PAH emission in the proplyd HST10: what is the mechanism behind photoevaporation?
Silvia Vicente, Olivier Berne, Alexander G. G. M. Tielens, Nuria Huelamo, Eric Pantin, Inga Kamp, Andres Carmona
aa r X i v : . [ a s t r o - ph . S R ] F e b D raft version F ebruary
12, 2018
Preprint typeset using L A TEX style emulateapj v. 5 / / PAH EMISSION IN THE PROPLYD HST10: WHAT IS THE MECHANISM BEHIND PHOTOEVAPORATION?
S. V icente , O. B ern ´ e , A. G. G. M. T ielens , N. H u ´ elamo , E. P antin , I. K amp , A. C armona Draft version February 12, 2018
ABSTRACTProplyds are photodissociation region (PDR)-like cometary cocoons around young stars which are thoughtto originate through photo-evaporation of the central protoplanetary disk by external UV radiation from thenearby OB stars. This letter presents spatially resolved mid-infrared imaging and spectroscopy of the proplydHST10 obtained with the VLT / VISIR instrument. These observations allow us to detect Polycyclic AromaticHydrocarbons (PAH) emission in the proplyd photodissociation region and to study the general properties ofPAHs in proplyds for the first time. We find that PAHs in HST10 are mostly neutral and at least 50 timesless abundant than typical values found for the di ff use ISM or the nearby Orion Bar. With such a low PAHabundance, photoelectric heating is significantly reduced. If this low abundance pertains also to the originaldisk material, gas heating rates could be too low to e ffi ciently drive photoevaporation unless other processescan be identified. Alternatively, the model behind the formation of proplyds as evaporating disks may have tobe revised. Subject headings: circumstellar matter — ISM: lines and bands — protoplanetary disks — stars: individual(HST10) — stars: winds, outflows INTRODUCTION
Most low mass stars are born in transient OB associa-tions (e.g., Lada & Lada 2003), and there is evidence that ourown young Solar System evolved near massive stars (e.g.,Hester et al. 2004). Externally illuminated protoplanetarydisks or proplyds (O’Dell et al. 1993) are young stellar ob-jects (YSOs) surrounded by Solar System-sized protoplane-tary disks and found embedded within or near a H ii region.In these extreme environments the disks are exposed to in-tense UV radiation fields and stellar winds from the OB stars,dynamical encounters with sibling stars and supernovae, ontimescales of planetary system formation and early evolution.Hence, the study of their properties may bring key constraintson the understanding of the general mechanism of planet for-mation and the origins of our Solar System.Proplyd morphology has been explained by models of evap-orating flows in externally illuminated disks or globules (e.g.,Henney et al. 1996; Sutherland 1997; Johnstone et al. 1998;Henney & Arthur 1998; St¨orzer & Hollenbach 1998, 1999;Richling & Yorke 1998, 2000; Vasconcelos et al. 2011) as aFUV (6 eV ≤ E < ≥ ∼ Kapteyn Astronomical Institute, Postbus 800, 9700 AV, Groningen,The Netherlands Universit´e de Toulouse; UPS-OMP; IRAP; Toulouse, France CNRS; IRAP; 9 Av. colonel Roche, BP 44346, F-31028 Toulousecedex 4, France Leiden Observatory, Leiden University, Niels Bohrweg 2, NL-2333CA Leiden, The Netherlands CAB (INTA-CSIC), LAEFF, P.O. Box 78, 28691 Villanueva de laCa˜nada, Madrid, Spain Laboratoire AIM, CEA / DSM - CNRS - Universit´e Paris Diderot,IRFU / SAP, 91191 sur Yvette, France UJF-Grenoble 1 / CNRS-INSU, Institut de Plan´etologie etd’Astrophysique de Grenoble (IPAG) UMR 5274, Grenoble, F-38041,France ical flow of neutral material escaping from the disk surfacewith velocities of 1-3 kms − . The mass-loss rate generatedby this photo-evaporation wind determines the lifetime of thegaseous disk and hence the timescale for the formation of gi-ant planets. At some distance from the disk, the EUV pho-tons ionize the neutral wind and form an ionization front (IF)with T ∼ K. The observed tails result from the di ff use UVradiation field (produced through the recombination of sur-rounding nebular gas) which drives an evaporation flow onthe shadowed side of the disk. These models consider theFUV photoelectric e ff ect on small dust grains and PolycyclicAromatic Hydrocarbons (PAHs, Joblin & Tielens 2011) as themain gas-heating mechanism at the disk surface. Recent ob-servations (Okada et al., submitted ) show that PAHs playa major role in the heating process, and theoretical studies(Kamp & Dullemond 2004) confirm this also for protoplane-tary disks. The proplyd models mentioned above are adaptedfrom classical PDR models (see Hollenbach & Tielens 1997for a review) which consider standard interstellar medium(ISM) PAH abundances. However, up to now, there is no ob-servational abundance determination for proplyds.HST10 (182-413; O’Dell & Wen 1994) is a teardrop-shaped proplyd (1 . ′′ × . ′′
