Measuring Organic Molecular Emission in Disks with Low Resolution Spitzer Spectroscopy
Johanna K. Teske, Joan R. Najita, John S. Carr, Ilaria Pascucci, Daniel Apai, Thomas Henning
aa r X i v : . [ a s t r o - ph . S R ] A p r Measuring Organic Molecular Emission in Disks with Low Resolution
Spitzer
Spectroscopy
Johanna K. Teske
Steward Observatory, University of Arizona, 933 N. Cherry Avenue, Tucson, AZ 85721, USA [email protected]
Joan R. Najita
National Optical Astronomy Observatory, 950 N. Cherry Avenue, Tucson, AZ 85716, USA [email protected]
John S. Carr
Naval Research Laboratory, Code 7211, Washington, DC 20375, USA [email protected]
Ilaria Pascucci
Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218, USA [email protected]
Daniel Apai
Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218, USA [email protected] andThomas Henning
Max-Planck-Institut f¨ur Astronomie, K¨onigstuhl 17, 69117 Heidelberg, Germany [email protected]
ABSTRACT
We explore the extent to which
Spitzer
IRS spectra taken at low spectral resolutioncan be used in quantitative studies of organic molecular emission from disks surround-ing low mass young stars. We use
Spitzer
IRS spectra taken in both the high and lowresolution modules for the same sources to investigate whether it is possible to define 2 –line indices that can measure trends in the strength of the molecular features in low res-olution data. We find that trends in HCN emission strength seen in the high resolutiondata can be recovered in low resolution data. In examining the factors that influencethe HCN emission strength, we find that the low resolution HCN flux is modestly cor-related with stellar accretion rate and X-ray luminosity. Correlations of this kind areperhaps expected based on recent observational and theoretical studies of inner diskatmospheres. Our results demonstrate the potential of using the large number of lowresolution disk spectra that reside in the
Spitzer archive to study the factors that in-fluence the strength of molecular emission from disks. Such studies would complementresults for the much smaller number of circumstellar disks that have been observed athigh resolution with IRS.
Subject headings: infrared: stars — (stars:) circumstellar matter — stars: pre-mainsequence — stars: formation — planetary systems: protoplanetary disks
1. Introduction
Circumstellar disks composed of gas and dust are ubiquitous around forming stars and arethe birthplace of planets. Since habitable planets are expected to form in warm inner disks ( < > ∼
10 AU of the star (e.g., Natta et al. 2007; Henning & Meeus 2009; Apai & Lauretta 2010).Observations of the warm gas within the inner disk are also necessary to fully understand thestructure and dynamics affecting disk evolution and planet formation (see Carr 2005; Najita et al.2007a; Millan-Gabet et al. 2007; Carmona 2010 for recent reviews). When such gas is viewed inemission from disks around T Tauri stars (TTS), which are optically thick in the continuum atsmall disk radii ( <
10 AU), the emission likely originates in a temperature inversion region at thedisk surface. The very inner regions of the gaseous disk ( < . O emission (e.g., Carr et al. 2004; Najita et al. 2000, 2009; Thi& Bik 2005). Observations of CO fundamental emission (e.g., Carr et al. 2001; Najita et al. 2003,2008; Blake & Boogert 2004; Brittain et al. 2007; Salyk et al. 2007, 2009; Pontoppidan et al. 2008)and UV transitions of H (e.g., Valenti et al. 2000; Ardila & Basri 2000; Herczeg et al. 2002, 2006;Bergin et al. 2004) have been used to probe larger disk radii (out to ∼ − R ∼ Spitzer Space Telescope (Houck et al. 2004) haverevealed that simple organic molecules (HCN, C H , CO ) and water (Lahuis et al. 2006; Carr &Najita 2008; Salyk et al. 2008) are present in the warm inner disk region ( . Spitzer Space Telescope depleted, it is no longer possible to obtainmore sensitive, mid-infrared spectroscopy of protoplanetary disks, making the
Spitzer archive theprimary source of new information on warm disk chemistry. However, with most of the data inthe archive taken in low-resolution mode, the question emerges: How much information regardingmolecular emission features can be extracted from the low-resolution observations? Pascucci et al.(2009) previously explored this question, showing that molecular emission could be detected in lowresolution IRS spectra of dozens TTS and lower-mass stars and brown dwarfs. They found thatHCN emission at 14 µ m was almost always brighter than C H emission at 13.7 µ m among T Tauristars, while only C H and no HCN was detected for lower mass stars and brown dwarfs. Thisled them to propose that there are differences in the relative abundance of molecular species as afunction of stellar mass.Here we build upon the work of Pascucci et al. (2009) by investigating the extent to whichwe can extract quantitative information from low resolution Spitzer
IRS spectra of inner T Tauridisks. To do this, we compare the molecular emission strength in a sample of high resolution IRSspectra of T Tauri stars with similar measurements of the same sources made in the low resolutionmode of IRS. If quantitative trends in the strength of molecular emission features can be recoveredfrom low resolution spectra, the archival
