First L-band Interferometric Observations of a Young Stellar Object: Probing the Circumstellar Environment of MWC 419
S. Ragland, R. L. Akeson, T. Armandroff, M. M. Colavita, W. C. Danchi, L. A. Hillenbrand, R. Millan-Gabet, S. T. Ridgway, W. A. Traub, P. L. Wizinowich
aa r X i v : . [ a s t r o - ph . S R ] J u l First L-band Interferometric Observations of a Young StellarObject: Probing the Circumstellar Environment of MWC 419
S. Ragland , R. L. Akeson , T. Armandroff , M. M. Colavita , W. C. Danchi , L. A.Hillenbrand , R. Millan-Gabet , S. T. Ridgway , W. A. Traub , P. L. Wizinowich Received ; accepted W. M. Keck Observatory, 65-1120 Mamalahoa Hwy, Kamuela, HI 96743;[email protected] NExScI, California Institute of Technology, 770 South Wilson Avenue, Pasadena, CA91125 Jet Propulsion Laboratory, California Institute of Technology, M/S 301-451, 4800 OakGrove Dr., Pasadena CA, 91109 NASA Goddard Space Flight Center, Exoplanets & Stellar Astrophysics, Code 667,Greenbelt, MD 20771 California Institute of Technology, Pasadena, CA 91125 National Optical Astronomy Observatories, P.O. Box 26732, Tucson, AZ 85726-6732 2 –
ABSTRACT
We present spatially-resolved K- and L-band spectra (at spectral resolutionR = 230 and R = 60, respectively) of MWC 419, a Herbig Ae/Be star. Thedata were obtained simultaneously with a new configuration of the 85-m baselineKeck Interferometer. Our observations are sensitive to the radial distribution oftemperature in the inner region of the disk of MWC 419. We fit the visibilitydata with both simple geometric and more physical disk models. The geometricmodels (uniform disk and Gaussian) show that the apparent size increases linearlywith wavelength in the 2-4 µ m wavelength region, suggesting that the disk isextended with a temperature gradient. A model having a power-law temperaturegradient with radius simultaneously fits our interferometric measurements andthe spectral energy distribution data from the literature. The slope of the power-law is close to that expected from an optically thick disk. Our spectrally dispersedinterferometric measurements include the Br γ emission line. The measured disksize at and around Br γ suggests that emitting hydrogen gas is located inside (orwithin the inner regions) of the dust disk.Keywords: stars: individual (MWC 419); stars: pre-main sequence; (stars:)circumstellar matter; stars: emission-line, Be; techniques: interferometric; instrumentation:interferometers
1. Introduction
Pre-main sequence (PMS) stars fall to the upper right of the main sequence on theHertzsprung-Russell (HR) diagram, journeying towards the main sequence via radialcontraction. For the first several million years, a PMS star is surrounded by a disk of gas 3 –and dust, left over from the early stage of star formation.The evolution of a disk is not well understood. Circumstellar disks provide theraw material for planet formation. Thus, understanding the evolution of a disk helps usunderstand planet formation. Clues to the physical conditions of planet formation andhence future suitability for life on planets other than Earth potentially can be inferred froma detailed characterization of inner young stellar object (YSO) disks.Herbig Ae/Be (HAeBe) stars are intermediate-mass PMS stars with spectral typesearlier than G0, located close to the zero-age main-sequence. Early spectral energydistribution (SED) modeling efforts (Hillenbrand et al. 1992) to explain the infraredexcess of HAeBe stars assumed (a) optically thick but geometrically thin circumstellardisks with inner optically thin holes, up to several stellar radii, in order to account forthe observed inflections in the 1-5 µ m region, and (b) relatively large accretion rates, onthe order of 10 − M ⊙ yr − , in order to account for the