6) containing a prominent nearlyedge-on disk (i ∼ ◦ , PA ∼ ◦ ) visible as a dark silhouette inH α , optical ionized species and the continuum, but glowingin [OI] λ µ m ro-vibrational line of H (Bally et al. 2000; Chen et al. 1998, Vicente et al., in prep. )This object is located at a projected distance of 56” from θ Ori C to the SE, the main ionizing star of the Trapezium Clus-ter (414 ± / VISIR observations of PAH emissionin HST10. We determine the PAH abundance in the neutralflow and discuss the implications for photo-evaporation mod-els. OBSERVATIONS AND DATA REDUCTION
Mid-IR imaging (75 mas / pix) and low-resolution spec-troscopy (R ∼ / pix) of the proplyd HST10were obtained on the 13th December 2005 with VISIR(Lagage et al. 2004), under good ambient conditions (see- Vicente, Bern´e, Tielens et al. Figure 1.
HST10 images (from left to right): H α (0.656 µ m) from HST / WFPC2 (45.6 mas / pix; HST archive), H (2.12 µ m) from VLT / NACO (27 mas / pix;Vicente et al., in prep. ), and PAH1 (8.6 µ m) and PAH2 (11.25 µ m) from VLT / VISIR (75 mas / pix; this work). The VISIR images show the 1 . ′′ µ m spectral settings respectively with the intensity scale given in mJy / arcsec . The arrow indicates the direction towards θ Ori C, the main ionizingO-star of the Trapezium cluster. North is up and east to the left. ing at 0.5 µ m between 0 . ′′ ′′ ). A spectro-photometricstandard (HD 35536) of similar airmass was observed imme-diately before and after the science target to allow for tel-luric absorption correction and calibration. The data were re-duced with a VISIR customary pipeline based on IDL scripts(Pantin et al. 2005) and corrected for the background using amulti-resolution inpainting scheme (Pantin 2010). The im-ages were collected with filters PAH1 (8.59 ± µ m), SIV(10.49 ± µ m), and PAH2 (11.25 ± µ m) for a total expo-sure time (on source) of 30min and using the parallel chop-ping / nodding jitter mode with a chopping throw of 7 ′′ . Thespatial resolution of the final images (19” × . ′′
35 for PAH1 and 0 . ′′ ff raction limited and reflectingthe improvement in the seeing . HST10 is spatially resolvedin the PAH1 and PAH2 images (0 . ′′ × . ′′ N -band spectroscopy consisted of three settings cen-tered at 8.5, 9.8 and 11.4 µ m (1h integration time) chosen forcovering the 10 µ m silicate feature and PAH emission bands at8.6 and 11.25 µ m. The 1 ′′ slit was placed along the proplydhead-tail, in the NS orientation, and the chop / nod was per-formed along the slit with a throw of 8 ′′ . The 1D-spectra wereextracted by integrating the flux for each wavelength over thespatial extension of the proplyd ( ∼ µ m (Fig. 2) but no detection above the noiselevel at 8.6 µ m. The slit position, estimated using the RA andDEC values from the headers, is overlaid on top of the VISIRimages in Fig. 1 and have an uncertainty of 0 . ′′ . ′′
3, corre-sponding to the accuracy of (small) relative o ff sets with theVLT. The RA o ff set of the slit at 8.6 µ m, relative to the pro-plyd nominal position in O’Dell & Wen (1994), is 0 . ′′ . ′′ ff setting per-formed to obtain the HST10 spectra. For the 11.4 µ m settingthe proplyd falls almost entirely inside the slit (see Fig. 1).The H µ m image (27 mas / pix) was collected withthe VLT adaptive optics instrument NACO under programID 076.C-0874 (PI, S. Vicente). This data is described andanalyzed in a forthcoming paper (Vicente et al., in prep. ). TheH α µ m image (45.6 mas / pix) from the Hubble Space The di ff raction limit at the VLT (8.2m) varies from 0 . ′′
26 to 0 . ′′
34 inthe range 8.6 – 11.25 µ m. As the size of a UT mirror is comparable to theturbulence outer scale, VISIR data are already di ff raction limited for opticalseeing below 0 . ′′ Figure 2.
