Near-Infrared Imaging Polarimetry of Young Stellar Objects in rho-Ophiuchi
A. F. Beckford, P. W. Lucas, A. C. Chrysostomou, T. M. Gledhill
aa r X i v : . [ a s t r o - ph ] F e b Mon. Not. R. Astron. Soc. , 1–26 (2008) Printed 26 October 2018 (MN L A TEX style file v2.2)
Near-Infrared Imaging Polarimetry of Young StellarOb jects in rho-Ophiuchi
A. F. Beckford, P. W. Lucas ⋆ , A. C. Chrysostomou and T. M. Gledhill Centre for Astrophysics Research, University of Hertfordshire, College Lane, Hatfield AL10 9AB, United Kingdom
Accepted 2007 . Received 2007 ; in original form Tuesday 17th April 2007
ABSTRACT
The results of a near-infrared (J H K L P ) imaging linear polarimetry survey of 20young stellar objects in ρ Ophiuchi are presented. The majority of the sources areunresolved, with K band polarizations, P K < K >
20% are seen overtheir envelopes. Correlations are observed between the degree of core polarization andthe evolutionary status inferred from the spectral energy distribution. K band corepolarizations >
6% are only observed in Class I YSOs.A 3-dimensional Monte Carlo model with oblate grains aligned with a magneticfield is used to investigate the flux distributions and polarization structures of threeof the ρ Oph young stellar objects with extended nebulae. A ρ ∝ r − . power lawfor the density is applied throughout the envelopes. The large scale centrosymmetricpolarisation structures are due to scattering. However, the polarization structure inthe bright core of the nebula appears to require dichroic extinction by aligned non-spherical dust grains. The position angle indicates a toroidal magnetic field in the innerpart of the envelope. Since the measured polarizations attributed to dichroic extinctionare usually µ m. Key words: polarization – (stars:) circumstellar matter – stars: formation
Multicolour imaging linear polarimetry at near-infrared(near-IR) wavelengths is a powerful tool for mapping thedusty discs and envelopes that surround young stellar ob-jects (YSOs). The observed flux distributions and the polar-ization patterns are both influenced by the distribution andproperties of the dust grains responsible for the extinctionand the scattering of the light. It can be used to determinewhich source in a given region dominates the illumination. Itcan also be used to distinguish between the different mech-anisms that produce polarization, since dichroic extinctionand scattering processes sometimes produce differing wave-length dependencies.Generally, the scattering pattern associated with spa-tially resolved YSOs is centrosymmetric with aligned vectorsin the core of the nebula. This pattern of aligned vectors iscommonly referred to as the polarization disc. Some YSOswith resolved nebulae differ from the traditionally expected ⋆ E-mail: [email protected] centrosymmetric pattern by displaying a much broader re-gion of aligned vectors. The most likely cause of this featureis dichroism due to magnetically aligned non-spherical dustgrains in the circumstellar disc. Other proposed methods forproducing the aligned vector patter include multiple scatter-ing (Whitney & Hartman 1993) and “illusory disc”, whichis an effect caused by limited spatial resolution (Whitney,Kenyon & Gomez 1997; Lucas & Roche 1998).The ρ Oph star-forming region, at a distance commonlyquoted as 160 parsecs, is one of the nearest sites of low-massstar formation. The cloud complex contains a number ofdistinct dark clouds (Lynds 1962) and filamentary clouds (orstreamers), with a total mass estimated to be 10 M ⊙ . It hasbeen studied extensively at wavelengths ranging from x-rayto radio (Sekimoto et al. 1997; Girart, Rodriguez & Curiel2000; Andre, Ward-Thompson & Barsony 1993; Kamazakiet al. 2001). It is known to contain a rich population ofyoung stars associated with circumstellar envelopes and/ordiscs (Grasdalen, Strom & Strom 1973; Vrba et al. 1975;Elias 1978; Wilking, Young & Lada 1989; Greene & Young1992; Barsony et al. 1997; Bontemps et al. 2001; Wilking et c (cid:13) A. F. Beckford, P. W. Lucas, A. C. Chrysostomou and T. M. Gledhill
Table 1.
Rho Oph sample.Name a Coordinates (2000) b IR Class c Flux (mags) d Alternative e RA DEC J H K L NamesVSSG1 16 26 21.5 -24 23 07 II 13.49 10.76 8.68 - EL20, YLW31GSS30 16 26 21.5 -24 23 07 I 13.89 10.83 8.32 6.1 GY6, WL15DOAR25 16 26 23.7 -24 43 13 II/D 9.78 9.83 7.73 7.28 GY17, YLW34WL16 16 27 02.5 -24 37 30 II ? 14.14 10.58 7.92 5.85 GY182, YLW5AEL29 16 27 09.6 -24 37 21 I 17.21 12.01 7.54 3.88 GY214, YLW7A, WL15WL20 E 16 27 15.9 -24 38 46 II 13.54 10.78 9.21 8.62 GY240, YLW11BWL20 W 16 27 15.69 -24 38 43.4 II - - - - -WL3 16 27 19.3 -24 28 45 II > > > > > > a VSSG: Vrba, Strom, Strom & Grasdalen 1975; GSS: Grasdalen, Strom & Strom 1973; DOAR: Dolidze & Arakelyn1959; WL: Wilking & Lada 1983; EL: Elias 1978; YLW: Young, Lada & Wilking 1986; WLY: Wilking, Lada & Young1989 Table 2 b Quoted from SMBAD c Wilking, Lada & Young 1989; Andre & Montmerle 1994; Bontemps et al. 2001 d Wilking, Lada & Young 1989; Greene et al. 1994; Barsony et al. 1997 e EL: Elias 1978; YLW: Young, Lada & Wilking 1986; GY: Greene & Young 1992; WL: Wilking & Lada 1983;WLY: Wilking, Lada & Young 1989 Table 2; GWAYL: Greene, Wilking, Andre, Young & Lada 1994 al. 2001). Current estimates based on infrared observationsput the number of YSOs in the region at approximately 200(Bontemps et al. 2001). The proximity of ρ Oph to our solarsystem and the wealth of identified YSOs make it a goodsite for a polarimetric study of the dusty discs and envelopessurrounding YSOs.In this paper we present the results of a near-infraredimaging polarimetry survey of young stellar objects in the ρ Oph star-forming region. We compare polarimetric datawith assumptions about the evolution of YSOs to determineif there are any correlations between the degree of polariza-tion observed and the evolutionary status determined fromthe IR SED, also looking for correlations with the objectcolour. We also present grain scattering models for three ofthe Class I objects.In § § § § § The sample contains 18 of the sources identified as ClassI, based on the shape of their IR SEDs and spectral in-dices, from the Wilking, Lada & Young (1989) IR survey of the ρ Oph region. Subsequent investigation combining sub-millimetre data with the IR data has led to 9 of the sourcesbeing re-classified as Class II (Andre & Montmerle 1994).It is these latter classifications that are typically quoted byother authors and are therefore adopted by this paper. TheClass II sources were retained in the sample to allow com-parisons to be made.In addition, data has been obtained for a further 2 ob-jects, the Class III object YLW13A and the Class II objectWLY47. Table 1 is a complete list of all 20 of the sources,their coordinates, IR classifications, and J, H, K and L mag-nitudes.The observations were made at the United Kingdom In-frared Telescope (UKIRT) in Mauna Kea, Hawaii during thenights of 1998 June 17-18, 1999 April 28-May 01, 2000 July01-03 and 2002 May 15-17. The instrument used was IR-CAM with the polarimeter module IRPOL2, designed andbuilt at the University of Hertfordshire. The J (1.2 µ m), H(1.6 µ m) and K (2.2 µ m) broadband filters were used dur-ing the 1998, 1999 and 2002 sessions, and the L P (3.8 µ m)band filter was used during the 2000 session. The instrumentoptics provided a plate scale of 0.286 arcseconds per pixel(0.143 arcseconds with the magnifier). This gave a typicalfield of view of 36 × c (cid:13) , 1–26 ear-Infrared Imaging Polarimetry of Young Stellar Objects in rho-Ophiuchi Table 2.
Rho Oph aperture linear polarimetry.Name P core a P max b θ disc c J H K L d J H KVSSG1 - 1.4 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± e - - - - 33.9EL29 - - 8.3 ± e - - - 35.3 ± e ± ± ± ± ± ± ± ± ± e - - - 11.6WL6 - - 3.3 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± e ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± a Evaluated in 2 arcsecond diameter apertures, except: GSS30, which is evaluated in a 3 arcsecond aperture (toenable comparison with Chrysostomou et al. 1996), and WLY47 arc which is evaluated in a 1 arcsecond aperture.The quoted errors are based on the standard deviation of the polarization within the aperture used to assess thedegree of polarization. b Evaluated in 0.5 arcsecond diameter apertures, the quoted errors are based on the standard deviation within theaperture. c Position angle of the long axis of the disc, evaluated from K waveband data where available d Evaluated in 1 arcsecond diameter apertures e Shift & Add Result a field of view of approximately 20 × ×
15 arcseconds. Typically the seeing wasbetween 0.5 and 1.0 arcsec.The Wollaston prism mounted in IRPOL2 splits theradiation into the orthogonally polarized beams, usually re-ferred to as the ordinary and extraordinary (o- and e-) beamsfor historical reasons. A focal plane mask is used to reducethe field of view into two strips, each about 260 ×
50 pix-els squared, to prevent the o- and e- beams from overlap-ping. The half-waveplate in the system is successively ro-tated through 0, 45, 22.5 and 67.5 degrees. The advantageof this technique is that the intensity of both the o- ande- beams are measured simultaneously, which improves thereliability of the data by greatly reducing the effects of vari-ations in atmospheric transparency and seeing, allowing anyvariations in atmospheric transparency to be accounted for.To remove bad pixels on the array a 3-point jitter pat-tern mosaic, with 5 arcsecond east-west offsets, was used.
