Stellar and Circumstellar Properties of the Pre-Main Sequence Binary GV Tau from Infrared Spectroscopy
aa r X i v : . [ a s t r o - ph ] M a y Accepted to ApJ, 15 May 2008
Stellar and Circumstellar Properties of the Pre-Main SequenceBinary GV Tau from Infrared Spectroscopy Greg W. Doppmann , Joan R. Najita , and John S. Carr [email protected]@[email protected] ABSTRACT
We report spatially resolved spectroscopy of both components of the low-mass pre-main-sequence binary GV Tau. High resolution spectroscopy in the K -and L -bands is used to characterize the stellar properties of the binary and toexplore the nature of the circumstellar environment. We find that the southerncomponent, GV Tau S, is a radial velocity variable, possibly as a result of anunseen low-mass companion. The strong warm gaseous HCN absorption reportedpreviously toward GV Tau S (Gibb et al. 2007) was not present during the epochof our observations. Instead, we detect warm ( ∼
500 K) molecular absorption withsimilar properties toward the northern infrared companion, GV Tau N. At theepoch of our observations, the absorbing gas toward GV Tau N was approximatelyat the radial velocity of the GV Tau molecular envelope, but it was redshiftedwith respect to the star by ∼
13 km s − . One interpretation of our results is thatGV Tau N is also a binary and that most of the warm molecular absorption arisesin a circumbinary disk viewed close to edge-on. Subject headings: infrared: stars – stars: formation, fundamental parameters,late-type, low-mass, pre–main sequence — stars: circumstellar matter, stars:individual (GV Tau)— techniques: spectroscopic, radial velocities Data presented herein were obtained at the W.M. Keck Observatory from telescope time allocated tothe National Aeronautics and Space Administration through the agency’s scientific partnership with theCalifornia Institute of Technology and the University of California. The Observatory was made possible bythe generous financial support of the W.M. Keck Foundation. NOAO, 950 North Cherry Avenue, Tucson, AZ 85719 Naval Research Laboratory, Code 7210, Washington, DC 20375
1. Introduction
GV Tau (Haro 6-10, Elias 7, HBC 389, IRAS 04263+2426) is an unusual young stellarobject (YSO) in the Taurus molecular cloud. While the spectral energy distribution of thesystem has the strongly rising 2–25 µ m continuum that is characteristic of Class I sources(e.g., Furlan et al. 2007), millimeter studies find comparatively weak dust and gas emissionfrom the source compared to other low-mass embedded YSOs (Hogerheijde et al. 1998).This, and the poorly defined molecular outflow structure from the source, suggest that thesystem lacks a significant envelope component (Hogerheijde et al. 1998) and may be a moreevolved Class I system.GV Tau is also a binary (projected separation 1 . ′′
2; Leinert & Haas 1989). It is one of asmall population of low-mass pre-main-sequence binaries in which one component is opticallyvisible, while the other is optically faint and radiates primarily at infrared wavelengths. Theoptically visible component in the GV Tau system is the southern component (GV Tau S), a TTauri star. Its stellar properties have been characterized previously based on high resolutioninfrared (Doppmann et al. 2005) and optical (White & Hillenbrand 2004) spectroscopy. GVTau S is unusual in that its radial velocity has been found to differ significantly (by more than3- σ ) from both the radial velocity of the surrounding molecular cloud (by > . − ) andthe radial velocity distribution of other Taurus sources (Doppmann et al. 2005; Covey et al.2006). In addition, L - and M -band absorption by gaseous warm organic molecules (HCN,C H , CO) has been reported toward GV Tau S (Gibb et al. 2007), making it one of the fewpre-main-sequence sources in which such absorption features have been detected.Less is known about the northern component, the infrared companion of the binarysystem (GV Tau N). It is approximately 100 times fainter than GV Tau S at optical wave-lengths (Stapelfeldt et al., in preparation), is bright at infrared wavelengths, and experiencessignificant extinction, as indicated by the strong water ice and silicate absorption observedtoward the source (Whittet et al. 1988; Van Cleve et al. 1994; Leinert et al. 2001). Extendedemission (on a 10 ′′ scale) associated with the GV Tau system is found to be highly polarized(6%) at infrared wavelengths, a result that is attributed to scattering by a flattened envelope(i.e., a shell with an aspect ratio of 10:1) or a disk viewed at high inclination (Menard et al.1993). Since strong silicate absorption is found only in the spectrum of GV Tau N andnot GV Tau S, this is taken as evidence for an edge-on disk associated with GV Tau N(Van Cleve et al. 1994).High angular resolution observations provide limited support for this interpretation.When studied at 0 . ′′
08 resolution at K , GV Tau N is found to be surrounded by a nebulawhose shape may be consistent with the presence of an edge-on disk, but the shape ofthe nebula is irregular and varies with time (Koresko et al. 1999). The irregular structure 3 –is in marked contrast to the symmetric lens-like shape of scattered light distribution seenin the HK Tau B system, a system that is also believed to possess a nearly edge-on disk(Stapelfeldt et al. 1998; Koresko 1998). Adding further complexity, significant photometricvariability is found for both components at 1 . − . µ m on timescales as short as a month(Leinert et al. 2001), with the 3 . µ m water ice band absorption strength also varying towardboth components.Here we use spatially resolved high resolution infrared spectroscopy of both binarycomponents to characterize the stellar and circumstellar components of the GV Tau system.Our K - and L -band spectroscopic observations are described in § §
3. We use these data to characterize the stellar and circumstellarproperties of the system ( § § §
2. Observations and Data Reduction2.1. Spectroscopic Observations K - and L -band infrared spectra were obtained on 2007 January 03 using the cryogenicechelle facility spectrometer, NIRSPEC (McLean et al. 1998), on the 10-m Keck II telescopeatop Mauna Kea, Hawaii. Spectra were acquired through the 0 . ′′
