An infrared view of (candidate accretion) disks around massive young stars
A. Bik, A. Lenorzer, W-.F. Thi, E. Puga-Antolin, L.B.F.M. Waters, L. Kaper, L.N. Martin-Hernandez
aa r X i v : . [ a s t r o - ph ] D ec **FULL TITLE**ASP Conference Series, Vol. **VOLUME**, **YEAR OF PUBLICATION****NAMES OF EDITORS** An infrared view of (candidate accretion) disks aroundmassive young stars.
A. Bik , A. Lenorzer , W.F. Thi , E. Puga Antol´ın , L.B.F.M.Waters , , L. Kaper , L. N. Mart´ın Hern´andez European Southern Observatory, Karl-Schwarzschild Strasse 2,Garching-bei-M¨unchen, D85748, Germany, Instituto de Astrof´ısica de Canarias, 38200 La Laguna, Tenerife, Spain Institute for Astronomy, The University of Edinburgh, RoyalObservatory, Blackford Hill, Edinburgh EH9 3HJ, United Kingdom Instituut voor Sterrenkunde, Celestijnenlaan 200D, B-3001 Leuven,Belgium Astronomical Institute ”Anton Pannekoek”, University of Amsterdam,Kruislaan 403, 1098 SJ Amsterdam, The Netherlands
Abstract.
Near-infrared surveys of high-mass star-forming regions start to shed lightonto their stellar content. A particular class of objects found in these regions,the so-called massive Young Stellar Objects (YSOs) are surrounded by densecircumstellar material. Several near- and mid-infrared diagnostic tools are usedto infer the physical characteristics and geometry of this circumstellar matter.Near-infrared hydrogen emission lines provide evidence for a disk-wind. Theprofiles of the first overtone of the CO band-heads, originating in the inner 10AU from the central star, are well fitted assuming a keplerian rotating disk.The mid-infrared spectral energy distribution requires the presence of a moreextended envelope containing dust at a temperature of about 200 K. CRIRESobservations of CO fundamental absorption lines confirm the presence of a coldenvelope. We discuss the evolutionary status of these objects.
1. Introduction
The observational study of high-mass star-forming regions is hampered by thefact that they are so unique, and thus, on average distant. Additionally, theactual formation happens deeply obscured inside molecular clouds behind hun-dreds of magnitudes of visual extinction. The actual formation sites are onlyobservable at radio and (sub)mm wavelengths. These observations, however,only provide indirect information about the forming stars. The Lyman contin-uum photon rate derived from the radio flux or the integrated infrared luminosityprovides an estimate on the properties of the massive star(s). Only after thestars are formed and start to clear out their environment, they become observ-able at shorter wavelengths. The near-infrared window is ideally suited to studythe massive stars in this evolutionary phase. The photosphere and close circum-stellar environment of the young massive stars can be detected as the extinction1
A. Bik et al is modest and the thermal emission of the surrounding molecular cloud is notdominant.Recent imaging and spectroscopic surveys have revealed the stellar con-tent of high-mass star-forming regions (Hanson et al. 2002; Kendall et al. 2003;Alvarez et al. 2004; Bik, 2004; Blum et al. 2004; Bik et al. 2005, 2006; Kaper et al.2007). Several types of objects were classified in these complex regions. A largenumber of OB stars have been identified by their photospheric lines (Watson & Hanson1997; Hanson et al. 2002; Bik et al. 2005) and their properties do not seem todiffer from OB main sequence field stars observed in the optical. Another class ofhigh-mass objects identified are the massive Young Stellar Objects. Their near-infrared colors indicate an infrared excess, and their near-infrared spectra aredominated by emission lines from ionized (hydrogen, helium) as well as neutralspecies, e.g. CO (Kendall et al. 2003; Blum et al. 2004; Bik et al. 2006).These spectroscopic features as well as the infrared excess can be used toderive the physical parameters of the cirumstellar material and try to understandthe evolutionary status of these objects. The detection and study of disks aroundyoung massive stars that are remnants of the formation process is key to ourunderstanding of the formation of massive stars (Zinnecker & Yorke 2007).
2. Near- and Mid-Infrared diagnostic tools
In this section the different diagnostic tools available in the near and mid-infraredwindow will be discussed. Line ratios of ionized hydrogen lines are used to tracethe physical conditions and geometry of the ionized circumstellar material. Theline profiles of the CO first overtone emission provide kinematic informationabout the dense, neutral gas in the inner 5 - 10 AU. Extended envelopes, locatedfurther away from the central star can be detected at longer wavelengths.