VLT / VISIR low-resolution spectrum (R ∼ µ m, and extracted from the2D spectrum on the top. The PAH2 filter range is indicated by the verticaldotted lines. A flux of 16 mJy for the PAH plateau was determined as themean flux in an interval of the same length as the PAH2 filter and is repre-sented by the horizontal dashed line. Over-plotted is the spectrum smoothedwith a boxcar = Telescope instrument WFPC2 was retrieved from the ESOarchive (program GO 6603, J. Bally). The optical and near-IRimages were rebinned to the same pixel size of the VISIRimages and aligned with sub-pixel precision (0.01 pix) withIMALIGN / IRAF. The NACO image was used as referencebecause it contained several point sources in the field, con-trary to the VISIR and HST images for which only one to twostars could be found. Although there is a perfect overlap ofthe centroids of the stars in the RGB image, the estimated er-ror in the alignment is ± ±
75 mas) from comparisonof the position of the silhouette disk seen in the H α image tothe H disk emission seen in the NACO image (Fig. 3). Thiserror results from the di ff erent Point Spread Functions (PSFs)observed at the di ff erent wavelengths (0 . ′′
06 for H α , 0 . ′′
08 forH , and 0 . ′′
35 for PAHs). OBSERVATIONAL RESULTS
Fig. 1 shows the optical, near-IR and mid-IR images ofHST10 tracing the di ff erent key elements of its morphology.The optical H α image traces the ionized gas at the ioniza-tion front with its peak surface brightness facing the Trapez-ium stars. This is a clear evidence that θ Ori C is the mainsource of UV radiation driving the photo-ionization and evap-oration in HST10. The H µ m emission traces the molec-ular gas at the disk surface. This was first discovered in aAHs in the proplyd HST10 3 Figure 3.
Color composite image of the proplyd HST10 applying a square-root stretch in intensities for better contrast, and showing the di ff erent regions tracedby the di ff erent wavelengths: H α tracing the ionized gas at the ionization front (blue), H the molecular gas at the surface of the nearly edge-on protoplanetarydisk (green), and the PAHs emission the atomic gas within the PDR (red). From left to right we increase progressively the intensity of the PAH2 emission andfind it to be coincident with the more extincted or less emitting regions in the H α optical image. The silhouette disk is indicated by the green ellipse. HST / NICMOS image (Chen et al. 1998) and more recentlyconfirmed with adaptive optics ground-based imaging (Vi-cente et al., in prep. ). The VISIR mid-IR images reported hereshow emission associated to the proplyd in both filters PAH1(8.6 µ m) and PAH2 (11.25 µ m). The latter detection is alsoconfirmed with spectroscopic observations (Fig. 2) and canbe attributed to (solo) C-H out-of-plane bending mode in PAHmolecules with long straight edges. No continuum is observedbelow 11 µ m, but beyond 11.3 µ m we see in Fig. 2 the 11-14 µ m PAH plateau resulting from the blend of C-H out-of-planebending modes (e.g. arising from PAH clusters). On the basisof the similarity of the morphologies in the PAH1 and PAH2images, and since the PAH1 filter is also situated on top ofa C-H vibration (in-plane bending mode), we conclude thePAH1 emission is also due to PAHs. The total density flux in-tegrated over the full proplyd extension (rectangular apertureof 25 ×
45 pix) in the PAH1 image is 26.6 ± σ ) whenconsidering only the photon noise. Given that the uncertaintyin the conversion factor used for calibrating the images is typi-cally 10%, we obtain 26.6 ± µ m feature in the spectrum, divided by the passband ofthe PAH2 imaging filter (10.95 − µ m), is 29.4 ± σ ) which is consistent, within the error, with the extractedphotometry in the PAH2 image, 38.8 ± ff setting. The 11.25 µ m emissionobserved in the 2D-spectrum in Fig. 2 is extended over ∼ . ′′ . ′′ / pix), similar to the head-to-tailsize of HST10 in the PAH images, and is about 50 mJy atthe peak position. The maximum surface brightness in thePAH1 and PAH2 filters ( I . and I . ) were derived by takingthe average of the four brightest pixels at the proplyd head ineach VISIR image. These values are given in Table 1. PROPERTIES OF PAHS IN HST10
Spatial distribution and ionization
In PDRs, PAHs emit mostly in the neutral gas where themajority of molecules are dissociated by FUV photons (e.g. inthe nearby Orion Bar, Tielens et al. 1994). Hence, PAHs area tracer of the atomic gas (for instance in the NGC 7023 neb-ula they follow the far infrared emission of C + , Joblin et al.2010), and in the particular case of proplyds, PAHs will tracethe morphology of the photo-evaporating flow. This is wellillustrated in Fig. 3 showing the ionized (H α ), neutral (PAHs)and molecular (H ) gas components of the proplyd HST10. The PAH emission is localized within the ionized envelopeand at the disk surface from where, according to PDR mod-els, the FUV-heated material evaporates generating the neu-tral wind and creating the proplyd PDR. Additionally, eventhought the PAH1 and PAH2 emission show similar head-to-tail extent, their distribution is di ff erent. PAH1 emissionat 8.6 µ m is fainter ( ∼ × in surface brightness) and morehomogeneously distributed within the cocoon, whereas thePAH2 emission at 11.25 µ m is sharper and brighter on the op-posite side to the direction of θ Ori C, and coincident withthe areas showing less emission or more heavily extinctedin the optical H α image (Fig. 3). Considering that most ofthe optical extinction is caused by ∼ µ m dust grains, andthese are expected to be depleted in the neutral flow (with A V = . − .