432 (3 pixel) wide slit,providing spectroscopic resolution R ≡ λ/δλ = 24 ,
000 (12.5 km s − ). The echelle and crossdisperser angles were oriented to obtain L -band spectra of transitions of HCN and C H (e.g.3.04 - 3.09 µ m, order 25). We used the KL blocking filter to image the L -band orders onto theinstrument’s 1024 × L -band observations, theechelle and cross disperser gratings were re-oriented to allow key K -band features (i.e., MgI, Al I, Na I, ν =2-0 CO, and Br γ ) to fall onto the detector. In this configuration, orders 32 –38 (non-contiguous) were imaged through the NIRSPEC-7 blocking filter onto the detectorarray.Both binary components of GV Tau were acquired in the slit simultaneously, at aposition angle on the sky of 356 ◦ (Leinert & Haas 1989). Initially, GV Tau N (the brightercomponent in our observations of λ > . µ m) was centered in the slit. During the L -and K -band observations the slit was held physically stationary to avoid slight grating angleshifts caused by vibrations from the slit rotator motor. During the time we integrated on thebinary pair ( <
25 min. total), the measured flux ratio of the components in the slit did notchange systematically. This is consistent with the negligible expected motion of the binarypair within the slit as the non-equatorially mounted telescope tracked. Despite some wind 4 –shake in the direction across the slit in the east-west direction, the individual components(separated by 1 . ′′
2) were well resolved in median seeing ( ∼ . ′′ L - and K -bands,respectively. The telescope was nodded ± ′′ from the slit center in an ABBA sequence alongthe 24 ′′ long slit so that each frame pair would contain object and sky in both nod positions.To correct for telluric absorption, we also obtained spectra of an early-type star located closeto GV Tau in the sky (HR 1412), before and after the GV Tau observations ( L - and K -bandsettings, respectively). Spectra of the internal NIRSPEC continuum lamp were taken forflat fields at the K - and L -band settings. Exposures of the Argon, Neon, Krypton, andXenon Arc lamps provided wavelength calibration for all the K -band orders, except in order33 where we used the telluric lines for calibration due to the scarcity of arc lamp lines inthis order. We also obtained short exposure (10 × K - and L -band images of the GVTau binary using the slit viewing camera detector (SCAM) in order to measure the relativebrightness of the two components. The spectra were reduced using standard IRAF packages (Massey et al. 1992; Massey1997). Sky subtracted beam pairs were divided by a normalized flatfield. Individual echelleorders were parsed from the multi-order, co-added images. Bad pixels (i.e., hot or lowresponsivity pixels) in each order were identified by inspection, and removed by interpolationusing fixpix in IRAF. To better remove telluric emission features which are more severe inthe L -band, we rectified the slit in our L -band data using 3rd order transform solutionderived from the brightest emission sky lines (typically 10-12 per order) traced along theslit length at each beam position before sky subtraction. Transformed and cleaned orderswere extracted using the apall task. Extractions for each binary component of GV Tauwere based on a signal profile down to 50% (FWHM) on both sides of the profile peak,to minimize the contaminating flux from the nearby companion. For the L -band spectra,residual background sky was subtracted in the extraction using selected regions of the profilecut that were well outside of the profile of the spatially double-peaked binary.Wavelength calibration in the L -band was achieved using selected telluric absorptionfeatures present in the spectra of the telluric standard star (HR 1412). Stronger unblendedtelluric lines were selected over weaker ones, while avoiding very saturated lines whose depthswent to zero. Rest wavelengths for the absorption features were obtained from the HITRANdatabase (Rothman et al. 1998). Most or all of these same lines were used to derive a 5 –wavelength solution for the target, with the exception of those lines that might fall veryclose in wavelength to emission or absorption lines present in the target spectrum.Telluric features present in the spectra of GV Tau at each nod position in the slit wereremoved by dividing by the spectrum of the telluric standard (HR 1412) observed at the samenod position. Several weak stellar absorption lines were present in the K-band orders of thestandard due to its relatively late spectral type (A7 III). These atomic lines were modeledand removed by dividing the standard spectrum by the best fit stellar synthetic spectrumbefore division into the GV Tau spectrum. The stellar synthetic spectra were generated withthe program MOOG (Sneden 1973) using the NextGen model atmospheres (Hauschildt et al.1999). The initial line list was taken from Kurucz (1993), and individual line parameterswere adjusted to fit the observed disk-center solar spectrum of Livingston & Wallace (1991).For HR 1412, we used a model with T eff = 7600 K and log g = 3 .
5. Small adjustments weremade in the elemental abundances to give the best empirical fit to HR 1412. We measured v sin i = 75 km s − , in agreement with published values (Royer et al. 2002).Br γ absorption at 2.166 µ m was also present in the telluric standard. We therefore usedthe synthetic telluric spectral modeling program, AT (Grossman, private communication) tomodel the telluric features that we observed in this region of the HR 1412 spectrum. Wefit the depths of the observed telluric lines and then used the best-fit synthetic spectrum todivide out the telluric lines in the GV Tau spectra in this order (order 35, Fig. 1c).In order to view the true continuum shape within each echelle order, we then multipliedthe resulting GV Tau spectra by the spectral slope of a 7600 K blackbody (Figs. 1a – 1e).Simple aperture photometry of the SCAM images showed that GV Tau N was 35-40%brighter than GV Tau S in the K+open, NIRSPEC7, and KL filters. The relative flux of theGV Tau N and S K -band spectra in orders 33 and 34 (Figs. 1a & 1b), which are located inthe middle of the K -band filters above, are consistent with the aperture photometry.