Spitzer
IRS data could be used to study the strength ofmolecular features in disks. Since, as we discuss below, a variety of physical and chemical processescan potentially affect the molecular emission strength, spectra of large samples of sources, suchas those available in the
Spitzer archive, are a valuable asset for demographic studies that seek toidentify the dominant processes influencing the molecular emission from disks. In § § §
2. Data Sets
For our comparison of high and low resolution data, we examined a small set of
Spitzer
IRSspectra of T Tauri stars in the Taurus-Auriga star-forming region. The higher resolution data weretaken with IRS in the short-high mode (SH, 10 − µ m, R ∼ − µ m, R ∼ . − M ⊙ yr − , to avoid the influence of highly energetic 4 –Fig. 1.— The 11 − µ m spectrum of AA Tau as observed in the SH ( R ∼ R ∼ < n − between − − n − between − n − and n − are the 6 − µ m and 13 − µ mcolors, respectively.To compare with the 8 SH spectra, we examined SL spectra of the same objects, originallyobserved as part of the Spitzer
GO Program 2 (P.I. Houck). We used the reduced SL spectra fromPascucci et al. (2009). The observations were originally published as part of a larger data set byFurlan et al. (2006) using an alternative reduction procedure that they detail. Since the molecularemission features were not the focus of the latter study, those spectra were not as reliable in the13 − µ m region.In order to determine the processes that might influence the strength of any observed molecularemission, we also examined SL spectra of an additional 10 sources from the Pascucci et al. (2009)sample that have stellar properties similar to those of the SH sample: accretion rates within anorder of magnitude of the typical T Tauri rate 10 − M ⊙ yr − (Hartmann et al. 1998), an absenceof close companions, and normal mid-infrared colors. While the full sample of 18 SL sources isrelatively uniform in infrared spectral shape, binarity, and spectral type, it exhibits more variety instellar accretion rate and X-ray luminosity (see Table 1). The stellar accretion rates in Table 1 arefrom Hartmann et al. (1998) and Najita et al. (2007b). Najita et al. adopted stellar accretion ratesfrom several literature sources and placed them on the same scale as Hartmann et al., providing aset of comparable, consistent values. The X-ray luminosities are from the recent reanalysis of G¨udelet al. (2010) of XMM-Newton and
Chandra observations of a large number of T Tauri stars. TheX-ray luminosities are for the 0.3 −
10 keV range and have been corrected for line-of-sight absorption(G¨udel et al. 2010). We also assume a distance of 140 pc. The properties of our full sample aredescribed in Table 1. 6 –Table 1. Our T Tauri SampleObject Spectral Type a log( ˙M ∗ /M ⊙ yr − ) c log(L X /erg s − ) e IRS ModeAA Tau K7 − − − · · · SLCX Tau M2.5 − · · · SLCY Tau M1 − · · · SLDK Tau K7 − − − − − · · · SLFZ Tau M0 b − · · · SLGI Tau K6 − d − − d − · · · SLIQ Tau M0.5 − − −
3. Analysis and Results3.1. SH vs. SL Measurements
As described in §
1, Pascucci et al. (2009) previously showed that the 14 µ m HCN feature isalmost always brighter than the 13 . µ m C H feature in T Tauri spectra, making it typically themost apparent feature at low spectral resolution. Thus, while we investigated the possibility ofdetecting the emission from several molecules (HCN, C H , H O) in the SL data, we chose to focusin this paper on HCN due to its greater detectability in our sample.To estimate the strength of the HCN feature, we defined a feature index based on the struc-ture seen in existing SH spectra and synthetic disk emission models (e.g., Carr & Najita 2008) toavoid contamination from neighboring molecular features. We selected the wavelengths 13 . µ mand 14 . µ m to define the boundaries of the HCN feature. To estimate the underlying con-tinuum, we found the average flux density in two neighboring regions, 13.776 − µ m and14.090 − µ m, assigned these values to the midpoint of each region, and performed a linear fitto these two midpoints. We subtracted the continuum estimate from the spectrum and summed theresulting spectrum within the wavelength boundaries of the feature to obtain the feature flux. Theequivalent width of the feature was calculated in a corresponding way. These values are reportedin Tables 2 & 3. In the SH spectra, the continuum regions each span three pixels and the HCNfeature spans fifteen pixels, while in the SL spectra the continuum regions each span less than onepixel and the HCN feature spans three pixels (see Figure 2).The errors on the SH spectra are described in Carr & Najita (2011). They are derived fromthe average RMS pixel variation around 14 µ m. To estimate the errors on the SL spectra, weperformed a linear fit to the continuum over ∼