Four of the sample sources, EL 29, WL 16, IRS 54 andIRS 48, were imaged in the K-band using the shift andadd image-sharpening technique. This technique involvesthe taking of very short exposures, which are shifted so thattheir peak pixels coincide and are then co-added. This is allperformed in real-time by the ALICE (Array Limited Con-trol Electronics) electronics system of IRCAM3. The mag-nifier was adopted during the shift and add runs providinga pixel scale of 0.143 arcseconds.The main advantage of the shift and add observing tech-nique is in the resolving of compact bipolar nebulae andsmall polarization discs. However, there are several majorlimitations of the method. The data are read noise limiteddue to the short integration times involved, which meansthat it is much less sensitive than the conventional back-ground limited polarimetry. It is not possible to select ordiscard frames in a given stack. The method can only beused to observe YSOs that have a prominent compact fluxpeak. c (cid:13) , 1–26 A. F. Beckford, P. W. Lucas, A. C. Chrysostomou and T. M. Gledhill
Table 3.
Rho Oph source properties.Name α IR a Polarization Vector Total Intensity DistributionPatternVSSG1 -0.49 - Point-likeBKLT b - - Point-likeJ162618-242818GSS30 1.20 Centrosymmetric Highly extended bipolar nebulaDOAR25 -1.58 Random Point-likeWL16 0.79 Aligned Point-likeEL29 0.98 Centrosymmetric Point-like, polarized intensity images- possibly faint bipolar nebulosityWL20 E -0.07 Aligned Point-like, member of a triple systemWL20 W -0.07 Aligned Point-like, member of a triple systemWL3 0.23 Aligned Point-likeYLW13A -2.48 Random Point-like, in the H band there is apossible second source to the southYLW13A S - Random Point-likeYLW13B 0.08 Aligned Point-likeWL6 0.59 Aligned Point-likeWLY43 0.98 - Point-like (slightly oval)GY263 c - - Point-likeYLW16A 1.57 Centrosymmetric Extended, bipolar nebulaVSSG18 -0.24 Centrosymmetric Highly extended cometary nebulaYLW16B 0.94 Aligned Point-likeWLY47 0.17 Aligned Point-like, arc of nebulosity to the NWWLY47 arc - Centrosymmetric Elongated, curved structureWLY48 0.18 Random Point-likeWLY51 -0.04 Aligned Point-like, possible closebinary companionWLY54 1.76 Centrosymmetric Extended, cometary nebulaWLY63 0.4 Aligned Point-likeWLY67 0.74 Centrosymmetric Extended, cometary nebula a Wilking, Lada, Young 1989; Bontemps et al. 2001 b The 2nd source in the VSSG1 field has been identified as possibly being BKLT J162618-242618 c The 2nd source in the WLY43 field has been identified as possiblly being GY263
The source WLY54 was also imaged during the 1998UKIRT run. Comparisons of the results of both imagingmethods reveals that the structure of the total flux distri-butions and the polarization vector patterns is comparable.
The initial reduction of the data was performed using theStarlink software package
CCDPACK . Each image was darksubtracted and then divided by a suitable normalized flat-field. Our flatfields for J, H and K are median filtered im-ages constructed from images of the sky taken at each half-waveplate position. For the L band data the flatfields aremedian filtered images constructed from source images atdifferent offsets. The sky subtraction and the extraction ofthe o- and e- beam images was performed using the Starlinksoftware package POLPACK.The images were combined with the same software togenerate the Stokes parameters, I, Q and U,The polarization is P = p Q + U − σ I (1) where σ is the variance on Q or U. The position angle ofthe polarization is θ = 0 . „ UQ « . (2) The majority of the sources are point-like, displaying no ob-vious extended structure. It is possible that these objectsmay be associated either with compact or faint nebulositythat the detecting instrument is not sensitive to. Five ofthe sources are clearly associated with extended nebulos-ity. Two of these appear to have bipolar flux distributions(GSS30 and YLW16A), i.e. they have two lobes of extendednebulosity (evidence for YLW16A having two lobes of ex-tended nebulosity can be seen in flux distribution maps thatcover more of the region than can be seen in the polarizationdata presented here). The remaining objects appear to havecometary morphologies (WLY54, WLY67 and VSSG18), i.e.there appears to be only one lobe of extended nebulosityand any counterlobe that may be present is not visible. Inaddition to these five objects, the Class II source WLY47 ap- c (cid:13) , 1–26 ear-Infrared Imaging Polarimetry of Young Stellar Objects in rho-Ophiuchi GSS30 GSS30YLW16A - H YLW16A - K (b)
Figure 1.
Contour plots of the flux distributions for (a) GSS30, (b) YLW16A. Both GSS30 and YLW16A are bipolar nebulae, although thebipolar nature of YLW16A is only apparent in previously published higher resolution images (see text) and in the polarised flux and degreeof polarisation maps (see Figure 4). Contours are normalised and spaced at 0.9,0.71,0.5,0.35,0.25,0.18,0.13,0.088,0.006,0.004,0.003,0.002. pears to be associated with a nebulous arc a few arcsecondsaway from the point source.The flux distributions of the five extended objects andWLY47 are shown in figures 1(a)–(g).
The results of aperture polarimetry on the 20 sources arepresented in Table 2. The core polarizations are evaluatedin 2 arcsecond apertures, centred on the flux peak. Themaximum polarizations are evaluated using 0.5 arcsec aper-tures, the exact positioning of the centre of these aperturesis source dependent but for each source the same position isused at each wavelength. Maximum polarizations are only shown for sources that are either clearly associated with ex-tended nebulosity, or, in the case of EL29, show a significantspatial variation in fractional polarization within the image,which is attributed to small scale nebulosity that is hidden inthe wings of the image profile. Where there is more than oneinfrared object in the field of view, the degree of polarizationof each is shown separately. In total, aperture polarizationresults are presented for 25 objects.The polarization vector patterns observed are dividedinto three categories dependent on their appearance: cen-trosymmetric, aligned, and random. The pattern is said tobe centrosymmetric when the polarization vectors are ar-ranged in a circle (or ellipse) about the main illuminatingsource. When the polarization vectors appear to be arrangedin “parallel lines” the pattern is aligned. In a random pat- c (cid:13) , 1–26 A. F. Beckford, P. W. Lucas, A. C. Chrysostomou and T. M. Gledhill
WLY54 - J WLY54 - HWLY54 - KWLY67 - H WLY67 - K (d)(c)
Figure 1. (c) WLY54, (d) WLY67. Both these YSOs display cometary nebulae. Contours are as in Figure 1(a-b) tern there is no apparent structure to the arrangement ofthe polarization vectors, indicating that no polarization wasdetected. In Table 3 the spectral index, α , for each source ispresented with a description of both its polarization vectorpattern and its structure in the total flux image (i.e. theStokes I parameter). The Class I objects YLW16B, WL6, WLY43, WLY48,WLY51 and WLY63, the Class II objects VSSG1, DOAR25,YLW13B, WL3, WL16 and WL20, and the Class III objectYLW13A all have point-like morphologies in both total fluxand polarised flux. Several of these objects were found to be associated with a companion. There is a second objectlocated approximately 1120 AU from the core of WLY43,which has been tentatively identified as GY263, the struc-ture of the companion is point-like. There is a small knotapproximately 2 arcseconds (320 AU) to the north of thecore of WLY51 at a position angle of approximately 20 ◦ ,which is surrounded by a horn of material. The appearanceof the knot is stronger in the H band, and is only suggestedby the K band data. There is a second object to the westof VSSG1. The second object has been identified as possi-bly being the infrared source BKLT J162618-242818. WL20has been identified as a triple source by previous authors(Ressler & Barsony 2001). In our polarimetry data two ofthe members of the system are visible. These are the ClassII objects WL20 east (WL20 E) and WL20 west (WL20 W).There is a bridge of material linking both sources. c (cid:13) , 1–26 ear-Infrared Imaging Polarimetry of Young Stellar Objects in rho-Ophiuchi VSSG18 - H
VSSG18 - KEL29 - K (g)(e)
Figure 1. (e) VSSG18, (f) WLY47, and (g) EL29. VSSG18 is a cometary nebula, while WLY47 is a point source which appears toilluminate a detached arc of nebulosity. EL29 is unresolved in this image of the total flux but is resolved in polarised flux (see Figure 8).Contours are as in Figure 1(a-b)
The degree of polarization over the cores of the Class Iobjects range from P H ∼
9% and P K ∼ K ∼ P K ∼ P H ∼
10% and P K ∼ α − ∼ -1.58, is the lowest displayed byany of the Class II objects; in some classification schemes itwould be considered a Class III object. The spectral indexfor YLW13B, α − ∼ θ K ∼ ◦ . A 1- σ upper limit onthe fractional polarization of <
1% is measured in both theH and K wavebands. A second source is visible to the southof YLW13A in the H band. The degree of polarization overYLW13A south is P H ∼ WLY47 displays an interesting structure (see Figure 1f).WLY47 appears to be point-like; the interesting feature isthe ’arc’ of nebulosity that can be seen to the northwest. The c (cid:13) , 1–26 A. F. Beckford, P. W. Lucas, A. C. Chrysostomou and T. M. Gledhill HK Figure 2.