3. Detected Spectral Features
Emission and absorption features in GV Tau N and GV Tau S were detected in the K -band spectra of each object (Figs. 1a – 1d). In order 33 (2.269 – 2.304 µ m, Fig. 1a), ν =2-0 CO overtone emission and absorption are present in both GV Tau N and GV TauS. The emission component dominates the spectrum in GV Tau N in this order, while theabsorption dominates in GV Tau S. Lines due to absorption of neutral atomic species thatare present in GV Tau S in this order (e.g., Fe I at 2.2747 µ m and Mg I at 2.2814 µ m) areabsent in GV Tau N. The K -band absorption features in both components are presumed to 6 –originate from the stellar photosphere.Emission features of Na I and H dominate the structure of the GV Tau N spectrum inorder 34 (2.203 – 2.236 µ m, Fig. 1b). Interestingly, the broad emission lines near 2.2065 and2.2090 µ m in GV Tau N appear to show absorption components possibly of a photosphericorigin. The S(0) 1-0 H emission at 2.2235 µ m is detected in both components, and isspatially extended (see § emission, the spectrum of GV Tau S inthis order displays neutral atomic absorption lines of photospheric origin (Fig. 1b).The spectra in order 35 (2.140 – 2.173 µ m, Fig. 1c) are characterized by strong H I Br γ emission in both GV Tau N and GV Tau S. The continuum level near the Brackett γ line is ∼
20% higher in GV Tau N, while the equivalent width of the emission line is 3 times greaterin GV Tau S.In order 36 (2.081 – 2.113 µ m, Fig. 1d), the spectrum of GV Tau N appears featureless.In contrast, GV Tau S exhibits photospheric absorption lines (e.g., Mg I at 2.1065 µ m andAl I at 2.1099 µ m), which are useful as diagnostics of the stellar effective temperature andsurface gravity.With our L -band spectra in order 25 (3.0450–3.0865 µ m, Fig. 1e), we detect strongHCN absorption ( ∼
10% deep) and weaker C H absorption in GV Tau N. In contrast, nomolecular absorption is detected in the GV Tau S spectrum. In §
4. Results4.1. Stellar and Circumstellar Properties
All four K -band orders of the GV Tau S spectrum display stellar photospheric absorptionby neutral atomic features (e.g., Na I, Si I, Sc I, Mg I, Al I) and ν =2-0 CO. We usedsynthetic stellar spectral models generated by the synthesis program MOOG (Sneden 1973)and using NextGen stellar atmosphere model structures (Hauschildt et al. 1999) to constrainthe stellar properties of GV Tau S. Our modeling focused on three spectral sub-regions: The“Na” region (2.204–2.210 µ m), the “Mg/Al” region (2.104–2.111 µ m), and the “CO” region(2.292–2.300 µ m). These regions of the K -band contain the strongest absorption featureswith which to measure the stellar properties of low-mass YSOs (Doppmann et al. 2005). Theabsorption lines of Na I and Mg I are particularly gravity and temperature sensitive, but 7 –in the opposite sense from each other. For example, at cool effective temperatures (3200–4500 K) and subdwarf surface gravities (3 . ≤ log g ≤ .
5) Na and Mg lines both growstronger as log g increases, but an increase in T eff causes Mg lines to grow while the Nalines weaken. A simultaneous fit of synthesis models to the Na and Mg/Al regions (Fig.2) constrains these key stellar properties in reducing or breaking the temperature-gravitydegeneracy displayed in a single absorbing species (Doppmann et al. 2005).The best model fit to these wavelength sub-regions yields T eff =3800 ±
100 K, log g =4.0 ± v sin i rotation= 24 ± − , and K -band veiling r K =2.5 ± T eff and log g derived above, we estimate a stellar luminosity, mass, and radius of L ∗ =0 . L ⊙ , M ∗ = 0 . M ⊙ , R ∗ = 1 . R ⊙ using the pre-main-sequence model tracks of Siess et al.(2000), where a 3 Myr isochrone was consistent with our values of temperature and gravity.The derived luminosity and effective temperature imply an estimated K -band extinctiontoward GV Tau S of A K = 0 . m K = 8 .
61) that was within1- σ of the average in the study over the time period from September 1988 to March 2000( m K = 8 . ± . K -band extinction is consistent with the average opticalextinction of A V = 5 . T eff , v sin i , and v LSR values are consistent with the high resolution opticalstudy of White & Hillenbrand (2004) given the low signal to noise (S/N <
10) in their 1999December 06 observation of GV Tau S. The moderately high veiling we derive ( r K = 2 . r K = 1.8, which includes a correction for a systematic effect seen in the bestfit synthesis models to observations of MK standards (see Eqn. 1 in § r K = 2.4,which is consistent with our measured value.The constraint on effective temperature and surface gravity is consistent with the re-sults from Doppmann et al. (2005). Doppmann et al. (2005) reported a somewhat warmereffective temperature ( T eff = 4500 K ) and higher surface gravity ( log g = 4 . µ m) at similar signal-to-noise and slightly lowerspectral resolution ( R =18,000). A comparably good fit to the 2001 data can be obtainedwith T eff = 4000 K and log g = 3 . v LSR of GV Tau S differs from previous measurements inthe literature for this source. In 2001, the radial velocity of GV Tau S was measured to be v LSR = –6.2 ± . − (Covey et al. 2006), blueshifted by ∼ − relative to theobservations of this study and that of White & Hillenbrand (2004, v LSR = +3 . ± . − ).As a result, this source was flagged as a radial velocity outlier, since its radial velocity was > σ from the mean of all the Taurus sources in the survey (Covey et al. 2006). Our 2007observations now place the radial velocity of GV Tau S ( v LSR = 9 . ± . − ) close tothe mean v LSR (4 . ± . − ) of the sources in the 2001 Taurus survey (Doppmann et al.2005; Covey et al. 2006). We discuss the radial velocity variations further in § §
3, the CO ν =2-0 bandhead of GV Tau S has both emission and ab-sorption components. Using the best fit stellar parameters obtained from the Na and Mg/Alregions of the GV Tau S spectrum as a constraint on the stellar photospheric properties,we fit the CO feature with a composite model of circumstellar disk emission and stellarphotospheric absorption (Fig. 3). The emission component is modeled as arising from adifferentially rotating disk that is in chemical equilibrium. The emission arises between aninner radius R i , at which the projected disk rotational velocity has a specified value of v sin i ,and an outer radius R o . The disk temperature and column density are modeled as simpleradial gradients that decrease as power laws (Carr 1993; Carr et al. 2004). The star anddisk are placed at the same radial velocity and the composite model spectrum is smoothedto the 13 km s − velocity resolution of NIRSPEC. The model fit parameters are given in thecaption to Fig. 3. The spectrum of GV Tau N shows emission and absorption components in both the COand Na regions (Figs. 4 & 5) of the K -band. We attribute the emission to a circumstellardisk; the absorption features are presumed to arise from the stellar photosphere, implying alate spectral type for the star. We assume a stellar mass of M ∗ =0.8 M ⊙ and an age of τ =3Myr for this object, consistent with the typical values found for low-mass YSOs in Taurus(Brice˜no et al. 2002). These assumptions imply an effective temperature of T eff =4100 K 9 –and a surface gravity of log g ∼ . v sin i = 15 km s − ) and heavy veiling( r K =12). Such a substantial infrared veiling greatly exceeds what has been measured fromnear–IR photospheric absorption lines in past studies (e.g., r K ≤ .