strength of the 3.8 µ m (L-band)excess. It was quickly pointed out (Hartmann et al. 1993) that at such a large massaccretion rate, the gas in the inner dust-free hole is expected to be optically-thick andto produce excess near-infrared emission, which is inconsistent with the dip seen inthe near-infrared SED. For the relatively low accretion rates later inferred for HAeBeobjects, disk models could not explain the observed SEDs in the near- infrared region.Puffed-up inner dust-rim models were introduced (Natta et al. 2001) and later refined(Dullemond et al. 2001; Isella & Natta 2005) to ameliorate the shortcomings of the classicaldisk models. These later models attribute the near-infrared excess to stellar radiationshining directly on the inner dust edge of the dust disk. Tuthill et al. (2001) directlyimaged the high luminosity YSO LkH α
101 and independently proposed that the bulk ofthe near-infrared emission arises in a hot ring located at the dust sublimation radius.Long-baseline interferometric observations provide the milliarcsecond angular 4 –resolution required to resolve the planet-forming structures immediately surroundingPMS stars. Such high-angular-resolution infrared observations are well suited for probingnear-circumstellar environments since the inner zones of the circumstellar disks andenvelopes emit primarily at near-infrared wavelengths. A large number of YSOs acrossthe luminosity range have now been spatially resolved at near-infrared wavelengths(Millan-Gabet et al. 2001; Colavita et al. 2003; Eisner et al. 2003; Eisner et al. 2004;Eisner et al. 2007; Monnier et al. 2005; Akeson et al. 2005; Millan-Gabet et al. 2007;Isella et al. 2008; Eisner et al. 2009; Tannirkulam et al. 2008), showing that indeed thecharacteristic sizes correlate strongly with central luminosity, lending support to the“puffed-up” inner dust rim paradigm, especially for Herbig Ae objects. However, it wasalso pointed out (Monnier & Millan-Gabet 2002; Eisner et al. 2004; Monnier et al. 2005)that most higher luminosity Herbig Be objects are considerably undersized compared tothe predictions of this model, and in better agreement with the “classical” models featuringoptically thick emission that extends inward very close to the central star.MWC 419 is a photometrically variable emission line star that is in many ways typicalof the HAeBe class, including illumination of a reflection nebula (Herbig 1960). MWC 419has spectral type B8 (Herbig 1960), luminosity 330 L ⊙ (Hillenbrand et al. 1992), anddistance 650 pc (Hillenbrand et al. 1992); it is also known as V594 Cas or BD+61 154.Narrow band H α imaging studies show unipolar large scale structures around MWC 419(Marston & McCollum 2008). Interestingly, these images show two lobes on the south-eastside of MWC 419 - the outer lobe extending to ∼ <
10 Myr pre-main sequence population.As an IRAS and ISO source, the infrared dust spectrum and spectral energydistribution of MWC 419 have been discussed by several authors (Brooke et al. 1993;Chen et al. 2000; Berrilli et al. 1992; Hillenbrand et al. 1992; Miroshnichenko et al. 1997;Pezzuto et al. 1997). No Spitzer data were obtained for this source but it has a weaksilicate emission at 10 µ m in ground-based data (Chen et al. 2000).MWC 419 was previously observed interferometrically using the Palomar TestbedInterferometer (PTI), and was spatially resolved in the K-band with a reported uniform-diskangular diameter of 3.34 ± ± ∼ µ m.The Keck Interferometer (KI), with ∼ µ m, can resolvethe inner disks ( ∼ µ m) and L-band (3.5-4.1 µ m). The simultaneous K- and L-band interferometricmeasurements enable us to probe different regions ( ∼ V ) data with variousYSO disk models. In Section 4 we discuss our results and in Section 5 we provide a briefsummary.