WLY47 polarization. Contour plots of the flux dis-tributions with polarization vectors overlaid for WLY47 in theH (upper) and K (lower) bands. The arc of nebulosity displaysa centrosymmetric polarisation pattern, demonstrating that it isilluminated by the adjacent bright YSO. The key is the samelength as a 100% vector. arc extends from the north to the west of the source and isapproximately twelve arcseconds (2000 AU) in length. Thereis evidence of a concentration of flux near the centre of thearc in both the H and K wavebands, possibly indicating thatthe arc contains a 2nd object. The core of WLY47 is approx-imately 5.5 arcseconds (880 AU) from the peak of the arc.There is nebulosity surrounding both WLY47 and the arcthat is apparent at low signal to noise. HST NICMOS im-ages also reveal the presence of the arc close to WLY47; thearc lies between WLY47 and VSSG18 (Allen et al. 2002).Figure 2 shows the H and K band polarization vectormaps for WLY47. At both the H and K wavebands the po-larization vectors across WLY47 are aligned, whilst the vec-tors over the ‘arc’ follow its structure, in the sense that thepolarization vectors over the ‘arc’ appear to be centrosym-metric with respect to WLY47. This indicates that the arcis a reflection nebula illuminated by WLY47. The positionangle of the polarization vectors over WLY47 and the peakof the arc are θ K ∼ ◦ and θ K ∼ ◦ , respectively. The levels of polarization observed are higher ( P H ∼
5% and P K ∼ P H ∼
11% and P K ∼ ◦ atthe southern end. The maximum levels of polarization givenin Table 2 are evaluated over the ‘arc’ and are shown to be P H ∼
28% and P K ∼ GSS30 has a bipolar morphology, with a northeast-southwest orientation, see Figure 1a and Figure 3. Thisstructure is interpreted as a reflection nebula, with thebright lobes corresponding to reflection from the walls of abipolar cavity in the circumstellar nebula, which is assumedhave been cleared by a bipolar outflow. The northern lobe isbrighter and more extensive than its southern counterpart.This has been previously used by Chrysostomou et al. (1996)to estimate the inclination of the system. Their investigationindicated that the system is inclined at an angle of approx-imately 25 ◦ - 30 ◦ to the plane of the sky, with the northernlobe tilted towards us. Near-IR data covering more of theGSS30 region has revealed an extensive bipolar nebula thatcontains three distinct sources (Chrysostomou et al. 1996).The main illuminating source is given the designation IRS1;the other two sources are IRS2 and IRS3, both lie in thenorthern lobe of nebulosity.The polarization vector map is shown in figure 3(a).The polarization vector pattern is centrosymmetric in theouter regions, becoming more elliptical towards the mainilluminating source (IRS1). There is evidence of a narrowpolarization disc over the core of IRS1 with a position angle θ K ∼ ◦ , at the J, H, K and L P wavebands. The po-larization disc runs perpendicular to the orientation of thebipolar extension. There is no apparent offset between theposition of the disc and the central flux peak. To the southof IRS1, along the polarization disc, the vectors experiencea 90 ◦ flip in their orientation, returning to the centrosym-metric pattern. This return to the centrosymmetric vectororientation is not observed at the northern end of the disc inthe J, H and K waveband maps, a feature previously notedby Chrysostomou et al.(1996). In the L P band map a returnto centrosymmetric pattern is seen at both ends of the po-larization disc. These reversals are predicted in Monte Carlomultiple scattering models that do not include dichroic ex-tinction effects (e.g. Figure 12 of Lucas & Roche 1998). Thereturn to a centrosymmetric pattern at the ends of the po-larization disc is caused by the transition from the opticallythick part of the disc to the optically thin region where sin-gle scattering dominates the pattern, as opposed to multi-ple scattering. It is likely that the absence of this reversalat the northern end of the disc in the shorter wavelengthdata is due the greater extinction at shorter wavelengths.Greater extinction would increase the size of the opticallythick, multiple scattering region and it would also strengthenany dichroic extinction effects which may also be contribut-ing to the observed polarisation disc, if the magnetic fieldhas a toroidal structure along this line of sight. c (cid:13) , 1–26 ear-Infrared Imaging Polarimetry of Young Stellar Objects in rho-Ophiuchi J HK Lp
Figure 3.
GSS30 polarization. (a) Contour plots of the flux distributions with polarization vectors overlaid in the J (top), H (uppermiddle), K (lower middle), and L P (bottom) bands (the key is the same length as a 110% vector). In figures 3(b) and (c) images of the degrees of polariza-tion and polarized flux at the H, K and L P bands are shown,respectively. (Polarized flux in each pixel is the product offlux and degree of polarisation). The H, K and L P band de-gree of polarization maps have a lower polarization regionrepresenting the polarization disc. The low polarization re-gion is surrounded by a higher polarization region, whichmarks the extent of the nebulosity shown in figure 3(a). Inthe L P band map there is also a lower polarization regionthat extends in the direction of the polarization disc axis, for the length of the bright north lobe. This region is narrowerthan the low polarization region that marks the polariza-tion disc. A similar feature is not seen in either the H orK band maps. The L band observations penetrate low den-sity matter and any knots of dust in the bipolar cavity toshow the influence of the dense material more clearly. It islikely that the low polarization region seen at L P band alongthe polarization disc axis is simply due to the smaller scat-tering angle for material in the walls of the bipolar cavitythat is projected along the axis, compared to off-axis loca- c (cid:13) , 1–26 A. F. Beckford, P. W. Lucas, A. C. Chrysostomou and T. M. Gledhill
Figure 3. (b). Degree of polarization maps for GSS30. The panelsare for the H (top), K (middle), and L P (bottom) bands. tions. At shorter wavelengths this effect (previously seen inIRAS 04302+2247 by Lucas & Roche 1997) may be obscuredby low density matter within the cavity. Future studies ofGSS30 should look at the object at radio and millimetrewavelengths to determine the appearance of the disc, seeZhang, Wootten & Ho (1997).In the L P band polarized flux map, there appear to bethree lobes. The smallest is the southern lobe seen in the fluxdistribution. The larger northern lobe that is apparent in thetotal flux distribution is seen as two lobes in the polarizedflux map. We interpret this as a simple consequence of theregion of low polarization located between the two lobes. Inthe H and K band maps (figure 3(c)) the structure revealedis similar to that seen in the flux distribution but the fluxdistribution appears pinched in the disk plane, revealing thebipolar nature of GSS30 more clearly. This is due to the lowpolarization in the disk plane, caused by multiple scatteringin this optically thick region. There are two lobes that extendnortheast and southwest. Close to IRS1, where the polarizedlight is brightest, the lobes appear to extend more along aneast-west axis. This inner east-west extension is not seen in Figure 3. (c). Maps of the polarized flux for GSS30 in the H(top), K (middle), and L P (bottom) bands. the total flux distribution, but it does roughly align with thehorn of nebulosity that extends to the east of the northernlobe. The eastern extension in polarized flux is due to aregion of high polarization shown in figure 3(b). This maybe caused by a region of low extinction inside the cavityto the east of IRS1, which would minimse the amount ofdepolarization due to multiple scattering. The data presented covers the central 7 × ×
800 AU) of the YLW16 region. The structure re-vealed by the polarization data in Figure 4 is a bipolarnebula, with an approximately east-west extension; in theH band the inner contours have a more northwest-southeastorientation. The bipolar structure is less obvious from theStokes I contours (see Figure 1b) but it is clear in the higherresolution HST NICMOS image shown in figure 15 (see alsoAllen et al.2002). Previous authors have shown that the neb-ula surrounding YLW16A has a radius of approximately3400 AU (Aspin, Casali & Walther 1989; Lucas & Roche c (cid:13) , 1–26 ear-Infrared Imaging Polarimetry of Young Stellar Objects in rho-Ophiuchi Figure 4.
YLW 16A polarization. (a) H and K band contour plotsof the flux distributions with polarization vectors overlaid, the keyis the same length as a 50% vector, (b) Degree of Polarizationmaps, and (c) Polarized flux distribution maps for YLW16A. Thedata are (b) scaled between 2% (black) and 45% (white), and (c)on a logarithmic scale between 1.6 (black) to 2.5 (white) at H and1.6 (black) to 3.1 (white) at K. µ m suggests that there is only one source.The polarization vector pattern is centrosymmetric inthe outer regions. In the core region there is a broad po-larization disc which is far more extended along the axis ofthe bipolar nebula than is typical and is evident in both theH and K band maps (see figure 4(a)). The strength of thisfeature strongly implies that it is produced by dichroic ex-tinction by grains whose short axis is preferentially alignedwith the disc plane., since multiple scattering models withspherical grains do not reproduce it, see section 5.3. Fur-thermore, the position angle of polarisation of the adjacentunresolved source YLW16B is very similar to that of the po- larization disc in YLW16A, which also suggests that dichroicextinction is operating.In order to test this hypothesis we removed the dichroiccomponent by subtracting the average Q/I and U/I valuesfor YLW16B from the Q/I and U/I frames for YLW16A.The resulting effect on the polarization vector pattern ofYLW16A is that the H band polarization disc is not as broadas shown in figure 3(a), and it has a position angle of θ H ∼ ◦ . In the K band the polarization disc is no longer clearlyvisible. This is what would be expected if dichroic extinc-tion is primarily responsible for the polarization disk, so theevidence supports this interpretation.There is no apparent offset between the polarizationdisc and the central flux peak. The polarization disc is alsoobserved to be approximately perpendicular to the directionof the inner contours, with a position angle of θ K ∼ ◦ and θ H ∼ ◦ . This is the largest variation in the positionangle of the polarization disc between wavebands shown byany of the sources. In the K band polarization vector map,following the length of the polarization disc to the south thevectors return to the centrosymmetric pattern. The sameis not seen to the northern end of the polarization disc ineither the H or K bands. The similarity in the degree andposition angle of the dichroic aspect of the polarization forYLW16A and YLW16B suggests that it is possible that thereis a strong feature at this point that is responsible.Figures 4(b) and 4(c) are images of the degree of polar-ization and the polarized flux in the H and K bands. In thedegree of polarization maps there is a region of low polar-ization that marks the position of the polarization disc, thelowest levels of polarization are observed at the ends of thepolarization disc. Higher polarization regions are observedto either side of the polarization disc.In the H band polarized flux map there are two lobesthat extend roughly northwest and southeast. The southernlobe is the largest. The polarized light is brightest in a cen-tral region that is roughly centred on the core. The polarizedlight is pinched to the south of the core of YLW16A; a sim-ilar pinching is not seen to the north. In the K band thestructure revealed gives the polarized light the appearanceof being an inverted V. The brightest region of polarizedlight is offset from the core, lying almost central to the east-ern lobe. A second less brilliant region is seen in the westernlobe.YLW16A has the lowest levels of maximum polariza-tion of all the extended sources, but it does not have thelowest core polarization. In Table 2 the core polarization ofYLW16A in the K band was shown to be P core ∼ max ∼ P max . The structure revealed in Figure 1c indicates that WLY54is a cometary nebula. The tail of nebulosity extends roughlyto the east of the core; the H band data indicates that thetail extends for at least 1500 AU. In the H and K band totalflux distribution maps there is a second ‘object’ (labelledS1) approximately 720 AU to the northwest of the core, thenature of this object is not known. There does not appear tobe any nebulosity extending from WLY54 to surround S1. c (cid:13) , 1–26 A. F. Beckford, P. W. Lucas, A. C. Chrysostomou and T. M. Gledhill
Figure 4. (b). Degree of polarization maps for YLW16A.