5, Luhman et al. 1998;Luhman & Rieke 1999; Doppmann et al. 2003, 2005). It also results in large uncertaintiesin the inferred K -band extinction.Broad Na emission has been observed in several Class I and flat spectrum protostars,always accompanied by ν =2-0 CO emission (Doppmann et al. 2005). The absorption linesin the Na region, apparent within the broader emission features, can be fit with the samestellar photospheric model and somewhat larger veiling ( r K = 15, Fig. 5) than was usedin fitting the CO region. If we subtract the stellar photospheric model from the observedspectrum, the resulting spectrum is similar to the Na emission seen in other T Tauri stars.The absence of detected spectral features in the Mg/Al region of GV Tau N in our data(signal-to-noise ∼ v LSR = − . − , see Table 2), and with the radial velocityof the emission component from which they arise. Given the assumed mass M ∗ = 0 . M ⊙ ,we estimate the stellar luminosity, radius, and K -band extinction (i.e., L ∗ = 0 . L ⊙ , R ∗ =1 . R ⊙ , and A K = 2 .
6) from the photospheric absorption we find in our spectra, usingSiess et al. (2000) evolutionary model tracks and Leinert et al. (2001) photometry from 2000March 07 ( m K = 8 . K -band extinction and veiling. If the apparent magnitudewas the historical average of m K = 9 . m K = 8 .
66 (Leinert et al. 2001), then the extinction ( A K ) would increaseby 1.2. If we had assumed a higher ( T eff =4300 K) or lower ( T eff =3900 K) temperature thanthe value we use here ( T eff =4100 K), which was based on the IMF peak (Brice˜no et al. 2002),this would imply a veiling ( r K ) of 11 or 13, and an extinction (∆ A K ) of +0.5 or -0.6 relativeto the derived value presented above ( A K =2.6). 10 – γ Emission
We detect H I Br γ emission in both components of GV Tau (Fig. 6). The broad emissionprofiles (i.e., 175 and 170 km s − FWHM, GV Tau S and N, respectively) are consistent withthe Br γ width of other active T Tauri stars (Najita et al. 1996; Folha & Emerson 2001). InGV Tau S, where the equivalent width is three times greater than in GV Tau N (EW South = − . § γ line inGV Tau N is blueshifted by 12.5 km s − relative to the CO absorption component ( § γ centroids, which are common among T Tauri stars, are consistent withan origin for the emission in gas infalling in a stellar magnetosphere (Najita et al. 1996;Muzerolle et al. 1998b; Folha & Emerson 2001).We can use the Br γ line strength to place a rough constraint on the contribution ofstellar accretion in GV Tau N and GV Tau S to the bolometric luminosity of the system.We first convert the emission equivalent width to a line luminosity using the 2000 March07 K -band photometry from Leinert et al. (2001), and the estimated K -band extinctionfor each component ( § § γ luminosities for GV Tau Nand S are 9 × − L ⊙ and 5 × − L ⊙ , respectively. Using the empirical relation givenby Muzerolle et al. (1998b) in their Figure 4, we infer accretion luminosities of 0.1 and 0.2 L ⊙ for GV Tau N and S, respectively. Thus, we estimate the total luminosity (stellar+hotaccretion) of the GV Tau system to be 1.2 L ⊙ , below what has been estimated for itsbolometric system luminosity ( L bol = 7 − L ⊙ ; Kenyon & Hartmann 1995; Furlan et al.2007). emission The S(0) 1-0 H emission (2.22329 µ m) that we detect from GV Tau is marginallyresolved in our spectra with a velocity width of 14 and 17 km s − for GV Tau S and N,respectively. Spatially, the bulk of the emission is coincident with GV Tau N and GV TauS (within the 0 . ′′ ◦ )ending at a bright knot 6 . ′′ emission is constant along theslit ( v LSR = 8 . − ), and agrees with the radial velocity of the HCN absorption ( § emission in both stellar components. The emissionis extended southward along the slit in the same way as the S(0) emission. The emission is2–3 times brighter than the S(0) line emission. As with the S(0) line, the S(2) equivalentwidth is ∼ line ratios (i.e., 1-0 S(2)/1-0 S(0)) in GV Tau N and S are consistentwith shock excitation, similar to most of the classical T Tauri stars observed by Beck et al.(2007). In order to characterize the molecular absorption detected in the L -band spectrum ofGV Tau N, we used a simple model of absorption by a slab at a single temperature andcolumn density. We adopted the HCN linelist from the HITRAN database (Rothman et al.1998). The data are well fit with a slab temperature T = 550 K and column density N HCN = 1 . × cm − , with a microturbulent line broadening of v turb = 3 km s − andradial velocity v LSR = 8 . ± . − (Fig. 8). The radial velocity of the molecular ab-sorption was measured using selected P-branch lines (Table 3) in regions of the spectrumthat had good telluric transmission and that were unblended with other species (such asC H ). We constructed an average HCN absorption profile from four selected HCN absorp-tion transitions (P11, P14, P15, and P16) to which we fit a Gaussian profile (Fig. 9). The3- σ error of 1 . − on the radial velocity was estimated by comparing the Gaussian fitof the average absorption profile to each of the P-branch line fits separately.
5. Discussion
To summarize, both GV Tau N and S appear to possess slowly rotating ( v sin i =15 −
24 km s − ) late type ( ∼ K7–M2) photospheres. While the stellar properties are consistentwith an age of ∼ K -band veiling ( r K = 12 − γ centroid. These properties are consistent with a young star undergoingactive accretion in a disk and stellar magnetosphere. The stellar radial velocity found for GVTau N ( v LSR = − . ± . − ) differs from the radial velocity of the GV Tau molecularenvelope ( v LSR = 7 . ± . − , Hogerheijde et al. 1998) and the radial velocity of thewarm ( ∼
550 K) HCN absorption detected in the L -band ( v LSR = 8 . ± . − ). Wediscuss the possible implications of these results in the rest of this section. The stellar radial velocity of v LSR = 9 . ± . − found for GV Tau S differs fromthe v LSR observed in November 2001 ( − . ± . − ; Covey et al. 2006), and the v LSR in December 1999 (3 . ± − , see Table 4; White & Hillenbrand 2004). The v LSR reported here differs slightly from the 6 . ± . − velocity of the Taurus molecular cloudin the vicinity of GV Tau, as measured in the CO J =1-0 transition (Dame et al. 2001)and the v LSR = 7 . ± . − measured for the gaseous envelope surrounding GV Tau(Hogerheijde et al. 1998).A plausible explanation for the radial velocity variations exhibited by GV Tau S is thatit is a spectroscopic binary with a secondary that is sufficiently faint that it is undetectablein our spectra. Assuming that the systemic velocity is that of the molecular cloud and thatthe primary has the stellar properties ( T eff , M ∗ ) derived earlier ( § ∼ . M ⊙ companion in a P orbit ∼
38 day circular orbit ( i = 90 , a = 0 .
