Organic Molecules and Water in the Inner Disks of T Tauri Stars
OOrganic Molecules and Water in the Inner Disks of T Tauri Stars
John S. CarrNaval Research Laboratory, Code 7211, Washington, DC 20375, [email protected] R. NajitaNational Optical Astronomy Observatory, 950 N. Cherry Avenue, Tucson, AZ 85716, [email protected]
ABSTRACT
We report high signal-to-noise
Spitzer
IRS spectra of a sample of eleven classical T Tauristars. Molecular emission from rotational transitions of H O and OH and ro-vibrational bands ofsimple organic molecules (CO , HCN, C H ) is common among the sources in the sample. Theemission shows a range in both flux and line-to-continuum ratio for each molecule and in theflux ratios of different molecular species. The gas temperatures (200–800 K) and emitting areaswe derive are consistent with the emission originating in a warm disk atmosphere in the innerplanet formation region at radii < 2 AU. The H O emission appears to form under a limitedrange of excitation conditions, as demonstrated by the similarity in relative strengths of H Ofeatures from star to star and the narrow range in derived temperature and column density.Emission from highly excited rotational levels of OH is present in all stars; the OH emission fluxincreases with the stellar accretion rate, and the OH/H O flux ratio shows a relatively smallscatter. We interpret these results as evidence for OH production via FUV photo-dissociation ofH O in the disk surface layers. No obvious explanation is found for the observed range in therelative emission strengths of different organic molecules or in their strength with respect towater. We put forward the possibility that these variations reflect a diversity in organicabundances due to star-to-star differences in the C/O ratio of the inner disk gas. Stars with thelargest HCN/H O flux ratios in our sample have the largest disk masses. While larger samples are required to confirm this, we speculate that such a trend could result if higher mass disks aremore efficient at planetesimal formation and sequestration of water in the outer disk, leading toenhanced C/O ratios and abundances of organic molecules in the inner disk. A comparison ofour derived HCN to H O column density ratio to comets, hot cores, and outer T Tauri star diskssuggests that the inner disks are chemically active.
Key words : accretion, accretion disks — circumstellar mater — infrared: stars — protoplanetarydisks — stars: pre-main-sequence
1. INTRODUCTION
Circumstellar disks around recently formed stars are the birthplaces of planets. Studies ofmolecules in these disks can provide insights into the planet formation process and the degree ofchemical processing in disks. For example, the abundance, distribution, and evolution of waterin disks are expected to have important effects on planetary system formation, including thegrowth of giant planets, the oxidation state of the inner disk, and the origin of water in terrestrialbodies. Protoplanetary disks could also be important sources for compounds of prebioticinterest, but the extent to which disks synthesize or preserve organic molecules is unknown. Thecold gaseous component of disks at large disk radii (> 40 AU) has been probed with millimeterspectroscopy (Dutrey et al. 2007). Hot molecular gas at small disk radii, including CO, H O andOH, have been studied with near-infrared spectroscopy (see Najita et al. 2007a for a review).The main planet formation region of the disk, however, is found at intermediate radii andtemperatures, where gas is best studied at mid- and far-infrared wavelengths. Relatively recentdevelopments have made gas at these radii accessible to study.Spectroscopy in the mid-infrared with the
Spitzer Space Telescope has opened a newwindow on gas in the inner planet formation region of disks by the detection of water andorganic molecules. Organic molecules were first detected in absorption. Lahuis et al. (2006a)reported absorption from warm (~300-700 K) HCN, C H , and CO in the spectrum of the youngstar IRS 46. The molecular absorption is considered to originate in the inner few AU of a diskviewed close to edge-on. Such absorption is rare (Lahuis et al. 2006a); additional examples include GV Tau (Gibb et al. 2007, 2008; Doppmann et al. 2008) and DG Tau B (Pontoppidan etal. 2008; Kruger et al. 2010).Carr & Najita (2008) were the first to discover emission from water, OH and organics(HCN, C H , CO ) in Spitzer spectra, as reported for the T Tauri star AA Tau. The temperaturesand emitting areas of the detected species were consistent with the emission arising in atemperature inversion region of the inner disk atmosphere within a few AU of the star, i.e., in theinner planet formation region of the disk. The relative column densities indicated a highabundance of water vapor in the inner disk, and comparisons to hot cores, comets, and chemicalmodels suggested that inner disks are chemically active. Unlike disk absorption, which requiresa special edge-on viewing geometry, the ability to detect molecules in emission means thatorganic molecules and water can potentially be studied in a large number of T Tauri disks.Other examples of mid-infrared molecular emission in T Tauri stars were reported fromsubsequent analysis of existing
Spitzer
IRS data. Salyk et al. (2008) presented the detection ofwater and OH emission from two T Tauri disks (DR Tau and AS 205). They also describedhigh-resolution near-infrared spectra of H O and OH for the same stars. The measured linevelocity widths were consistent with the emission originating within a few AU of the stars, andLTE analysis indicated large columns of warm gas. Pascucci et al. (2009) reported HCN andC H emission from several T Tauri stars, based on spectra taken in the low-resolution mode of Spitzer
IRS. They also presented observations of young brown dwarfs and the first detections ofC H emission from their disks.Studies of larger samples of young stars have fully established that mid-infrared emissionfrom water and other molecules is a common property of T Tauri stars. Pontoppidan et al.(2010a) presented a search for molecular emission in 73 stars. The sample included T Tauri starsin Chamaeleon, Ophiuchus and Lupus that are part of a dedicated Spitzer cycle-5 program onmolecular emission (PID 50641, J. Carr, PI), re-processed spectra from the c2d Legacy program(Evans et al. 2003), and archival spectra of Herbig Ae/Be stars. Emission from H O, HCN, CO ,C H , and OH is shown to be widespread among the T Tauri stars. On the other hand, the HerbigAe/Be stars lack detectable molecular emission, with upper limits on line/continuum ratios 3-10times lower than measured in T Tauri stars. However, hints of H O/OH emission are seen in the spectra of some Herbig Ae/Be stars and one shows CO emission. Detailed analysis of thissample of stars is presented in Salyk et al. (2011).In this paper we analyze in detail high-quality Spitzer spectra of a sample of eleven T Tauristars. These stars constitute the initial cycle-2
Spitzer program that carried out a deep search forH O and other gas emission lines in T Tauri stars, which included the observations of AA Taureported in Carr & Najita (2008). A simple LTE slab model is used to decompose the spectrainto contributions from water and organics and determine the emitting gas properties for eachmolecular species. We describe the characteristics and discuss the origins of the molecularemission. Ratios of our derived molecular column densities are used as estimates of relativeabundances to compare to results for hot cores, comets, outer T Tauri disks and protostellarenvelopes, and chemical models of disk atmospheres.
2. OBSERVATIONS
Spitzer
IRS (Houck et al.2004) using the Short-High (10–19 µ m) module, with a nominal resolving power of ~ 600. Allspectra were obtained in staring mode, which places the target at two nod positions along thelength of the 11x4.7” slit. The total integration times were set using the SpectroscopyPerformance Estimation Tool to achieve a continuum signal-to-noise ratio (S/N) ≥ Spitzer
GO program 2300. Theseobservations were designed primarily to search for H O emission from disks, before suchemission was known to be common among T Tauri stars. Results from this program for AA Tauwere first reported in Carr & Najita (2008). The program also obtained spectra for all targets inthe Long-High (19–37 µ m) module; analyses of those data will be presented in a future paper. The main part of this sample are 9 well-studied classical T Tauri stars in the Taurus-Aurigastar forming region, selected to cover a range in mass accretion rate. Spectral types range fromK3 to M0. The program included two additional sources, AS353A and V1331 Cyg, both ofwhich are extremely active stars known to have CO overtone, H O or OH emission in the near-infrared (Najita et al. 2007a, 2009; Doppmann et al. 2010; Prato et al. 2003). V1331 Cyg isoften presumed to have a spectral type and mass intermediate between T Tauri stars and HerbigAe stars due to the non-detection of photospheric features and a higher luminosity than typical TTauri stars (e.g., Hamann & Persson 1992). Of the Taurus sources, DG Tau has also shown COovertone and near-infrared water emission (Carr 1989; Najita et al. 2000), but the CO overtoneemission varies and is not always present (Greene & Lada 1996; Biscaya et al. 1997). V1331Cyg, AS353A, and DG Tau also have known jets and/or outflows.The star UY Aur is a 0.9” binary whose components have similar accretion rates (Hartigan& Kenyon 2003) and comparable brightness in the infrared; the binary flux ratio at 10 µ m hasvaried from 1.2 to 2.0 over a decade (Skemer et al. 2010). The 10 µ m silicate spectra of thecomponents have been spatially resolved by Skemer et al. (2010). However, we cannotdetermine the contribution of the components to the line emission spectrum in the unresolved Spitzer
IRS spectra. Two other stars have companions within the SH slit, DK Tau and RW Aur,with separations of 2.4 and 1.4”, respectively. The primary dominates the mid-infrared flux ineach case, with flux ratios ~ 10 (McCabe et al. 2006).Some properties of the sample that are used in plots later in this paper are listed in Table 1.The stellar accretion rates for most stars are those reported in Najita et al. (2007b), which placedaccretion rates from different literature sources on the same scale as Gullbring et al. (1998).Accretion rates are probably no better than a few tenths of a dex due to intrinsic variability. ForGI Tau, we adopted the accretion rate from Hartmann et al. (1998), which is on the same scale asGullbring et al. We scaled down the accretion rate for AS353A (Hartigan et al. 1995) by 0.12,based on the prescription of Najita et al. (2007b). We also made a rough estimate of theaccretion rate of V1331 Cyg based on the accretion flux derived by Eisner et al. (2007).Assuming a distance of 550 pc (as in Najita et al. 2009), the corresponding accretion rate is ~10 -6 M sun yr -1 . The disk masses are from the submillimeter measurements in Andrews & Williams(2005). the frame time. For eachtarget, frames are combined to form a mean image for the “A” slit position, the “B” slit position,and the combined background positions. When combining frames, the noise is calculated foreach pixel and any highly deviant values are rejected. In addition, pixels in the backgroundimage that have a standard deviation much higher than the distribution for all pixels are flaggedas being unstable; these are similar to the original definition for “rogue pixels” for the IRSarrays. These flagged pixels are used to make an unstable pixel mask, which is combined withknown permanently dead pixels for a total bad pixel mask. The mean background image is thensubtracted from the mean A and B images, which removes any background signal, dark current,and hot (but stable) pixels. These images are next divided by a flat-field. We use the pipelineflat-field that has been normalized in both the spectral and spatial dimensions using the IRAFtask apflatten. This avoids artificial increases in the noise where the flat drops quickly near theends of the slit and preserves the noise statistics. After division by the flat-field, the bad pixelmask is applied, replacing bad pixels by interpolation in the spectral direction. A visualinspection is also carried out and any missed and obviously bad pixels are fixed by hand. Themulti-order spectra are traced and extracted with optimal weighting using the IRAF task apall,making use of the noise statistics determined from the data. The wavelength scale for theextracted spectra was obtained by fitting a polynomial function to the pipeline wavelengthcalibration table.The extracted orders were divided by a spectral response function to remove theinstrumental spectral shape and fringing and to flux calibrate the spectra. The spectral responsefunction was customized for each slit position of each target using a suite of spectra for the IRSstandard stars ξ Dra and δ Dra. These data were obtained from the
Spitzer archive and processedin the same way as the target spectra, with an additional step of dividing the spectra by a corresponding model template provided on the
Spitzer
IRS website. The quality of fringecorrection is very sensitive to the choice of spectrum in the suite of calibrators; most likely this isdue to small differences in centering of stars in the slit. We divided the target spectrum by eachof the calibrator spectra to find those that minimized the residual fringing. For the SH module,we examined order 20, which usually shows the largest residual fringing and showed fewintrinsic emission features in the T Tauri stars. Those standard star spectra (typically 1–4) thatgave the smallest residual fringing for a target and slit position were averaged for the spectralresponse function. Standard star spectra that minimized the residual fringing also eliminatedsignificant curvature in the divided spectra at the ends of the orders.While this procedure provides a good correction for fringing in our sample of stars, there isoften some residual fringing seen in the highest (shortest wavelength) orders. To correct for thisremaining residual we used the routine IRSFRINGE (Lahuis et al. 2006b) to search for andremove any significant fringes near the main fringe frequencies of 3500 and 7800 cycles.However, no correction was applied for the 3500 cycle fringe in orders near the HCN and C H bands, because the R and P-branch spacing for these molecules is close to this frequency.Finally, the individual spectral orders were combined using appropriate weighting in thewavelength overlap regions. No scaling of individual orders was required because the match influx levels between adjacent orders was excellent. The A and B spectra were then averaged. Acomparison of the A and B spectra was used to estimate the noise in the mean spectrum. Thecontinuum S/N at 14 µ m ranged from 200 to 400.
