Detections of Ro-Vibrational H Emission from the Disks of 2 T Tauri Stars Jeffrey S. Bary1, David A. Weintraub1 and Joel H. Kastner2 Received ; accepted 3 0 0 2 n Version date: November 12, 2002 a J 7 1 v 4 2 1 1 0 3 0 / h p - o r t s a : v i X r a 1Department of Physics & Astronomy, Vanderbilt University, P.O. Box 1807 Station B, Nashville, TN 37235; jeff[email protected], [email protected] 2Carlson Center for Imaging Science, RIT, 54 Lomb Memorial Drive, Rochester, NY 14623; [email protected] – 2 – ABSTRACT We report the detection of quiescent H emission in the v=1→0 S(1) line 2 at 2.12183 µm in the circumstellar environment of two classical T Tauri stars, GG Tau A and LkCa 15, in high-resolution (R ≃ 60,000) spectra, bringing to four, including TW Hya and the weak-lined T Tauri star DoAr 21, the number of T Tauri stars showing such emission. The equivalent widths of the H emission 2 line lie in the range 0.02-0.10 ˚A and, in each case, the central velocity of the emission line is centered at the star’s systemic velocity. The line widths range from 9 to 14 km s−1, in agreement with those expected from gas in Keplerian orbits in circumstellar disks surrounding K-type stars at distances ≥ 10 AU from the sources. UV fluorescence and X-ray heating are likely candidate mechanisms responsible forproducing theobserved emission. We present mass estimates from the measured line fluxes and show that the estimated masses are consistent with those expected from the possible mechanisms responsible for stimulating the ob- served emission. The high temperatures and low densities required for significant emission in the v=1→0 S(1) line suggests that we have detected reservoirs of hot H gas located in the low density, upper atmospheres of circumstellar disks of 2 these stars. Subject headings: circumstellar matter – infrared: stars – solar system: formation – stars: open clusters and associations – stars: individual (GG Tau, LkCa 15, DoAr 21, TW Hya) — stars: pre-main-sequence – 3 – 1. Introduction The study of circumstellar disks around young stars may provide insight into how and when planets form. Theoretical models suggest different mechanisms and timescales for planet formation. The core accretion model requires that a rocky core of ∼ 10M aggregate ⊕ before substantial gaseous accretion can form a Jupiter-like gas giant. This theory predicts that gas giants will form a few AU or more from the star due to the enormously greater mass of solids available beyond the snow line. A time period of roughly 107 yr is predicted for giant planet formation to occur through runaway accretion in a greater than minimum mass solar nebula (Lissauer 1993; Pollack et al. 1996). A second though less thoroughly investigated model, which may require a time period as brief as 103 yr for the formation of gas giant planets, proposes that such planets may form as a consequence of gravitational instabilities in disks (Boss 2000). Typically, astronomers have resorted to emission from trace molecules and dust for insight into the physical conditions and masses of gas in circumstellar disks. Once these tracers disappear, the disks appear to have been removed from the system. Two decades of such observations have shown that young stellar objects (YSOs) appear to be surrounded by thick envelopes composed of dust and gas that evolve into circumstellar disks on the order of 106 yr (Strom et al. 1995). The absence of thermal dust emission, as measured through millimeter and infrared continuum excesses, as well as the non-detection of CO line emission towards most T Tauri stars (TTS) over the ages of 3-5 Myr old, suggests that the disks become depleted in gas and dust grains rather quickly. Two viable explanations may account for these observations: in one model, when the disk becomes undetectable it no longer exists, while in the second the disk is present but difficult to detect. According to the first of the these scenarios, circumstellar material has been dissipated through one or more of several processes: photo-evaporation, accretion onto the star, tidal stripping – 4 – by nearby stars, outflows, or strong stellar winds (Hollenbach et al. 2000). The second scenario, which describes the evolution of a circumstellar disk into a planetary system according to the core accretion theory of planet formation, suggests that larger structures (i.e., planetesimals) could have formed through accretion, collecting both solids and some gases like CO, reducing the emitting surface area of the dust and pushing the emission from both dust grains and gaseous CO below detectable levels. In order to know whether the disks are dispersed or accretionally evolved, we must find a method to observe the part of the disk that does not disappear quickly during accretion. Since the bulk of the gaseous component of the disk, hydrogen and helium gas, will be the last to be collected into protoplanets, assuming that it avoids being removed through the various processes listed above, the H and He should remain in the disk and in the gas phase even after the dust and gaseous CO has become undetectable. Therefore, we need to study the H gas directly. 