Spectro-interferometry of the Be star delta Sco: Near-Infrared Continuum and Gas Emission Region Sizes in 2007
R. Millan-Gabet, J. D. Monnier, Y. Touhami, D. Gies, E. Hesselbach, E. Pedretti, N. Thureau, M. Zhao, T. ten Brummelaar
SSpectro-interferometry of the Be star δ Sco: Near-InfraredContinuum and Gas Emission Region Sizes in 2007
R. Millan-GabetCalifornia Institute of Technology, NASA Exoplanet Science Institute, Pasadena, CA91125, USA
J. D. MonnierDepartment of Astronomy, University of Michigan, MI 48109, USAY. Touhami and D. GiesCenter for High Angular Resolution Astronomy and Department of Physics andAstronomy, Georgia State University, Atlanta, GA 30302, USAE. HesselbachUniversity of Toledo, Department of Physics & Astronomy, Toledo, OH 43606, USAE. Pedretti and N. ThureauSchool of Physics and Astronomy University of St Andrews, St Andrews, Fife, KY16 9SS,Scotland, United KingdomM. ZhaoJet Propulsion Laboratory, 4800 Oak Grove Drive, M.S. 169-327, Pasadena, California91101, USAandT. ten Brummelaar and the CHARA GroupCenter for High Angular Resolution Astronomy, Georgia State University, PO Box 3969,Atlanta, Georgia 30302-3969, USA 2 –Received ; acceptedAccepted by ApJ September 2 2010 3 –
ABSTRACT
We present near–infrared H and K –band spectro–interferometric observationsof the gaseous disk around the primary Be star in the δ Sco binary system,obtained in 2007 (between periastron passages in 2000 and 2011). Observationsusing the CHARA/MIRC instrument at H –band resolve an elongated disk witha Gaussian FWHM 1 . × .
91 mas. Using the Keck Interferometer, the sourceof the K –band continuum emission is only marginally spatially resolved, andconsequently we estimate a relatively uncertain K –band continuum disk FWHMof 0 . ± . I λ . µ m and Br γλ . µ m, is clearly detected, with ∼
10% lower visibilities than those ofthe continuum. When taking into account the continuum/line flux ratio thistranslates into much larger sizes for the line emission regions: 2 . ± . . ± . I and Br γ respectively. Our KI data also reveal a relativelyflat spectral differential phase response, ruling out significant off–center emission.We expect these new measurements will help constrain dynamical models beingactively developed in order to explain the disk formation process in the δ Scosystem and Be stars in general.
Subject headings: techniques: high angular resolution — stars: emission-line, Be —stars: individual ( δ Sco)
1. Introduction
Be stars are massive, rapidly rotating stars, on or near the main–sequence, which, atleast at some time during their evolution, exhibit peculiar observational characteristics:infrared excesses compared to photospheric levels, emission lines, partial polarization ofradiated light, and variability of all those properties on many time–scales, from weeks todecades. It is generally understood that the observational characteristics that define Bestars arise from the development of a thin disk of gas in the equatorial plane. The precisephysical processes that govern its formation are, however, not well understood. Processesthat are believed to dictate or participate in the formation of a gaseous equatorial diskinclude: equatorial mass–loss due to near–critical rotation, radiatively drive winds, andphotospheric pulsations (see e.g. review by Porter & Rivinus (2003). The role, if any, thatbinarity plays in the Be phenomenon is also unclear, although many Be stars may havebeen spun up in the past by mass exchange (Peters et al. 2008).The star δ Sco (HD 143275, HIP 78401, HR 5953; V ∼ ∼ ∼ ∼ . δ Sco are similar to those of typical Be stars, and are satisfactorily explained with models ofan equatorial gaseous disk (Banerjee et al. 2001; Miroshnichenko et al. 2001; Carciofi et al.2006). Many intriguing and unexplained features remain however, in particular pertaining 5 –to the variability, over several distinct time–scales, of the photometric and spectroscopicactivity indicators.Here we present new observation of the δ Sco primary, obtained in 2007, seven yearsafter the 2000 periastron passage and three years before the following one in 2011. Ourspectro–interferometric observations spatially resolve distinct sources of emission: the H and K –band continuum (star + gaseous disk), and regions of He I and Br γ emission.