3. OBSERVED SPECTRA
The complete Short-High spectra are presented in Figure 1. The low-frequency variations inthe continuum are due to dust emission features, in particular the amorphous silicate bandscentered at 10 and 18 µ m. Crystalline silicates are present in many objects, with forsteriteemission centered at 11.2 µ m, 16 µ m, and 11.9 µ m, and silica emission near 12.5 µ m. We alsofind weak and shallow emission features centered roughly at 13.7, 14.5 and (less discernible)15.5 µ m that sometimes underlay the forest of emission lines. These peaks correspond inwavelength with pyroxenes (Jager et al. 1998; Chihara et al. 2002), but curiously the corresponding peaks of the 10 µ m complex are not apparent in the spectra. No broadbandabsorption features are identified in the spectra.As was described by Carr & Najita (2008), rotational lines of H O dominate the forest ofemission lines seen above the dust continuum. Some of the rotational transitions that make upthe water emission features are identified in Figure 4 of Pontoppidan et al. (2010a). Relative tothe continuum, the water emission ranges from very prominent (e.g., RW Aur) to barelydiscernable (e.g., DO Tau). Water emission is not detected in three stars: DG Tau, V1331 Cygand AS 353A. The other molecules contributing to the emission spectrum are rotationaltransitions of OH and the ro-vibrational fundamental bending modes of HCN ( ν ), C H ( ν ) andCO ( ν ), with Q-branches at 14.0, 13.7, and 14.97 µ m, respectively.A continuum was subtracted from each spectrum in order to facilitate modeling andmeasurement of the emission features. The definition of the continuum is challenging becausethere are few true continuum regions due to the density of emission lines and the low spectralresolution. This is further complicated by the presence of the dust emission features, whichincreases the uncertainty in the continuum level for water lines near 12.5, 14.5, and 16 µ m. Inaddition, the HCN and C H bands overlay the wide and shallow 13.7 µ m dust feature. We usedour spectral modeling (§4) to provide a guide to the best continuum or pseudo-continuumregions. Polynomials were fit for 6 to 8 separate intervals over the SH range, and theseindividual continuum subtracted regions were joined into a final spectrum. All the continuumsubtracted spectra are presented in Figure 2 for the 12-16 µ m region, which contains the organicmolecular bands.The complexity of the molecular spectrum in this region is demonstrated in Figure 3, whichshows synthetic emission spectra for the individual molecules calculated using the methodsdescribed in §4. Figure 3(a) shows that there is significant blending of the molecular emissionfrom different species at the resolution of the IRS SH module. Failure to account for thisblending can lead to erroneous fluxes or false detections for a molecule, depending on therelative emission strength of different species. Furthermore, nearly all observed molecularfeatures are blends of individual rotational or ro-vibrational transitions (compare to Figure 3(b),the same model at high-resolution). Emission from OH is detected from every star in the sample, including those in which H Oemission is not detected. The most isolated and easily measured OH feature is at 14.64 µ m (seeFig. 2). This feature is a blend of unresolved components of the ground vibrational state withtotal angular momentum quantum number N=19. Other rotational OH emission features, rangingfrom N = 14 to as high as N = 29, can be seen in the spectra over the SH spectral range (seeNajita et al. 2010 for a list of OH wavelengths in this spectral region).Prominent HCN emission is clearly present in 5 stars of the sample (including the previouslyreported case of AA Tau). HCN emission may be present in UY Aur, but our modeling (§4)shows that a large part of the 14.0 µ m feature in UY Aur is due to H O and OH. C H emissionis present in at least 3 of these stars, but modeling is required to account for the H O and HCNemission in the 13.7 µ m feature in others.The most commonly observed organic molecule is CO . It is detected in all stars except DGTau, with a questionable detection in DO Tau. CO is observed in some stars without HCN orC H emission and in two that lack H O emission (V1331 Cyg and AS 353A). Other starsshowing only CO emission were reported by Pontoppidan et al. (2010a). Note that a pair ofH O emission features is blended with the blue wing of the CO Q-branch (see Figure 2).Najita et al. (2010) reported the identification of a Q-branch of HCO + at 12.06 µ m in the Spitzer spectrum of the transitional disk object TW Hya. We do not detect this HCO + band inany star in our sample of T Tauri stars. HCO + emission with the distance-corrected fluxmeasured in TW Hya would likely have escaped detection in our sample of Taurus objects,especially since the HCO + Q-branch is close to an observed water feature at 12.08 µ m. Thedetection of HCO+ emission from TW Hya was simplified by the notable lack of strong wateremission.One notable result is that the relative intensities of different molecular species differ fromobject to object. The spectra in Figure 2 show that the flux ratios of organic molecules to water,and the flux ratios of different organic bands, are not constant. Compare, for example, therelative strength of CO to HCN in UY Aur and BP Tau, or the relative strength of HCN to waterlines near 14.4 µ m in AA Tau and GK Tau. We return to this point in §5, where we make use ofrepresentative flux measurements for the emission from each molecule. There are also some atomic emission lines observed in the SH spectra. The [NeII] line at12.81 µ m is very common in this sample, but it is often quite weak and is close in wavelength toa number of weak molecular lines. It appears most prominently in AA Tau, UY Aur, and DGTau. Atomic hydrogen has a number of transitions in this wavelength region (see Najita et al.2010), with the strongest line at 12.37 µ m line. However, this line is partially blended with H Oemission features. The 17.93 µ m line of [Fe II] is the most prominent emission line in V1331Cyg and AS353A and is also present in DG Tau. In other stars, this [Fe II] line is coincidentwith a water emission feature.