2. Observations and Data Reduction
KI is a near- and mid-infrared long-baseline interferometer consisting of two 10-mdiameter apertures separated by a B = 85-m baseline at a position angle of ∼ o east ofnorth. Both Keck telescopes are equipped with adaptive optics. The maximum resolution ofour KI observations is λ /2 B ∼ ∼ V measurements since 2002 (Colavita et al. 2003). We obtained first light with this new 7 –instrument in April 2008. The L-band limiting magnitude of 6 is set by the requirementfor a broadband phase signal-to-noise ratio (SNR) of 10 in the 10 ms fringe-tracker frame.The total integration time on source is typically 200 sec, which provides good SNR for V measurements in the 10 spectral channels.The L-band instrument uses a PICNIC focal plane array detector with a 5 µ m cutoffwavelength, while FATCAT uses a HAWAII array. The detector electronics are similar forboth instruments. Details of this L-band instrument are given in Ragland et al. (2008).For the current work a prototype multi-wavelength observing capability was used,enabling simultaneous K- and L-band observations of our science target. In thisconfiguration, the telescope pupil is split into left and right-halves at the dual-star modulesof both telescopes and routed through separate coude paths utilizing the existing beam-traininfrastructure for the nulling mode. The left-half pupils of both telescopes are combined bythe K-band instrument and the right-half pupils are combined by L-band instrument.The K- and L-band science instruments each have 2 complementary interferometricoutputs with different spectrometers on each. For L-band, these are a (pseudo) broad-band,and a low dispersed mode (10 channels across the L-band; R = 60). For K band, these area broad-band, and a medium dispersion mode (42 channels across the K-band; R = 230).The field-of-view of the instrument, defined by the single mode fibers that couple thatlight to the detector array, is ∼
55 mas for the K-band and ∼
93 mas for the L-bandobservations. These field restrictions were enforced in our modeling work.The ZABCD algorithm (Colavita et al. 2003) is used for fringe tracking and sciencemeasurements. In this procedure five reads are made while the fast delay line scans over onewavelength (2.2 µ m for the K-band and 3.7 µ m for the L-band); explicitly, the reads are thereset pedestal (z), followed by four non-destructive reads (a, b, c, d) spaced at λ /4 intervals 8 –as the detector integrates up. For each quarter-wave bin we then calculate A = a − z , B = b − a , etc. These values are used to estimate the square of the fringe visibility ( V )for science, and the fringe phase and SNR for fringe tracking. We apply corrections to thebin data to account for detector bias and for differences between the length of the delayscan and the actual wavelength. Residual instrument bias and atmospheric seeing effectsare corrected by observing a calibrator of known V under similar observing conditions.Essentially, we estimate instrument transfer function using the bracketing observations ofcalibrator stars with known angular diameters using a weighted averaging scheme thatconsiders the time and sky proximity of calibrators relative to the target .The observations reported here were taken on the night of UT 19 August 2008. Weobserved two calibrators - HD 1843 and HD 6210 - under similar observing conditions asthe science target to calibrate the science data. The adopted angular sizes of the calibratorsare 1.0 ± ± o to 33 o ) and projected baselines (76.1m to 77.8m), and hence no attempt was made toderive an inclination angle and position angle of the disk. For the purpose of this article,we assume that the disk is face-on. Thus, the sizes reported here are the values along theposition angle of about 29 o , and the actual size of the disk could be larger depending onthe inclination angle and position angle. Our K-band measurements, taken at the positionangle of about 29 o , gives similar disk size as that of previous PTI observations, taken at theposition angle of about 83 o . Hence, a face-on disk for MWC 419 is a resonable assumptionin the absence of necessary high spatial resolution observations. 9 –
3. Analysis