35 AU). A companion with thismass at the typical age of Taurus sources ( ∼ − K continuum. Its K -band spectral features would appear only ∼ K -band spectrum arealso consistent with smaller companion masses and orbital separations. A larger stellar massfor the companion in a more face-on orbit is ruled out since it would be detectable in ourhigh signal-to-noise spectra. Gibb et al. (2007) previously reported absorption toward GV Tau S in the HCN low- J R and P lines near 3 . µ m, based on observations made on 2006 February 17 and 18. TheHCN absorption was characterized by equivalent widths ∼ .
015 cm − , a radial velocity 13 – v LSR = 9 . ± . − , and a rotational temperature of T = 115 K. No HCN absorptionwas reported toward GV Tau N in the Gibb et al. (2007) study. The wavelength coverage ofour observations includes many of the same lines that were observed by Gibb et al. (2007)as well as higher rotational lines.At the epoch of our observations, no HCN absorption was present toward GV Tau S.However, HCN absorption was detected toward GV Tau N. The absorption lines we detecttoward GV Tau N have equivalent widths and radial velocities similar to those reported byGibb et al. (2007) toward GV Tau S.We find a higher temperature ( ∼
550 K) for the absorbing gas toward GV Tau N, butthis temperature is consistent with the relative HCN line strengths observed by Gibb et al.(2007). Their lower temperature was based on analysis of only low- J lines (i.e., R1–R6,P2–P7); warmer temperatures are indicated when the higher- J lines used in our analysisare included. Consistent with our results, HCN absorption with properties similar to thosereported here was also detected toward GV Tau N (but not GV Tau S) in mid-infraredobservations made with TEXES on Gemini-North at an intermediate epoch (November 2006;Najita et al., in preparation).These results may indicate that the molecular absorption toward both components isvariable. Indeed, both GV Tau N and S are found to experience significant photometricvariability in the H - through M -bands and in the 3 . µ m ice feature (Leinert et al. 2001).The high temperature of the absorbing gas suggests that the gas is located close to thestar rather than in a distant circumbinary envelope or the molecular cloud. So if variabilityaccounts for the difference in the absorption properties, we are likely to be observing thevariations in the near-circumstellar environment of each source, rather than the motion ofa distant absorber across our line-of-sight. Such a distant absorber is believed to accountfor the photometric variations in GV Tau S and perhaps variations in the ice absorption(Leinert et al. 2001).Another, more mundane, possibility is that the components of the binary pair weremisidentified by Gibb et al. (2007). The interpretation that the warm molecular absorptionis entirely associated with GV Tau N would be consistent with the systematically stronger3 . µ m ice feature measured toward this component, the strong silicate absorption thatis detected only toward GV Tau N (Van Cleve et al. 1994), and the hypothesis from theliterature that GV Tau N is surrounded by an edge-on disk. Additional observations of theGV Tau system would be useful in resolving this issue.Setting aside for now the possibility that the HCN absorption is time variable, we canexplore the nature of the absorbing gas that we observed based on the stellar and circum- 14 –stellar properties that we have found for GV Tau N. As described at the beginning of thissection, the stellar velocity is blueshifted from the envelope velocity by ∼
11 km s − . Thissuggests that the detected stellar component of GV Tau N possesses an orbiting companion.If it does not, at a velocity of 11 km s − relative to the envelope, the star would travel 0.1pc in 10 yr, escaping its molecular envelope on a timescale much less than the age of thesystem. If GV Tau N is then a binary, the envelope velocity is the more appropriate velocityreference frame for the system. Another possibility is that the stellar light is not seen di-rectly, but rather in reflection against material on the far side of the envelope that is infallingtoward us (S. Strom, private communication). In either interpretation, the measured stellarvelocity is not the systemic velocity.As also described at the beginning of this section, the radial velocity of the HCN absorp-tion is marginally different from that of the GV Tau molecular envelope by 1 . ± . − .The small velocity difference allows for the possibility that some of the absorption arises ingas that is infalling with respect to the envelope. Higher spectral resolution observations(e.g., with TEXES; Najita et al., in preparation) can better determine whether this is thecase.In the meantime, we can consider two possible scenarios for the origin of the HCNabsorption: (1) Some fraction of the absorption arises in gas infalling from the molecularenvelope toward the star at ∼ − and/or (2) the absorption arises in warm gas in orbitaround the star. Origin in an Envelope.