2. Observations and Observables2.1. CHARA/MIRC
Observations of δ Sco were made on UT 2007 May 10–12 (JD 2,454,231–233) atthe CHARA interferometer (ten Brummelaar et al. 2008), using the Michigan InfraRedCombiner instrument (MIRC, Monnier et al. (2008)). MIRC operates in H –band( λ = 1 . µ m, ∆ λ = 0 . µ m). For these observations we used a mode which disperses theimage plane fringes over 8 spectral pixels, in order to avoid bandwidth–smearing coherencelosses We used four CHARA telescopes resulting in six simultaneous baselines, with thefollowing average projected lengths and position angles (measured East of North): E1W1(301.7 m, 203 . . ◦ . . ◦ . . ◦ . . ◦ . . ◦ . . ◦ δ Sco were interleaved with observationsof calibrator stars used to monitor the instrument’s transfer function. The calibrators usedwere: HD190327 (V=5.5, H=3.3, K0III, θ = 1 . ± .
014 mas), HD149757 (V=2.6, H=2.7, Note that HD149757 = zeta Oph is sometimes a Be star with H-Balmer emission. How-ever, according to the Database of Be Star Spectra (http://basebe.obspm.fr/basebe/) ob- 6 –O9V, θ = 0 . ± .
05 mas), and HD164259 (V=4.6, H=3.7, F2IV, θ = 0 . ± .
03 mas);where their uniform disk angular diameters ( θ ) and their errors were estimated using thesurface brightness relations of Barnes et al. (1978).Reduction and calibration of the MIRC data were performed using its standardpipeline, as described in Monnier et al. (2007). While an analysis of multi–epoch closurephases is under way (including the effects of the companion orbital motion and investigationof possible off–center disk emission), for the purposes of the work presented here we onlyconcern ourselves with the visibility amplitudes, in order to measure the H -band size of thedisk surrounding the primary star.Using the most recent orbital solutions (Mason et al. 2009; Tango et al. 2009), wecompute that the angular separation of the δ Sco companion to be 182 mas at this epoch.This is inside the MIRC field of view (FOV ∼ . < Observations of δ Sco were made on UT 2007 July 03 (JD 2,454,285) at the KeckInterferometer (KI, Colavita et al. (2004)). We used a low spectral dispersion mode, whichprovides 42 spectral channels across the near–infrared K –band ( λ = 2 . µ m, ∆ λ = 0 . µ m, R = 200, see Eisner et al. (2007)). The width of each spectral channel is therefore ∼ . ◦ θ = 1 . ± .
07 mas) and HD135742 (V=2.6, K=2.9, B8V, θ = 0 . ± .
05 mas).Reduction and calibration of the interferometer data (visibility amplitudes, differentialphases and spectral fluxes) were done using standard packages provided by the NExScI .The flux spectra (also obtained from the interferometer data) were normalized to the fluxin the longest wavelength channel (2 . µ m), and corrected by dividing by the averageof the normalized calibrator spectra. The spectral differental phase data are referencedto the phase of the broadband K –band channel (used to “fringe track”, i.e., to stabilizethe interference fringes against the optical path differences caused by the atmosphere andinstrument). Two independent observations of δ Sco yield V that agree to better than 1%.Nevertheless, per project documentation describing the performance of the KI instrument,we added an external error of 0.03 to the formal statistical errors in the absolute calibration i.e., Kvis and nbCalib ; http://nexsci.caltech.edu/software/KISupport/ 8 –of V .The angular separation of the δ Sco companion was 180 mas at the epoch of the KIobservations, much larger than the 50 mas field FOV of the KI near–infrared instrument(set by the acceptance angle of the single–mode fiber which feeds the fringe tracker camera,and also matched to the KI telescopes diffraction limit) . The companion therefore adds nocoherent or incoherent flux, and we may ignore its existence for the purposes of interpretingthe KI data.The calibrated visibility amplitude, differential phase and flux data are shown inFigure 2. The following are the main features one can see in these data: (1) Spectralfeatures are clearly seen in the visibility and flux data, attributed to He I (2 . µ m)and Br γ (2 . µ m) gaseous line emission. We note that He I and Br γ emission werealso observed in K -band spectra made shortly after after the last periastron brightening(Banerjee et al. 2001). (2) The visibilities at the wavelengths of these lines are lower than that at wavelengths of continuum–only emission (by ∼ larger than the continuum emission region (actually, by a largefactor, as will be seen in the next section). (3) The K –band continuum is only slightlyresolved ( V = 0 . ± . − , we expect a FWHM(Br γ ) ∼ × . http://olbin.jpl.nasa.gov/data/ . 9 –
3. Sizes of the Continuum and Line Emission Regions3.1. H –band Continuum (CHARA/MIRC) We fit the CHARA/MIRC observations with a model consisting of a central primarystar, an elliptical Gaussian brightness (representing the gaseous circumstellar disk), and acompanion star. For equations describing the interferometer’s response to these standardmorphologies, see e.g. Berger & Segransan (2007). Here and in the following sections wechoose a Gaussian brightness because it has been shown in simulations (e.g. Stee et al.(1995)) and experimentally (Tycner et al. 2006) to be a good approximation for the diskbrightness (and much better than e.g. a uniform disk or ring).For the primary photosphere, we adopt a uniform disk angular diameter of0 . ± .