4. SYNTHETIC SPECTRA e asthe radius of a circular region with this emitting area. The temperature and column density aredetermined by matching the relative strengths of spectral features (usually blends of lines) in thecase of H O and OH, or by fitting the shape of the Q-branch in the case of HCN, C H and CO .For each combination of temperature and column density, R e is adjusted to match the observedabsolute flux. We neglect continuum dust opacity within the line-emitting layer. The local linewidth was taken to be due to thermal broadening without assuming any additional turbulentmotions. The synthetic spectrum was first calculated at a resolution sufficient to sample the localline width and then smoothed to the resolution of IRS. The result was binned to the samewavelength sampling as the observed spectra to allow easy direct comparison.The molecular line data are taken from the HITRAN database (Rothman et al. 2004), withthe exception of HCN. For HCN we used a theoretical linelist from Harris et al. (2006), whichincludes higher vibrational bands than are present in the HITRAN linelist. The Harris et al. linestrengths must be increased by a factor of 6 to account for the state independent degeneracy factor (Fischer et al. 2003) as used in the HITRAN database and its associated partitionfunctions. Because the calculated line strengths from Harris et al. are somewhat larger than thecorresponding ones in HITRAN, we further scaled them by a factor of 0.88 in order to beconsistent with synthetic spectra that use the HITRAN data for HCN. For H O, we also ranmodels with the BT2 linelist (Barber et al. 2006). We found no significant differences betweenspectra computed with the BT2 and HITRAN linelists for the range of temperatures that fit theIRS water spectra. However, we caution that the HITRAN linelist will be inadequate atsignificantly higher (> 1000 K) temperatures. The partition functions were calculated usingavailable FORTRAN routines from the work of Fischer et al. (2003), which are consistent withthe definition of statistical weights in the HITRAN database.4.1.1 H O and OHIn modeling the water emission, we focused on obtaining the best fit to the H O spectrum inthe 12–16 µ m region. The energy levels for the H O transitions have no overall trend withwavelength and the lines in the 12–16 µ m interval are representative of the range of upperenergy levels within the wavelength coverage of the IRS. In our experience, this approachprovides a reasonable characterization of the average water emission properties in the full Spitzer spectrum (but see §4.2.2). By focusing on this wavelength region we can also optimize theremoval of the water emission spectrum in the region of the organic emission bands, which aidsin modeling these features. We also fit and remove an OH emission spectrum for the samepurpose.Synthetic H O spectra were calculated over a range in temperature and column density andcompared to the observed spectrum. The final H O model parameters were based on a best fit byeye to the spectrum. The approach used for the organic bands in §4.1.2, of calculating χ fromthe difference of the model and observed spectrum, did not work as well over the largewavelength interval for H O; variations in the zero continuum level and intervals with weak orno lines had undue influence on the fit. Table 2 gives the model parameters adopted for H O.An example synthetic spectrum is over-plotted on the observed spectrum of BP Tau in Figure 4.For GK Tau and DO Tau, H O is detected but the S/N in the emission lines is insufficient toconstrain the parameters well. For the purpose of subtracting a synthetic H O spectrum from the data for these two stars, scaled synthetic spectra were calculated using the mean temperature andcolumn density in Table 2.Emission from high rotational transitions of OH are present throughout the SH spectralregion in all of the stars in our sample. After subtraction of the synthetic H O spectrum, a modelOH spectrum was calculated to match the OH spectrum (see Fig. 4 for an example). We subtractthis OH synthetic spectrum from the data, though the only feature that has some impact on ourresults is the OH feature at 14.06 µ m (N=20), which blends with the red wing of the HCN band(see Fig. 3). 4.1.2 The Organics: HCN, C H and CO In modeling the HCN, C H and CO bands, we first subtract the best H O + OH model foreach object from the observed spectrum. H O can contribute significantly to the flux within theQ-branches of HCN and C H . In addition, both HCN and C H contribute emission to the Q-branch of the other, which can be important when the emission from one molecule issignificantly greater than the other. BP Tau is a good example where this blending produces anapparent C H feature. As can be seen in Figure 4, BP Tau shows emission near 13.7 µ m thatcould be mistaken for the C H band; however, the model in Figure 4 shows that H O emissionaccounts for most of this flux. Further modeling shows HCN emission can account for theremaining residual.The shapes of the Q-branches are determined by both temperature and column density.Synthetic spectra were calculated on a grid in T and N, scaling R e to match the model flux to theobserved flux integrated over the band. χ was calculated at each grid point from the differenceof the model and observed spectrum, and contour plots of χ were produced to determine thebest-fit values and confidence intervals. There is significant degeneracy between T and N infitting each of the three molecular bands, with increasing T offset by decreasing N.When calculating the grid of HCN models, a synthetic C H spectrum was subtracted fromthe observed spectrum to account for C H contribution to the HCN band. This fixed C H modelspectrum was calculated from a combined HCN and C H model that was close to the finalsolution based on a first round of fitting. Two examples of contour plots of χ for HCN are shown in Figure 5. The heavy contour is the 90% confidence limit, which we adopt as theuncertainty range.A similar procedure was followed for fitting the CO Q-branch. Before fitting the CO band, both the best H O+OH model and the final HCN+C H model were subtracted from thedata. Example χ contour plots for CO are shown in Figure 6.For C H , we assumed that HCN and C H have the same excitation temperature andemitting area, and adjusted the C H abundance relative to HCN to fit the C H band. Thereason for this approach is that T and N were not generally constrained at a useful level ofconfidence, probably because of the lower flux in the C H Q-branch and greater contaminationfrom H O and HCN. An exception is RW Aur, which has the largest C H to HCN flux ratio inthe sample (see §4.2.1). 4.2 Modeling Results4.2.1 Organic EmissionThe best-fit (minimum χ ) synthetic spectra for HCN and C H are shown in Figure 7 , over-plotted on the observed spectra with the best H O+OH model subtracted. In some cases (e.g.,DK Tau), significant residuals remain after fitting the HCN+C H model; many of theseresiduals coincide with H O features that are not well matched by the best-fit water spectrum.We were able to determine HCN parameters (Table 3) for 5 of the 6 stars in which HCN wasdetected. For UY Aur, the HCN band is only marginally detected (Fig. 7), but its measuredstrength is sensitive to the choice of the continuum fit. Emission from C H in UY Aur can onlybe considered an upper limit.The observed HCN Q-branches are a blend of multiple ro-vibrational bands, with significantcontribution to the emission from vibrational levels up to v=3 or 4 (depending on thetemperature) and rotational levels up to J=24 to 30. In the best cases (AA Tau and BP Tau inFig. 7), the R-branch lines can be discerned extending to shorter wavelengths to at least theJ=22–21 transition. Hence, HCN is both rotationally and vibrationally hot, and the fits in Figure7 are consistent with a similar rotational and vibrational excitation temperature. A possibleexception is GI Tau, which shows an excess at 13.85 µ m relative to the model. This could beexcess emission from the v=4 and 5 vibrational bands, but an unidentified spectral feature is also a possibility. The definition of the continuum is also an issue at this spectral resolution andwavelength, especially since HCN and C H sit atop the 13.7 µ m solid state feature which has anunknown spectral structure.The best fit model for the HCN band is always optically thick or marginally optically thick,with τ =0.4 to 4 in the Q(10) line of the v=1-0 band. The HCN column densities for the 5 starshave a range of about 3, with temperatures between 550 and 850 K. The contour plots in Figure5 show that as the temperature is decreased (or increased) a respective increase (or decrease) inthe column density can produce acceptably good fits over a range in parameters. Within the 90%confidence interval, the lower limit on the temperature corresponds to the upper limit on thecolumn density; these parameters are given in Table 3 as the ‘T minimum’ solution. The upperlimits on the HCN column densities are 2-5 times greater than the minimum χ solution.For each object there also exists a warmer, optically thin solution within the 90% confidenceinterval. This is where the contours in Figure 5 continue to lower column density at roughlyconstant temperature. Hence, while the fits place an upper limit on the temperature thatcorresponds to the optically thin case, there is no formal lower limit on the column density, aslong as an arbitrarily large area (but constant number of molecules) can be invoked to producethe total emission flux. A lower limit can be placed on the HCN column by assuming that theemitting area for HCN is no larger than that for H O. This is a reasonable assumption given thatthe derived temperatures for HCN are similar to or greater than those found for H O. This limit,with R e (HCN) = R e (H O), is given as the ‘optically thin’ solution in Table 3.The best-fit HCN solution for AA Tau in Table 3 is very close to that reported in Carr &Najita (2008), although the uncertainty in column density and temperature is much larger in thispaper. The reason for this difference is that the uncertainty given by Najita & Carr consideredeach parameter independently rather than looking at the two-dimensional confidence contours.This also meant that the optically thin solution was not considered as a possibility in Carr &Najita.As discussed in §4.1.2, the parameters for C H are not determined independently, but theC H column density relative to HCN is derived by assuming that C H and HCN have the sametemperature and emitting area. The ratio of the C H to HCN column densities is given in Table H column density ranges from 0.04 to0.4 that of HCN for the best-fit HCN solution.The assumption that C H has the same excitation temperature as HCN produces a good fitto the C H Q-branch in all three cases where the emission is well detected (see Fig. 7), but smalltemperatures differences (~ 100-200 K) cannot be ruled out. For RW Aur, we can independentlyfit the C H band. This gave T=615 K for C H , compared to 690 K for HCN; the C H columndensity is 3 times larger than that obtained assuming the same temperature and area. The χ contour plots overlap significantly within their 90 % confidence intervals. For the optically thinsolution, the temperature is 715 K for C H vs. 745 K for HCN. Hence, the modelinguncertainties are consistent with the same temperature for HCN and C H in RW Aur, thoughdifferences in temperature or emitting area would affect the derived relative column densities.A sampling of best-fit models for the CO emission is shown over-plotted on the data inFigure 8. The observed spectra in Figure 8 show differences in the width of the CO Q-branch,indicating a range in temperature and/or column density from object to object. However, themodeling is unable to place constraints on the CO column density. The examples in Figure 6show that the column density is not bounded within the 90% confidence contour in either theoptically thin or the very optically thick limits. For this reason, we do not report columndensities for CO .The models can place some constraint on the CO excitation temperature. In Table 5, wegive the CO temperature at the minimum χ solution and the allowable range for thetemperature. At the lower temperature end, the 90% confidence contour often extends totemperatures below 75 K (e.g., Fig. 6b), which is the minimum temperature at which theHITRAN partition functions can be applied. We also give the minimum radius for the emittingarea that was found within the 90% confidence contour. This radius corresponds to a marginallyoptically thick solution at the high temperature edge of the confidence area. The average of thebest-fit temperatures in Table 5 is 340 K, but there is a considerable range between objects and inthe allowable range for each object, and some stars (e.g., BP Tau) are very poorly constrained.The most noteworthy result is that the CO temperature is usually lower, often substantially less,than the temperatures derived for H O and HCN. O and OHThe single temperature and column density LTE model provides a good first order match tothe observed H O spectrum, though there are noticeable discrepancies. With the parameters forour best fits in the 12–16 µ m region, the largest disagreements are seen for a number of strong,high excitation lines (E u > 5000 K) whose strengths are over predicted by the model, particularlyin the 16-18 µ m region. Lower temperatures can provide a better fit to these lines, but at theexpense of poor fits in the 12-16 µ m interval. Meijerink et al. (2009) show in their non-LTEwater calculations that transitions with higher energy levels and larger Einstein A-values will bethe farthest from LTE. This strongly suggests that we are seeing the effect of sub-thermalpopulations in many higher rotational levels. Our modeling also shows that the water spectra aredominated by rotational transitions in the ground vibrational state, with only minor contributionsfrom transitions in the first vibrational state. The absence of excess emission from transitions inthe first vibrational state, compared to the thermal LTE model, implies that infrared pumping inthe fundamental ro-vibrational water bands is not significant.The spectra in Figure 2 are remarkably similar in the relative strengths of H O features fromstar to star, suggesting similar excitation conditions and optical depth. This impression isconfirmed by the best-fit models, which have a relatively narrow range in temperature andcolumn density (Table 2). In fact, within the uncertainties, the results are consistent with a singletemperature near 600 K. The H O column density shows some variation from star to star, but itsrange is less than a factor of 5. For each of the objects modeled, the main H O lines are opticallythick. The narrow parameter range for the H O lines is a more robust result than the absolutevalues for the parameters, because the actual values will be subject to systematic modelingeffects, such as assumptions about LTE and local line broadening. Salyk et al. (2011) also find asmall parameter range for H O, but with a somewhat different mean temperature.Despite the similar temperature of the water emission among the sources in the sample, theemission fluxes vary by a factor of 5 (Table 6). Based on the modeling shown in Table 2, this isprimarily the result of the projected emitting areas. The range of projected emitting radii is lessthan a factor of 2, with a mean on the order of 1 AU. Thus, the water emission is likely arisingfrom a similar range of disk radii. The typical temperature required to match the rotational intensity distribution of the OHlines in the SH spectra was 4000 K. This result is similar to the high rotational temperaturefound from the OH spectrum of the transition disk TW Hya (Najita et al. 2010). For TW Hya,Najita et al. suggested the possibility of prompt OH emission from the photo-dissociation ofH O. We will discuss this further in §6.3. A detailed analysis of the OH emission in thissample, which combines data from the SH and LH spectra, will be presented in a subsequentpaper.