Since the bolometric luminosity of the system is 7–9 L ⊙ (Kenyon & Hartmann1995; Furlan et al. 2007) and the stellar contribution (photospheric and stellar accretion, § § ∼ . L ⊙ , the residual disk and envelope accretion is L acc ∼ − L ⊙ . The large residual accretion luminosity compared to the stellar accretionluminosity for GV Tau N inferred from the Br γ emission ( ∼ . L ⊙ ; § R in , the disk accretion luminosity L d = GM ∗ ˙ M d / R in of 7 L ⊙ corresponds to a disk accretion rate of ˙ M d ∼ − M ⊙ yr − for R in = 2 . R ⊙ and astellar mass of M ∗ = 0 . M ⊙ . In comparison, the spectral energy distributions of most of theClass I sources in Taurus are fit with envelope accretion rates up to an order of magnitudelarger (Furlan et al. 2007). A lower accretion rate ∼ − M ⊙ yr − is more consistent withthe weak envelope emission and outflow activity associated with GV Tau.For the estimated mass of the stellar component(s) of GV Tau N (0.8 M ⊙ ), the envelopeinfall velocity v r = (2 GM ∗ /r ) / is expected to be ∼ − at distances of 360 AU. Incontrast, models of gas in collapsing protostellar envelopes suggest that the observed HCN 15 –temperature of ∼
500 K can be achieved only close to the star, within 2 AU given theaccretion luminosity of the system (Ceccarelli et al. 1996, their figure 4), i.e., well within thedistance of 360 AU that would be inferred for the infalling gas based on its velocity relativeto the molecular envelope. It therefore appears unlikely that most of the absorption couldarise in an infalling envelope unless there are additional heating processes for the infallinggas beyond those considered in Ceccarelli et al. (1996). If other heating processes, such asoblique shocks, can heat distant material to ∼
500 K, the warm molecular absorption that weobserve could also arise in a distant, non-infalling envelope. However, the lack of molecularabsorption observed toward GV Tau S argues against the gas originating from a distantenvelope which would surround both GV Tau N and S.An alternative scenario is that GV Tau N is a single star and its measured stellar velocityis the appropriate systemic velocity. In this case, at a distance 2 AU from the star, the radialcomponent of the infall velocity would be (
G M ∗ /r ) / ∼
20 km s − (Cassen & Moosman1981) for the estimated stellar mass of GV Tau N. This is similar to the 13 km s − velocitydifference between the HCN absorption and GV Tau N stellar photosphere (Table 4). If GVTau N is a single star, it is then moving relative to the molecular envelope and, as alreadydiscussed above, it will travel ∼ . < yr. It then seems unlikely that the star andthe envelope are physically associated, which makes it difficult to account for the origin ofthe infalling gas in this scenario. Origin in a Circumstellar Disk.
Another possibility is that the HCN absorption arisesin the heated atmosphere of a gaseous disk (see also Gibb et al. 2007). Models of the chem-istry of the inner regions of circumstellar disks predict that HCN will be abundant at radialdistances of a few AU (Markwick et al. 2002). Models of the thermal structure of disk atmo-spheres further predict that the disk surface will reach temperatures &
500 K at distances ofa few AU (e.g., Glassgold et al. 2004). If the disk in the GV Tau N system is viewed edge-on,as hypothesized in the literature (Menard et al. 1993; Van Cleve et al. 1994), it seems plau-sible that the warm HCN absorption could arise in the disk atmosphere seen in absorptionagainst the warmer, inner region of the disk that produces the L -band continuum.The rarity of such a line–of–sight through a disk may also explain another unusualcharacteristic of GV Tau: the low CO ice absorption optical depth toward the sourcecompared to its silicate optical depth. The data of Furlan et al. (2007) show that the peak10 µ m silicate optical depth is ∼ µ m CO iceabsorption. All other Taurus Class I sources in the Furlan sample have τ silicate /τ CO .
4, withthe exception of DG Tau B, another possible edge-on source. A high ratio of τ silicate /τ CO may occur if a substantial fraction of the silicate optical depth is produced in a warm diskregion in which CO ice has sublimated. 16 –GV Tau N would be a spectroscopic binary in this scenario, in order to account forthe difference in radial velocity between the stellar photosphere and the warm molecularabsorption. Because the warm HCN absorption is approximately at rest relative to themolecular envelope velocity, in this scenario it would arise in a circumbinary disk seen edge-on. In contrast, the CO overtone emission, which is found to share the radial velocity of thestellar photosphere, arises from the circumstellar disk of the primary component. Strong absorption by gaseous warm organic molecules appears to occur rarely amonglow mass young stars. GV Tau and another source, IRS 46, are the only systems in whichsuch absorption has been reported to date. IRS 46 is a low-mass YSO with a spectral energydistribution that indicates either a Class I source or a Class II source viewed nearly edge-on(Lahuis et al. 2006). IRS 46 is similar to GV Tau N in that it also shows Br γ and 2.3 µ mCO overtone emission (see YLW 16B in Doppmann et al. 2005) and strong absorption bygaseous warm organic molecules in the L -band and at mid-infared wavelengths (CO, HCN,C H , CO ; Lahuis et al. 2006). The temperature ( ∼
400 K) and equivalent width of theHCN absorption are similar to that observed toward GV Tau N. Unlike GV Tau N, thewarm HCN absorption in IRS 46 ( v LSR = −
20 km s − ) shows a large velocity offset fromthe surrounding molecular cloud ( v LSR = 4 . − , Table 4). In contrast to the situationfor GV Tau N, the stellar radial velocity of IRS 46 is unknown. If the system is at thecloud velocity, the warm molecular absorption is blueshifted from the system by 24 km s − .Lahuis et al. (2006) speculated that the warm molecular absorption in IRS 46 arises in theinner region of a circumstellar disk, possibly at the footpoint of an outflowing disk wind inorder to account for the presumed blueshift of the absorption. This scenario is consistentwith the interpretation of the spectral energy distribution as indicating a disk system viewededge-on. Our “disk origin” scenario for GV Tau N is similar to the explanation given byLahuis et al. (2006) for the HCN absorption in IRS 46. In both cases, the warm molecularabsorption arises in a disk, although there is no outflow component to the flow in the caseof GV Tau N.
6. Summary and Conclusions
We have used spatially resolved K - and L -band spectroscopy to characterize the stellarand circumstellar properties of the GV Tau binary system. We find that GV Tau S is aradial velocity variable, possibly the result of an unseen low-mass companion. The radial 17 –velocities measured here and in the literature, when combined with the apparent absence ofthe spectral signature of a companion in our K -band spectra, are consistent with a companionwith M ∗ < . M ⊙ and a < .