06 mas. This value represents an average of estimates obtained by two differentmethods: the pre–active primary stellar parameters of Carciofi et al. (2006), and the surfacebrightness–color relations of Barnes et al. (1978), and the 12% uncertainty representsthe scatter among the various estimates. For the companion, we estimate a uniformdisk angular diameter of 0 . ± .
03 mas, by following Carciofi et al. (2006) in adoptingan early B spectral type and a secondary to primary flux ratio at all wavelengths of F s /F p = 0 . − .
25 (based on the measured visual magnitude difference; Bedding (1993);Tango et al. (2009)), and again using the Barnes et al. (1978) relations.Thus, the free parameters in our fit are the primary to total H –band flux ratio( F p /F T ) H , the elliptical Gaussian brightness FWHM along the major and minor axis(FWHM M and FWHM m ), its orientation on the sky (position angle, PA M ), and thecompanion coordinate offsets. The baseline–dependent bandwidth smearing effect on thesignature form the companion is fully accounted for in the modelling. The best–fit solutionplaces the companion at a separation of ∼
190 mas, consistent with expectations from 10 –the orbital solution. However, because the long baselines contain no information aboutthe companion, the data cannot constrain its position angle (as indicated by the fact thatfitting the different epochs separately yields very different results). We note however thatthis does not affect the robustness of the fit to the disk parameters.The best fit solution (Table 1) has ( F p /F T ) H = 0 . ± .
05, FWHM M = 1 . ± .
16 masand FWHM m = 0 . ± .
12 mas. We note that in the thin disk approximation the ratioof minor to major axis sizes yield the inclination, and the MIRC fit implies i = 39 ± ◦ ,consistent with i = 38 ◦ from Carciofi et al. (2006). As can be seen in Figure 1, the ( u, v )coverage is limited along most directions (due to the Southern declination), resulting in arather uncertain disk position angle PA M = 25 ± ◦ . K –band Continuum (KI) In order to compare the angular size for the region of K –band continuum emissionwith that of the primary star alone, we define the spectral pixels outside the lines and fitthe observed visibilities with a model consisting of a single uniform disk brightness. Weobtain a mean continuum uniform disk diameter of 0 . ± . ∼ σ detection of a K –band “continuum size” in excess of what is expectedfrom the primary photosphere alone ( θ = 0 . ± .
06 mas, see also Figure 2). However, thedetection is strengthened by the fact that the uniform disk diameters fit to the individualspectral pixels increase monotonically across the K –band, indicating that we resolve anunderlying brightness of wavelength dependent radius. If the underlying brightness weresimply that of the primary photosphere, and given that for this type of star and wavelengthlimb–darkening would be undetectable at this spatial resolution, the uniform disk diameterswould be the same for all continuum spectral channels. 11 –This “continuum size” includes both the central primary star and K –band continuumemission from the gaseous disk. In order to obtain a characteristic size for the K –band disk emission, we fit the continuum visibilities to a model consisting the central primary starplus a Gaussian brightness. The KI measurement was made at essentially a single spatialfrequency; therefore we cannot constrain the disk geometry (inclination and position angle)and we use a face–on model (circularly symmetric). For the same reason, in order to fitthe Gaussian FWHM we require an independent estimate of the primary to total K –bandflux ratio ( F p /F T ) K via, for example, spectral energy distribution (SED) decomposition.Unfortunately, there exists no near–contemporaneous SED in the published or un–publishedliterature (to our knowledge). We thus form an estimate using the K –band flux seen bythe interferometer itself, calibrated by that measured during the calibrator observations.We obtain the total K –band magnitude K total = 2 . ± .