5. MOLECULAR FLUXES O the mostcommon contaminant. In order to account for this blending, we used the synthetic spectrummodeling described in §4. The fluxes for all species other than H O were measured aftersubtracting the best H O model from the spectrum. Synthetic spectra for additional species weresubtracted as necessary, as described below. Measured fluxes for the molecular species aregiven in Table 6 , and fluxes for the atomic lines are in Table 7. The uncertainties in the fluxesinclude both the pixel-to-pixel noise in the spectrum and an uncertainty on the continuum level.The reported H O flux is the combined flux of the three H O features centered at 17.09,17.22 and 17.34 µ m. Each of these spectral features is composed of multiple H O transitions.The three H O features are in close proximity to each other, well separated from other knownmolecular or atomic species, and among the strongest H O features in the SH spectral region.The H O flux was measured between 17.075-17.385 µ m. For OH, we measured the spectralfeature at 14.64 µ m. This OH feature is the cleanest in the SH spectrum. It is dominated by ablend of the 4 components of the N=19 transitions of the ground vibrational state of OH.The Q-branches of HCN and C H each contain some contribution from transitions of theother. Within this data set, this has a substantial effect only when C H is weak or undetectablerelative to HCN, which is the situation for BP Tau and DK Tau. The HCN flux was measuredover the interval from 13.90 to 14.05 µ m, after subtracting the best-fit models for H O, OH and C H . Similarly, the C H flux was measured over the interval 13.65 to 13.735 µ m, aftersubtraction of the best-fit models for H O, OH, and HCN. The CO Q-branch has a pair ofprominent H O features that blend with the blue wing of the band. The CO flux was measuredfrom 14.93 to 14.997 µ m, with the blue edge of the interval chosen to minimize uncertainty fromthese H O features. A combined synthetic model for H O, OH, HCN and C H was subtractedbefore measuring the CO flux.The last molecular feature included in Table 6 is the S(1) transition of molecular hydrogen at17.03 µ m, which sits at a clean location between H O lines. The S(2) rotational line at 12.28 µ m, on the other hand, is coincident with both H O and OH emission features (see Fig. 3).These features normally dominate over any H emission in our sample and make it impossible tomeasure the S(2) line.Fluxes for atomic lines are given in Table 7. Emission from the [NeII] line at 12.81 µ m ispresent in nearly every object, but it is often not prominent in the spectrum. Because weakfeatures of H O, OH, HCN and C H are present near this line, we subtracted the combinedsynthetic molecular spectrum before measuring the [NeII] flux. The strongest HI line in thespectrum is at 12.37 µ m. However, this line is blended with H O features that can contribute upto one-half the combined strength of the feature, based on the synthetic H O spectrum. Thisblending could add additional, but unknown, uncertainty to the HI strengths in Table 7. Linefluxes for the [Fe II] line at 17.93 µ m are only given for the three stars that lack H O emission.For the rest of the objects, the strength of the 17.93 µ m water feature predicted by the syntheticspectrum is equal to or greater then the observed strength, and hence it is not possible to extractany information on [Fe II] emission for these stars.5.2 Absolute and Relative FluxesAs noted in §3, the spectra in Figure 2 show a significant star-to-star variation in the relativefluxes of different molecular species. Figure 9 plots ratios of the measured molecular fluxesfrom Table 6 against one another. The first two panels show the ratios of HCN, C H , CO , andOH with respect to H O. The last panel compares the organic bands among themselves byplotting the ratio of C H to HCN against CO to HCN. These plots clearly establish that there isa real range in the relative fluxes of molecular species, about one order of magnitude in mostcases. The range in the ratio CO /H O is somewhat smaller, a factor of four. The OH to H O flux ratio for all but one of the stars falls within a relatively narrow band, about a factor of two inwidth.The absolute fluxes for each of the molecular species in Table 6 are also diverse, showing arange of about one order of magnitude for the Taurus sample. There is also an order ofmagnitude range seen in the line to continuum ratios (or equivalent widths) of the emissionfeatures.
6. DISCUSSION O and CO are 89%. Such highdetection rates are comparable to the frequency of CO fundamental in classical T Tauri stars(Najita et al. 2003; Salyk et al. 2011). Emission in the Q-branches of HCN and C H is lesscommon, at 67 and 44 %. These molecular detection rates are comparable to those found inlarger samples of classical T Tauri stars (Pontoppidan et al. 2010a).As we initially reported for the star AA Tau, the analysis of the mid-infrared molecularemission in this larger sample of classical T Tauri stars yields emitting areas and temperaturesthat are consistent with an origin in the inner few AU of the circumstellar disk. We find that themolecule with the best constrained parameters, H O, has projected emitting radii of ~ 1 AU,implying de-projected radii < 2 AU. The best-fit values for HCN give smaller emitting areas, 0.1to 0.4 that of H O, and projected radii of 0.4–0.6 AU. However, equal areas for HCN and H Oare also consistent with the errors (§4.2.1). The temperatures for H O, HCN, and C H areconstrained to be in the range of 400 to 800 K. The temperature for CO , while often moreuncertain, is generally lower and in the range of 100 to 600 K. These temperatures are consistentwith models that calculate the vertical thermal structure of the molecular gas at radii within acouple of AU of the star (Glassgold et al. 2004; Nomura et al. 2007; Glassgold et al. 2009;Woitke et al. 2009). The interpretation of a disk origin for the molecular emission is corroborated by the velocityprofiles of H O lines that have been measured for a small number of objects with high-resolutionmid-infrared spectroscopy (Pontoppidan et al. 2010b; Knez et al. 2007). The line widths areconsistent with disk radii of ~ 1 AU and similar to those measured for the CO fundamentaltransitions. The high critical densities required for thermal excitation of the mid-infrared watertransitions (up to 10 cm -3 , Meijerink et al. 2009) provide additional support that the lines formin a disk atmosphere.Shocks associated with unresolved outflows within the IRS slit are not likely to explain themolecular emission, because the mid-infrared spectra for the T Tauri stars are dissimilar fromthose observed for outflows. The Spitzer
IRS spectra of protostellar outflows are generallydominated by emission in the rotational transitions of H and the atomic fine-structure [S I] 25 µ m, [Fe II] 26 µ m, and [Si II] 35 µ m lines (e.g., Neufeld et al. 2009; Dionatos et al. 2010).When hot water has been detected (Melnick et al. 2008), relatively few H O transitions are seen,all in the LH module at λ > 29 µ m. This water emission indicates low gas densities ~ 10 cm -3 and is consistent with non-dissociative shocks, in agreement with conclusions reached for wateremission observed from outflows in the far-infrared (e.g., Giannini et al. 2001; Benedettini et al2002; Nisini et al. 2010). A potential exception to the above is the possibility of a shockcontribution to the OH emission seen in the T Tauri stars, given that similarly high rotationaltransitions of OH have been detected in one Herbig-Haro object (Tappe et al 2008; see §6.3).Dissociative shocks could explain some of the atomic lines observed in our sample. Emissionfrom [Fe II] 26 µ m, and sometimes [Si II] 35 µ m, is present in stars with strong jets; the [Fe II]17.9 µ m transition is also strong in some of these stars, though this line is not apparent inoutflows. In addition, the exceptionally strong [Ne II] emission from DG Tau shows a blueshiftof 130 km s -1 in our data and is likely to originate from a jet (see Guedel et al. 2010 on jetcontributions to [Ne II] emission). 6.2 Water EmissionWe find that the properties of the water emission are fairly similar among the sources in oursample, with a narrow range in the LTE temperature (with a mean of ~ 600 K) and H O columndensities ~ 10 cm -2 (see also Salyk et al. 2011). The uniformity in the water emission spectra(optically thick emission with similar excitation temperature) is remarkable. Given possible variations in the grain properties, UV and X-ray irradiation, accretion heating, transportprocesses of disk atmospheres, one might imagine that inner disk atmospheres and the wateremission from them might be highly diverse. This does not appear to be the case.The narrow temperature range suggests that H O could play a role in regulating thetemperature of the gaseous disk atmosphere. The large H O line luminosities determined fromthe mid-infrared H O spectra of disks demonstrate that water is a major gas coolant of the diskatmosphere (Pontoppidan et al. 2010a). At the same time, water can contribute to the gas heatingthrough absorption of FUV radiation from the accretion shocks and near-infrared flux from thestar. If absorption in the FUV bands of water dominates the gas heating, then water self-shielding will place an upper limit on the total column of warm H O, at ~ 10 cm -2 (Bethell &Bergin 2009), a value that is similar to our derived water column densities. The thermal-chemical disk atmosphere models of Glassgold et al. (2009) reproduce well thewater temperatures and column densities that we derive. Particularly interesting is their findingthat the molecular emitting layer has a temperature structure that does not change