However, in the K band there is a second knot of nebulositybetween WLY54 and S1.The polarization vector maps in the J, H and K bandare presented in figure 5(a). The vector pattern is centrosym-metric. There is evidence in the J and H band maps of a nar-row polarization disc with a position angle θ ∼ ◦ , which liesover the central flux peak. The polarization discs observedfor other objects appear to be perpendicular to the orien-tation of the extended nebulosity; this is not the case forWLY54. In WLY54 the angle between the polarization discand the direction of the extension is approximately 109 ◦ .The polarization vector pattern indicates that WLY54might be a bipolar system that is at an inclination angle thatminimizes the amount of counterlobe visible. The positionangle of the polarization disc relative to the direction ofthe extension would make sense if there were a source offoreground extinction that is obscuring the northern regionsof WLY54.The core polarization levels seen are approximately 2%,5%, and 9% at the K, H, and J wavebands respectively.This is at the lower end of the levels observed for ClassI objects, and is much lower than the levels for the otherextended sources. The maximum polarizations for WLY54are assessed over the cometary tail at a distance of approx-imately 6 arcseconds (1000 AU) from the flux peak. The Figure 4. (c). Maps of the polarized flux for YLW16A. maximum polarizations at H and K are approximately 30%and 43% respectively.The degree of polarization maps in the H and K bands(see figures 5(b) and (c)) reveal a low polarization regionover the core of WLY54. A higher polarization region is seenover the cometary extension, and a smaller higher polariza-tion region is seen to the west of the core. The knot S1 isvisible in the H band degree of polarization map; in the Kband map S1 and S2 appear as a single “stream”. In figures5(b) and (c) the polarized flux maps at the H and K bandsare presented. The structure of the polarized light is similarto the total flux distribution. The structure of the knots isnot seen clearly in either the H or K band polarized lightmap. In the H band there is a slight bipolar pinching to thesouth of the core; the pinching is not seen in the K band.
WLY67 has a cometary morphology, see Figure 1d and Fig-ure 6. The cometary tail extends to the north of the core forat least 7 arcseconds (1120 AU). In the K band the struc-ture of the tail appears to narrow with distance from thecore, a similar narrowing is seen in the H band data. To thesouth of the core in the K band there is a small broad hornof nebulosity that is not visible in the H band. c (cid:13) , 1–26 ear-Infrared Imaging Polarimetry of Young Stellar Objects in rho-Ophiuchi (a)JHK Figure 5.
WLY54 polarization. (a) J, H and K band contour plotsof the flux distributions with polarization vectors overlaid, the keyis the same length as a 100% vector, (b) Degree of Polarizationmaps, and (c) Polarized flux distribution maps, for WLY54.
The polarization vector map presented in figure 6 re-veals that the vectors over the cometary extension have acentrosymmetric orientation. The vectors to the south ofthe core also appear to be centrosymmetric. In the H bandthere is evidence of a narrow polarization disc with a posi-tion angle of θ ∼ ◦ over the core. The polarization disc isnot seen in the K band.In polarized light the structure revealed for WLY67 issimilar to that observed for the total flux distribution. Thereis no significant pinching towards the ends of the polariza-tion disc defined by the vector map. VSSG18 is the only Class II source in the sample that is asso-ciated with a highly extended nebula (see Figure 1e, Figure7). All other non-Class I sources appear to be point-like innature; suggesting that any associated extended nebulosityhas already dispersed leaving only a more compact struc-ture. This is consistent with star formation models, whichpredict that by the Class II stage of evolution an object willhave lost its extended envelope. The extensive nebulosityof VSSG18 suggests that it’s evolutionary status is actuallysimilar to that usually associated with Class I sources. Ifthis is so, the weak mid-IR and far-IR emission could beexplained if the disc axis is oriented close to pole-on (seeWhitney et al.2003) but we would then expect the centralprotostar to appear much brighter relative to the surround-ing nebula. A possible alternative explanation is a gap in theinner accretion disc, perhaps due to binarity, which wouldreduce the amount of warm dust emitting at the IRAS wave-lengths.The nebulosity associated with VSSG18 has a cometarymorphology, which extends roughly to the northwest. Thenebulosity broadens with distance from the core, giving thematerial a fan-like appearance. The H and K band flux dis-tribution maps show that the cometary nebulosity extendsfor at least 2500 AU. In the J band the structure of the neb-ulosity differs from that observed in the H and K bands.The nebulosity appears to extend to the north; approxi-mately 320AU from the core the nebulosity is “pinched”, be-fore broadening. HST NICMOS data reveals a large sigma-shaped nebula; the nebulosity extends for at least 3800 AU,the exact location of the main illuminating source is notidentified (Allen et al. 2002).The polarization vector maps in the H and K bands arepresented in figure 7. The vectors over the extended nebulos-ity are centrosymmetric. Atypically, the vectors over the coreare aligned with a position angle of θ ∼ ◦ , placing themparallel to the direction of the extension. The other objectsthat have a centrosymmetric polarization vector structureshow aligned vectors that are approximately perpendicularto the direction of the extended nebulosity. To the south-ern end of the aligned vectors there is no clear evidence of areturn to the centrosymmetric pattern. The amount of infor-mation in the J band polarization vector map (not shown)is limited, but it can be seen that vectors over the core areat a position angle of 0 ◦ and the vectors over the extendednebulosity are centrosymmetric.The structure of the polarized light reveals a bright re-gion approximately centred over the core surrounded by afaint region that marks the extended nebulosity. There is noevidence of the typical bipolar pinching at the ends of thepolarization disc. EL29 (see Figure 1g, Figure 8) is one of the four objectsthat were imaged using the shift and add image sharpeningtechnique during the UKIRT 1999 observing run. AlthoughEL29 has been studied before at near-IR wavebands this isthe first time it has been looked at using high resolution lin-ear polarimetry. The structure of the total flux distributionrevealed indicates that EL29 is a point-like object. Previous c (cid:13) , 1–26 A. F. Beckford, P. W. Lucas, A. C. Chrysostomou and T. M. Gledhill
Figure 5. (b). Degree of polarization maps for WLY54. studies have indicated that El29 may be associated with ex-tended nebulosity (Elias 1978; Duchene et al. 2004).EL29 differs from the other point-like objects discussedpreviously (see § θ K ∼ ◦ . To the northern end ofthe disc the level of polarization decreases to approximately P ∼ Figure 5. (c). Maps of the polarized flux for WLY54. gion. At the ends of the disc, two polarization nulls are seen.The structure revealed in the polarized intensity map is notthe point-like object seen in the total flux distribution. In-stead in polarized light EL29 appears to be elongated, witha southeast-northwest orientation. There is a slight ‘pinch-ing’ of the polarized light that is characteristic of a bipolarobject.This result for the polarised flux image has recentlybeen independently confirmed by the conference report ofHu´elamo et al.(2007), which contains higher resolution datafrom the VLT. Deep infrared imaging by Ybarra et al.(2006)also reveals the extended nebulosity and shows that thereare H emission features within the outflow cavities, whichare oriented southeast-northwest as our polarised flux imageimplied. c (cid:13) , 1–26 ear-Infrared Imaging Polarimetry of Young Stellar Objects in rho-Ophiuchi Figure 6.
WLY67 polarization (K band). A contour plot of theflux distribution with polarization vectors is overlaid. The keylength is the same as a 50% vector.
Figure 7.
VSSG18 polarization. Contour plots of the flux distri-butions with polarization vectors overlaid for VSSG18. The keyis the same length as a 50% vector.
Figure 8.
EL29 polarization.(a) Contour plot of the flux dis-tribution with polarization vectors overlaid, the key is the samelength as a 50% vector, (b) Degree of Polarization map, scaledbetween 0% (black) to 40% (white) and (c) Polarized flux distri-bution map for EL29, plotted on a logarithmic scale between 0(black) to 2.4 (white).c (cid:13) , 1–26 A. F. Beckford, P. W. Lucas, A. C. Chrysostomou and T. M. Gledhill
Figure 9.
Wavelength dependence of the core polarizations. Solidlines are class I YSOs, dashed lines are class II and dotted linesclass III. The behaviour of the polarization with wavelength indi-cates that for the majority of the sources dichroism is importantin the core regions.