2 H , at the low temperatures typically found in circumstellar environments, 2 predominantly populates the ground vibrational level and only low rotational excitation levels and remains difficult to stimulate due to its lack of a dipole moment. In the small, inner regions of disks where temperatures may be high enough to excite H gas into 2 the first vibrational state such that 2.12183 µm emission might be possible, densities often are too high to permit detectable levels of H line emission. Therefore, the bulk 2 of the disk material has remained nearly impossible to detect directly. Assuming that most of the gas in these environments maintains a temperature much less than 1000 K, few molecules will be excited into the first vibrational state; thus, the flux of thermally excited line emission from H at 2.12183 µm will be extremely low and consequently 2 impossible to detect. In the case of T Tau, the detected emission from shocked H traces 2 approximately 10−7 M (Herbst et al. 1996), a mere fraction of the total mass in the system ⊙ – 5 – (Weintraub et al. 1989a; Weintraub et al. 1989b; Weintraub et al. 1992). Pure rotational H line emission found in the mid-infrared at 17 and 28 µm corresponds 2 to excitation temperatures on the order of ∼100 K. These lines therefore should act as tracers of the bulk of the unshocked gas which resides in the circumstellar disks. Of these two lines, only the 17 µm line can be successfully observed from ground based telescopes; however, Thi et al. (1999,2001a,b) reported the detection of emission from H at 17 and 28 2 µm using a low spatial and spectral resolution mid-infrared spectrometer on the Infrared Space Observatory (ISO) towards GG Tau, more evolved sources with Vega-type debris disks, and Herbig Ae/Be stars. However, in much more sensitive, higher spatial and spectral resolution mid-infrared spectra obtained by Richter et al. (2002), centered on the 17 µm H 2 emission line, no emission was observed towards GG Tau, HD 163296, and AB Aur. The Richter et al. results not only did not confirm the detections made toward these stars, but also call into question other ISO detections. We have taken a different approach. Motivated by models predicting that X-rays produced by TTS may be sufficient to ionize a small fraction of the gas and indirectly stimulate observable near-infrared H emission in circumstellar environments through 2 collisions between H molecules and nonthermal electrons (Gredel & Dalgarno 1995; 2 Maloney et al. 1996; Tine et al. 1997), we have begun a study of TTS searching for near-infrared H emission. We reported our initial detections of quiescent H emission in 2 2 the near-infrared towards TW Hya (Weintraub et al. 2000), an X-ray bright classical TTS (cTTS), and DoAr 21 (Bary et al. 2002), an X-ray bright weak-lined TTS (wTTS). We previously have suggested that the most plausible explanation for the stimulation of the H 2 emission from these sources is the X-ray ionization process. However, preliminary results from our modeling of UV and X-ray photons originating from the source and incident on the disk at varying angles (Bary, in prep) suggest that UV radiation from the source may – 6 – be an equally likely stimulus for the observed emission. Therefore, these results indicate that gaseous disks of H surrounding other TTS possessing substantial X-ray and UV fluxes 2 could be detected at 2.12183 µm using high-resolution, near-infrared spectroscopy. Inthispaper, wepresent thefirst resultsofahigh-resolution, near-infraredspectroscopic survey of cTTS and wTTS in the Taurus-Auriga and ρ Ophiuchus star forming regions. We report new detections of H towards two sources: GG Tau A and LkCa 15. We use the new 2 data and our previously reported detections of H emission toward TW Hya and DoAr 21 2 to estimate disk masses for these four stars. 