2, Henney & O’Dell 1999) due to grain growthand settling in the protoplanetary disk, the 11.25 µ m emis-sion is tracing the regions of higher density in atomic gas andpossibly very small particles of dust. Additionally, while the11.25 µ m feature is dominant for neutral PAH molecules, the8.6 µ m feature is stronger for ionized PAHs mostly present inhigh UV irradiated regions (Joblin & Tielens 2011, and refer-ences there in). Hence, the di ff erence in spatial distributionobserved in the two filters may reflect a charge e ff ect associ-ated to the proplyd location in the foreground of the nebula,that is, in between θ Ori C and the observer. Positively ion-ized PAHs are expected to be more abundant on the irradi-ated side of the proplyd opposite to us, whereas the bulk ofthe PAHs reservoir, as seen in the proplyd “shadowed” side,seem to be in the form of neutral molecules. The I . / I . band ratio, at the position where we extracted brightnesses inHST10, is of the order of 0.5, in accordance to the astronom-ical template of Pilleri et al. (2012) for which a value around0.4 is found for neutral PAHs and 1.45 for PAH + . For highradiation fields, PAHs can be neutral if the density of the gasin the flow is high ( ∼ cm − , Tielens 2005), allowing fore ffi cient recombination of PAH cations with slow electrons.As we will see, this is most likely the case (Sect. 4.2). Finally,we note that there is no PAH emission beyond the ionizationfront, suggesting that they are largely destroyed beyond thispoint as seen in the Orion Bar (Giard et al. 1994). PAH abundance
Proplyd models consider disk photoevaporation to be duemainly to e ffi cient heating of the gas by energetic photo-electrons provided by small grains and PAH molecules. How-ever, these models assume PAHs in proplyds are as abundant Vicente, Bern´e, Tielens et al. Table 1
Parameters of NGC 7023 and HST10 used to derive the PAH abundance inequation 1Parameter ReferenceNGC 7023 I . ± − this paper I . ± − this paper G . × Joblin et al. 2010N H × cm − Joblin et al. 2010 f PAHC × − Bern´e & Tielens 2012HST 10 I . ±
111 MJy sr − this paper I . ±
111 MJy sr − this paper I . / I . G . × St¨orzer & Hollenbach 1999N H . × cm − St¨orzer & Hollenbach 1999 f PAHC × − this paper as in the interstellar medium (ISM), an assumption which hasnot yet been verified with observations. PAHs have been de-tected in, at most, 15% of the observed disks around isolatedTTauri stars and they are under-abundant by a factor of 25when compared to the ISM (Geers et al. 2007; Oliveira et al.2010). The VISIR images of HST10 presented in this papercan be used to estimate the abundance of PAHs in a proplydPDR for the first time.The mid-IR PAH emission at a given wavelength I λ is pro-portional to the number of carbon atoms locked in PAHs inthe line of sight (e.g. Joblin et al. 2010), and on the intensityof the UV radiation field, G . Hence, we can write I λ = ǫ λ f PAHC [ C ][ H ] N H G , (1)where f PAHC is the fraction of elemental carbon locked inPAHs, N H is the column density of hydrogen atoms in theline of sight, [ C ] / [ H ] is the abundance of carbon relativeto hydrogen atoms (1 . × − ), and ǫ λ is the PAH emis-sivity at the given wavelength. The latter parameter, ǫ λ canbe derived for sources where all the other parameters givenin Eq. 1 can be determined independently as in the caseof the reflection nebula NGC 7023. The values adopted for f PAHC , N H and G were taken from the literature (Table 1)while the intensities I . and I . for NGC 7023 were mea-sured directly in the Spitzer