The wavelength dependence can reveal information aboutthe nature of the mechanisms responsible for the scatteringand absorption of light in an YSO. The wavelength depen-dence in the core regions for the 18 objects that were ob-served at multiple wavebands is shown in figure 9. The keyindicates the identities of individual sources. The wavelengthdependence seems to be independent of the IR classification;similar behaviour is observed for Class I, II and III YSOs.Most of the objects show a change of a few percent, typically3 – 4%, between 1.6 µ m and 2.2 µ m.In the majority of the objects the degree of polarizationdecreases with increasing wavelength. This is the expectedwavalength dependence if dichroism is the main mechanismresponsible for the polarization of light. However it is alsopossible for scattering to produce the same wavelength de-pendence if: (i) the polarization within the adopted apertureintegrates over a complicated polarization structure (Whit-ney et al.1997); or (ii) the measurement is of a polariza-tion disc of a bipolar source where most photons have beenmultiply scattered (Lucas & Roche 1998). We think it verylikely that the observed wavelength dependence in the pointsources is due to dichroic extinction in most cases. The al-ternative explanation, that the measured polarisation is dueto scattering in spatially unresolved reflection nebulae withcomplicated polarization structures, is not likely to be sucha common occurrence.Four of the YSOs investigated show an increase in thelevel of polarization with increasing wavelength: VSSG1,BKLT J162618-242818, WLY43, and the WLY47 ‘arc’. Thisindicates that scattering is important in these sources, since Figure 10.
Wavelength dependence of the maximum polariza-tions, for spatially resolved sources. Class I and Class II YSOsare represented with solid and dashed lines, respectively. The be-haviour of the maximum polarization suggests that in the outerregions of the YSOs scattering is important. a rising or flat polarization with increasing wavelength is pre-dicted by the Mie theory for single scattering by sub-microngrains with a range of sizes. In VSSG18 there is a decreasein the polarization between 1.2 µ m and 2.2 µ m, between2.2 µ m and 3.8 µ m the degree of polarization increases. Thesame behaviour is observed for WLY54. Three of the objectsshow no significant change in polarization with wavelength.The first is the Class III object YLW13A, the second is theClass II object DOAR25, and the third is the Class II ob-ject WLY47. The polarization levels for both YLW13A andDOAR25 are very low, P ∼ P ∼ µ m and 2.2 µ m, P H ∼
11% and P K ∼ µ m and 3.8 µ m, P Lp ∼ µ m and 1.6 µ m; only a one percent change in polarizationis observed, with larger decreases seen after 1.6 µ m.Figure 10 shows the wavelength dependence for themaximum polarizations for the 6 objects with extended neb-ulae. The degree of polarization is seen to increase withincreasing wavelength for three of the sources. One of theYSOs shows little change in polarization with wavelength;the remaining two objects display a decrease in polarizationwith wavelength. Of the three sources showing a positiveslope two are Class I and extended, the third is the extendedarc of nebulosity that is located close to the Class II objectWLY47. WLY47 is typically given a Class II designation,however the spectral index α = 0.17 puts WLY47 in thetransition object regime and the aperture is across the arcof nebulosity. The first object that shows a definite negativeslope is the Class II object VSSG18. The other object thathas a negative slope is WLY67; it should be noted that the c (cid:13) , 1–26 ear-Infrared Imaging Polarimetry of Young Stellar Objects in rho-Ophiuchi Figure 11.
A comparison of the degree of core polarization inthe K waveband for each object with its spectral index assessedin the wavelength range 2 µ m to 14 µ m. Error bars are shownfor the degree of K polarization; where no error bar is visible theerror is too small to be seen outside the point. The higher thedegree of core polarization the more positive the spectral index. polarization data in the K band is poor quality. YLW16Ashows only a 1% decrease in polarization. This could be ex-plained as a result of the proximity of the aperture centre tothe polarization disc; the data for YLW16A only cover theinner region of the envelope.Figures 9 and 10 suggest that in the inner regions ofYSOs dichroism is an important mechanism in the produc-tion of polarized light. However, we caution that this conclu-sion might be false if there are spatially unresolved scatteredlight structures with complicated polarization patterns. Inthe outer envelopes scattering is usually dominant. Withthe exception of the arc of nebulosity close to WLY47 allthe sources that display core polarization levels above P core ∼
4% have negative slopes between 1.2 µ m and 2.2 µ m. Thesources that display negative slopes to their maximum po-larizations have P max < The current method used to determine the evolutionarystage of an YSO is based on a combination of the shapeof its IR SED and its spectral index; both are dependent onthe distribution of circumstellar material. Generally, the de-gree of core polarization is seen to decrease as the age of theYSO, as indicated by the IR classification, increases. TheClass I sources typically display polarizations in the range1% < P K < < P K < P K < Figure 12.
A plot of the maximum degree of polarization in theK waveband for each object with extended nebulosity against thespectral index assessed in the wavelength range 2 µ m to 14 µ m. of polarization, contrasting with scattering in the Class Isources. The main difference in the dust properties of theClass I and Class II stages of evolution is thought to be onlythat the circumstellar envelope has dissipated, leaving onlythe disc.Again we caution that that Class designations are em-pirical and are influenced by the system inclination as wellas the evolutionary status. Phyiscally, the spectral index de-pends as much on the optical depth in the line of sight tothe protostar as the actual structure of the circumstellarmatter. Systems viewed in the equatorial plane (i.e. with anedge-on disc) will display a more steeply rising SED, higherpolarization and higher optical depth, than pole-on systems,which may lead to mis-classification.Figure 11 compares the degree of polarization in theK band for all the sources against their spectral indices, α − . As discussed in § µ m to 10-25 µ m. The subscript (2-14) indicates that the spectral indicesplotted are assessed in the wavelength range 2 µ m to 14 µ m,where the longer wavelength is based on the ISO 14 µ m filter.The degree of polarization tends to increase with increasing α − ; this trend is also observed in the H and Lp wavebandsand is suggested by the limited J band data available. It isimportant to note that the majority of the sources in thesample have α − >
0, only four of the sources observedhave α − < α − in figure 12.The maximum polarization of an YSO with extended neb-ulosity does not appear to be dependent on its IR evolu-tionary stage. The same is also seen with the H band polar-ization. α − maps the warm gas component, whereas theextended nebulosity represents the cold gas component.Figures 11 and 12 show that the degree of polarizationcannot be used to independently determine the evolutionarystage of an object, but it can be used to provide an approx- c (cid:13) , 1–26 A. F. Beckford, P. W. Lucas, A. C. Chrysostomou and T. M. Gledhill
Figure 13.
A comparison of the degree of core polarization ofeach object in the K waveband with its H-K colour. The opentriangles indicate sources with P K < > P K > K > imate guide. The older objects typically display lower coreand maximum polarizations. Polarizations of P >
6% are onlyseen for Class I YSOs.
One of the advantages of conducting the survey in the ρ Oph region is that complete near-IR J, H, K, and L spectralinformation is available for each source (see Wilking, Lada& Young 1989; Greene et al. 1994; Barsony et al. 1997). Thisenables comparisons between the polarimetric and spectraldata. In figure 13 the degree of polarization in the K bandfor each source is compared with its H-K colour. The H-Kcolour is a measure of the redness of the source. The rednessof an YSO decreases as it evolves. Therefore, the H-K colouris an indicator of the age of an YSO. There is a positivetrend observed between the polarization and H-K colour.The sources that display the highest levels of polarizationhave H-K > ≈ ρ Oph sample sources are presented with K band polar-ization measurements. The open triangles indicate sourceswith polarization less than 1.5%, filled squares representsources with polarization between 1.5% and 5%, and opensquares indicate sources with polarization greater than 5%.The colour-colour plots indicate that the polarization corre-lates with near-IR excess emission, as measured by the H-K
Figure 14.
Near-IR colour-colour diagrams with K core polar-ization measurements. The open triangles indicate sources withP K < > P K > K > and K-L colour index. All sources with K-L > P K > > P K > The shadow.f code (Lucas et al. 2004; Lucas 2003) can beused to represent a YSO with either 2-D, axisymmetricmodel or a 3-D, non-axisymmetric model. The distributionof matter is based on a simple star, disc, envelope system.The grains in the system are oblate spheroids; their appear-ance does not change with rotation about the short axis. It isassumed that the short axis of the grains is perfectly alignedwith the magnetic field and that the grains spin about thisaxis. Large numbers of photon packets are generated at thesurface of the protostar and these then move through thesystem, suffering modifications to their Stokes vectors dueto dichroic extinction and scattering events.The model results produced by the shadow.f code areoutput as Stokes I, Q and U images for comparison with the c (cid:13) , 1–26 ear-Infrared Imaging Polarimetry of Young Stellar Objects in rho-Ophiuchi observational data. The pixel scale applied to the Stokesparameters is the same as that for the UKIRT polarizationdata; 0.286 arcseconds pixel − or 0.143 arcseconds pixel − with the magnifier in place (the distance to ρ Oph is assumedto be 160 parsecs). The Stokes I, Q, U are convolved usinga Moffat profile with the form1 “ ` Rα ´ ” β (3)where α and β are two constants determined from the point-spread-function (PSF) of a standard star. The POLPACKpackage is then used to produce the degree of polarizationmap and the vector catalogue. The shadow.f code uses numerous input variables, which arelisted in Table 4. The recommended minimum number ofinput photons is 100,000. The radius of the system (i.e. thelarge scale envelope), the outer radius of the disc, the baseradius and the opening angle of the cavity manipulate thephysical dimensions of the system. All four variables canbe estimated from the observational results. The polar andazimuthal viewing angles can also be estimated from thedata. A polar viewing angle of 0 or 180 degrees means thatthe observer is looking towards the pole of the object, anangle of 90 degrees provides an edge on view of the system.The density coefficient and the vertical density gradient areused to manipulate the structure of the envelope and itsoptical depth.