19. In order to estimate theprimary star K –band flux, we use the pre–outburst photometry of The et al. (1986) whichgive a total K magnitude (primary plus secondary) of 2.70. Thus the magnitude of theprimary alone is K primary = 2 . ± .
04 (considering the error in the secondary/primaryflux ratio, as described in the previous section). Thus, we obtain ( F p /F T ) K = 0 . ± . . ± . . ± . K –band. The mean K –band FWHM is0 . ± . K –band flux ratio and angular size of the primary star) introduce an additional 10% errorin the continuum disk size. At wavelengths of line emission, there is also continuum emission. Assuming nooff-center emission (consistent with our DP data), the measured visibilities as a function of 12 –wavelength may be decomposed as: V λi = ( F λi,c · V λi,c ( θ λi,c ) + F λi,l · V λi,l ( θ λi,l ))( F λi,c + F λi,l ) (1)where V is the visibility modulus, θ represents a characteristic angular size, F is theflux, and λi refer to the KI spectral bins. The wavelength subscripts “ c ” and “ l ” refer tothe regions where the continuum or line emission originate. From this, it can be seen thatif significant flux arises in compact continuum regions, as is the case here, even a modestdecrease in the measured visibility (e.g. ∼
10% in Figure 2) can correspond to a largedifference between the sizes of the continuum and line regions. For example, in the limitingcase that the system consists of an unresolved central star surrounded by a very large,completely resolved, line–emitting region ( V λl ( θ λl ) = 0 . V λ = F λc / ( F λc + F λl ). Within each spectral bin, we use Equation 1 monochromatically, atthe center wavelength of each bin.In the above equation, θ λi,c is the continuum size derived in the previous section; andbefore we can fit the line emission region sizes, we must estimate the contribution to thetotal flux from the continuum at each wavelength. This can be readily done using theflux spectrum measured by KI and interpolating the continuum flux under the lines. Weemphasize that this is all that is required in order to obtain absolute sizes for the lineemitting regions: determining the line sizes does not depend on the decomposition of thecontinuum emission into star+disk presented in the previous section, and therefore does notdepend on our estimates of the primary star diameter or its contribution to the total flux.As before, the line emitting gaseous disk is represented by a circularly symmetricGaussian brightness distribution and we fit the FWHM. The results are shown in Figure 3,which shows the disk FWHM as a function of wavelength. As noted above, the line emission 13 –is much larger than the continuum size, × . × . I and Br γ , respectively.Table 1 also summarizes these results.These line region sizes may be considered upper limits, because we may be under–estimating the emission line fluxes if there is some line absorption due to the the primarystar photosphere. We note that there appears to be no Br γ absorption in the spectrumshown in Banerjee et al. (2001) and indeed it is possible that the disk component completelyobscures the photospheric component in δ Sco. Nevertheless, in order to quantify this effect,we use a Kurucz model (Kurucz 1979) for the primary star, for the stellar parameters ofCarciofi et al. (2006). The photospheric Br γ absorption spans ∼ . ∼ .
5% of the stellar continuum flux; or ∼ .
25% of the total K –bandcontinuum (recall that we estimated above that the primary star contributes 62% ofthe K –band continuum seen by KI). When this maximum correction is applied to theline/continuum flux ratio, we obtain a Br γ FWHM that is 4% smaller; a relatively smalleffect compared to the ∼
15% errors in the sizes that result from the errors in the visibilitydata. The He I photospheric absorption is weaker, and the effect would be even smaller.The errors in the KI spectrum fluxes translate into ∼
10% errors in the line/continuum fluxfractions, which in turn translate into ∼
30% errors in the fitted Gaussian FWHM.