2. Observations We obtained high-resolution (R ≃ 60,000), near-infrared spectra of selected TTS (Table 1) in Taurus-Auriga and ρ Ophiuchus on 1999 December 26-29, and 2000 June 20-23, UT, using the Phoenix spectrometer (Hinkle et al. 1998) on NOAO’s 4-m telescope atop Kitt Peak. In spectroscopic mode, Phoenix used a 256 × 1024 section of a 512 × 1024 Aladdin InSb detector array. Our observations were made using a 30′′ long, 0′.′8 (4 pixel) wide, north-south oriented slit, resulting in instrumental spatial and velocity resolutions of 0′.′11 and 5 km s−1, respectively. Our actual seeing limited spatial resolution was ∼ 1′.′4. The spectra were centered at 2.1218 µm, providing spectral coverage from 2.1167 to 2.1257 µm. This spectral region includes three telluric OH lines, at 2.11766, 2.12325, and 2.12497 µm, that provide an absolute wavelength calibration. Integration times for our program stars, with K magnitudes 5.9 ≤ K ≤ 9.6 (Table 1), ranged from 3600 s to 7200 s. Nightly observations were obtained of an A0V star, either Vega or HD 2315, for telluric calibration, with on-source integration times of 600 sec. Flat-field images were made using a tungsten filament lamp internal to Phoenix. – 7 – Observations were made by nodding the telescope 14′′ along the slit, producing image pairs that were then subtracted to remove the sky background and dark current. The spectra were extracted from columns of the array covering the region from 0′.′7 east to 0′.′7 west of each source for total beam widths of 1′.′4, beyond which no emission was detected. Spectra were then divided by the continuum to produce normalized spectra and ratioed with that of a star with a featureless spectrum in this spectral region in order to remove telluric absorption features from the spectra of the program stars. SU Aur, a program star classified as an X-ray bright G2III wTTS produced a high signal to noise ratio, featureless spectrum and therefore, also was utilized as a telluric calibrator. 3. Results We detected line emission very nearly at 2.12183 µm from two cTTS, GG Tau A and LkCa 15 and, as reported previously (Bary et al. 2002), the wTTS DoAr 21 and the cTTS TW Hya (Weintraub et al. 2000)3. Full spectra for all our target stars obtained with Phoenix are presented in Figure 1. Most of the absorption features in these spectra are telluric. Asbest ascouldbedonegiven tens-of-minutes timescale fluctuations inatmospheric conditions, the spectra presented in Figure 2 have been corrected for telluric absorption features, extracted from the total spectra presented in Fig. 1 and shifted in velocity using previously measured values of V (Skrutskie et al. 1993; Kamazaki et al. 2001), placing lsr the spectra in the rest frames of the stars. In no cases in which H emission was observed 2 was any emission detected in the spectra beyond 0′.′7 from the central positions of the star. 3Note that LkCa 15 had been classified as a wTTS (with an Hα EW of 13 ˚A) in the Herbig Bell Catalog; however, Wolk & Walter (1996) measured W (Hα) = 21.9 ˚A, leading λ some authors to reclassify this star as a cTTS. – 8 – The central wavelengths of the emission lines were determined by fitting a Gaussian to the emission features. In each case, the central wavelength of the observed emission line matches the rest wavelength of the 2.12183 µm v=1→0 S(1) transition of H to within 2 errors and the line is narrow (< 14 km s−1) yet spectrally resolved (> 9 km s−1). Therefore, we conclude that the observed emission is from gaseous H molecules within ∼ 100 AU of 2 the stars. The central velocities of the observed H reservoirs are neither red-shifted nor 2 blue-shifted with respect to the stars and therefore experience no net line-of-sight motion. H equivalent widths and line strengths for all four stars now detected in the H line, 2 2 corrected for extinction, are presented in Table 1, along with 3-σ upper limits for the line strengths of the non-detections. A was determined using previously reported values for the k visual extinction (Table 5) and taking A /A = 0.1 (Becklin et al. 1978). k v Several of our sources, notably V836 Tau, V819 Tau, and IP Tau show possible emission peaks at wavelengths just shortward of 2.122 µm. The low signal to noise levels of these stars’ spectra, however, preclude drawing any positive conclusions about these features in these data. Two absorption features centered at 2.11699 µm and 2.12137 µm remain after the telluric corrections were made. The line at 2.12137 µm labeled ‘Al’ is indicated in Figure 2. The line at 2.11699 can be seen in many of the Fig. 1 spectra, in most cases slightly red-shifted to & 2.117 µm. We have identified these lines as photospheric absorption features produced by Al (Wallace & Hinkle 1996). Each