One of the spectral characteristics of the massive YSOs is that they show linesemitted by the recombination of hydrogen and helium. In the K-band, notonly Br γ , but also weaker lines of the hydrogen Pfund series (Pf23 - Pf 33) areobserved, as well as HeI lines in some cases (Bik et al. 2006). The HeI linesindicate the presence of a hot ionizing source with a spectral type of at leastlate - mid O (Hanson et al. 2002).The hydrogen lines can be used to obtain an estimate on the density of theemitting gas. The Pfund lines are high-density tracers which are only emitted invery dense gas (N e up to 10 cm − ). These lines likely originate in a circumstellardisk. This is, moreover, supported by the very broad line profiles (300 - 500km s − ) or by double peaked profiles in a few objects. Br γ , on the other hand,can be emitted in a large range of environments with different densities. It canbe associated to diffuse HII regions, stellar winds, disk-winds, or very densegaseous disks around Be stars. Also here, the observed spectral profile can helpto discriminate between possible scenarios. The observed velocities of Br γ arein the range of 100 - 200 km s − . This excludes the possibility that the lines areof nebular origin. The recombination lines in an HII regions have a broadeningsimilar to the local sound speed which is on the order of 20 km s − for ionizedgas. Also, the origin in a stellar wind can be excluded as the velocity profiles n infrared view of (candidate accretion) disks around massive young stars. N o r m a li s e d F l ux + C -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4log(Hu14/Pfg)-2.0-1.5 -1.0-0.50.0 l og ( H u14 / B r a ) LBV B[e] Be
Figure 1. left 2 pannels:
L-band spectra taken with ISAAC at the VLT of12 massive YSOs. The broad absorption are molecular solid-state transitionfeatures caused by material in the diffuse ISM or molecular clouds. Theseabsorption features can be used to derive an accurate line-of-sight extinc-tion value. The emission lines are mainly hydrogen emission lines. The linesneeded to construct the diagnostic diagramme are Br α (4.05 µ m), Pf γ (3.754 µ m) and Hu14 (4.02 µ m). right: The L-band diagramme as described inthe text. The gray circles represent the loci of the massive stars with knowncirumstellar geometry for which Lenorzer et al. (2002) constructed the dia-gramme: LBVs, B[e] and Be stars. Overplotted are the massive YSOs. of stellar winds show very broad lines (500 - 2000 km s − ). Double peaked Br γ emission in some massive YSOs clearly favors the disk-scenario (Blum et al.2004). However, not all the Br γ lines show this profile and a contribution of thedisk-wind cannot be excluded for these. The observed velocity of the Br γ lineis consistent with the disk-wind scenario.A disk-wind is caused by the interaction between the UV photons and theionized upper layer of the disk. This results in a radiation driven, outflowingdisk-wind with velocities up to 200 km s − (Drew et al. 1998). In a disk-wind,the average density of the emitting ionized gas is lower than in a geometri-cally thin circumstellar disk. This will affect the relative strength of the hy-drogen lines. While in low density regions (e.g. in an HII region), the relativestrength of hydrogen lines follows the optically thin (case B) approximation(Storey & Hummer 1995); in high density regions, where the gas is opticallythick, the relative strength of the lines is determined by the emitting surface ofthe gas.Lenorzer et al. (2002) constructed a diagnostic diagram using 3 specific hy-drogen lines in the L-band (Br α , Hu14 and Pf γ ) to demonstrate this method.The ratio of Hu14 over Br α spans a range of 2 orders of magnitude (from ∼ A. Bik et al
Figure 2. Observed CO bandhead profiles from 2 massive YSOs (Bik & Thi2004). The top spectra are the observed spectra, the bottom spectra arethe best fitting models. These models indicate that the CO is located in acircumstellar disk within 5 - 10 AU from the central star. in case of an optically thin HII region to unity in the case of an optically thickemitting medium). In order to calibrate the diagram and to relate it to differentcircumstellar environments, Lenorzer et al. (2002) used the observed line ratiosof massive stars with a known circumstellar geometry (Fig 1b). Three differenttypes of objects were chosen; Luminous Blue Variables (LBVs), evolved massivestars with a very strong stellar wind, B[e] stars with an equatorially flatteneddense stellar wind or disk and the Be stars with an optically thick, geometricallythin ionized disk. The location of these respective objects is plotted in Fig 1b ascircles. The Be stars fall as expected close to the optically thick location whilethe LBV stars are located more in the direction of the optically thin material.We observed a sample of 12 massive YSOs in the L-band with ISAAC at theVLT (Fig. 1a) and measure the line ratios to compare them with the locationof the well studied massive stars in the diagram. In Fig 1b the massive YSOswhich have all the 3 hydrogen lines in their spectrum are overplotted on thediagram.The location of the massive YSOs overlaps that of the B[e] and Be stars.First this confirms that the ionized emitting region of the massive YSOs is, atleast partially, confined into a disk. The span in location implies large differencesfrom object to object for the averaged densities of the ionized circumstellar gas.A possible explanation for this, is that some objects are surrounded by ageometrically thin and confined disk (similar to Be disk) while others have disksthat are more puffed up as expected when the disk is under the process of beingphoto-evaporated.