15 pixels between 13 µ m and 14.2 µ m, excludingthe regions around HCN (13.885 µ m − µ m) and C H (13.609 µ m − µ m), and usedthe standard deviation of the difference between the observed spectrum and the fit as a measureof the pixel-to-pixel noise. We quote this measurement as our 1 σ errors in Table 2. These errorsare generally smaller than those reported by Pascucci et al. (2009), who adopted an error for eachpixel based on the difference in flux observed in a small number (2) of nod positions. In Table 2 we show the SL fluxes, equivalent widths, and errors determined using the featureand continuum regions defined above. Objects for which we have SH data are listed in Table 3 This latter error estimate can be affected by flux differences in the two beam positions if the object is notequally centered in the slit in each beam position. Some of the spectra appeared to suffer from this effect as theestimated errors were often larger than the pixel-to-pixel differences in the final spectrum (e.g., CW Tau, CY Tau,DN Tau, GI Tau, GK Tau, IP Tau). While our errors are generally smaller than the Pascucci et al. (2009) errors, ouradopted errors may still overestimate the true error. That is because our approach assumes that the true spectrum isfeatureless in the region used to estimate the pixel-to-pixel variation (i.e., in the regions around the HCN and C H features), whereas the spectra may in fact have a rich spectrum of weaker emission features (Fig. 1). We return tothis issue below. µ m HCN feature. The dotted vertical lines indicate theleft and right continuum regions, and the vertical lines define the HCN feature, as listed in §
3. 9 –along with their flux and equivalent width measurements.To understand any difference between these two data sets, we first smoothed the SH spectrato the approximate resolution of the SL spectra ( R ∼ ∼
50% of those measured for the SH data (Fig. 3a).The lower values for the smoothed/resampled data are the result of the neighboring line emissionfrom water and other features (Fig. 1; Carr & Najita 2008, 2011; Pontoppidan et al. 2010), whichblends into a pseudo-continuum at lower spectral resolution, diluting the HCN flux and equivalentwidth. Because the neighboring line emission can vary from source to source in both shape andstrength relative to HCN (stronger or weaker neighboring emission lines), there is dispersion aboutthe ∼
50% average value.We would expect that the effect of the lower spectral resolution would lead to a similar dif-ference between the SH measurements and those made on the real SL data. An additional factorin comparing the SH data with the (non-contemporaneous) SL data is the possibility of time vari-ability in the HCN and/or non-HCN line emission spectrum, which would increase the dispersionbeyond that arising from the lower resolution alone. This is indeed the case. The comparison ofthe SL equivalent widths shows more dispersion than the smoothed/resampled data when com-pared against the SH data (Fig. 3b). Figure 3c shows that the lower average equivalent width ofthe smoothed/resampled data does indeed capture the trend of the reduction in the SL equivalentwidth. Similar results are found for the HCN fluxes of the SH, smoothed/resampled, and SL datasets. The HCN equivalent width and flux measurements from the SH and SL data are well correlated(Figure 3 and Table 4). To assess the significance of the apparent trends, we use two correlationcoefficients, Kendall’s rank correlation coefficient, τ Kendall , and Pearson’s linear correlation coeffi-cient, r . The former, τ Kendall , is a non-parametric statistic that measures the degree of correlationbetween two variables; values close to unity signify a tighter correlation, while values close to 0signify no correlation. Our calculated τ Kendall -values are all ≥ P values thatcorrespond to τ Kendall , P τ , represent the confidence levels of the coefficient – a smaller P valueindicates a lower probability of a false conclusion. Pearson’s r -value measures how closely twoparameters fit a linear relationship (assuming the parameter distributions are normal). The closer | r | is to unity, the more linear the relationship. Our calculated r -values are all ≥ p rand (as a %), the probability that our measurementsare randomly distributed (and thus uncorrelated). The calculated p rand values, ∼ µ m) (10 − µ m)AA Tau 4.00 ± ± ± ± ± ± − ± − ± − ± − ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± − ± − ± ± ± − ± − ± ± ± ± ± ± ± µ m) (10 − µ m) (mJy- µ m) (10 − µ m)AA Tau 4.43 ± ± ± ± ± ± ± ± ± ± ± ± ± ± − ± − ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Parameters n a r b τ Kendall c P τ d p rand (%) e SH vs. SL EW 8 0.904 0.714 0.019 1.77SH vs. SL Flux 8 0.797 0.590 0.108 7.01Smoothed vs. SL EW 8 0.908 0.714 0.019 1.64 a The number of objects used for calculation of the statistic. b Pearson’s r linear correlation coefficient, a measure of how closelytwo variables fit a linear relationship. | r | values closer to 1 indicatebetter correlation. c Kendall’s τ rank statistic, a measure of the degree of correlationbetween two parameters that does not assume normally distributeddata. The closer | τ | is to 1, the better the correlation. d Two-sided P value, the probability (assuming no correlation) ofobtaining a result at least as extreme as the result that is actuallyobserved. The lower the P value, the higher the probability of corre-lation.. e Probability of getting r from a random distribution of size n.