4. Discussion
In Figure 4 we represent schematically the sizes measured for the different emissionregions. Interestingly, within errors, we do not detect a significant difference between the H and K –band continuum sizes. It remains to be seen whether this is generally the casefor Be disks (preliminary results indicate that this is also the case for ζ Tauri, G. Schaefer,private communication). 14 –In order to compare our new NIR measurements with predictions at otherwavelengths, we make use of the relationship between H α luminosity and disk major axisempirically established by Tycner et al. (2005). We used near-contemporaneous spectra(UT 2007 July 12) obtained at the Ritter Observatory. We measure an EW( Hα ) = − . . Hα ) = − . α luminosity of 106 . × W and a disk semi-majorFWHM of 72 . × m, or 14 . R (cid:63) (for R (cid:63) = 7 R (cid:12) ). This size scale is × . × . γ and He I FWHM measured here along the KI position angle (44 . ◦ ± ◦ ) to be similar to that of the KI baseline, suchthat the size difference cannot be attributed solely to the projection of the KI baseline.Indeed, the optical depth in H α is likely to be much larger than in either of He I or Br γ ,and therefore the disk will look much larger in H α (Gies et al. 2007). Furthermore, Carciofiet al. (2009) argue that Br γ forms closer in than H α and is modulated in a different way bythe one–armed spiral asymmetry in the disk.Our measured line emission sizes in K–band are also ∼ × R disk = 7 R (cid:63) ). This radius howeverrepresents the outer boundary of the disk. The optical depth in Br γ will be larger in theinner part of the disk where the density is higher, thus it is not surprising that we measurea smaller radius than the disk outer boundary. In any case, as pointed out by those authors,in those models disk size and density are degenerate. Our measurements provide directconstraints on the size, albeit for a later epoch.We note that photometric light curves indicate that the state of the disk was less active(lower density) at the time of our observations compared to earlier years (see e.g. S. Otero’s 15 –web site http://varsao.com.ar/delta Sco.htm ); and our estimate of the primary + disk K –band magnitude (2 . ± .
19) appears to connect with the 2005 K –band decline reportedin Carciofi et al. (2006). The H α emission was also on the decline (Pollmann 2009).As pointed out above, several groups find that the δ Sco primary shows on averagethe observational signatures of a typical Be disk. But it has also been found that thissystem exhibits very unusual characteristics (e.g. Carciofi et al. 2006) which make it a veryinteresting case to study in detail. Observations near the periastron passages will offer usthe opportunity to observe the onset of emission line activity, which can reveal fundamentalinformation about the building of stellar envelopes, the causes of the Be phenomenon,and the role of binarity. Ultimately, dynamical models are needed and are being activelypursued by several groups. Our results presented here offer crucial observational constraintsto those models.The Keck Interferometer is funded by the National Aeronautics and SpaceAdministration as part of its Navigator program. The data presented herein were obtainedat the W.M. Keck Observatory, which is operated as a scientific partnership among theCalifornia Institute of Technology, the University of California and the National Aeronauticsand Space Administration. The Observatory was made possible by the generous financialsupport of the W.M. Keck Foundation. The authors wish to recognize and acknowledge thevery significant cultural role and reverence that the summit of Mauna Kea has always hadwithin the indigenous Hawaiian community. We are most fortunate to have the opportunityto conduct observations from this mountain. The CHARA Array is funded by the NationalScience Foundation through NSF grants AST-0307562 and AST-0606958 and by theGeorgia State University. J.D.M. acknowledges support from NSF grants AST-0352723 andAST-0707927.
Facilities:
Keck Interferometer, CHARA interferometer, Ritter Observatory. 16 –
REFERENCES