35 AU. Further spectroscopic monitoring of this source wouldbe useful to confirm our interpretation and to better constrain the companion properties.The other component of the binary system, GV Tau N, is found to show strong COovertone emission, strong K -band veiling ( r K = 12 − γ emission with a blue-shifted emission centroid; signatures consistent with a young actively accreting star. Fromthe presence of apparent stellar absorption features in the 2-0 CO overtone bandhead andthe 2.2 µ m Na emission features, we infer a late spectral type ( T eff ∼ v LSR = − . − . The warm ( T = 550 K) HCN absorption foundtoward GV Tau N in the L -band spectrum of this source is offset in velocity by ∼
13 km s − relative to the star and ∼ − relative to the GV Tau molecular envelope. The largevelocity of the star relative to its molecular envelope would cause it to escape the molecularenvelope on a timescale much shorter than its age unless the star is a spectroscopic binary.The small redshift (1 . ± . − ) of the warm molecular absorption relative to themolecular envelope velocity (assumed to be the systemic velocity) suggests that most of theabsorption arises in a disk atmosphere viewed close to edge on, although an origin in themolecular envelope is not ruled out completely. This interpretation can be tested with furtherobservations of GV Tau N. Observations at higher spectral resolution can better constrainthe radial velocity of the absorbing gas relative to the systemic velocity. Observations thatexplore the molecular abundances of the absorbing gas may also be useful in testing a diskorigin for the absorption.The authors wish to recognize and acknowledge the very significant cultural role andreverence that the summit of Mauna Kea has always had within the indigenous Hawaiiancommunity. We are most fortunate to have the opportunity to conduct observations fromthis mountain. We thank Al Conrad and other Keck Observatory staff who provided supportand assistance during our NIRSPEC run. We thank Sean Brittain for sharing results fromGibb et al. (2007) in advance of publication. We thank Elise Furlan for information on thesilicate and CO ice optical depths of Taurus Class I sources in advance of publication.Nathan Crockett contributed helpful advice on data-reduction strategies. Financial supportfor this work was provided by the NASA Origins of Solar Systems program and the NASAAstrobiology Institute under Cooperative Agreement No. CAN-02-OSS-02 issued through theOffice of Space Science. This work was also supported by the Life and Planets AstrobiologyCenter (LAPLACE). Facility:
Keck:II(NIRSPEC). 18 –
REFERENCES
Adams, F. C., Lada, C. J., & Shu, F. H. 1987, ApJ, 312, 788Baraffe, I., Chabrier, G., Allard, F., & Hauschildt, P. H. 1998, A&A, 337, 403Beck, T. L., McGregor, P.J., Takami, M., & Pyo, T.S. 2007, astroph
IRAF , National Optical AstronomyObservatory 20 –Massey, P., Valdes, F., Barnes, J. 1992 A Users’s Guide to Reducing Slit Spectra with
IRAF ,National Optical Astronomy ObservatoryMenard, F., Monin, J.-L., Angelucci, F., & Rouan, D. 1993, ApJ, 414, L117Movsessian, T. A., & Magakian, T. Y. 1999, A&A, 347, 266Muzerolle, J., Calvet, N., & Hartmann, L. 1998a, ApJ, 492, 743Muzerolle, J., Hartmann, L., & Calvet, N. 1998b, AJ, 116, 2965Najita, J., Carr, J. S., Glassgold, A. E., Shu, F. H., & Tokunaga, A. T. 1996, ApJ, 462, 919Reipurth, B., Rodr´ıguez, L. F., Anglada, G., & Bally, J. 2004, AJ, 127, 1736Rothman, L. S., et al. 1998, Journal of Quantitative Spectroscopy and Radiative Transfer,60, 665Royer, F., Grenier, S., Baylac, M.-O., G´omez, A. E., & Zorec, J. 2002, A&A, 393, 897Scoville, N., Kleinmann, S. G., Hall, D. N. B., & Ridgway, S. T. 1983, ApJ, 275, 201Siess, L., Dufour, E., & Forestini, M. 2000, A&A, 358, 593Sneden, C. 1973 PhD Thesis, University of Texas at AustinStapelfeldt, K. R., Krist, J. E., Menard, F., Bouvier, J., Padgett, D. L., & Burrows, C. J.1998, ApJ, 502, L65Stojimirovi´c, I., Narayanan, G., & Snell, R. L. 2007, ApJ, 660, 418Strom, K. M., Strom, S. E., Wolff, S. C., Morgan, J., & Wenz, M. 1986, ApJS, 62, 39Terebey, S., Shu, F. H., & Cassen, P. 1984, ApJ, 286, 529Van Cleve, J. E., Hayward, T. L., Miles, J. W., Gull, G. E., Schoenwald, J., & Houck, J. R.1994, Ap&SS, 212, 231White, R. J., & Hillenbrand, L. A. 2004, ApJ, 616, 998Whitney, B. A., Kenyon, S. J., & Gomez, M. 1997, ApJ, 485, 703Whittet, D. C. B., Bode, M. F., Longmore, A. J., Adamson, A. J., McFadzean, A. D.,Aitken, D. K., & Roche, P. F. 1988, MNRAS, 233, 321
This preprint was prepared with the AAS L A TEX macros v5.2.
21 –Table 1. GV Tau South: Radial Velocity MeasurementsOrder/Wavelength Lines(s) absorption emission v LSR notes( km s − )Order 37, 2.03 µ m S(2) 1-0 H no yes 10.3 line centroidOrder 36, 2.11 µ m Mg I, Al I yes no ... radial velocity poorly constrained in this regionOrder 35, 2.16 µ m Br γ no yes 3.7 line centroidOrder 34, 2.21 µ m Na I, Ti I, Fe I yes no 9.1 ± µ m S(0) 1-0 H no yes 8.5 emission extends south of the starOrder 33, 2.29 µ m Fe I, Ca I, Ti I yes no 10.0 ± µ m ν =2-0 CO yes yes ... radial velocity poorly constrained in this regionOrder 25, 3.07 µ m HCN, C H no no ...Table 2. GV Tau North: Radial Velocity MeasurementsOrder/Wavelength Lines(s) absorption emission v LSR notes( km s − )Order 37, 2.03 µ m S(2) 1-0 H no yes 5.9 line centroidOrder 36, 2.11 µ m Mg I, Al I no no ...Order 35, 2.16 µ m Br γ no yes -17.0 line centroidOrder 34, 2.21 µ m Na I, Si I, Sc I yes yes -4.5 ± µ m S(0) 1-0 H no yes 8.5 line centroidOrder 33, 2.29 µ m ν =2-0 CO yes yes -4.5 ± µ m HCN, C H yes no 8.7 ± λ rest λ topo v LSR % Transmission Equivalent Width µ m µ m km s − ˚AP11 3.0508033 3.0511622 8.73 0.75 0.12P12 3.0537670 ... ... 0.35 0.21P13 3.0567549 ... ... 0.20 0.12P14 3.0597676 3.0601268 8.65 0.98 0.14P15 3.0628043 3.0631660 8.86 0.70 0.15P16 3.0658660 3.0662214 8.21 0.85 0.19P17 3.0689521 ... ... 0.25 0.24P18 3.0720633 ... ... 0.45 0.15P19 3.0751988 ... ... 0.65 0.20P20 3.0783594 3.0787095 7.56 0.95 0.14P11,14,15,16 8.7 ± . v LSR