The code has been adapted so that it can use two differentenvelope density equations. The first is an empirical distri-bution that uses a power law index to fix the vertical densitygradient. ρ env = CR − / „ µ k + 0 . « (4)where C is a free parameter whose value is is the densitycoefficient in units of kg m − / ). In most cases the line ofsight to the protostar passes through the envelope but notthe accretion disc, so C is simply proportional to τ K , theoptical depth at K band, which is one of the parameterslisted in Table 4. R is the 3-dimensional radius for the nat-ural system of cylindrical polar coordinates defined by thedisk rotation axis ( R = √ r + z ); the azimuthal coordinateis Λ), µ =z/r, and k is the power law index (increasing thevalue of k increases the degree of flattening of the envelope).The second is based on the density distribution of Tere-bey, Shu and Cassen (1984), ρ env = CR − / „ µµ « / „ µµ + 2 r c µ R « − (5)where r c and µ are as described in that paper. Modelsthat employ Equation 5 do not use the power law index, k ,that is listed in Table 4 but replace this parameter with the“centrifugal radius” r c . The bipolar nature of this sytem can be inferred from thepolarization maps and polarized flux maps in figure 4(b-c).However, the image in figure 4(a) does not resolve resolvethe structure, so we illustrate it better with additional datashown in figure 15(a-b). Figure 15(a) shows a NICMOS im-age at 0.15 arcsec resolution from the HST archive. Thecentral regions of the system are seen to be double peaked.Figure 15(b) shows a deep ground based image of a widerarea reproduced from Lucas & Roche (1998), with slightlyhigher spatial resolution than the polarimetric data in figure4. This image shows the arrowhead structure mentioned in § Models with an evacuated cavity along the system axis (thedefault assumption) failed to qualitatively reproduce eitherthe arrowhead structure or the double-peaked nature of thecore. We therefore attempted to reproduce these features bymodelling YLW16A as a bipolar source with a dusty jet in-side the bipolar cavity. Dust was introduced into both lobesof the bipolar cavity. Low density dust in the western lobeof the cavity (the part tilted away from the line of sight)reflects light from the protostar that would otherwise es-cape the system unseen, and serves to increase the promi-nence of the western peak. High density dust in the easternlobe is used to increase the obscuration of the protostar andthe eastern part of the reflection nebula so that the jet fea-ture produced by dust in the western lobe appears relativelybright. The eastern lobe is slightly curved in the parabolicsense with cavity walls at: r = R c + | z | tan ( θ c ) „ R c R « cc (6)where θ c is a parameter equivalent to the opening angle closeto the disc plane, R c is the radius of the cavity in the discplane, and cc is the parabolic curvature parameter. The dustin the eastern lobe is located throughout the cavity fromthe stellar surface to a radius, R =1000 AU, but only for az-imuthal angles | Λ | < ◦ (Λ=0 is the azimuth of the line ofsight). This increases the extinction toward the inner regionsof the eastern lobe, while allowing some light through to illu-minate the large scale reflection nebulosity that is observed.In order to better reproduce the narrow jet-like feature seenin figure 15) in the western lobe, the cavity on that side hasa narrow cylindrical structure described by r < R c and dustfills the region at: R CD < R < AU. (7)with density ρ cav = 3 × − G CD . The density in the easternlobe of the cavity is: ρ cav = 3 × − G CD „ RR c « ! − kgm − (8)where G CD and G CD are free parameters, with CD refer-ring to cavity dust.The Moffat profile used to convolve the Stokes elements c (cid:13) , 1–26 A. F. Beckford, P. W. Lucas, A. C. Chrysostomou and T. M. Gledhill
F160W - NICMOSYLW16A H band - WHTYLW16A ✻ NE Figure 15.
Images of YLW16A. (top) NICMOS image in theF160W filter, showing the bipolar nature of the system. We in-terpret the small scale narrow extension in the western lobe as re-flection from a dusty jet. Data from the HST archive (programmexxxx, PI S.Terebey). (bottom) Wider field H band image repro-duced from Lucas & Roche 1998. This image shows the extendednebulosity and the arrowhead structure more clearly than the po-larimetric data in figure 4. The figures have the same orientationas the data in figures 1 and 4, i.e. north is up and east is to theleft. was described by α = 6.49 and β = 3.35. These values weredetermined from the point spread function (PSF) for thepoint-like young star YLW16B imaged on the same night asYLW16A. The final model provides a quantitative fit to the positionangle of the polarization vectors and degree of polarizationover the core, which was assessed in a 2 arcsecond aper-ture and is centred at the point between the two peaks. Inaddition it provides a quantitative fit to the maximum po-larization assessed in a 0.5 arcsecond aperture centred ap-proximately 3.5 arcseconds from the core.The model fit to YLW16A is shown in figure 16, the fitparameters are summarised in table 4.The model provides a qualitative fit to the total flux dis-tribution and the polarized intensity distribution. By placing
YLW16A polarization modelYLW16A high resolution modelYLW16A low resolution model
Figure 16.
The model fit for YLW16A. (top) polarization vectormap, smoothed to the approximate resolution of the data in figure4. The toroidal field structure reproduces the prominent polar-ization disc.(middle) flux distribution, shown at the approximateresolution of the NICMOS image. Note that we do not attempt toreproduce the northwest-southeast orientation of the inner con-tours seen in figure 15. (bottom) flux distribution shown at theapproximate resolution of the ground based data. The figure hasthe same orientation as figure 15.c (cid:13) , 1–26 ear-Infrared Imaging Polarimetry of Young Stellar Objects in rho-Ophiuchi Table 4.
Parameters for model fit to YLW16A at the K band.Parameter ValueSystem Radius 3400 AUOuter Radius of Disc, r o
100 AUEnvelope Optical Depth to protostar at K, τ K k θ ◦ .4Azimuthal Viewing Angle, Λ 0 ◦ Base Radius of Cavity, R c
25 AUHalf-opening Angle of Cavity at base, θ c ◦ Grain Albedo 0.4Grain Axis Ratio 1.015 > gr > µ mMinimum Radius of Cavity Dust, R CD
25 AULower cavity Density Coefficient, G CD CD dust in both cavities it was possible to reproduce the sec-ond peak seen in the NICMOS data. If dust was placed inonly the western (right hand) lobe it was not possible to re-produce the second peak for any value of G CD investigated,hence the need for the dense dust in the eastern lobe. Thefit is achieved for values of the upper and lower density co-efficients of G CD = 6 ± < G CD < ◦ θ ◦ . In thefinal model a polar viewing angle of θ = 66 ◦ (cos( θ )=0.4)was found to provide the best fit to the data. The openingangle of the cavity for the final model was θ c = 25 ◦ (fromthe disc axis to the cavity wall), and the value of R c was 25AU. The Terebey, Shu & Cassen envelope density equation(5) and the empirical envelope density equation (4) wereboth investigated. Both equations were capable of producingsimilar results. The final model used the empirical equationwith a power law index of k = 0. More positive values ofthe power law index failed to reproduce the double peakedstructure. The Terebey, Shu & Cassen density equation wascapable of providing a qualitative fit to the arrowhead struc-ture for a centrifugal radius of R c
100 AU, which approx-imates to the simple R − / distribution as the final modelas R c → τ K =18, so the protostar itself is entirely obscured from our lineof sight. The optical depth must be lower along other linesof sight to permit the illumination of the large reflectionnebula, which is why the cavity dust in the eastern lobe isrestricted to a certain range of azimuthal angles, as notedearlier.The optical depth through the envelope to the protostarin the K band was investigated for values of the envelopedensity coefficient C that correspond to optical depth 4.6 τ K τ K = 8.5 ± τ k τ K abovethis range resulted in a less prominent western peak and nodistinct arrowhead structure.In the final model the dust grains were aligned with atoroidal magnetic field. An axial magnetic field was able toreproduce the polarization vector structure in the extendednebula, but failed to reproduce the core polarization struc-ture. The outer and inner polarization vector structure isquite successfully reproduced by either a toroidal field orsimply a uniform interstellar field oriented parallel to thedisc plane. The actual field structure may well be more com-plicated than this: as noted in § µ m. Smaller max-imum grain sizes were considered, however they resulted incore polarizations that were higher than required. The grainaxial ratio adopted was 1.015 +0 . − . . Grain axial ratios aboveor below this range led to core polarizations that were sig-nificantly higher or lower than required, respectively. Notethat since these grains are almost spherical, the magneticfield direction direction only affects the core polarizationand has negligible effect on the other model parameters.It is of course possible that the core polarization is due tograins which are highly aspherical but only weakly alignedwith the magnetic field. In Figure 5 there is an oblique angle between the normalto the polarization disc plane and the extended nebula. Wesuggest that this could be explained if WLY54 is a bipo-lar nebula with a foreground extinction cloud. The vectorsreveal the possible existence of a small counterlobe to thesouthwest of the core. The position angle of the polarizationdisc relative to the direction of the extended nebula would beexplained if there were a source of extinction to the north c (cid:13) , 1–26 A. F. Beckford, P. W. Lucas, A. C. Chrysostomou and T. M. Gledhill of WLY54 that obscures part of the nebula. Further sup-port for the bipolar nature of WLY54 comes from the east- west orientation of the degree of polarization map, whichshows a low polarization region running north - south, withtwo higher polarization regions to either side. The bipolarnature of WLY54 is further supported by the appearance ofthe less prominent (western) lobe in an H-K colour map (notshown) which appears to be redder than the more prominent(eastern) lobe. However, there is no direct evidence for theforeground extinction cloud that is used in the model to re-produce the unusual total flux and polarization structures.If there is no foreground extinction cloud then somethingelse must be responsible for the misalignment between thedisc axis and the extended nebulosity. Two possible expla-nations are that the orientation of the extended nebulosity isinfluenced by source movement through a dense medium, orextinction within the cavity. Alternatively, the polarizationdisc may not be a feature of the system at all, but mightsimply by due to foreground dichroic extinction if the po-larization intinsic to the WLY54 system is very low in thatregion. We note that most of the sources in this ρ Ophiuchisample have redder colours than the sources in the Taurussample of Whitney et al.(1997), which implies higher fore-ground extinction. This is particularly obvious for the ClassII systems, which have less intrinsic extinction.The location of the foreground extinction cloud alongthe line of sight is defined by somewhat arbitrary limits inthe model (see Table 5). The structure of the extinctioncloud in the plane of the sky is described by a gradient thatbegins 200AU to the north of the core. Two extinction gra-dients were investigated. The first is an exponential profiledescribed by ρ screen = Cτ s e Λ (9)where C is a constant (here given the value 4.55 × − kgm − ), τ s is a variable that can be used to manipulate theoptical depth of the extinction cloud, and Λ describes theazimuthal angle. The second is a linear profile ρ screen = Cτ s rsin Λ (10)where the constant C is here given the value 3 × − kgm − , and r sin(Λ) is the distance north of the protostar.It was assumed that the density distribution of the enve-lope is adequately described by the empirical density equa-tion (10).The Moffat profile values are determined from the PSFfor the point-like YSO YLW13A imaged during the sameobserving run as WLY54 ( α = 3.71 and β = 3.49). The parameters that provide the fit to the H band data aresummarized in Table 5 and the fit is illustrated in figure 17.The model provides a quantitative fit to the maximumdegree of polarization in WLY54 (evaluated in a 0.5 arcsec-ond aperture), and the position angle and degree of polar-ization of the polarization disc (evaluated in a 2 arcsecondaperture).The model provides a qualitative fit to the shape of
Figure 17.