In this section, we fit our measurements with different models of increasing complexityand increasing physical realism.The measured V includes contributions from the central star (V ∗ ) and the brightcircumstellar disk. The visibility-square (V disk ) of the circumstellar disk of MWC 419 isobtained from the measured data by removing the contributions from the central star(Millan-Gabet et al. 2001) using V measured = (cid:18) F ∗ V ∗ + F disk V disk F ∗ + F disk (cid:19) (1)The adopted disk-to-star flux ratio (F disk /F ∗ ) is 12 and 40 in K and L based on ourSED analysis discussed in Section 3.3. The central star is assumed to be unresolved for ourobservations (i.e., V ∗ = 1.0), which is a reasonable assumption for a B8 star at a distanceof 650pc. These corrections yield values of V disk that are smaller than the total V by0.04 ( ∼ ∼ V measurementsare consistent with earlier K broadband measurements and provide additional spectrallyresolved information within the K-band spectral region.The visibilities of the disk models (Section 3.2 and 3.3) are computed by numericallysumming the contributions from annular rings of infinitesimally small widths and weightingthem by their respective flux contributions; for this purpose, we divided the disk radiallyinto 5000 annular rings, in logarithmic scale. The normalized visibility for a uniformly brightannular ring with an inner diameter of θ in , an outer diameter of θ out and θ in . θ . θ out , canbe written as V ann ( B, λ, θ ) = (cid:18) λπB ( θ out − θ in ) (cid:19) ( θ out J ( πBθ out /λ ) − θ in J ( πBθ in /λ )) (2) 10 –where J is the first-order Bessel function, B is the projected baseline and λ is thewavelength of observation.The flux from such an annular uniform ring can be written as F ann ( λ, θ ) = π P ( T, λ ) ( θ out − θ in ) (3)where P ( T, λ ) is the Planck blackbody function and T is the mean temperature of theannular ring.Now, the visibility of the total disk can be written as V disk ( B, λ ) =
X (cid:18) V ann ( B, λ, θ ) × F ann ( λ, θ ) P F ann ( λ, θ ) (cid:19) (4)where P F ann ( λ, θ ) is the total disk flux derived by summing all the annular rings of thedisk in order to normalize the flux. We adopted a nonlinear least-squares fitting method tominimize the chi-square for all our modeling work.For our SED analysis, photometric measurements from the literature are used, afterinterstellar extinction corrections using the extinction law of Cardelli et al. (1989). Theextinction for MWC 419 in the V-band is assumed to be A v = 2.1 (Hillenbrand et al. 1992).The models presented in this article assume face-on geometry for the disk. A disk withan inclination angle φ could predict a larger inner disk radius depending on the inclinationangle and the position angle of the projected interferometer baseline on the sky. In thiscase, SED models would underestimate the flux at all wavelengths by a factor of cos ( φ ). Inother words, classical accretion disk models would have a factor of cos ( φ ) larger accretionrate to explain the SED data and power-law models would have a relatively larger disktemperature at all radial distances. 11 – The measurements presented in this article consist of spectrally dispersed data withinthe K- and L-bands. As stated above, the K and L visibilities corrected for the stellarcontribution are different, suggesting a wavelength dependence to the size of the spatiallyresolved emission. We can use the spectrally dispersed data to investigate in more detailthis wavelength dependency. We choose three geometrical models, namely, uniform-disk,Gaussian distribution and ring models, to fit to our measurements.The complex visibilities for a normalized pole-on uniform-disk and a Gaussiandistribution can be written as follows: V UD ( B, λ ) = 2 J [ πBθ UD /λ ] πBθ UD /λ (5) V Gaussian ( B, λ ) = exp (cid:18) − ( πBθ F W HM /λ ) ln (cid:19) (6)Here θ UD is the uniform-disk angular diameter of the circumstellar disk, θ F W HM is thefull-width at half maximum (FWHM) of the Gaussian distribution, and other terms asabove. See Eqn. 2 for the ring model.To start, we fit the data with a wavelength-independent geometrical model (Figure 1).The derived uniform-disk, Gaussian and ring angular sizes, from simultaneous fits to Kand L band data, are 3.59 ± ± ± χ R ) values are 36.1, 35.0 and 38.3respectively. As indicated by the very poor χ R values, the uniform disk, Gaussian and ringmodels fit neither the K and L data in the mean, nor the slope of the visibilities throughouteither the K- or L-bands.As the wavelength-independent size models fail to fit the measurements, we fit the 12 –Fig. 1.— Wavelength-independent model fits. The multi-wavelength measurements (dia-mond symbols) are plotted against spatial frequency. Also shown are geometrical model fitsto the data; the dotted-line, dashed-line and continuous line refer to uniform-disk, Gaussianand ring models respectively. The poor model fits indicate that these simple models are aninadequate description of the data. 13 –data with wavelength-dependent sizes. The results of these geometrical model fits to themeasured data points are shown in Figure 2.Fig. 2.— Wavelength-dependent linear fits. Left:
The derived uniform-disk angular sizes asa function of wavelength are shown in diamond symbol along with error bars. A linear fitto these sizes (dotted line) is also shown here. The dip seen at around 2.17 µ m is due to thepresence of compact Br γ emission line (see Section 4 for details). Middle:
Same as the leftfigure, but a zoomed view of the Br γ emission line region. Right:
Same as the left figureexcept for a Gaussian distribution.The observed wavelength dependence in the 2-4 µ m region has a simple linearrelationship to first order. The parameters of a linear fit to the derived apparent wavelengthdependent diameters are given in Figure 2. The χ R value for this model fit is 2.1. Themeasured uniform-disk diameter in the center of the L-band ( θ L = 5.04 mas) is ∼ θ K = 3.49 mas). The steep slopes of these linearrelations suggest that the 2-4 µ m emission source must be extended with strong radialtemperature dependence. Earlier multi-color (near- and mid-infrared) interferometry of twoYSO disks have shown similar wavelength dependence (Acke et al. 2008; Kraus et al. 2008).In the following section we explore more complex extended disk models to explain ourobservations. 14 – We compared our interferometric data with an accretion disk model by Hillenbrand etal. (1992) based on an SED analysis. In this classical accretion disk model, the temperaturedistribution of dust is derived by combining the contributions from a radiatively heatedreprocessing component T rep and a viscously heated accretion component T acc . Theparameters of the disk are the following: The inner disk (hole) radius is R in = 0.22 AU (10R ∗ ), the outer disk radius is 62 AU and the mass accretion rate ( ˙ M ) is 1.98 × − M ⊙ yr − . The stellar radius is 4.8R ⊙ , the stellar effective temperature is 11,220K, the distanceto the star is 650pc and the stellar mass is 5.3M ⊙ . The inner disk temperature is 2480Kand the outer disk temperature is 40K. The resulting flat accretion disk model highlyoverestimates V in the K- and L-bands. We get a reduced-chi-square ( χ R ) value of 664 forthe interferometric data. The model results are given in Table 1. The physical reason thatthe classical accretion disk model fails is that it creates too much flux in the inner region ofthe disk because of the added accretion luminosity, making the angular size much smallerthan observed.We have also fit visibility data with a similar accretion disk model (Figure 3) bytreating the inner disk (hole) radius as a free model parameter. In addition, the massaccretion rate was increased by a factor of 1.5 (i.e. 2.97 × − M ⊙ yr − ) in orderto fit SED data. The χ R value for this model fit to interferometric measurements is2.59. The derived inner disk (hole) radius is 1.66 ± ∼ ∗ ) andthe inner disk temperature is 1457K. The outer disk radius is fixed at 50 AU wherethe disk temperature is 51K. This model fits the K band data well, but is about3 σ too high in V compared to the L band data. The extremely large accretionluminosity required by the SED data poses challenges to these two accretion diskmodels. Thus the classical model that could fit the SED fails here, just like in the previous 15 –interferometric studies of Herbig Ae/Be disks (Millan-Gabet et al. 2001; Eisner et al. 2004).Moreover, it is not sufficient to adjust the inner dust radius and temperature alone. Amodification of the disk temperature profile is needed as in the case of more recent studies(Eisner et al. 2007; Acke et al. 2008; Kraus et al. 2008); this is discussed in the followingsection.Fig. 3.— Classical accretion disk model fits. Interferometric data points are shown witherror bars. Shown in continuous line is an accretion disk model with the inner radius of thedisk (hole size) treated as a free model parameter. The outer radius of the disk is fixed at50 AU. Right:
