has been characterized very well. Analysis of the COrotational ladder revealed unusually high excitation (Weiß et al.2007). However, a previous search for the ortho ground-state11,0–10,1 H2O line (upper level energy Eu/k = 61 K) wasunsuccessful (Wagg et al. 2006). We therefore targeted lines ofhigher excitation, with Eu/k from 101 to 454 K.

2. OBSERVATIONS AND RESULTS

We used the Institut de Radioastronomie Millimetrique(IRAM) Plateau de Bure Interferometer (Guilloteau et al. 1992)with six antennas in 2010 December and 2011 February to ob-serve the four H2O lines listed in Table 1 (which also lists allrelevant observational parameters) in APM 08279+5255. Ob-serving times (including overheads) varied from 2 to 5.2 hr perline. The WIDEX back end was used, providing 3.6 GHz instan-taneous bandwidth in dual polarization. Data reduction using theGILDAS package included the standard steps of data flagging,amplitude, phase and bandpass calibration, and conversion intodatacubes. The flux density scale is accurate to within 15%. Thedatacubes were deconvolved using a CLEAN algorithm (Clark1980).

All four lines were detected and an image of the fluxdistribution of the 21,1 − 20,2 line is shown in Figure 1. Fitsto the continuum uv-data were used to calculate the continuumfluxes (reported in Table 1) and source sizes. In the high angularresolution observations the emission was found to be slightly

Figure 1. H2O emission from APM 08279+5255. The contoured pseudocolormap shows the distribution of the velocity-integrated continuum-subtractedH2O 21,1 − 20,2 line flux, with contour levels of 3σ , 5σ , 7σ , and 9σ , whereσ = 0.5 Jy km s−1 beam−1. The dashed white ellipse indicates the synthesizedbeam. The inset presents the NICMOS F110W image (Ibata et al. 1999) andshows that the gravitationally lensed images (the brightest of which are 0.′′378apart) are fully covered by the synthesized beam.(A color version of this figure is available in the online journal.)

spatially resolved, with source sizes of approximately 0.′′5,in agreement with high-resolution observations of the 1 mmcontinuum (Krips et al. 2007), and CO 1−0 (Riechers et al.2009b). Integrated spectra of the four H2O lines are shown inFigure 2 and parameters of the detected lines are presented in

2

The Astrophysical Journal Letters, 741:L38 (5pp), 2011 November 10 Van der Werf et al.

Figure 2. Spectra of H2O emission lines from APM 08279+5255 at a common frequency resolution of 40 MHz. The horizontal axis shows velocity relative to redshiftz = 3.911. The red curves indicate Gaussian fits to each spectrum.(A color version of this figure is available in the online journal.)

Table 1. The most sensitive detection is that of the 21,1 − 20,2line, which displays a symmetric line profile with a full widthat half-maximum (FWHM) of 480 ± 58 km s−1, excellentlymatching the FWHM values of 445–520 km s−1 of the CO lines(Weiß et al. 2007). Recently, a low spectral resolution 1 mmspectrum of APM 081279+5255 obtained with Z-Spec has beenreleased by Bradford et al. (2011). In the Z-Spec spectrum, the32,1 − 31,2 line is the only H2O line detected at the 3σ level,but our flux measurement for this line is a factor two lower.This discrepancy is similar to that between the CO line fluxesmeasured by Z-Spec and by the IRAM 30 m telescope, several ofwhich have been independently confirmed by the IRAM Plateaude Bure Interferometer (Weiß et al. 2007). A similar discrepancyis found between our flux measurement of the 42,2 − 41,3 lineand that by Bradford et al. (2011), although the latter had asignificance of only 2.5σ . The continuum flux densities matchvery well between the two data sets. Finally, we note that theserendipitous detection of another H2O line, the 22,0 − 21,1 line,using the IRAM Plateau de Bure Interferometer, was recentlyreported by Lis et al. (2011).

3. FAR-IR PUMPED H2O EMISSION

The distribution of the detected line flux as a function of theenergy of the upper level of the transition is shown in Figure 3.The flux distribution shows considerable emission out to the42,2 − 41,3 line, implying that for purely collisional excitationthe kinetic temperature must be of the order of at least 400 K.Critical densities are approximately 108 cm−3 for the lowestlines (20,2 − 11,1 and 21,1 − 20,2) and higher than 108 cm−3

for the higher lines. Therefore, any purely collisionally excitedmodel that can account for the observed 32,1−31,2 and 42,2−41,3

Figure 3. Flux in H2O emission lines from APM 08279+5255 as a functionof upper level energy Eu. Black symbols and error bars indicate the observedline fluxes and the previously published upper limit on 11,0 − 10,1 (indicatedby a downward arrow). Red crosses indicate values produced by our model asdescribed in the text.(A color version of this figure is available in the online journal.)

fluxes would produce much stronger emission in the lower lines,including the 11,0 − 10,1 level, for which a sensitive upper limitexists (Wagg et al. 2006). We therefore rule out purely collisional

3

van der Werf et al. 2011

H2O emission at z=3.911

Conclusions

• Water traces shocks in protostars: mass less important than evolution

• Hot water toward HL Tau originates in wind, not inner disk: implications for hot water in disks?

• Cluster-in-a-box models provide access to low-mass populations in embedded high-mass clusters

colloquium given at nasa/jpl

Science