The model fit for WLY54 (top) flux distribution,and (bottom) polarization vector map. the cometary nebula as revealed by the total flux distri-bution. In addition the model successfully reproduced theangle between the orientation of the polarization disc andthe direction of extension. These features are reproduced byintroducing a foreground extinction cloud to the north of thesystem. The structure of the total flux distribution was bestreproduced when the influence of the foreground extinctioncloud was described by the linear profile defined in equ.10.The exponential profile did not have a significant influenceon the structure of the total flux of WLY54. It was alsofound that parabolic cavity walls produced a better fit thanconical cavity walls. The parabolic cavity walls are definedby equ.6, which introduces a curvature parameter cc to theconical profile.The cometary appearance of WLY54 suggested a polarviewing angle that minimises the amount of counterlobe thatwas visible. The possible presence of the foreground extinc-tion cloud in the system means that the polar viewing angleis not tightly constrained since acceptable results were foundfor θ = 45 ◦ .6 +14 − . The adopted polar viewing angle for thefinal model provided the best fit to both the total intensitydistribution and the polarization data. c (cid:13) , 1–26 ear-Infrared Imaging Polarimetry of Young Stellar Objects in rho-Ophiuchi Table 5.
Parameters for model fit to WLY54 at the H band.Parameter ValueSystem Radius 1560 AUOuter Radius of Disc, r o
100 AUOptical Depth to protostar at H, τ H ± ± θ ◦ .6Azimuthal Viewing Angle, Λ 0 ◦ Base Radius of Cavity, R c
30 AUHalf-opening Angle of Cavity at base, θ c ◦ Parabolic Curvature Parameter, cc 0.2Grain Albedo 0.45Grain Axis Ratio 1.015Maximum Grain Size 0.35 µ mInner Radius of Extinction Cloud 2180 AUOuter Radius of Extinction Cloud 4680 AUDensity parameter of Extinction Cloud, τ s A suitable qualitative fit to the total flux distributionwas found for a power law index of k = 1.0 ± < τ H < τ H < τ H > τ H = 6.5, whereasthe best fit to the polarization data is achieved for an opticaldepth of τ H = 5.5.The dust grain albedo was treated as a free parameter.The grains were a mixture of silicates and amorphous car-bons. For the polar viewing angle adopted, θ = 45 ◦ .6 +14 − ,the best fit to the data was found for a grain albedo of ap-proximately 0.45 at H.The position angle of the polarization disc and the outerpolarization vector structure were reproduced by dust grainsaligned with a toroidal magnetic field. An axial magneticfield led to the position angle of the polarization vectors overthe core being at 90 ◦ to the requirement due to dichroic ex-tinction. We note that a return to an axial field structure isof course possible on larger spatial scales, and is expectedfrom basic star formation theory. To fit the core 5.3% polar-ization observed at UKIRT in the H band with the adoptedmagnetic field structure and maximum grain size of 0.35 µ mrequired that the grain axial ratio was not larger than 1.015.Increasing the grain axial ratio resulted in core polarizationlevels that were significantly larger than required. A fit tothe degree of core polarization and position angle of the po-larization disc was also found using larger grains. However,the models with larger grains were not able to reproduce thehigh polarization observed over the more extended envelope.Spherical grain models were capable of producing a rea-sonable approximation to the WLY54 data. However, thereare some key problems. The model fails to reproduce thepolarization vector structure over the core region, and thedegree of core polarization is approximately one tenth of the requirement. The model does manage to reproduce themaximum polarization and provides a qualitative fit to thetotal flux distribution. To fit the 5.3% polarization requiresan optical depth of τ = 8.7 ± τ s
50, as it was not known how its presence would affectthe system. Increasing the value of τ s causes a reduction inthe flux from both lobes to the north of the core. The finalmodel required that the density parameter was τ s = 30 ± > k = 0.6 and θ = 53 ◦ .1 a small westernlobe is visible.Increasing the cavity curvature parameter results in anarrowing of the extended nebulosity, a qualitative fit to thetotal flux distribution is found for cc = 0.2 when combinedwith the influence of the foreground extinction cloud. EL29 was modelled as a bipolar source. The density dis-tribution of the envelope was assumed to be adequately de-scribed by the empirical density equation, equ. 4. The struc-ture of the cavity was assumed to be conical (tan( θ c ) = (( R - R c )/ | z | ). The base radius and opening angle of the cavitywere estimated from the polarized flux data for EL29.A system radius of 1035 AU was estimated from theextent of the degree of polarization map, due to the point-like appearance of the total flux distribution. The point-likenature of the flux distribution makes it difficult to determinethe inclination angle of the system. For this reason EL29was investigated as both a pole-on and edge-on object. Anobject that is near pole-on would have a more point-likestructure similar to that seen in the data. However, if EL29is a source whose true nature is obscured either due to thefaintness of the nebulosity or the limited sensitivity of theobserving technique then it is also possible that EL29 iscloser to edge-on.The radius of the inner accretion disc, R a , was investi-gated for values between 50 and 600 AU, which is the rangeof possible values suggested by Ceccarelli et al. (2002).The dust grains are oblate spheroids that are perfectlyaligned by a magnetic field. The magnetic field was investi-gated with both an axial and toroidal structure. The albedoof the grains was treated as a free parameter.The final Stokes images (I, Q and U) were convolved us-ing a Moffat profile described by α = 4.08 and β = 1.86, forthe pole-on models. The values of α and β are determinedfrom the PSF for the point-like UKIRT photometric stan-dard star FS140 (s587-t) imaged during the same observingrun as EL29. c (cid:13) , 1–26 A. F. Beckford, P. W. Lucas, A. C. Chrysostomou and T. M. Gledhill
Figure 18.
The model fit for EL29 (a) polarized flux distribution,and (b) polarization vector map.
The parameters for the final model are given in Table 6. Thepolarized flux distribution and polarization vector maps forEL29 and the final model are shown in figure 18. The fi-nal model provides a quantitative fit to the degree of corepolarization, the structure of the polarization vectors andthe maximum polarization measured in each lobe. A quali-tative fit is provided to the polarized flux distribution. Thefit is achieved by modelling EL29 as a system that is neitherpole-on nor edge-on and has an outer disc radius of 50 AU.The inclination of the system was investigated for polarviewing angles in the ranges 0 ◦ < θ < ◦ (near pole-on)and from 80 ◦ < θ < ◦ (edge-on). The point-like nature ofEL29 in the Stokes I image was not successfully modelled;instead the focus was turned to modelling the polarized fluxdistribution, the polarization vector structure and the frac-tional polarization. The inclination angle of the system wasdetermined from previous observations of EL29 that haveindicated the presence of a flared disc with i ◦ (Boogert Table 6.