IRS spectrum (Pilleri et al. 2012).This yields for the PAH emissivities ǫ . = . × − and ǫ . = . × − MJy sr − cm − G − which, when insertedin Eq. 1 combined with the parameters in Table 1, give thePAH abundance in the PDR of HST10, f PAHC , if a good esti-mate of the column density of atomic gas in the line of sight N H is provided. This parameter can be estimated from theelectron density at the ionization front which can be mea-sured with a fair accuracy. Bally et al. (1998) obtain a valueof n e = × cm − from the H α surface brightness, whileSt¨orzer & Hollenbach (1999) find n e = . × cm − , when expressed in terms of the Habing field which corresponds to an integratedintensity between 91.2 and 240nm of 1 . × − ergs cm − s − (Habing 1968). correcting for the extinction to the Orion Nebula. Assum-ing pressure equilibrium at the ionization front, the densityof H atoms in the neutral flow n H , must be of the order of10 × n e or n H ∼ cm − . For a head width of ∼ × cm,measured in the PAH2 image, and assuming this value forthe neutral flow lenght along the line of sight (symmetry inHST10), we obtain N H ∼ × cm − . From models andusing the same parameters, St¨orzer & Hollenbach (1999) find N H = . × cm − . Adopting their value for N H , and us-ing the brightnesses I . and I . (Table 1) measured in theVISIR PAH1 and PAH2 images, we find similar PAH abun-dances in HST10 of f PAHC = . × − and f PAHC = . × − .These values are nevertheless extremely low: they correspondto an abundance of PAHs 90 times lower than in NGC 7023(Bern´e & Tielens 2012), or about 50 times less than the valuesfound in the Orion Bar or in the di ff use ISM (Tielens 2005).And, since the brightnesses I λ have been measured for thebrightest pixels at the proplyd head (Sect. 3), our estimationgives the maximum abundance of PAHs in the evaporatingflow toward θ Ori C. DISCUSSION
As other studies of TTauri stars (Geers et al. 2007;Oliveira et al. 2010) we find PAHs to be under-abundant inthe PDR of the proplyd HST10, by a factor of 50 or more rel-ative to the di ff use ISM. The origin of this under-abundancecannot be readily explained, but some proposed hypothesesinclude clustering of PAHs followed by sedimentation insidethe disk, destruction by FUV photons in the PDR, or de-struction by X-rays emitted by the low-mass central star .Nevertheless, more important than the causes for PAH under-abundance are the implications this result has on our under-standing of the physical processes shaping morphology anddriving mass-loss in proplyds. Photo-evaporation of proplydshas been explained so far by the photo-electric heating of thegas which is known to have a much reduced photo-electrice ffi ciency for grains larger than 100Å (Tielens 2005). Giventhe high column density N H = . × cm − and low ex-tinction A V = . − . ffi ciency for PAHs(molecules of a few Å) and the small end of the very smallgrains (up to a few tens of Å). But according to Pilleri et al.(2012) (Fig. 6 in their paper), for high radiation fields as thosefound in proplyds ( > G ), the very small grains are de-stroyed and evaporated into free-flying PAHs. Therefore, wedo expect PAHs to be the main agents of the photo-electricheating in proplyds.Assuming the disk surface has the same PAH abundanceas the proplyd PDR (they are lifted from the disk surface bythe evaporative wind), the low fraction of PAHs found in thisletter will have a profound impact on the disk surface gas tem-perature. And hence, the current hypothesis of an evaporatingdisk creating the proplyd PDR and morphology may have tobe revised. Two possibilities arise: 1) other gas heating mech-anisms are relevant for disk evaporation, such as collisionalde-excitation of UV pumped H and H photo-dissociationfollowed by reformation on grain surfaces; 2) the PAH emis-sion from the proplyd cocoon is associated to remnant atomicgas from the protostellar envelope or the surrounding neb-ula. By combining detailed modeling with upcoming Her- The majority of stars in the Orion Nebula cluster are in the mass rangeof 0 . − . ⊙ (Hillenbrand & Hartmann 1998) AHs in the proplyd HST10 5schel (Bern´e et al., in prep. ) and VLT data (Vicente et al., inprep. ), the heating-cooling mechanisms at the disk surface canbe assessed allowing to test each one of the possible scenarioscreating the puzzling proplyd morphology.), the heating-cooling mechanisms at the disk surface canbe assessed allowing to test each one of the possible scenarioscreating the puzzling proplyd morphology.