Referencekm s − GV Tau Cloud 6.2 ± ± ± ± ± ± ± |Fe |Mg |Ca |Fe |Ti |Ti|(2-0)|(2-0) Fig. 1.— Simultaneous spectra of GV Tau N (top) and GV Tau S (bottom) in the K -bandin order 33 (Fig. 1a), order 34 (Fig. 1b), order 35 (Fig. 1c), order 36 (Fig. 1d), and in the L -band order 25 (Fig. 1e). The relative flux ratio (f South /f North ) of the two components is ∼ ∼ K -band orders show emission features in bothobjects, while absorption lines dominate the spectra in GV Tau S. The L -band spectrumof GV Tau N (Fig. 1e) shows strong P-branch absorption lines of HCN (red tick marks,labeled) and several weaker C H lines (blue tick marks), while these features are absent inGV Tau S. Spectral regions with poor telluric transmission have been excised from Fig. 1e. 24 – |Sc |Na|Si|Sc |Na |Ti | |Ti |Fe |Ti |Ti|Na|Si? |Na | Fig. 1b. — Continued. 25 – |Mg |Fe ||
Fig. 1c. — Continued. 26 – |Si |Si |Mg |Al
Fig. 1d. — Continued. 27 –Fig. 1e. — Continued. 28 –
Topocentric Wavelength ( μ m) Fig. 2.— Two regions of the K -band of GV Tau S (black histogram) that show stellarphotospheric absorption lines. A spectral synthesis model (dotted blue) with T eff =3800 K,log g = 4.0, v sin i = 24 km s − (includes instrumental broadening), along with a veiling of r K = 2.5 fits the spectrum. The model fit in the top panel is at a radial velocity of v LSR =9.1 km s − (see Table 1). 29 – N o r m a li z e d F l u x GV Tau S
Topocentric Wavelength ( μ m) Fig. 3.— The CO overtone spectral region of GV Tau S (black histogram) showing bothemission and absorption components. We fit the observed spectrum using a model of diskemission and stellar photospheric absorption components at the same radial velocity. Ourdisk emission model (dashed red line) has the properties: v sin i (inner edge) = 88 km s − , R o /R i = 7, T = 3500 ( r/R i ) − . K, Σ = 0 . r/R i ) − . g cm − , where r is the disk radius.The stellar model (dotted blue line) has a temperature T eff =3800 K, gravity log g =4.0,rotation v sin i = 24 km s − , and veiling r K = 2 .
6. The radial velocity of the model fit( v LSR = 9 . − ) is consistent with the stellar radial velocity of GV Tau S reported inTable 4. The stellar model used in the fit is consistent with the fit to the Na and Mg/Alregions (Fig. 2). The combined disk emission and photospheric absorption model (solidgreen) is a reasonable fit to the observed spectrum. 30 – GV Tau N
Topocentric Wavelength ( μ m) Fig. 4.— The observed ν =2-0 CO spectrum of GV Tau N (black histogram) also show-ing both emission and absorption components. A model of CO emission from a rotatingdisk (dashed red line) is constructed from the following parameters: v sin i (inner edge) =95 km s − , R o /R i = 7, T = 3000 ( r/R i ) − . K, Σ = 2 . r/R i ) − . g cm − , where r is thedisk radius. The disk model is combined with a model of stellar photospheric CO absorp-tion (dotted blue line) at the same radial velocity having the physical parameters T eff =4100 K, log g =4.0, v sin i =15 km s − , and r K = 12.0. At v LSR = − . − (Table 4),the combination (solid green), is a reasonable fit to the observed CO feature in GV Tau N. 31 – N o r m a li z e d F l u x Fig. 5.— High resolution (R = 24,000) spectrum of GV Tau N (black histogram) with astellar photospheric model fit (dotted blue) to the absorption component that appears tobe present within the broad emission observed in the Na region of the K -band. The Naabsorption component is reasonably fit with the same stellar parameters used in modelingthe CO region (Fig. 4; T eff =4100 K, log g =4.0, v sin i = 15 km s − , and v LSR = − . − )and with veiling r K = 15. The emission spectrum (red histogram), obtained by subtractingthe model absorption from the observed spectrum, reveals broad emission in the Na linesand possibly the Si line. 32 –Fig. 6.— Resolved Br γ emission profiles for GV Tau S (top), and GV Tau N (bottom). Zerovelocity is defined as the stellar velocity, traced by the radial velocity of K -band absorptionlines (Table 4). The vertical dashed lines show the respective velocity centroids of the Br γ feature, while the solid horizontal lines illustrate the feature width (FWHM). The velocitycentroid relative to the stellar velocity is blueshifted by 5.7 and 12.5 km s − for GV Tau Sand N, respectively (Table 1 and 2). 33 – Topocentric Wavelength ( μ m) SN1”.2 GV Tau 6.5”
Extended S(0) 1-0 H emission Topocentric Wavelength ( μ m) SN1”.2 GV Tau 6.5”
Extended S(2) 1-0 H emission Fig. 7.— Spectral images of both components of GV Tau showing extended H emission.The S(0) 1-0 emission (2.2233 µ m, top) and the S(2) 1-0 emission (2.0338 µ m, bottom) areobserved at the positions of both GV Tau S and GV Tau N. In both images, a bright knotof emission is also clearly present 6 . ′′ L -band spectra of GV Tau N and the telluric standard, HR 1412 (black histograms)in the topocentric velocity frame where the observed HCN lines (red tickmarks) are redshiftedby 35 . − . Spectral regions with poor telluric cancelation have been excised from theGV Tau N spectrum. The blue tickmarks denote the expected positions of C H lines ( ν and ν + ( ν + ν ) , lower and upper sets, respectively). A synthetic model of HCN absorptionin a disk with temperature T = 550 K and column density N = 1 . × cm − (green line)provides a good fit to the observations. 35 – v LSR (km s -1 ) N o r m a li z e d F l u x Fig. 9.— HCN absorption lines in GV Tau N in spectral regions with atmospheric transmis-sion ≥