much over arange in disk radii (0.25-2AU), while maintaining water column densities of 10 - 10 cm -2 .Water heating and cooling were not included in their models; it would be interesting to add theseprocesses to determine their effect on the temperature structure of the disk atmosphere.6.3 OH EmissionOur detection of high-rotational OH transitions suggests the role of UV irradiation inproducing the OH emission. The 14.65 micron feature, along with the other OH features in theSH spectrum, have upper energy levels several 1000 K above ground and relative intensitiesindicating a characteristic rotational temperature of ~4000 K (§ 4.2.2). Hot OH emission of thiskind is prominent in the SH spectrum of the transition object TW Hya and has been suggested toresult from the photo-dissociation of water by stellar UV (Najita et al. 2010). Photo-dissociationby FUV photons in the second absorption band of water ( λ = 1200-1400 Å) produces OHmolecules in highly excited rotational levels but predominantly in the ground vibrational state(Harich et al. 2000).The hot OH emission we measure in the SH module differs from the warm ( ≤ al. 2011). The presence of both hot and warm OH components can be understood if the OHmolecules produced by photo-dissociation first emit hot (prompt) OH emission and thenthermally relax to the temperature of the surrounding gas, producing cooler thermal emission(Najita et al. 2010).Glassgold et al. (2009) suggested the possible role of UV irradiation in dissociating diskwater vapor to produce abundant OH, noting that the OH column densities predicted by theirthermal-chemical models fell far short of those reported for the warm (~500 K) OH emission forAA Tau (Carr & Najita 2008). Bethell & Bergin (2009) have shown that the column densities ofthe warm OH emission agree well with the predictions of a simple UV irradiated slab model inwhich OH is produced by the photo-dissociation of water in the disk surface layers. In theirmodel, the maximum OH column is limited by the FUV opacity due to OH and H O self-shielding.Consistent with a role for UV radiation, we find that the 14.65 µ m OH flux increases withstellar accretion rate (Figure 10), which is a measure of the UV flux of the system. We alsofound that the flux ratio of the 14.65 µ m OH feature (non-thermal) and the 17 µ m water feature(thermal) is relatively constant among a majority of sources in our sample (Figure 9), suggestinga close connection. One possibility is that they are both driven by stellar accretion, with the OHproduced by UV dissociation and the water heated by UV radiation and/or accretion heating inthe disk. We note that the highest accretion rate objects in our sample show OH emission but notH O emission. This suggests that at sufficiently high FUV fluxes, H O may be destroyed infavor of OH in the warm disk atmosphere (Bethell & Bergin 2009). Extending such an effect tothe higher FUV fluxes for Herbig Ae/Be stars may explain the detection of OH and but not H Oemission from Herbig Ae/Be stars (Mandell et al. 2008; Pontoppidan et al. 2010a).An alternative interpretation is that some fraction of the OH detected from the higheraccretion rate objects in our sample may arise in shocks close to the star that are produced by theoutflows in these systems (DG Tau, AS353A, V1331 Cyg). Tappe et al. (2008) detected highlyexcited OH emission in the
Spitzer spectrum of young stellar outflow HH 211. That emission iscomparable, but with higher rotational levels, to that observed here. They attributed the origin ofthe OH emission to H O that is the photodesorbed from grains and then photodissociated by theUV radiation generated in the terminal outflow shock. H , CO ) isdifficult to interpret. The relative fluxes of the organic features and their strengths with respectto H O vary, often by an order of magnitude or more (Fig. 9). The ratio of the derived columndensity of C H to HCN shows a similar range (Table 4). A number of factors could potentiallyimpact the chemistry or structure of the disk atmosphere and produce variations in the relativemolecular fluxes. Some obvious physical processes include UV and X-ray irradiation of thedisk, accretion heating, and grain settling. We examined the position of points in Figure 9 withrespect to measured system parameters that track these processes to search for any patterns (e.g.,whether all stars with strong C H relative to H O have high X-ray fluxes). We considered themass accretion rate (Table 1) as a measure of UV radiation and accretion heating, the un-absorbed X-ray flux (Guedel et al. 2010), and the mid-infrared spectral index n as a measureof grain settling (Furlan et al. 2006).We did not find any obvious patterns that predict the behavior of the relative fluxes of theorganic molecules with respect to these parameters. Perhaps larger datasets are required toreveal the influence of a particular parameter, particularly if multiple parameters play a role.Alternatively, the lack of an obvious predictor could indicate the importance of anotherparameter that has not been explored.One possible factor, which lacks a direct observable, is variations in the C/O ratio of the gas.Such variations are expected to arise in the inner disk as a consequence of the growth andtransport of icy material over sizes ranging from grains to planetesimals and protoplanets. Asdescribed by Ciesla & Cuzzi (2006 and references therein), when icy grains in the outer diskgrow into approximately meter-sized bodies, they experience torques that lead them to migrateinward more rapidly than the surrounding gas. As they pass the “snow line” on their inwardjourney, they vaporize , pumping water (and oxygen nuclei) into the gas phase. If these bodiesinstead collide and grow to planetesimal size while they are still beyond the snow line, they dropout of the accretion flow and can thereby sequester water (and oxygen) in the outer disk.In the former case, the hydration of the gaseous inner disk leads to a reduction in the C/Oratio. In the latter case, the sequestration of water in the outer disk leads to an enhancement inthe C/O ratio. The detailed models of Ciesla & Cuzzi (2006) illustrate how the water abundance in the inner disk (and hence the C/O ratio) can be enhanced or reduced by an order of magnituderelative to its initial value, depending on the efficiency with which large bodies grow. HigherC/O ratios would tend to enhance the formation of organic molecules.As the efficiency with which large bodies form may depend on the initial conditions (e.g.,disk mass), we could expect a range in the C/O ratios of the inner disks among T Tauri stars ofsimilar age, such as in our sample. Higher mass disks, which may have experienced moreextensive planetesimal and protoplanet formation, might be expected to have a higher C/O ratioand enhanced abundances of organic molecules in the inner disk. Interestingly, the two sourcesin our sample with the largest HCN/H O flux ratios (AA Tau and BP Tau) are the two sourceswith the largest disk masses (Figure 11 plots HCN/H O vs. disk mass, using the disk masses inTable 1). Studies of larger samples are needed to explore the reality and implications of arelationship between HCN/H O and disk mass.
7. ABUNDANCES
Spitzer
IRS spectra only provide a global average ofthe disk emission, and it is not known to what extent different molecules occupy the samevolume, given the absence of spatial information. The LTE modeling implies comparabletemperatures for H O, HCN and C H , suggesting that they are similar in their distributions, butthe CO emission may arise from regions of cooler gas. Eventually, line profiles obtained withhigh-resolution spectrographs can provide information on the relative radial distribution of theemission by using velocity and assumed Keplerian motion as a proxy for radius. However,various molecular species, even those arising from the same disk radii, could still have differentvertical distributions in the atmosphere. Thermal-chemical models of disk atmospheres mayprovide insights into whether certain molecular species are likely to be co-spatial. The value of such insights depends critically on vetting the models by comparing their predicted structureswith observations of disks. The data presented here provide a small step in that direction.Secondly, it is worth noting that the infrared molecular emission is likely to probe only theatmosphere of the disk. The inner disks of classical T Tauri stars are optically thick at mid-infrared wavelengths, and the observed emission lines probably arise in a temperature inversionabove the disk photosphere. A key question then is whether the composition in the lineformation region is representative of the vertically integrated column of the disk. The chemistryin the upper disk atmosphere is likely to be different than that in deeper layers toward the mid-plane. Irradiation by UV and X-rays, in particular, is expected to be important; in addition, thedensities are lower, and gas to dust ratios will be higher than in the midplane, if grain settling hasoccurred. Vertical chemical gradients could, however, be smoothed or reduced by mixing, if thetimescale for vertical transport is short with respect to the key chemical timescales.In contrast to the likely true physical situation, the model used to determine the columndensity and temperature from the spectra assumes a plane parallel layer for each molecule that ishomogeneous in both vertical and radial directions and that the rotational and vibrational levelpopulations are described by a single Boltzman distribution. In spite of its simplicity, the LTEmodel does a surprisingly reasonable job of fitting the observed spectra. There are differencesbetween the calculated and observed H O spectra, which