Table 5. Correlations Between Stellar Parameters & SL HCN Emission
Parameters Points Rejected n a r b τ Kendall c P τ d p rand (%) e χ q log( ˙M ∗ /M ⊙ yr − ) vs. SL Flux – initial fit none 18 0.534 0.386 0.028 8.42 1.17 0.280log( ˙M ∗ /M ⊙ yr − ) vs. SL Flux – final fit 2 16 0.655 0.567 0.178 2.87 0.753 0.721log(L X /erg s − ) vs. SL Flux – initial fit none 12 0.403 0.382 0.099 34.19 1.37 0.186log(L X /erg s − ) vs. SL Flux – final fit 1 11 0.648 0.587 0.016 9.80 0.676 0.731Spectral Type vs. SL Flux – initial fit none 18 − − − − X /erg s − ) none 12 − − ∗ /M ⊙ yr − ) none 18 − − X /erg s − ) vs. log( ˙M ∗ /M ⊙ yr − ) none 12 −
12 –Fig. 3.— The comparison SH, smoothed/resampled and SL HCN equivalent widths (Tables 2 and3). A unity line is shown for reference in each plot.To summarize, while the SL measurements do not recover the HCN flux of the SH spectra, ourresults suggest that studies using SL spectra can recover quantitative trends in molecular emission 13 –strength seen in higher resolution observations. The SL HCN measurements may therefore trackthe SH HCN measurements well enough to reveal interesting trends when compared with other TTauri properties. We explore this possibility in the next section.Although we refer to our SL measurements as “fluxes” and “equivalent widths”, it is moreuseful to think of these quantities as line indices. The index can be positive (e.g., if there is HCNemission) or negative. The latter could occur either if there is either true absorption (e.g., as inIRS 46; Lahuis et al. 2006) or emission from other features in the “continuum” regions that areused to define the index.In addition to the HCN emission feature, we also attempted a similar analysis for C H ( ∼ µ m) and an H O feature at ∼ µ m. We were unable to recover with the SL data emissionstrength trends seen in the SH data for these features, probably because they are weaker than HCNin spectra of T Tauri stars (Pascucci et al. 2009). We note that greater success may be possiblewith data analysis techniques more sophisticated than those used here. We also note that when weperformed the same analysis using the Furlan et al. (2006) reduction of the SL data we did not finda significant correlation between SL and SH emission strengths, demonstrating that the specificdata reduction procedure for the SL data can influence the ability to recover trends in SH data. In our sample of SL spectra, the HCN flux varies from non-detections (below ∼ µ m) toover 5 mJy- µ m, and the HCN equivalent width varies over approximately an order of magnitude(see Table 2). What causes the strength of the HCN feature to differ in these systems? Althoughthe sources have many similar properties (e.g., they have similar stellar masses and spectral types),the stellar accretion rate ( ˙M ∗ ) and X-ray luminosity (L X ) do vary across the sample, as may otherphysical properties not described here. To investigate whether stellar accretion rate and X-rayluminosity play a role in determining the HCN emission strength, we compared the HCN fluxes ofthe sources in the SL sample with their values of ˙M ∗ and L X from the literature (Table 1).In Figure 4, panels (a), (b), and (c) plot SL HCN flux against stellar accretion rate, stellarX-ray luminosity, and spectra type, respectively. Panels (d), (e), and (f) plot these three quantities– accretion rate, X-ray luminosity, and spectral type – against each other. The distribution ofpoints suggests possible trends between SL HCN flux and the quantities in Fig. 4a, b, c, althoughthese trends, if they exist, are not extremely tight. The lack of a tight correlation is perhaps notsurprising since many physical and chemical processes (e.g., heating that is unrelated to accretion,chemical synthesis, photodestruction, excitation conditions) can potentially affect the strength ofany given molecular emission feature. As a result, outliers in any trend are to be expected, e.g., ifsome systems have managed to synthesize more or less HCN. We therefore employed the followingsimple rejection scheme when examining our data for possible trends: we performed a weightedlinear fit, including uncertainties in both the x- and y-directions, to all of the data in Fig. 4a, b, and 14 –c and iteratively rejected the top one to two outliers, all of which were above 3.3- σ . The outliersare plotted as open triangles in Fig. 4, and a summary of the fit statistics is given in Table 5. Table5 also reports the reduced χ of the linear fit and q , the probability that a correct model wouldgive a χ value equal to or larger than the observed χ .In the case of Fig. 4a, where we plot SL HCN flux versus stellar accretion rate, the Pearson’s r -value associated with all of the data points shown is 0.53 and the τ Kendall value is 0.39 (see Table5). Rejection of the top two outliers at 3.6 σ and 3.8 σ (open symbols) resulted in a Pearson’s r -value associated with the remaining data points of 0.66 and the τ Kendall value of 0.57 (see alsoTable 5), suggesting a potential positive correlation between stellar accretion rate and HCN flux.Even with outlier rejection, there is still significant scatter, which is perhaps to be expected, asdiscussed above. In addition, the difficulty in determining precise veiling and bolometric correctionslikely introduces systematic uncertainty in stellar accretion rate measurements, as discussed byHartigan et al. (1991) and Gullbring et al. (1998). These authors also note that time variability,as a result of intrinsic fluctuation