Photometric data taken from the literature are shown along with the SEDmodel for the same accretion disk model. The dashed line shows the blackbody spectraldistribution of the central star and the dashed-dotted line shows the blackbody emissionfrom the disk. 16 –
The wavelength dependence of the measured visibilities and SED data shown aboveimplies an inverse T ( r ) relationship. We fit our data with a simple model with a power-lawtemperature gradient of functional form T ( r ) ∝ r − α , where r is the radial distance fromthe central star and α is the power-law parameter. The model V is derived using Eqn. 2and Eqn. 3 with a power-law temperature distribution. The radius of the inner disk and α are treated as free parameters. The temperature of the inner disk is fixed at 1800K andthe optical depth at 1 for the entire disk, in order to satisfy the SED data; a larger valuefor the optical depth overestimates the disk flux. The resultant model fits are shown inFigure 4. The derived inner disk radius is 1.47 ± ∼ ∗ ) and thederived value for α is 0.71 ± χ R for this model fit to our interferometric data is 0.73, and the sensitivityof χ R to the model parameters is shown in Figure 5. As the power-law disk model is asatisfactory fit, we consider it the simplest model that explains the data. However, one canalso conceive of more complex and physically plausible models such as optically thin innerdisk regions (holes) surrounded by optically thick outer disks, perhaps including puffed upinner rims. Our initial exploration of such models suggests that they can also fit our data,but the interferometric data themselves do not drive us to such complex models. In this section we investigate whether the data can be explained, without invokinghot/warm circumstellar dust, by the effect of a companion star. The visibility model for abinary system can be written as 17 –Fig. 4.— Power-law disk model fits.
Left:
Interferometric data points are shown witherror bars. The solid line is a power-law temperature gradient disk model. The power-lawparameter, the inner disk radius and the inner disk temperature are free parameters. Theouter radius is fixed at 20AU.
Right:
Photometric data taken from the literature are shownwith error bars. The solid line is the SED model for the same accretion disk model. Thedashed and dashed-dotted lines are the SEDs of the star and disk. 18 –Model R in T in α ˙ M χ R (AU) (K) ( × − M ⊙ yr − )Classical accretion disk 0.22 2480 0.75 1.98 664(parameters fixed at published values)Classical accretion disk 0.538 ± in and ˙ M varied)Modified power-law disk 0.477 ± ± χ R values are for the interferometer data (only). http://nexsci.caltech.edu/software/V2calib/ V binary ( λ ) = q V p + R V s + 2 | V p | | V s | R cos ((2 π/λ ) B · s )1 + R (7)where R is secondary-to-primary flux ratio, V p and V s are visibilities of primary andsecondary components, and s is the binary separation.The derived parameters from a binary model fit to our measurements are the following:projected binary separation along the position angle of 29 o (east of north) = 5.12 ± ± ± χ R valueis 0.94. In this scenario, the secondary component would have the same K-L color of 1.34as the primary YSO disk suggesting that the potential companion were a YSO disk withsignificant IR excess. In this case, our observations would have resolved the disk around 19 –Fig. 5.— Contour map of χ R as a function of the power-law parameter α and the inner diskradius. The contour lines - from inner to the outer - refer to χ R values of 1.0, 1.2, 1.4, 1.6,1.8 and 2.0.the companion. Hence, we also fit our measurements by fixing the companion size to be 2mas in order to explore this possibility. The derived parameters for this binary model arethe following: projected binary separation along the position angle of 29 o (east of north)= 5.23 ± ± ± χ R value is 1.57, but additional observations at otherbaseline orientations would provide significant additional constraints on the binary model.The derived flux ratio of ∼
12 suggests a cooler stellar companion with a smaller inner diskradius compared to the primary star. However, such a companion model would overestimateSED at the shorter wavelengths because of the flux from the companion photosphere.Hence, a binary scenario is not favored as an explanation for our measurements. 20 –Fig. 6.— Binary model fits.
Left:
The multi-wavelength measurements are plotted againstspatial frequency. Also shown is the best fit binary model (continuous line) assuming anunresolved companion.