of the stars with observed H 2 emission have both of these photospheric features, as do many of the sources not detected in H emission; however, four of our sources do not. The strength of these absorption features 2 decreases toward earlier spectral types and disappears entirely for the F and G type stars in our sample (Table 1) in agreement with the Wallace & Hinkle (1996) data. – 9 – 3.1. The Location of the Emitting H Gas 2 The spectrum of LkCa 15 is presented in Figure 2. The double-peaked H emission line 2 profile for LkCa 15 is consistent with the double-peaked CO(J=2→1) and HCO+ emission profiles reported by Duvert et al. (2000). The presence of blue-shifted and red-shifted emission peaks for each of these emission features, including the H , is indicative of gas 2 revolving in a circumstellar disk (Beckwith & Sargent 1993; Mannings & Sargent 1997). The peak-to-peak velocity separation for the H is resolved at ∆v ≈ 10 ± 1.5 km s−1, 2 which is considerably larger than the peak-to-peak separation of ∆v ≈ 2 km s−1 found from both the CO and HCO+ spectra. For a K5V star with a disk inclination angle of 34◦± 10◦ as determined by Duvert et al., the H emission would be produced at radial distances 2 of between 10 and 30 AU while the CO emission should arise from molecular gas located ∼ 600 AU from the source. The combination of the CO and H data indicate that the 2 near-infrared H emission is sampling a different reservoir of gas than is the CO(J=2→1) 2 emission and that the H is from regions of the circumstellar disk in which gas and ice giant 2 planets might form. In addition, the mechanism responsible for exciting the H apparently 2 is most efficient in doing so for gas at these intermediate distances from LkCa 15. As all four stars sharing H emission have similar H line widths, fluxes and relative velocities, it 2 2 is plausible that the excitation mechanism is similar in all four stellar environments. We suggest, therefore, that the H emission emerges from the 10-30 AU region around each 2 of these four stars. Only LkCa 15, however, has a disk inclined to our line of sight such that we can distinguish the double-peaked profile (TW Hya, in fact, is known to be viewed nearly pole-on). The spatial resolution for the spectra centered on each source probes emission from the disks out to radii of ∼ 30 AU for TW Hya and ∼ 110 AU for LkCa 15, GG Tau, and DoAr 21. Thus, the spatial resolution in our data is consistent with our spectroscopically derived conclusion about the location of the emitting H . 2 – 10 – In concluding that the gas is located in the regions 10-30 AU from the sources, we are ruling out the possibility that the emission arises in extended halos surrounding the sources. The fact that in all four cases the H line emission is spatially unresolved is inconsistent 2 with emission from a halo that should be many hundreds or thousands of AU in extent. In addition, since the H emission lines appear to be spectrally resolved, the velocity line 2 width of the gas also is inconsistent with a halo interpretation for the observed emission, for which we would expect line profiles with FWHM of only ∼ 1-2 km s−1. Line widths of this size would be unresolved at R ≃ 60,000. 4. H Mass Calculation 2 While detection of H is important in establishing the continued presence of gas in the 2 evolving circumstellar environment of TTS, a determination of the exact amount of H gas 2 still present has further reaching implications concerning the timescale for planet formation and/or the dispersal of the disk. Making such a calculation requires understanding in what environments and under what conditions the H molecule can be excited into the v = 1, J = 2 3 state and what excitation mechanism(s) is active in this environment. The v=1→0 S(1) transition energy corresponds to a temperature of ∼ 5000 K above the ground state. Therefore, excitation temperatures between 1000 K and 2000 K typically are required in all types of astrophysical environments in order for H gas to populate the first vibrationally 2 excited level and produce detectable levels of emission at 2.12183 µm (Tanaka et al. 1989). According to the disk model discussed in Glassgold et al. (2000), which includes X-ray heating, the outer surface layers of the disk may be heated to temperatures of 1000 K out to several AU, while the midplane remains cool at ∼ 100 K. Hollenbach et al. (2000) show that temperatures of 1000 K produced by incident UV radiation are high enough to create photoevaporative flows that deplete disks on million year timescales. Increasing the