in the accretion rate or the modulation of a nonaxisymmetricmagnetosphere, can contribute to the uncertainty; they suggest a cumulative uncertainty of ∼ r -value for all of the data points is 0.40 and the τ Kendall value is 0.38 (see Table 5). Rejection ofthe top outlier at 3.3 σ (open symbol) resulted in a Pearson’s r -value associated with the remainingdata points of 0.65 and the τ Kendall value of 0.59 (see also Table 5), suggesting a potential positivecorrelation between stellar X-ray luminosity and HCN flux. The larger p rand and P τ for these data(compared to those shown in Fig. 4a or 4c; see Table 5) are partly a result of the smaller samplesize n (12 versus 18 objects). Some of the scatter in this plot is likely the result of variability inL X . G¨udel et al. 2010 notes that the range of uncertainty in X-ray flux determination is dominatedby variability on various time scales, and (apart from singular flares) is typically characterized byflux variations within a factor of two from low to high levels. We represent this uncertainty by thehorizontal error bar in the lower left corner Fig. 4b.Fig. 4c shows SL HCN flux versus stellar spectral type. The associated Pearson’s r -value for allof the data points is − τ Kendall value is − σ (open symbol) resulted in a Pearson’s r -value associated with the remaining data points of − τ Kendall value of − ∼ r -value for all the objectsplotted is − τ Kendall value is − X and L ∗ in pre-main sequence stars,with L X /L ∗ ∼ − − − (Telleschi et al. 2007; Preibisch et al. 2005). Among stars in Myr-oldpopulations such as those in our sample, L ∗ also decreases with later spectral type (Stelzer &Neuh¨auser 2001; Preibisch et al. 2005; Winston et al. 2010), so L X would also be expected todecrease with later spectral type in our sample. Thus, the trend in Fig. 4c may not reflect afundamental relationship between HCN flux and spectral type, but instead results from the twounderlying relations between L X and HCN flux (Fig. 4b) and L X decreasing with later spectral type(Fig. 4d).Another possibility is that the luminosity associated with accretion (L acc ) is decreasing withlater spectral type and this is what drives the trend of HCN flux with spectral type. The averageaccretion rate is known to decrease with decreasing mass (later spectral type), but the spread atany given mass is ∼ two orders of magnitude (Muzerolle et al. 2005). In Fig. 4e, we plot stellaraccretion rate versus spectral type. There is no strong correlation (see Table 5) within the narrowrange of spectral type of our sample, consistent with Muzerolle et al. (2005). In Fig. 4f, we plot thestellar X-ray luminosity versus the stellar accretion rate. The comparison also shows no correlation(see Table 5).Because our data set is small (and our analysis methods explorative), larger samples of IRSspectra are needed to confirm that any trends exist and test whether any of the fits proposed arereasonable representations of the trend. Our sample is artificially sparse at high accretion rates dueto the difficulty in measuring HCN emission from low resolution spectra of high-accretion sources;their enhanced continuum flux reduces the contrast of emission features above the continuum. Thusit would be useful to expand the sample to include more sources covering the same range of stellaraccretion rates as well as a larger range of accretion rates. If HCN flux and stellar accretion rateare correlated, we would expect that sources with accretion rates < − M ⊙ yr − would have lowto undetectable HCN fluxes. Similarly, we would expect that sources with X-ray luminosities below ∼ . × erg s − would not show detectable HCN, and that sources with X-ray luminositiesabove ∼ . × erg s − might continue to show enhanced HCN emission with increasing X-rayflux. 16 –Fig. 4.— The comparison of stellar parameters and SL HCN flux. Open triangles designate outliersidentified by iterative rejection. (a) SL HCN flux versus stellar accretion rate ( ˙M ∗ ). (b) SL HCNflux versus stellar X-ray luminosity (L X ). (c) SL HCN flux versus spectral type. (d) L X versusspectral type. (e) ˙M ∗ versus spectral type. (f) L X versus ˙M ∗ . 17 –
4. Discussion
We find that SH and SL measurements of the 14 µ m HCN feature are correlated in our smallsample of T Tauri stars. Our results support the work of Pascucci et al. (2009), who used these SLspectra as part of their larger sample to deduce the differences between gaseous disks surroundingT Tauri stars and those surrounding lower mass stars and brown dwarfs. That study showed aprominent difference in the relative detection rates of HCN and C H between the two samples,with HCN detected more commonly in TTS than in the lower mass objects. The median spectrathey created of samples of T Tauri stars and the lower mass objects showed that the flux ratioof HCN to C H is ∼ ∼ β emission,which is moderately correlated with the accretion diagnostic H α . Accretion-related processes couldstrengthen the HCN emission by enhancing the temperature, and/or the HCN abundance, in thedisk atmosphere.The effect of accretion-related heating on disk molecular emission has been studied by Glassgoldet al. (2004, 2009). They proposed two sources of mechanical heating in the disk atmosphere:viscous accretion, possibly generated by the magnetorotational instability (MRI; Stone et al. 2000),and stellar wind interaction with the disk surface (Glassgold et al. 2004). Glassgold et al. (2009)invoked mechanical heating, due to one or both