Right:
Same as the previous caption but for a companion of 2 masdiameter. 21 –
4. Discussion
The KI measurements presented here agree with previous observations from PTI ofthe K-band size of MWC 419 (Wilkin & Akeson 2003). The addition of the spectrallydispersed K-band information and new L-band observations, also spectrally dispersed,provide powerful new constraints on the physical structure of the material surrounding thecentral star. As shown the left-most panel of Figure 2, the wavelength dependence of thesize is apparent within both the K and L bands; the combination of the two bands allowsan even tighter constraint on the temperature power-law than possible using either dataset alone. The radial temperature profiles of the disk models presented in Sections 3.2 and3.3 are shown in Figure 7. The temperature profile of the best fit power-law model is verysimilar to that of a classical, geometrically thin accretion/reprocessing disk ( α = − . − M ⊙ yr − in order to fit theSED data. Simple estimates of the accretion rate for this star range from 5 × − to10 − M ⊙ yr − based on mass loss data and upper limits to radio emission at 3.6 cm(Kurchakov et al. 2007; Boehm & Catala 1995; Skinner et al. 1993). The necessity of anunrealistically large accretion rate in the accretion disk model is probably an artifact of thephysically unrealistic assumption of a zero-thickness disk which causes the re-processing ofradiation to be very inefficient. Hence, we do not favor an accretion disk model for theMWC 419 disk.A power law temperature profile with radius in which the radial exponent is afree parameter is a very good fit to the data, because such a simple model roughlyaccounts for the vertical scale height as a function of radius. The derived slope of -0.71is consistent with a “flat” disk geometry that one tends to find for Herbig Be stars(Acke et al. 2008; Kraus et al. 2008). 22 –Fig. 7.— The radial temperature profiles of various disk models. The dotted and dashedlines refer to the two cases of the classical-accretion disk model (Section 3.2; Table 1). Thedotted-dashed line refers to the power-law temperature gradient disk model (Section 3.3;Table 1). 23 –Earlier studies (Monnier & Millan-Gabet 2002; Eisner et al. 2004) using broadband,usually single-wavelength, interferometric data, recognized a difference in the near-infraredsize vs. luminosity behavior of high luminosity objects (pre-main sequence Be) comparedto lower luminosity ones (pre-main sequence Ae), the former being more consistent with“classical disk” models. This has been revisited most recently by Vinkovi´c & Jurki´c(2007), who use a model-independent comparison of visibility to scaled baseline and finda distinction between low-luminosity ( . L ⊙ ) and high-luminosity ( & L ⊙ ) YSOdisks where the luminosity break point corresponds to approximate spectral type B3-B5.These authors then modelled this observable with a ring and halo model for low luminosityHerbigs, a halo alone for T Tauris and an accretion disk for high luminosity Herbigs.However, multi-wavelength interferometric studies have not always supported theseconclusions when objects are modeled in detail. Kraus et al. (2008) observed the B6 starMWC 147 in H, K and N bands. They performed Monte Carlo modeling and found thatthe interferometric and SED date were not well fit with a standard, irradiated accretiondisk alone, but were well fit with the standard disk plus emission from hot, optically thickgas within the innermost radius of the dust disk. In contrast, Acke et al. (2008) observedthe B1.5 star MWC 297 in H, K and N, and found that they could not fit the data witha single accretion disk, even with the radial temperature exponent as a free parameter.Instead, they used a three component geometric model, with characteristic blackbodytemperatures of 1700, 920 and 520 K. Our results show that MWC 419 (B8, 330 L ⊙ ) has thedisk characteristics of a high-luminosity object in the categories of Monnier & Millan-Gabet(2002) but a luminosity lower than the break point of 10 L ⊙ identified by Vinkovi´c & Jurki´c(2007). Examining Figure 2 of Vinkovi´c & Jurki´c (2007) clearly shows MWC 419 fits wellwithin the population of lower luminosity Herbigs in their model-independent comparison.Their physical interpretation of the low-luminosity Herbig group is an optically thick diskwith an optically thin dust sublimation cavity and an optically thin dusty outflow. Our 24 –multi-wavelength results do not support grouping MWC 419 in that physical model, butperhaps incorporating multi-wavelength data in the model-independent visibility groupingswould have produced a