suggest that some higher energy levelsare sub-thermally populated (§ 4.2.2). This is not surprising given the real possibility that gasdensities in the upper atmosphere could be too low to thermalize all the level populations.Radiative pumping may also contribute to the excitation, though there is no clear evidence thatthis is occurring (§ 4.2). Finally, it should be noted that the adopted fitting procedures and othermodeling assumptions can also have considerable impact on the results (cf. Salyk et al. 2011).With these known caveats, we examine the relative column densities of H O, HCN, andC H for the T Tauri stars in our sample. In Figure 12 is plotted the column density ratio of C H to HCN vs. the ratio of HCN to H O for the five stars where we could model the HCN emission.The large error bars on the HCN/H O ratio reflect the large parameter range allowed by the fitson the HCN column density. The lower end on the error bars corresponds to the optically thinsolution with the assumption of equal HCN and H O emitting areas. The best-fit points for theHCN column are a few percent that of H O, with an average ratio of 0.04. If we treat the upper and lower errors bars as the range of possible systematic effects in fitting the HCN band, then thelikely range for HCN/H O is 0.005 to 0.15. The smaller errors bars for the C H /HCN columndensity ratio result from the assumption that C H and HCN have the same temperature andemitting areas. Under these assumptions, there is a real range in the C H to HCN ratio, from0.04 to 0.4, with an average of ~ 0.1.7.2 Comparison to Envelopes, Outer Disks, Comets, and Hot CoresWe can compare the inner disk abundances described above with the abundances of gaseousprotostellar envelopes, outer T Tauri disks, and comets. Since these systems probe the gas andsolid phase abundances at larger radii than the inner disk, they represent chemical precursors thatthe inner disk ultimately inherits as a consequence of envelope infall, disk accretion, and solid-body migration. Such a comparison may lead to insights into the extent of chemical processingin disks.The HCN/CO abundance ratios for inner disks appear to be a few orders of magnitude largerthan the values for low-mass protostellar envelopes and the outer disks of T Tauri stars.Prestellar cores, Class 0 sources, and Class 1 sources, which are the evolutionary precursors of TTauri stars, typically have HCN/CO mass ratios of 10 -4 or smaller in the gas phase, as measuredin millimeter emission lines (Jørgensen et al. 2004). The outer disks surrounding T Tauri starsare also found to have HCN/CO column density ratios of a few times 10 -4 or smaller (Thi et al.2004; Dutrey et al. 1997). In comparison, our inferred HCN abundance relative to H O for innerdisks is approximately a few percent. If we assume that the disks in our sample have CO/H Ocolumn density ratios ~1, as found for AA Tau and other T Tauri stars (Carr & Najita 2008;Salyk et al. 2011), then the HCN/CO ratios for inner disks are also a few percent. Similarly highHCN/CO ratios are derived from infrared absorption measurements towards two low-massyoung stars where the absorption is interpreted as arising in the inner region of a disk seen closeto edge-on: HCN/CO is 0.025 in IRS 46 (Lahuis et al. 2006a) and 0.014 in GV Tau N (Gibb etal. 2007, 2008). Some of the difference between the gas phase HCN/CO ratio of inner disks andthose of envelopes and outer disks may be accounted for by freeze out onto grains, which isthought to be substantial (e.g., Jørgensen et al. 2004).Cometary abundances provide insight into what was frozen out onto grains, because cometsare made up of icy material that was originally present in the Jupiter-Saturn region and in the outer solar nebula (the Kuiper Belt) during the formation of the solar system. Analyses of near-infrared spectroscopy of comets find a mean HCN/H O abundance ratio of 0.0026 and a meanC H /H O abundance ratio of 0.0024; thus the C H /HCN ratio is typically ~ 1 (DiSanti et al.2009; DiSanti & Mumma 2008). In comparison, the HCN/H O ratios we find for inner disks areall systematically higher than those found for comets, by approximately an order of magnitudeon average (see Figure 12). The C H /HCN abundance ratios for inner disks are alsosystematically lower than those of comets by a factor of 2–20, or approximately an order ofmagnitude on average.A larger HCN/H O ratio for gaseous inner disks, compared to both comets and gaseousouter disks, would suggest that the abundances of inner disks are not simply the result of gas andicy material migrating from the outer to the inner disk. Instead, further gas-phase processingmay occur within the snow line following the evaporation of icy material.Our results can also be compared to the abundances of hot cores, the molecular envelopessurrounding high-mass protostars. Hot cores have long been known to show a rich chemistrywith complex organic molecules. The same molecules we observe in emission in the T Tauristars have also been detected in absorption in mid-infrared bands towards several hot cores using
ISO (Lahuis & van Dishoeck 2000; Boonman & van Dishoeck 2003). The derived gastemperatures for the absorbing gas are in the same range as we found for the T Tauri stars. Theratios of reported absorption columns are plotted in Figure 12. Typical values are HCN/H O ~0.01, larger than those of comets and within the range found here for inner disks, and C H /HCN~ 0.5 (at the upper end of the range inferred here for inner disks).Thus the average inner disk atmosphere appears to have similar or slightly higher HCNabundance with respect to H O than do hot cores. However, the HCN abundance would be atleast an order of magnitude larger if CO were taken as the reference molecule. This is becausethe infrared absorption columns towards hot cores have an average CO/H O column density ratioof 10, whereas the disks in our sample are more likely to have CO/H O column density ratios ~1,(Carr & Najita 2008; Salyk et al. 2011).Similar chemical processes could be in operation in hot cores and inner disks. In hot cores,the warm gas phase follows the sublimation of molecules from icy grain mantles, where the richchemistry that is observed is the result of some combination of grain-surface and warm gas- phase reactions. Inner disks present an analogous situation, in that the warm gas-phasemolecules we observe represent material that was desorbed from grains as it migrated inwardfrom the outer disk. Because of these similarities, hot core chemistry may serve as a roughguide to the kind of chemical synthesis that may be occurring in inner disks. Numerous complexmolecules are detected in hot cores (e.g., Herbst & van Dishoeck 2009). Similar complexspecies could be present in inner disks.7.3 Comparison to chemical models of inner disksSeveral features of the molecular emission we observe are reproduced in recent chemicalmodels of inner disks. The thermal-chemical calculation of Glassgold et al. (2009) is able toaccount for the large column densities (~10 cm -2 ) and warm temperatures (~ 500 K) of thewater emission by invoking H formation on warm grains and (accretion-related) mechanicalheating in the disk atmosphere. Grain settling also plays a role in achieving large columndensities of warm molecular gas, by reducing the gas-grain cooling in the atmosphere (Glassgoldet al. 2004, 2009).In the thermal-chemical protoplanetary disk model of Woitke et al. (2009; PRODIMO),water is also present in the inner disk atmosphere although at much lower column densities(<10 cm -2 above 200 K at 1 AU) than we measure. Willacy & Woods (2009) also calculate thegas temperature in their model of disk atmospheres. They find very large columns of warm(>1000 K) H O (~10 cm -2 ) at 1 AU, much larger than those measured here. The differentdominant sources of heating and assumptions regarding grain abundances in these models maybe partly responsible for the very different predictions.The disk chemistry calculation of Agúndez et al. (2008) considers the formation of organicmolecules as well as water. In lieu of a thermal calculation, they adopt an existing model for thetemperature and density structure of the disk (D'Alessio et al. 1998, 1999), which is thenirradiated by a (primarily) interstellar FUV field. At 1 AU, the resulting HCN vertical columndensity at the disk surface is ~ 2x10 cm -2 , with a C H column density that is ~ 10 timessmaller and H O and CO columns ~2x10 cm -2 and ~8x10 cm -2 . These values are similar tothe values we measure for inner disk atmospheres. At 2 AU, the HCN vertical column density atthe disk surface is much smaller, ~ 5 x 10 cm -2 , with a C H column density that is ~ 5 timessmaller. If these results are a guide to what to expect in disk atmospheres, the declining HCN and C H abundances beyond 1 AU suggest that HCN and C H emission would be limited to theinner 1-2 AU of the disk, consistent with our results.An important difference between the observations and the Agúndez et al. disk atmospheremodel is the gas temperature. Because the gas temperature in the calculation is taken to equalthe dust temperature, the surface molecular layers are much cooler ( ≤
300 K) than are observed(~ 500–700 K). Warmer disk atmospheres would tend to further enhance the molecularabundances (Fig. 1 of Agúndez et al.).Willacy & Woods (2009) also study HCN and C H . At 1 AU, they find vertical columns ofwarm HCN (~4x10 cm -2 ) that are comparable to those measured here, accompanied by muchsmaller columns of C H (~9x10 cm -2 ) than we measure. Both species are present at warmertemperatures (>1000 K) than we measure. The differences between these models and theobservations may help to guide the future development of disk atmosphere models.