Banerjee, D. P. K., et al. 2001, A&A, 380, L13Barnes, T. G., Evans, D. S., & Moffett, T. J. 1978, MNRAS, 183, 285Bedding, T. 1993, AJ, 106, 2Berger, J. P. & Segransan, D. 2007, New Astronomy Reviews, Volume 51, Issue 8–9,Proceedings of the EuroSummer School ”Observation and Data Reduction with withthe VLT Interferometer”ten Brummelaar, T. et al. 2008, Proc. SPIE, 7013, 701308-701308-12Carciofi, A. C. et al. 2009, A&A, 504, 915Carciofi, A. C. et al. 2006, ApJ, 652, 1617Colavita, M. M. et al. SPIE Proceedings, 5491, 454Cot´e, J. & van Kerkwijk, M. H. 1993, A&A, 274, 870Eisner, J. A. et al. 2007, ApJ, 654L, 77Gies, D. R. et al. 2007, ApJ, 654, 527Halonen, R. J. et al. 2008, PASP, 120, 498Kurucz, R. L. 1979, ApJS, 40, 1Pauls, T. A. et al. 2005, PASP, 117, 837, 1255Peters, G. J., Gies, D. R., Grundstrom, E. D., & McSwain, M. V. 2008, ApJ, 686, 1280 17 –Lawson, P. R. 2000, “Principles of Long Baseline Stellar Interferometry”, Ed. PeterR. Lawson. Published by National Aeronautics and Space Administration, JetPropulsion Laboratory, California Institute of Technology, Pasadena, CA, 2000Mason, B. D. et al. 2009, AJ, 137, 3358Miroshnichekno, A. S. et al. 2001, A&A, 377, 485Monnier, J. D. et al. 2007, Science, 317, 342Monnier, J. D. et al. 2008, Proc. SPIE, 7013, 701308-701302-11Otero, S. et al. 2001, IBVS, 5026Pollmann, E. 2009, BAV Rundbrief, 58, 151Porter, J. M. & Rivinus, T. 2003, PASP, 115, 1153Stee, Ph. et al., 1995 A&A, 300, 219Tango, W. J. et al. 2009, MNRAS, 396, 842The, P. S., Wesselius, P. R. & Janssenm I. M. H. H. 1986, Astronomy & AstrophysicsSupplemet Series, 66, 63Tycner, C. et al. 2005, AJ, 624, 359Tycner, C. et al. 2006, AJ, 131, 127This manuscript was prepared with the AAS L A TEX macros v5.2. 18 –
200 100 0 -100 -200U coordinate (10 radians -1 )-200-1000100200 V c oo r d i na t e ( r ad i an s - ) CHARA/MIRC UV Coveragefor d Sco 0 50 100 150 200 250Spatial Frequency (10 radians -1 )0.00.20.40.60.81.0 V i s i b ili t y Along Disk Major AxisAlong Disk Minor Axis
Fig. 1.— Observations made in the H–band using CHARA/MIRC. The left panel showsthe uv coverage. The right panel shows the calibrated visibility data along with the best–fitvisibility curve along the major and minor axes of the disk. Although the stellar companionis fully accounted for in the modelling, for clarity in this plot, we have removed it from thebest–fit model shown. As described in the text, the variability in the visibilities seen atshort baselines is due to the effect of the companion, the effect at longer baselines disappearsbecause the coherence envelope becomes smaller than its angular separation. 19 – m m)0.800.850.900.951.001.05 V HeI Br g m m)0.70.80.91.01.11.21.3 F l u x ( a r b ) m m)-1.0-0.50.00.51.0 D P ( deg ) Fig. 2.— Keck Interferometer calibrated K–band data as a function of wavelength: V (top),relative flux (middle), and differential phase (DP, bottom). The dashed line in the top panelrepresents the V of the primary star photosphere (uniform disk diameter 0 . ± .
06 mas). 20 – m m)0123 D i sk F W H M ( m a s ) Fig. 3.— Gaussian FWHM fitted to the Keck Interferometer data for the continuum and linewavelengths. The dotted line represents the flux fraction (continuum/total), for the case thatphotospheric absorption is ignored. As discussed in the text, corrections for photosphericline absorption lead to line sizes that are at most 5% smaller. 21 – -10 -5 0 5 10X(mas)-10-50510 Y ( m a s ) H continuum disk K continuum diskBr g disk He I disk H a disk KI Fig. 4.— Representation of the disk FWHM for the various disk emission regions. Theprimary star is represented by the central dashed circles. For each emission region, we plota Gaussian of the measured disk FWHM (bright circles). For H α , the size is not directlymeasured but inferred from the measured H α spectrum, as described in the text. North isup and East is to the right. 22 –Table 1: Results. H–band (CHARA/MIRC)Fixed parameters: θ p = 0 . ± .
06 mas, θ s = 0 . ± .
03 mas, F p /F s = 0 . − . . ( F p /F T ) H a FWHM M FWHM m FWHM M FWHM m PA M (mas) (mas) ( D (cid:63) b ) ( D (cid:63) ) (deg, East of North)0.57 ± ± ± ± ± ± c : θ p = 0 . ± .
06 mas, ( F p /F T ) K = 0 . ± . a Disk Region FWHM FWHM(mas) ( D (cid:63) )Continuum 0 . ± . . ± . I . ± . . ± . γ . ± . . ± . a Consistent with the companion location relative to the FOV for each instrument, in the above flux ratiosthe H–band total flux ( F T ) includes the secondary star, but the K–band total flux does not. b D (cid:63) = 14 R (cid:12) . cc