of these sources, in addition to the formationof H on warm grains, to explain the large column densities of warm H O that are observed inemission in disk atmospheres. Glassgold et al. (2009) determined that these processes can increasethe thickness of the warm water column to the extent reported by Carr & Najita (2008) and Salyket al. (2008). If mechanical heating does affect the thermal-chemical structure of disk atmospheresin this way, and if higher accretion rates and higher rates of mechanical heating derive from thesame physical mechanism, we would expect to see a correlation between accretion rate and H Ofeature strength. Accretion rate may play a similar role in enhancing HCN emission strength, i.e.by increasing the column density of warm HCN in the disk atmosphere.There may be an additional chemical connection between H O and HCN emission, with efficientwater formation possibly leading to an enhanced HCN abundance. As described by Lahuis & vanDishoeck (2000), efficient H O formation will drive most of the available oxygen into H O, resulting 18 –in a lower abundance of gaseous O . Since O would otherwise react with atomic carbon, the lackof O could lead to an enhanced atomic C abundance and in turn a larger HCN abundance (e.g.,via the reaction scheme described by Agundez et al. 2008). Perhaps for this reason, hot coresthat are found to have the highest gas phase H O abundances are also those with the highestHCN abundances (e.g., van Dishoeck 1998; Lahuis & van Dishoeck 2000). Thus accretion-relatedmechanical heating in disks may enhance disk HCN emission both thermally, by producing a deepertemperature inversion at the disk surface, and chemically, by enhancing the HCN abundance as aconsequence of efficient water formation. Detailed modeling is needed to explore these possibilities.Increased UV irradiation produced by higher stellar accretion may also enhance the HCNabundance. Using Ag´undez et al. (2008) as a guide, Pascucci et al. (2009) argued that the HCNabundance in disk atmospheres may be limited by the availability of atomic nitrogen and that theatomic nitrogen abundance depends primarily on the dissociation of N via UV-dissociation. Thus,HCN would be brighter for sources with more energetic UV flux (i.e., higher accretion rate), whileC H (not requiring nitrogen to form) would not vary. This may explain their finding that T Tauristars have stronger HCN emission relative to C H than lower mass stars and brown dwarfs, asthese lower mass objects would have lower photospheric UV emission and lower accretion ratesthan TTS. The range in stellar accretion rate among T Tauri stars may induce a range in theirHCN abundances for similar reasons.Another factor that may play a role in setting the HCN flux from the disk is X-ray irradiation,based on Fig. 4b. The effect of X-ray irradiation on the thermal-chemical structure of disks hasbeen investigated previously by Glassgold et al. (2004, 2009), although they did not specificallystudy HCN. X-ray irradiation may enhance the abundance of molecular ions and radicals that leadto enhanced HCN emission. Further modeling is needed to investigate the relative roles of X-rayand UV irradiation in this context.We find a possible trend of HCN flux decreasing with stellar spectral type (Fig. 4c). While thisis in the spirit of the trend found by Pascucci et al. (2009), it is unlikely that stellar spectral typeitself (i.e., stellar temperature) is affecting the HCN flux for this small sample of TTS. The othertwo processes we examined, stellar accretion rate and stellar X-ray flux (and/or other processesnot yet identified) are likely to have a more direct influence on the HCN flux. Stellar accretionrate is not well correlated with spectral type (see Fig. 4e) and the TTS in our sample span a smallmass range, so the resulting accretion luminosity seems unlikely to be correlated over the range ofspectral types that we studied. In comparison, L X shows a possible correlation with spectral type(Fig. 4d), so it may be responsible for the moderate correlation of HCN flux with spectral type.Several of the objects in our sample (plotted as open triangles in Fig. 4a, b, and c) appearto deviate from the possible trends we identify here. The dispersion we observe could arise fromdifferences in disk structure (e.g., flaring) and composition that may originate from the natalenvironment as well as the dynamic processing that occurs within the disk lifetime. This makesthe objects that deviate from our observed trends not only expected, but of particular interest. For 19 –example, while variations in stellar accretion rate are typically factors of ∼ ∼ ) on grains influences disk chemical synthesis, variations in grain properties may lead tovariations in observable molecular features (Glassgold et al. 2009). In the panels of Figure 4, thereare several outlying points whose HCN flux index is enhanced or depleted relative to the rest ofthe points. These might be ideal systems in which to look for additional chemical peculiarity orheating mechanisms that could be affecting the molecular emission strength.The trends described here require a larger sample to confirm. In tandem, it may be possibleto expand the wavelength range we analyze by considering observations from Spitzer
IRS modulesthat cover a wider wavelength range (i.e., Long-High, 20 µ m − µ m) and more molecular species.Additional high resolution data would also help verify the technique of using SL spectra to recoverreal trends.