Additional to the ionized lines, lines emitted by neutral species are observed inthe spectra of massive YSOs. Among the most prominent features in the near-infrared are the CO first-overtone emission lines. These bands are observed inabout 25% of all the massive YSOs (e.g Chandler et al. 1993, 1995; Bik & Thi n infrared view of (candidate accretion) disks around massive young stars. Figure 3. Spectral energy distributions of the candidate massive YSOs. Thedata points are fit by a simple model. Panels a,b,c and e show a model fitwith a 2 temperature black body, the data points of panel d are fitted with apowerlaw and the fit in panel f is a single temperature black body. cm − .This hot neutral gas is located in the inner, dust-free regions of the circumstellarenvironment, relatively close to the star.The kinematic information provided by high resolution spectra of the band-heads can be used to constrain the location of the emitting material in the cir-cumstellar matter. Moreover, intrinsically, the blue side of the bandhead is verysteep and different velocity profiles give rise to a different shape of the bandhead.Keplerian disks create a blue wing on top of the profile (Fig 2). This blue wingcannot be explained by a gaussian velocity distribution. Bik & Thi (2004) andBlum et al. (2004) succesfully fit the high-resolution CO spectra with a emissionprofile from a rotating disk, assuming optically thin emission.This model also shows that the CO is emitted very close to the central star,in the inner 5-10 AU of the circumstellar environment. At these distances fromthe central star, the CO molecules need to be shielded from the UV photons inorder not to be photo-dissociated. Therefore, this gas cannot be located at thesurface of the disk, that would be ionized by the stellar radiation. The opticalthickness rises steeply in the radial direction, shielding the midplane of the diskfrom the UV radiation, allowing the material to become neutral and the COmolecules to survive. One of the main characteristics of massive YSOs is that they have an infraredexcess. Their Spectral Energy Distribution (SED) is not dominated by pho-tospheric emission, but by the emission of circumstellar matter, mainly dustemission. The near-infrared emission lines, as discussed in the previous subsec-tions trace the ionized or hot neutral gas in the circumstellar environment. The
A. Bik et al
Figure 4. Absorption profiles of 3 CO lines observed with CRIRES at highspectral resolution (6 km s − ). Two absorption components are detected, thestrongest component at +5.5 km s − is saturated, the weaker component at-5.5 km s − is only detected in the R(4) to the R(0) lines, suggesting a verylow temperature. infrared SED traces the warm material (down to a few 100 K) in the cirmstellarenvironment located further away from the central star.We have performed photometry of a sample of massive YSO covering thewavelength range from 1 to 20 µ m using the ESO instruments SOFI, TIMM2and VISIR (see Fig. 3 for some examples). The observed SEDs are dereddenedusing extinction values determined by Bik et al. (2006). The SEDs show a largerange in spectral slopes, suggesting large differences in the properties of thecircumstellar environment. They vary from blue slopes (Fig 3, panel d) to veryred slopes (panels b,c and e).The SED of the bluest object in our sample (11097nr693) cannot be ex-plained by dust emission. The blue slope can only be produced by opticallythick free-free emission from e.g. a gaseous disk like in Be stars. The SED canbe fitted with a powerlaw, with a slope typical for Be stars (Waters et al. 1991).Surprisingly, the SED around 10 µ m suggests the presence of silicate emission,not expected in a gaseous disk. This could be caused by a contribution of thehigh background from the surrounding HII region.The other objects have a much redder SED, and their SED can be explainedby dust emission. Overplotted to the observed data in Fig. 3 is the best fit usinga one or two temperature blackbody. The objects fitted with a 2 temperatureblackbody show that a colder component (T ∼