8. SUMMARY
One notable discovery of the
Spitzer mission is that molecular emission is common in themid-infrared spectra of T Tauri stars. In the sample of 11 classical T Tauri stars studied in thisreport, we find high detection rates of H O and OH rotational lines and ro-vibrational bands ofCO , HCN, and C H . Spitzer spectroscopy of larger samples of T Tauri stars (Pontoppidan et al.2010a) produce similar results. Our sample shows significant star-to-star variation (up to anorder of magnitude) in the absolute molecular fluxes, the equivalent widths, and the relative fluxratios of different molecules.We estimated gas temperatures, column densities and emitting areas using an LTE slabmodel. For H O, we find a similar temperature for each object, about 600 K, a narrow range forline-of-sight column density, ~ 10 cm -2 , and projected emitting areas of radius ~ 1 AU.Comparable or potentially higher temperatures are derived for HCN, in the range of 500-800 K.The C H Q-branch is consistent with having the same temperature as HCN. The best fits to theHCN spectra require a smaller emitting area for HCN than for H O, though similar emittingareas are not ruled out. Fits to the CO Q-branch are more uncertain, but imply lowertemperatures than the other molecules (100-600 K), with an average best-fit temperature of 350 K. This suggests that the CO emission arises in a cooler volume in the disk, either at larger radiior at a different vertical height.The gas temperatures, emitting areas, and high critical densities for the observed H Otransitions are consistent with gas in a disk atmosphere at radii within 2 AU of the star. Hence,the observed mid-infrared molecular emission traces gas in the inner planet formation region,radii that correspond to the terrestrial planet region in the Solar System and that fall within thegenerally accepted location of the snowline in protoplanetary disks.We discuss how the HCN/H O or HCN/CO column density ratios are roughly a few percent,which is about an order of magnitude larger than the average ratio derived for comets and at leasttwo orders of magnitude larger than those measured for the outer disks of T Tauri stars and low-mass protostellar cores. The HCN/H O ratio for inner disks is comparable to the ratios derivedfor hot molecular cores around massive protostars, but the HCN/CO ratios could be an order ofmagnitude larger. These comparisons suggest a picture where the gas-phase abundances in theinner disk are the result of chemically active disks, rather than simple infall of material from theprotostellar envelope and inward migration through the outer disk.The H O emission spectra are remarkably similar from star to star, which is reflected in thenarrow range of model parameters and implies limited variation in H O excitation conditions andoptical depth. High-rotational transitions of OH are present in all spectra and are likely due toprompt emission of OH following FUV photo-dissociation of water. The role of UV radiationand a connection between OH and H O are supported by the increase in OH emission flux withincreasing stellar accretion rate and the relatively small scatter in the OH/H O flux ratio. Wepropose that some of the wide variation in the relative emission strengths of the organicmolecules with respect to H O could be related to star-to-star differences in the C/O ratio (theoxygen fugacity ) of the inner disk gas that affects the organic chemistry. Differing C/O ratioscould result from enhancements or depletions of water in the inner disk depending on theefficiency of the growth of large bodies beyond the snowline (Ciesla & Cuzzi 2006). Theobjects in our sample with the largest HCN/H O flux ratios also have the largest disk masses.We speculate that this can be understood if higher mass disks are more efficient at formingplanetesimals and sequestering water in the outer disk, leading to higher C/O ratios and enhancedabundances of organic molecules in the inner disk. This work is based on observations made with the Spitzer Space Telescope, which isoperated by the Jet Propulsion Laboratory, California Institute of Technology under a contractwith NASA. Support for this work was provided by NASA. Basic research in infraredastrophysics at the Naval Research Laboratory is supported by 6.1 base funding. REFERENCES
Agúndez, M., Cernicharo, J., & Goicoechea, J. R. 2008, A&A, 483, 831Andrews, S. M. & Williams, J. P. 2005, ApJ, 631, 1134Biscaya, A. M., Rieke, G. H., Narayanan, G., Luhman, K. L. & Young, E. T. 1997, ApJ, 491,359Barber, R. J., Tennyson, J., Harris, G. J., & Tolchenov, R. N. 2006, MNRAS, 368, 1087Benedettini, M., Viti, S., Giannini, T., Nisini, B., Goldsmith, P. F., & Saraceno, P. 2002, A&A,395, 657Bethell, T., & Bergin, E. 2009, Science, 326, 1675Boonman, A. M. S. & van Dishoeck, E. F. 2003, A&A, 403, 1003Carr, J. S. 1989, ApJ, 345, 522Carr, J. S., & Najita, J. R. 2008, Science, 319, 1504Chihara, H., Koike, C., Tsuchiyama, A., Tachibana, S., & Sakamoto, D. 2002, A&A, 391, 267Ciesla, F. J. & Cuzzi, J. N. 2006, Icarus, 181, 178D'Alessio, P., Canto, J., Hartmann, L., Calvet, N., & Lizano, S. 1998, ApJ, 500, 411D'Alessio, P., Calvet, N., Hartmann, L., Lizano, S., & Canto, J. 1999, ApJ, 527, 893DiSanti, M. A., et al. 2009, Icarus, 203, 589DiSanti, M. A. & Mumma, M. J. 2008, Space Sci. Rev., 138, 127Dionatos, O., Nisini, B., Cabrit, S., Kristensen, L., & Pineau Des Forêts, G. 2010, A&A, 521, 7Doppmann, G. W., Najita, J. R., & Carr, J. S. 2008, ApJ, 685, 298Doppmann, G. W., Najita, J. R., Carr, J. S., & Graham, J. R. 2010, ApJ, submittedDutrey, A., Guilloteau, S., & Ho, P. 2007, in Protostars and Planets V, eds. B. Reipurth, D.Jewitt, & K. Keil, (Tucson: University of Arizona Press), 495Dutrey, A., Guilloteau, S., & Guelin, M. 1997, A&A, 317, L55Eisner, J. A., Hillenbrand, L. A., White, R. J., Bloom, J. S., Akeson, R. L., & Blake, C. H. 2007,ApJ, 669, 1072Evans, N. J., II, et al. 2003, PASP, 115, 965Fischer, J., Gamache, R. R., Goldman, A., Rothman, L. S., & Perrin, A. 2003, JQSRT, 82, 401Furlan, E. et al. 2006, ApJS, 165, 568Giannini, T, Nisini, B., & Lorenzetti, D. 2001, ApJ, 555, 40 Gibb, E. L., Van Brunt, K. A., Brittain, S. D., & Rettig, T. W. 2007, ApJ, 660, 1572Gibb, E. L., Van Brunt, K. A., Brittain, S. D., & Rettig, T. W. 2008, ApJ, 686, 748Glassgold, A. E., Meijerink, R., & Najita, J. R. 2009, ApJ, 701, 142Glassgold, A. E., Najita, J., & Igea, J. 2004, ApJ, 615, 972Greene, T. P. & Lada, C. J. 1996, AJ, 112, 2184Guedel, M. et al. 2010, A&A, 519, 113Gullbring, E., Hartmann L., Briceno C., Calvet N. 1998, ApJ, 492, 323Hamann, F., & Persson, S. E.1992, ApJ, 394, 628Harich, S. A., Hwang, D. W. H., Yang, X., Lin, J. J., Yang, X., Dixon, R. N. 2000, JChemPhys,113, 10073Harris, G. J., Tennyson, J., Kaminsky, B. M., Pavlenko, Ya. V., & Jones, H. R. A. 2006,MNRAS, 367, 400Hartigan P., Edwards S., Ghandour L., 1995, ApJ, 452, 736Hartigan, P., & & Kenyon, S. J. 2003, ApJ, 583, 334Hartmann L., Calvet N., Gullbring E., D'Alessio P., 1998, ApJ, 495, 385Herbst, E. & van Dishoeck, E. F. 2009, ARAA, 47, 427Houck, J. R. et al. 2004, ApJS, 154,18Jaeger, C., Molster, F. J., Dorschner, J., Henning, Th., Mutschke, H., Waters, L. B. F. M. 1998,A&A, 339, 904Jørgensen, J. K., Schoier, F. L., & van Dishoeck, E. F. 2004, A&A, 603Knez, C. et al. 2007, BAAS, 211, 5005Kruger, A. J., Richter, M. J., Carr, J. S., Najita, J. R., Doppmann, G. W., & Seifahrt, A. 2010,ApJ, submittedLahuis, F. & van Dishoeck, E. F. 2000, A&A, 355, 699Lahuis, F. et al. 2006a, ApJ, 636, L145Lahuis, F. et al. 2006b, C2d Spectroscopy Explanatory Suppl. (Spitzer Science Center:Pasadena).Mandell, A. M., Mumma, M. J., Blake, G. A., Bonev, B. P., Villanueva, G. L., & Salyk, C. 2008,ApJ, 681, L25McCabe, C., Ghez, A. M., Prato, L., Duchêne, G., Fisher, R. S., & Telesco, C. 2006, ApJ, 636,932 Meijerink, R., Pontoppidan, K. M., Blake, G. A., Poelman, D. R., & Dullemond, C. P. 2009,ApJ, 704, 1471Melnick, G. J., Tolls, V., Neufeld, D. A., Yuan, Y., Sonnentrucker, P., Watson, D. M., Bergin, E.A., & Kaufman, M. J. 2008, ApJ, 683, 876Najita, J. R., Carr, J. S., Strom, S. E., Watson, D. M., Pascucci, I., Hollenbach, D., Gorti, U., &Keller, L. 2010, ApJ, 712, 274Najita, J. R., Carr, J. S., Glassgold, A. E., & Valenti, J. A. 2007a, in Protostars and Planets V,eds. B. Reipurth, D. Jewitt, & K. Keil, (Tucson: University of Arizona Press), 507Najita, J. R., Carr, J. S., & Mathieu, R. D. 2003, ApJ, 589, 913Najita, J. R., Doppmann, G. W., Carr, J. S., Graham, J. R., & Eisner, J. A. 2009, ApJ, 691, 738Najita, J. R., Edwards, S., Basri, G., & Carr, J. 2000, in Protostars and Planets IV, eds. V.Najita, J. R., Strom, S. E., & Muzerolle, J. 2007b, MNRAS, 378, 369Neufeld, D. A., Nisini, B., Giannini, T., Melnick, G. J., Bergin, E. A., Yuan, Y., Maret, S., Tolls,V., Güsten, R., & Kaufman, M. J. 2009, ApJ, 706, 170Nisini, B. et al. 2010, A&A, 518, L120Nomura, H., Aikawa, Y., Tsujimoto, M., Nakagawa, Y., & Millar, T. J. 2007, ApJ, 661, 334Pascucci, I., et al. 2009, ApJ, 696, 143Pontoppidan, K. M., et al. 2008, ApJ, 678, 1005Pontoppidan, K. M., Salyk, C., Blake, G. A., Meijerink, R., Carr, J. S., & Najita, J. 2010a, ApJ,720, 887Pontoppidan, K. M., Salyk, C., Blake, G. A., Kaufl, H. K. 2010b, ApJ, 722, L173Prato, L., Greene, T. P., & Simon, M. 2003, ApJ, 584, 853Rothman, L. S. et al. 2004, J. Quant. Spectrosc. Radiat. Transf., 96, 139Salyk, C., Pontoppidan, K. M., Blake, G. A., Lahuis, F., van Dishoeck, E. F., & Evans, N. J., II.2008, ApJ, 676, L49Salyk, C., Pontoppidan, K. M., Blake, G. A., Najita, J. R., & Carr, J. C. 2011, ApJ, in pressSkemer, A. J., Close, L. M., Hinz, P. M., Hoffmann, W. F., Greene, T. P., Males, J. R., & Beck,T. L. 2010, ApJ, 711, 1280Tappe, A., Lada, C. J., Black, J. H., & Muench, A. A. 2008, ApJ, 680, L117Thi, W.-F., van Zadelhoff, G.-J., & van Dishoeck, E. F. 2004, A&A, 425, 955Willacy, K. & Woods, P. M. 2009, ApJ, 703, 479 Woitke, P., Kamp, I., & Thi, W.F. 2009, A&A, 501, 383 Fig. 1.—