5. Summary & Conclusions
Our goal was to investigate the extent to which lower resolution
Spitzer
IRS data can be used torecover quantitative molecular emission trends seen in higher resolution
Spitzer
IRS data. We haveshown that a simple prescription for measuring the strength of the 14 µ m HCN emission feature,when applied to low resolution Spitzer data, can recover trends in HCN emission strength that areseen in high resolution
Spitzer data. Additionally, we report possible correlations between HCNflux and stellar accretion rate, and HCN flux and stellar X-ray luminosity, that may originate fromaccretion-driven mechanical heating and/or photochemistry at work in the inner disk atmosphere.While qualitative comparisons of the presence of line emission were possible and successful earlier(e.g., Pascucci et al. 2009), our results demonstrate that quantitative comparisons of the line 20 –intensities can also be carried out.What controls the presence and strength of organic molecular features such as HCN in theplanet-forming regions around young stars? One challenge in addressing this question is the largenumber of physical and chemical processes that can potentially affect the molecular emissionstrength, as discussed in §
4. Our methods and results show that the large number of low reso-lution disk spectra that reside in the
Spitzer archive could be used in future demographic studiesto attempt to identify the relevant processes.
Facilities:
Spitzer () 21 –
REFERENCES
Ag´undez, M., Cernicharo, J., & Goicoechea, J. R. 2008, A&A, 483, 831Alencar, S. H. P., & Batalha, C. 2002, ApJ, 571, 378Alexander, R. D., Clarke, C. J., & Pringle, J. E. 2004, MNRAS, 354, 71Apai, D., & Lauretta, D. S. 2010, Protoplanetary Dust: Astrophysical and Cosmochemical Per-spectives, 128Ardila, D. R., & Basri, G. 2000, ApJ, 539, 834Bergin, E. A., Aikawa, Y., Blake, G. A., & van Dishoeck, E. F. 2007, Protostars and Planets V,751Bergin, E., et al. 2004, ApJ, 614, L133Berthoud, M. G., Keller, L. D., Herter, T. L., Richter, M. J., & Whelan, D. G. 2007, ApJ, 660, 461Blake, G. A., & Boogert, A. C. A. 2004, ApJ, 606, L73Blum, R. D., Barbosa, C. L., Damineli, A., Conti, P. S., & Ridgway, S. 2004, ApJ, 617, 1167Brittain, S. D., Simon, T., Najita, J. R., & Rettig, T. W. 2007, ApJ, 659, 685Carmona, A. 2010, Earth Moon and Planets, 106, 71Carr, J.S., & Najita, J.R. 2011, ApJ, submittedCarr, J.S., & Najita, J.R. 2008, Science, 319, 1504Carr, J. 2005, High Resolution Infrared Spectroscopy in Astronomy, 203Carr, J. S., Tokunaga, A. T., & Najita, J. 2004, ApJ, 603, 213Carr, J. S., Mathieu, R. D., & Najita, J. R. 2001, ApJ, 551, 454Carr, J. S., Tokunaga, A. T., Najita, J., Shu, F. H., & Glassgold, A. E. 1993, ApJ, 411, L37Chandler, C. J., Carlstrom, J. E., Scoville, N. Z., Dent, W. R. F., & Geballe, T. R. 1993, ApJ, 412,L71Dullemond, C. P., Hollenbach, D., Kamp, I., & D’Alessio, P. 2007, Protostars and Planets V, 555Dutrey, A., Guilloteau, S., & Ho, P. 2007, Protostars and Planets V, 495Dutrey, A., Guilloteau, S., Prato, L., Simon, M., Duvert, G., Schuster, K., & Menard, F. 1998,A&A, 338, L63 22 –Dutrey, A., Guilloteau, S., Duvert, G., Prato, L., Simon, M., Schuster, K., & Menard, F. 1996,A&A, 309, 493Ercolano, B., Clarke, C. J., & Drake, J. J. 2009, ApJ, 699, 1639Ercolano, B., Drake, J. J., Raymond, J. C., & Clarke, C. C. 2008, ApJ, 688, 398Furlan, E., et al. 2006, ApJS, 165, 568Glassgold, A. E., Meijerink, R., & Najita, J. R. 2009, ApJ, 701, 142Glassgold, A. E., Najita, J., & Igea, J. 2004, ApJ, 615, 972Gorti, U., & Hollenbach, D. 2009, ApJ, 690, 1539Gorti, U., & Hollenbach, D. 2008, ApJ, 683, 287G¨udel, M., et al. 2010, A&A, 