different result. The new multi-wavelength data set presented hereprovides significant new information to aid in determining the physical conditions of theseyoung stars, and our data show that the disk surrounding the B8 star MWC 419 is closerin physical characteristics to the more massive Be stars than to the Herbig Ae and T Tauristars.Another result from the spectrally dispersed data is that the measured V is marginallyhigher at 2.17 µ m than at adjacent wavelengths. This V feature is seen in all 4 independentdata sets. As this is the wavelength of Br γ , and Brackett recombination emission linessuch as Br γ and Br 10-20 are observed in the spectrum of MWC 419 (Harvey 1984), weattribute the difference in line and continuum visibilities to a more compact Br γ emittingregion. Similar behavior in the apparent size in the Br γ line compared to the neighboringcontinuum has been seen before in spectrally resolved interferometry from both VLTI andKI (Kraus et al. 2008; Eisner et al. 2009). The derived uniform-disk diameter for the Br γ emitting region is 3.33 ± ∼
4% less than the continuum size (3.46 mas)around Br γ (Figure 2). The significance level of this detection of the Br γ emitting regionwith respect to the continuum region is ∼ σ . The actual size of the Br γ emission lineregion could be much smaller than the value reported here because of the coarse spectralresolution of our measurements. Detection of Br γ emission inside the innermost dust radiussuggests that the disk is optically thin in the inner region where atomic hydrogen gas exists.
5. Summary
This article reports the first milliarcsecond angular resolution observations of aHAeBe star (MWC 419) providing L-band, as well as simultaneously obtained K-band 25 –data, both spectrally dispersed. This multi-wavelength observational capability is wellsuited to probing the temperature distribution in the inner regions of YSO disks, whichis very important for distinguishing between models and gaining insight into the threedimensional geometry of the inner disk. Such measurements could distinguish discretespatial distributions, such as dust-rims, from relatively-smooth spatial distributions, suchas classical accretion disks, based on their distinct wavelength dependent disk sizes. Inaddition, interferometric measurements in the relatively unexplored L-band provide neededconstraints to the disk/envelope geometry and temperature structure.Simple geometrical pole-on disk models are used to infer a linear relationship betweenthe derived object size and wavelength in the 2-4 µ m region, suggesting a simple physicalmodel for the disk. The steep slopes of these linear relations imply that the disk is extendedwith a radial temperature gradient. We find that the accretion disk model of Hillenbrand(1992) derived from SED analysis does not fit our interferometric measurements. Anupdated accretion disk model with accretion rate 1.5 times larger and inner cavity 2.4 timeslarger fits the K band data well, but lies 3 σ above the L band data. However, both of theseclassical accretion disk models predict an unrealistically large accretion rate of ∼ × − M ⊙ yr − to fit the SED data. A power law temperature profile with a slightly shallowerslope of -0.71 fits both the spectrally dispersed interferometric measurements and the SEDsatisfactorily, suggesting a relatively flat disk geometry for MWC 419. The measured disksize at Br γ reveals the presence of compact emitting hydrogen gas in the inner regionsof the disk. A more complete sample of YSO disk observations with adequate wavelengthand (u,v) coverage, plus detailed radiative transfer modeling, are required to address theintriguing inner disk geometry in these sources 26 – Acknowledgment
Keck Interferometer is funded by the National Aeronautics and Space Administration(NASA). Observations presented were obtained at the W. M. Keck Observatory, whichis operated as a scientific partnership among the California Institute of Technology, theUniversity of California, and NASA. The Observatory was made possible by the generousfinancial support of the W. M. Keck Foundation. We thank E. Appleby, B. Berkey, A.Booth, A. Cooper, S. Crawford, W. Dahl, C. Felizardo, J. Garcia-Gathright, J. Herstein, R.Ligon, D. Medeiros, D. Morrison, T. Panteleeva, B. Smith, K. Summers, K. Tsubota, G.Vasisht, E. Wetherell, for their contributions to the instrument development, integrationand operations. S. Ragland also thanks M. Hrynevych, M. Kassis and J. Woillez for usefuldiscussions. We thank the referee for constructively critical comments that have helped usto significantly improve the paper. 27 –
REFERENCES