Spitzer short-high (SH) IRS spectra for the entire sample. For the purpose of plotting,offsets have been applied to the spectra of several stars. The offsets in Jy are: -0.25 for GI Tau,-0.20 for GK Tau, +0.05 for DK Tau, -0.35 for RW Aur, -0.40 for UY Aur, +0.85 for V1331Cyg, and +0.60 for AS 353A. For DG Tau, the spectrum was also divided by a linear function inorder to reduce the spectral slope and allow plotting on the same figure; hence, the relative linefluxes and the spectral slope across the SH band are reduced by nearly a factor of two from thetrue spectrum for DG Tau. Fig. 2.— The SH spectra in the 12–16 µ m region after subtraction of the continuum. The spectrafor RW Aur and DG Tau have been divided by 2. The wavelength positions are indicated for theQ-branches of HCN, C H , and CO , the S(2) rotational transition of H , the 12.37 µ m line of HI, the 12.81 µ m [Ne II] line, and the 14.64 µ m rotational feature of OH. The remaining emissionfeatures are mostly due to rotational transitions of H O with some additional OH features. Fig. 3.— Synthetic spectrum for the 12–16 µ m region, showing the individual contributionsfrom each of the detected molecules. The temperatures, column densities and areas arecharacteristic of those derived for T Tauri stars in this paper. The H O spectrum is plotted as ablack histogram, HCN as red, C H as blue, CO as green, and OH as magenta. (a) For theresolution and sampling of the IRS SH module. (b) For a resolving power of R=30,000. TheH O flux was divided by 3 in order to display it adequately with respect to the other molecules. Fig. 4.— The synthetic H O model spectrum for BP Tau (dashed red line) over-plotted on theobserved continuum subtracted spectrum. The model parameters are given in Table 2. The solidbold (blue) line shows the model OH spectrum used for BP Tau. Fig. 5.— Contour plots of χ as a function of N(HCN) and T for fits to the HCN Q-branch in BPTau and DK Tau. The bold line is the 90 % confidence contour, the dashed line indicates thelocation of minimum χ , and the thin solid lines correspond to 1, 2, 3, and 4 σ . Fig. 6.— Contour plots of χ as a function of N(CO ) and T for fits to the’ CO Q-branch in AATau and UY Aur. The bold line is the 90 % confidence contour, the dashed line indicates thelocation of minimum χ , and the thin solid lines correspond to 1, 2, and 3 σ . Fig. 7.— Synthetic spectra for HCN and C H over-plotted on the data. The data (thick blackhistogram) are the continuum-subtracted spectra (Fig. 1) with the best H O + OH modelsubtracted. The model spectra (thin red histogram) are the best-fit parameters for HCN and C H in Tables 3 and 4. The residual spectrum for UY Aur is plotted here to show the marginaldetection of HCN, but a model spectrum was not calculated. Fig. 8.— Synthetic spectra for CO over-plotted on the data. The data (thick black histogram) arethe continuum-subtracted spectra (Fig. 1) with the best H O+OH+HCN+C H model subtracted.The model spectra (thin red histogram) use the minimum χ temperature solution in Table 5. Fig. 9.— Log-log plots of different flux ratios against one another for measured molecules. Alllimits (arrows) are 2- σ . (a) Flux ratio of C H /H O vs. HCN/H O. (b) Flux ratio of OH/H O vs.CO /H O. (c) Flux ratio of C H /HCN vs. CO /HCN. Fig. 10.— The OH emission flux vs. log of accretion rate. The accretion rate is in M sun yr -1 .The open square is for the binary UY Aur. The error bar in the lower right suggests a possiblerange in variability in the accretion rate. Fig. 11.— The flux ratio of HCN to H O plotted against disk mass. Inverted triangles are upperlimits (2- σ ) and correspond to stars with non-detections of HCN emission. The open square isfor the binary UY Aur, which also has a marginal detection for HCN. Fig. 12.— Line-of-sight column density of C H to HCN plotted against the column density ofHCN to H O. The squares are the column density ratios for T Tauri star disks from the LTE slabmodels in this paper. BP Tau, plotted as an open square, has an upper limit on the C H /HCNratio. The lower error-bar on HCN/H O for each star corresponds to the optically thin solutionwith equal emitting areas for HCN and H O. The triangles are column density ratios for hotmolecular cores around massive protostars derived from mid-infrared absorption measurements.The diamonds are ratios of the abundances in comets as determined from near-infraredspectroscopy. Data for hot cores and comets are from references given in text. Table 1Observations and Sample Parameters
Object FrameTime a (sec) Cycles b Flux c (Jy) SpT log M acc (M sun yr -1 ) Disk Mass(log M sun )AATau 31.5 24 0.32 K7 -8.5 -1.89BPTau 31.5 24 0.37 K7 -7.5 -1.74DGTau 6.3 24 5.25 K7 -6.3 -1.62DKTau d d e Notes . a Total SH integration time on source = 2x(Frame Time)xCycles. b Number of on-source cycles. Number of cycles in off position is 1/2 that on-source,except for GI Tau and GK Tau, where the on and off source cycles are equal. c Continuum flux at 14 µ m measured in this work. d Unresolved binary within the IRS slit, with secondary contributing < 10% of thebroadband mid-infrared flux. e UY Aur is an unresolved binary within the IRS slit with both components contributingsimilar amounts to the broadband mid-infrared flux. Table 2Results for H O Modeling a Line-of-sight column density b Radius of projected emitting areaObject T(K) N(H O) a (10 cm -2 ) R eb (AU)AA Tau 575 (50) 78 (20) 0.85 (0.12)BP Tau 650 (50) 78 (20) 0.83 (0.09)DK Tau 650 (50) 60 (12) 1.27 (0.15)GI Tau 575 (50) 42 (12) 1.16 (0.20)RW Aur 600 (80) 155 (37) 1.49 (0.24)UY Aur 600 (100) 179 (119) 1.04 (0.45) Table 3Results for HCN Modeling
Best Fit a Optically thick b Optically Thin c Object T(K) N(HCN)(10 cm -2 ) R e (AU) T(K) N(HCN)(10 cm -2 ) R e (AU) T(K) N(HCN)(10 cm -2 )AA Tau 690 6.5 0.28 570 13.0 0.28 840 0.41BP Tau 550 2.2 0.50 490 5.1 0.47 605 0.56DK Tau 540 6.0 0.37 350 31.0 0.54 705 0.20GI Tau 850 1.8 0.42 680 8.4 0.27 890 0.22RW Aur 690 2.1 0.59 500 8.6 0.51 745 0.27 Notes . Column densities are line-of-sight, and radii are that of the projected emitting area. a Minimum chi-square solution. b Lower bound on T and upper bound on N(HCN). c Optically thin solution with R e = R(H O) from Table 2. Table 4N(C H ) / N(HCN) Ratio Object Best Fit Opticallythick OpticallyThinAA Tau 0.14 0.07 0.26BP Tau <0.035 <0.017 <0.051DK Tau 0.04 0.006 0.10GI Tau 0.25 0.14 0.29RW Aur 0.38 0.19 0.43
Notes . Ratio of C H to HCN columndensity for the same cases in Table 3. Table 5Results for CO Modeling a Minimum chi-square solution. b Temperature range within confidence interval. c Minimum emitting radius within confidence interval.Object T a (K) T rangeb (K) R minc (AU)AA Tau 325 120-490 0.8BP Tau 425 75-800 0.1DK Tau 220 75-300 2.0GI Tau 100 75-360 0.9RW Aur 510 140-680 0.6UY Aur 200 75-390 1.5AS353A 600 160-800 0.4 Table 6Measured Molecular Fluxes
Object H O HCN C H CO OH H S(1)AA Tau 4.8 (0.4) 6.6 (0.2) 1.5 (0.2) 1.3 (0.1) 0.6 (0.1) 0.30 (0.06)BP Tau 6.6 (0.4) 5.4 (0.2) <0.5 0.9 (0.1) 0.8 (0.1) 0.31 (0.09)DG Tau <2.6 <2.8 <1.9 <1.7 6.3 (0.7) <0.70DK Tau 15.4 (0.9) 5.9 (0.5) 0.5 (0.2) 1.4 (0.2) 2.0 (0.3) <0.20DO Tau 4.9 (0.8) <1.4 <1.0 1.3 (0.4) 4.1 (0.4) <0.38GI Tau 6.0 (0.7) 7.0 (0.5) 2.2 (0.2) 0.7 (0.1) 1.0 (0.1) 0.59 (0.09)GK Tau 4.6 (0.8) <0.6 <0.5 0.4 (0.1) 1.3 (0.1) 0.35 (0.09)RW Aur 21.7 (2.1) 9.6 (0.8) 6.7 (0.5) 6.8 (0.4) 2.8 (0.5) 1.31 (0.25)UY Aur 12.1 (1.4) :3.7 (2.0) <0.7 (0.3) 4.5 (0.4) 3.1 (0.4) <0.59AS353A <0.4 <0.5 <0.4 1.7 (0.1) 1.0 (0.1) 0.48 (0.09)V1331Cyg <1.0 <0.6 <0.5 0.5 (0.1) 0.9 (0.1) <0.21
Notes . Fluxes are 10 -17
W m -2 . Upper limits are 2 sigma. Table 7Measured Atomic Fluxes
Object [NeII] a H I b [FeII] c AA Tau 1.25 (0.08) 0.20 (0.06) …BP Tau 0.28 (0.05) 0.90 (0.05) …DG Tau 25.7 (0.3) 7.4 (0.05) 6.4 (0.8)DK Tau 0.73 (0.13) 1.4 (0.1) …DO Tau 0.91 (0.18) 2.7 (0.4) …GI Tau 0.73 (0.08) 1.0 (0.1) …GK Tau 0.53 (0.07) 1.1 (0.1) …RW Aur 0.5: (0.2) 5.3 (0.4) …UY Aur 2.41 (0.44) 1.5 (0.3) …AS353A 1.10 (0.11) 3.1 (0.1) 3.7 (0.2)V1331CYg <0.2 2.2 (0.2) 8.3 (0.3)
Notes . Fluxes are 10 -17
W m -2 . Upper limits are 2 sigma.a The 12.81 µ m [Ne II] line.b The 12.37 µ m H I line.c The 17.93 µ m [Fe II] line is coincident with a H2