519, A113Guilloteau, S., & Dutrey, A. 1998, A&A, 339, 467Gullbring, E., Hartmann, L., Briceno, C., & Calvet, N. 1998, ApJ, 492, 323Hartigan, P., Strom, K. M., & Strom, S. E. 1994, ApJ, 427, 961Hartigan, P., Kenyon, S. J., Hartmann, L., Strom, S. E., Edwards, S., Welty, A. D., & Stauffer, J.1991, ApJ, 382, 617Hartmann, L., Calvet, N., Gullbring, E., & D’Alessio, P. 1998, ApJ, 495, 385Henning, T., & Meeus, G. 2009, arXiv:0911.1010Herczeg, G. J., Linsky, J. L., Walter, F. M., Gahm, G. F., & Johns-Krull, C. M. 2006, ApJS, 165,256Herczeg, G. J., Linsky, J. L., Valenti, J. A., Johns-Krull, C. M., & Wood, B. E. 2002, ApJ, 572,310Houck, J. R., et al. 2004, ApJS, 154, 18Kastner, J. H., Zuckerman, B., Weintraub, D. A., & Forveille, T. 1997, Science, 277, 67Kenyon, S.J., & Hartmann, L. 1995, ApJS, 101, 117Lahuis, F., et al. 2006, ApJ, 636, L145Lahuis, F., & van Dishoeck, E. F. 2000, A&A, 355, 699Millan-Gabet, R., Malbet, F., Akeson, R., Leinert, C., Monnier, J., & Waters, R. 2007, Protostarsand Planets V, 539 23 –Muzerolle, J., Luhman, K. L., Brice˜no, C., Hartmann, L., & Calvet, N. 2005, ApJ, 625, 906Najita, J. R., Doppmann, G. W., Carr, J. S., Graham, J. R., & Eisner, J. A. 2009, ApJ, 691, 738Najita, J. R., Crockett, N., & Carr, J. S. 2008, ApJ, 687, 1168Najita, J. R., Carr, J. S., Glassgold, A. E., & Valenti, J. A. 2007a, Protostars and Planets V, 507Najita, J. R., Strom, S. E., & Muzerolle, J. 2007b, MNRAS, 378, 369Najita, J., Carr, J. S., & Mathieu, R. D. 2003, ApJ, 589, 931Najita, J. R., Edwards, S., Basri, G., & Carr, J. 2000, Protostars and Planets IV, 457Najita, J., Carr, J. S., Glassgold, A. E., Shu, F. H., & Tokunaga, A. T. 1996, ApJ, 462, 919Natta, A., Testi, L., Calvet, N., Henning, T., Waters, R., & Wilner, D. 2007, Protostars and PlanetsV, 767Pascucci, I., Apai, D., Luhman, K., Henning, T., Bouwman, J., Meyer, M. R., Lahuis, F., & Natta,A. 2009, ApJ, 696, 143Preibisch, T., et al. 2005, ApJS, 160, 401Pontoppidan, K. M., Salyk, C., Blake, G. A., Meijerink, R., Carr, J. S., & Najita, J. 2010,arXiv:1006.4189Pontoppidan, K. M., Blake, G. A., van Dishoeck, E. F., Smette, A., Ireland, M. J., & Brown, J.2008, ApJ, 684, 1323Qi, C., Wilner, D. J., Aikawa, Y., Blake, G. A., & Hogerheijde, M. R. 2008, ApJ, 681, 1396Salyk, C., Pontoppidan, K. M., Blake, G. A., Najita, J., & Carr, J. 2011, ApJ, acceptedSalyk, C., Blake, G. A., Boogert, A. C. A., & Brown, J. M. 2009, ApJ, 699, 330Salyk, C., Pontoppidan, K. M., Blake, G. A., Lahuis, F., van Dishoeck, E. F., & Evans, N. J., II2008, ApJ, 676, L49Salyk, C., Blake, G. A., Boogert, A. C. A., & Brown, J. M. 2007, ApJ, 655, L105Semenov, D., Wiebe, D., & Henning, T. 2006, ApJ, 647, L57Semenov, D., Pavlyuchenkov, Y., Schreyer, K., Henning, T., Dullemond, C., & Bacmann, A. 2005,ApJ, 621, 853Stelzer, B., & Neuh¨auser, R. 2001, A&A, 377, 538Stone, J. M., Gammie, C. F., Balbus, S. A., & Hawley, J. F. 2000, Protostars and Planets IV, 589 24 –Telleschi, A., G¨udel, M., Briggs, K. R., Audard, M., & Palla, F. 2007, A&A, 468, 425Thi, W.-F., & Bik, A. 2005, A&A, 438, 557Thi, W.-F., van Dalen, B., Bik, A., & Waters, L. B. F. M. 2005, A&A, 430, L61Thi, W.-F., van Zadelhoff, G.-J., & van Dishoeck, E. F. 2004, A&A, 425, 955Turner, N. J., Willacy, K., Bryden, G., & Yorke, H. W. 2006, ApJ, 639, 1218Valenti, J. A., Johns-Krull, C. M., & Linsky, J. L. 2000, ApJS, 129, 399van Dishoeck, E. F. 1998, The Molecular Astrophysics of Stars and Galaxies, edited by ThomasW. Hartquist and David A. Williams. Clarendon Press, Oxford, 1998., p.53, 4, 53Watson, D. M., et al. 2009, ApJS, 180, 84Willacy, K., Langer, W., Allen, M., & Bryden, G. 2006, ApJ, 644, 1202Winston, E., et al. 2010, AJ, 140, 266