Parameters of the model fit to EL29 at the K band.Parameter ValueSystem Radius 1035 AUOuter Radius of Disc, r o
50 AUOptical Depth to protostar at H, τ K θ ◦ .1Azimuthal Viewing Angle, Λ 0 ◦ Base Radius of Cavity, R c
20 AUHalf-opening Angle of Cavity at base, θ c ◦ Grain Albedo 0.2Grain Axis Ratio 1.015Maximum Grain Size 0.70 µ m et al. 2002). Acceptable results for the polarized flux distri-bution and to the maximum polarizations in each lobe werefound for an inclination angle of i = 53 ◦ .1 +7 − .The values of the cavity base radius and half-openingangle investigated were based on the structure of the polar-ized flux distribution. A reasonable fit to the polarized fluxdistribution can be found for cavity base radii of and half-opening angles in the ranges 20 AU R c
30 AU and 20 ◦ θ c ◦ .The optical depth to the protostar was investigated be-tween 4 τ K
11. The final model requires that the opticaldepth is τ K = 7.75 +0 . − . Optical depths outside this rangewere not able to reproduce a reasonable fit to the polarizedflux distribution.The empirical density equation was used to determinethe density distribution. The power law index was capable ofproducing acceptable results for k = 0.0 +0 . . Increasing thepower law index (i.e. the vertical density gradient) resultedin a decrease in the prominence of the receding lobe. A moresevere flattening of the envelope was not able to reproducethe polarization structure of EL29.The position angle of the polarization disc was repro-duced using dust grains that were aligned with a toroidalmagnetic field. Grains aligned with an axial magnetic fieldwere capable of reproducing the vector structure in the outerparts of the polarised flux image but they were not able toreproduce the position angle of the polarization disc. To fitthe degree of core polarization, P K = 8.3%, required thatfor a maximum grain size of 0.70 µ m the grain axial ratio benot more than 1.03 for the adopted magnetic field structure.The larger grain axial ratio models and the spherical grainmodels resulted in core polarizations that were too high ortoo low, respectively.The dust grains were a mixture of silicates and amor-phous carbons. For the adopted system inclination the dustgrain albedo needed to be ω K ω K = 0.2, which provided the best fit to the po-larization data. Increasing the albedo results in an increasein the prominence of the western (receding) lobe.This model fails to reproduce the point-like flux dis-tribution that we observed for EL29, though we noted in § c (cid:13) , 1–26 ear-Infrared Imaging Polarimetry of Young Stellar Objects in rho-Ophiuchi the strongly peaked structure of EL29 suggests that theremay be a “hole” in the material along the line of sight tothe protostar. The near-infrared linear polarization data for a sample ofyoung stellar objects in the ρ Ophiuchi star-forming regionhave been analysed. The majority of the objects were spa-tially unresolved. Five of the objects were clearly associatedwith extended nebulosity: two of these are bipolar nebulaeand three have cometary morphologies.The extended objects have centrosymmetric vector pat-terns with a polarization disc over the core. A few of theunresolved objects have polarizations that were too small todetect, with upper limits of 1-2%.The wavelength dependence of the degree of polariza-tion suggests that dichroism is the dominant mechanismresponsible for the polarization of light in the unresolvedsources, which constitute the majority of the sample. (Thealternative is that these sources have complicated but unre-solved polarization structures). In the envelopes of the ex-tended sources the wavelength dependence indicates thatscattering dominates the generation of polarized light.The results of the linear polarimetry survey of ρ Ophwere compared to the infrared evolutionary status of theobjects. It was found that: • The distribution of the material around the samplesources does not seem to be strongly correlated with theevolutionary scheme determined by the infrared spectral en-ergy distributions. Point-like structures were seen for objectswith Class I, Class II and Class III designations. Five objectswere identified as being associated with extended nebulosity.The majority of the extended objects were Class I but onehad a Class II designation. • The near-infrared colour-colour diagrams show that thedegree of core polarization is correlated with the H - K andK - L colours. • A weak correlation is observed between the size of theextended nebulae and the core polarizations. • There is a positive correlation between the degree ofcore polarization and the evolutionary stage indicated by theinfrared spectral index. Typically, Class I objects show corepolarizations in the range 2% < P K < < P K <
6% are seen for theClass II objects, and the only Class III object in the sampleshows PK < • A weaker positive correlation between the maximumpolarization assessed over the envelopes and the infraredspectral index was also observed. • Generally, redder sources are associated with higher de-grees of core polarization. Those objects that have P K > > P K >
5% have the greatestrange of (H - K) colours (1 < (H - K) < K < K ).Similar correlations have been previously identified for young stellar objects in the Taurus region (Whitney, Kenyon& Gomez 1997). The results presented in this work for ρ Ophand by previous authors for Taurus reveal that the polariza-tion data appears to support the evolution of circumstelarmatter predicted in models of low-mass star formation.The shadow.f code was used to model three Class Iobjects in ρ Oph that represented a cross-section of themorphologies observed. The first was the bipolar objectYLW16A, the second was the cometary object WLY54, andthe third was the point-like object EL29. In each case a fitto the polarization data was found by assuming that theobjects could all be modelled as a bipolar system.The point-like source EL29 displays a bipolar polariza-tion pattern across the image profile. This was assumed tobe the result of diffuse bipolar nebulosity that was only de-tectable in polarized light. The presence of diffuse nebulosityhas been confirmed by more recent observations. Our modelfails to reproduce the point-like total flux distribution butsuccessfully fits the polarized flux distribution and the po-larization vector structure.The results of the modelling revealed that to reproducethe observed total flux distributions and polarization datathere were several properties common to the three systemsinvestigated. These properties are: • The dust grains that are responsible for the polarizationof the light are required to be smaller than the wavelengthof observation. The maximum grain size was found to be1.05 µ m. • The dust grains are either near-spherical or very weaklyaligned. More strongly dichroic grains can only be presentif the field structure varies along the line of sight to theprotostar in a manner that weakens the measured dichroicextinction effects. Our models assume perfect grain align-ment with the magnetic field. We find that grains with anaxial ratio of more than 1.03 led to significantly higher polar-izations than the data. In addition, grains with axial ratiosof 1.1 were found to result in little spatial variation in thepolarization across the source for parameters that were ca-pable of producing acceptable fits to the flux distributionsinvestigated. • Spherical grain models were successful at reproducingthe flux distributions and polarization vector patterns, butthey could not successfully reproduce the fractional polar-izations measured in the cores of the YSOs. Typically, thecore polarization provided by a spherical grain model was atenth of the observed value. • Given that some dichroic extinction is required to repro-duce the core polarizations, a toroidal magnetic field struc-ture is necessary in the central regions of the nebula in orderto reproduce the orientation of the polarization disc parallelto the accretion disc.
ACKNOWLEDGMENTS
We wish to thank the staff of UKIRT, which is operated bythe Joint Astronomy Centre on behalf of the UK Scienceand Technology Facilities Council (STFC). STFC has takenover the former astronomy functions of PPARC, the ParticlePhysics and Astronomy Research Council. This research wassupported by a PPARC PhD studentship awarded to AFB. c (cid:13) , 1–26 A. F. Beckford, P. W. Lucas, A. C. Chrysostomou and T. M. Gledhill
PWL is supported by the STFC via an Advanced Fellowshipat the University of Hertfordshire.
REFERENCES
Allen L.E., Myers P.C., Francesco J.D, Mathieu R., ChenH., Young E., 2002, APJ, , 993Andre P., Ward-Thompson D., Barsony M., 1993, APJ, , 122Andre P., Montmerle T., 1994, ApJ, 420, 837Aspin C., Casali M.M., Walther D.M., 1989, in ReipurthB., ed., ESO Conf. Work. Proc. 33, Low Mass Star For-mation and Pre-Main Sequence Objects, p.349, publishedby European Southern Observatory, Garching, GermanyBarsony M., Kenyon S.J., Lada E.A., Teuben P.J., 1997,ApJSS, 112, 109Bontemps S., Andre P., Terebey S., Cabrit S., 1996, A&A, , 858Bontemps S., Andre P., Kaas A.A., Nordh L., Olofson G.,Huldtgren M., Abergel A., Blommaert J., Boulanger F.,Burgdorf M., Cesarky C.J., Cesarky D., Copet E., DaviesJ., Falgrone E., Lagache G., Montmerle T., Perault M.,Persi P., Prusti T., Puget J.L., Sibille F., 2001, A&A, 372,173Boogert A.C.C., Hogerheijde M.R., Ceccarelli C., TielensA.G.G.M., van Dishoeck E.F., Blake G.A., Latter W.B.,Motte F., 2002, APJ, , 708Ceccarelli C., Boogert A.C.A., Tielens A.G.G.M., Caux E.,Hogerheijde M.R., Parise B., 2002, A&A, , 863Chrysostomou A.A., Clark S.G., Hough J.H., GledhillT.M., McCall A., Tamura M., 1996, MNRAS, 278, 449Chrysostomou A.A., Menard F., Gledhill T.M., Clark S.,Hough J.H., McCall A., Tamura M., 1997, MNRAS, 285,750Dolidze M.V., Arakeylan M.A., 1959, SvA, 3, 434Duchene G., Bouvier J., Bontemps S., Andre P., Motte F.,2004, A&A, , 651Elias J.H., 1978, ApJ, 224, 453Girart J.M., Rodriguez L.F., Curiel S., 2000, APJL, ,153Grasdalen G.L., Strom K.M., Strom S.E., 1973, ApJ, 184,L53Greene T.P., Young E.T., 1992, ApJ, 395, 516Greene T.P., Wilking B.A., Andre P., Young E.T., LadaC.J., 1994, APJ, 434, 614Hu´elamo N., Brandner W., Wolf S., 2007, Rev.Mex.A.A.Conf. Series., 29, 149Kamazaki T., Saito M., Hirano N., Kawabe R., 2001, APJ, , 278Lucas P.W., Roche P.F., 1997, MNRAS, 286, 895Lucas P.W., Roche P.F., 1998, MNRAS, 299, 699Lucas P.W., 2003, J. Quantitative Spectroscopy & Radia-tive Transfer, 79-80, 921Lucas P.W., Fukagawa M., Tamura M., Beckford A.F., ItohY., Murakawa K., Suto H., Hayashi S., Oasa Y., Naoi T.,Dori Y., Ebizuka N., Kaifu N., 2004, MNRAS, 352, 1347Lynds B., 1962, APJS, 7, 1Ressler M.E., Barsony M., 2001, AJ, 121, 1098Sekimoto Y., Tatematsu K., Umemoto T., Koyama K.,Tsuboi Y., Hirano N., Yamamoto S., 1997, APJL, ,63 Tamura M., Sato S., Suzuki H., Kaifu N., Hough J.H., 1990,ApJ, 350, 728Terebey S., Shu F.H., Cassen P., 1984, ApJ, 286, 529Vrba F.J., Strom K.M., Strom S.E., Grasdalen G.L., 1975,APJ, 197, 77VWhitney B.A., Hartman L., 1993, ApJ, 402, 605Whitney B.A., Kenyon S.J., Gomez M., 1997, ApJ, ,703Whitney B.A., Wood, K., Bjorkman J.E., Cohen M., 2003,ApJ, 598, 1079Wilking B.A., Lada C.J., 1983, APJ, 274, 698Wilking B.A., Lada C.J., Young E.T., 1989, ApJ, 340, 823Wilking B.A., Bontemps S., Schuler R.E., Greene T.P., An-dre P., 2001, APJ, , 357WYbarra J.E., Barsony M., Haisch K.E., Jarett T.H., SahaiR., Weinberger A.J., 2006, ApJ 647, L159Young E.T., Lada C.J., Wilking B.A., 1986, ApJ, 304, L45Zhang Q., Wootten A., Ho P.T.P., 1997, ApJ, 475, 713This paper has been typeset from a TEX/ L A TEX file preparedby the author. c (cid:13)000