200 K) tracing dust furtheraway is needed to explain the shape of the SED. These objects also are spatiallyresolved at mid-infrared wavelengths and their observed size becomes larger withwavelength. For one of the reddest objects, 16164nr3636, the size of the 10 and20 µ m emission extends up to 7500 AU away from the central star. Far-infraredobservations of this source reveal a strong point source with a derived mass ofabout 200M ⊙ (Karnik et al. 2001). Another way to study the outer regions of the circumstellar material is theanalysis of absorption lines of e.g. CO towards a bright background. As partof the Science Verification observations (Siebenmorgen et al. 2007) for CRIRES,one of the latest VLT instruments, we observed one massive YSO (IRAS 16164- n infrared view of (candidate accretion) disks around massive young stars. CO bandheadd = 5 − 10 AU windPf23−30T = 2000 − 5000 K I o n i ze d s u r f a ce l ay er T < 1500 K d= 0 AUHeI T ~ K T = 10 − 500 Kd = 7500 AU Stellar B r g Ionized disk windNeutral Mid−PlaneDust emission BrgDust emission
Envelope, traced by mid−infrared emissionInner disk, traced by NIR line emission photo−evaporated by the central star
Figure 5. A schematic view of the different line-forming regions in the cir-cumstellar disk surrounding a massive YSO. The CO is located in the densemid plane of the disk, while the hydrogen lines are emitted by the ionizedsurface layer and/or disk wind. µ m to detect the CO lines of the fundamental transitions. Thespectral resolution of these observations is R = 50,000 (6 km s − ). The COlines we observe are in absorption and show multiple components. In Fig. 4,the R(7), R(4) and R(0) lines of CO lines are plotted. In all the spectra,a saturated absorption profile is present at -5.5 km s − . In the spectrum ofthe R(4) line a second absorption is present at +5.5 km s − . This line is notpresent in the R(7) spectrum but it appears even stronger in the R(0) line. Thissuggests that this line is absorbed in very cold material of a few 10 K. The coldenvelope detected in the far-infrared observations of Karnik et al. (2001) is likelythe material that causes the saturated absorption profile of the CO lines. Theother component could be related to the source, but also to a cold molecularcloud in the line of sight.
3. Summary and Evolutionary status
Several different diagnostic tools are discussed to obtain information about thephysical and kinematical properties of the circumstellar material. The differentdiagnostics trace different physical conditions and therefore different locationsin the circumstellar environment of the massive YSOs. Based on the collectionof observations a sketch can be drawn on how the circumstellar environment ofthe massive YSOs could look like. In Fig. 5 the different environments and theirdistance to the central star are sketched. Note that not all the objects have allthe characteristics discussed here, some only show a subset of them.Stellar photons ionise the surface layer of the inner disk, which recombina-tion is observed in the hydrogen and HeI lines; the radiation pressure exercedonto the surface layer also drives matter away in the form of a disk wind. In
A. Bik et al some objects the inner parts of the disk are so dense that the mid-plane canbecome neutral as it is shielded for the UV photons. In these regions (the inner5 - 10 AU) originate the CO bandhead emission showing evidence for keplerianrotation. Further out, the midplane of the disk cools even more and dust isable to survive giving rise to a red SED observed in the mid-infrared. A fewobjects show evidence for a cold dust component suggesting the presence of amore extended envelope. Most objects, however, are not detected in the mmregime, suggesting that they lack a large reservoir of very cold dust or gas.This illustrates the difference with the objects discussed by Zhang (2007)where large disks are detected in the (sub) mm regime. Still in the process offormation, these objects are too embedded to be detected in the near-infrared.Likely these large and cold outer regions of the circumstellar material are (being)photo-evaporated away. Photo-evaporation first leads to the disruption of theouter disk, while the inner disk can survive much longer (Hollenbach et al.1994). The inner disk is what is detected at near- and mid-infrared wavelengths.Objects for which the observations indicate the presence of an envelope mightbe the younger objects of our sample. Another important factor in the photo-evaporation process is the amount of UV photons emitted by the central star.This can drastically influence the dispersion time scales. A larger sample wouldbe needed to disentangle these effects.