ANOMALOUS ACCRETION ACTIVITY AND THE SPOTTED NATURE OF THE DQ TAU BINARY SYSTEM

The DQ Tau system, the subject of the study presented here, is a tight binary (P ≈ 15.8 days; Huerta et al. 2005) located in the nearby Taurus-Auriga star forming region (D ∼ 140 pc; age ∼ 1–3 Myr). DQ Tau possesses two nearly equal mass stars (M_a + M_b ∼ 1.3 M_☉) with a high orbital eccentricity (e = 0.556; Mathieu et al. 1997). For this binary system, possessing a semi-major axis of ∼0.13 AU, the inner truncation radius of the circumbinary disk should reside at ∼0.4 AU. The spectral energy distribution (SED) shows both near- and mid-infrared excesses suggesting the presence of both a circumbinary disk and warm dust in the environment of the circusmtellar disks (Strom et al. 1989; Skrutskie et al. 1990). Recent Keck Interferometer observations and modeling of the K-band excess emission confirm that the warm dust must exist in the region where circumstellar disk material should reside with an extent of 0.1–0.2 AU (Boden et al. 2009). Combined with previous detections of warm CO line emission (T ∼ 1200 K; R ∼ 0.1 AU; Carr et al. 2001), these observations suggest that the dynamically cleared region located within the circumbinary disk contains dust and gas, leading to the conclusion that material is continually drawn inward from the outer disk.

For high-eccentricity binary systems such as DQ Tau, hydrodynamic models predict that orbitally modulated accretion activity should result in accretion flares occurring at periastron passes of the stellar companions (Günther & Kley 2002; de Val-Borro et al. 2011). Quasi-periodic flares phased with periastron in the DQ Tau system were first observed, using broadband photometry, by Mathieu et al. (1997). Subsequent multi-epoch optical and near-infrared spectroscopic observations detected increases in the fluxes of several standard accretion signatures similarly phased with periastron passages in DQ Tau and two other tight binaries, UZ Tau E and AK Sco (Basri et al. 1997; Huerta et al. 2005; Jensen et al. 2007; Bary et al. 2008). For these systems, in which the stellar companions come within a few stellar radii of one another, the flaring could be explained by interacting magnetospheres and are likely the cause of millimeter and X-ray flares observed in the DQ Tau and UZ Tau E systems (Salter et al. 2008, 2010; Getman et al. 2011; Kóspál et al. 2011). As the stars recede, the strength of the interaction between the magnetospheres diminishes rapidly with increasing distance. The length of the millimeter and X-ray flares is on the order of hours, while the optical photometric and spectral feature flares last for several days or as much as a third of the orbit. Noting that the longer flares occur relatively far in orbital phase from periastron passage, Jensen & Mathieu (1997) argue that accretion activity is most likely responsible for this flaring. Therefore, these observations likely confirm the hydrodynamical model predictions that material from the circumbinary disk can flow into the unstable region, replenishing the inner circumstellar disks.

For widely separated young binary systems, accretion streams provide a mechanism for increasing the lifetimes of the individual circumstellar disks, potentially impacting the formation of planets in such systems. While it may be unreasonable to expect planets to form on stable orbits in the tidally truncated circumstellar disks associated with the stellar companions in the DQ Tau system, the recent Kepler detections of circumbinary planets clearly indicate that the circumbinary disk is a viable site for planet formation. While star–disk interactions play an important role in the evolution of planet-forming disks in single star systems, star–star–circumbinary disk interactions are presumed to be equally important in the evolution of circumbinary planetary systems. In addition, as described by Artymowicz & Lubow (1996) and Rozyczka & Laughlin (1997), the structure of a tight binary that is in the process of clearing a gap in the inner region of its circumbinary disk is analogous to the structure of a transitional disk system in which a gas giant planet is suspected of clearing a gap in the inner region of a circumstellar disk. The potential for analogous orbitally modulated accretion activity in transitional disk systems underscores the importance of studying young binary systems and their star–star–disk interactions.

In this paper, we present a multi-epoch spectroscopic study of pulsed-accretion activity in the DQ Tau system. We use near-infrared (IR) spectral accretion signatures (i.e., H i, Ca ii, and He i) to establish an average mass accretion rate for the system in its quiescent accretion phase. We report for the first time the detection of an anomalous accretion flare well outside of the usual periastron related activity, timed more closely in orbital phase with an apastron passage of the system. The detection is discussed in the context of current hydrodynamic models of star–star–disk–disk interactions. The moderate-resolution spectra allow for the determination of the infrared spectral type of the DQ Tau system based on the spectral shape, molecular absorption features, and metallic absorption lines. Similar to previous studies of near-IR spectra of T Tauri stars (TTSs), we find a discrepancy between the spectral types previously derived from optical spectroscopy and the near-IR spectral type. Other authors have also noted a color anomaly at IR wavelengths for TTSs and have suggested that both discrepancies may be related to the existence of large cool spots on the surfaces of these stars (Gullbring et al. 1998a, 1998b). We construct simple synthetic spectra to represent a spotted star from spectral standards and perform a fit to a representative DQ Tau spectra. We solve simultaneously for visual extinction, photospheric temperature, spot temperature, and spot filling factor by fitting the strong TiO and FeH bands and the shape of the near-IR spectrum. The best fits cover a small region of parameter space and demonstrate that the existence of large, cool spots are reasonable explanations for the observed optical/IR spectral mismatch and the color anomaly.

Optical and near-IR hydrogen emission features (i.e., Balmer, Paschen, and Brackett series) are defining characteristics of the classical T Tauri classification of pre-main-sequence stars. These features are accepted as signposts for the presence of circumstellar material and indicators of accretion activity. In the paradigm of magnetospherically guided accretion, matter free-falls from the circumstellar disk toward the stellar surface guided by the large-scale dipole-like stellar magnetic field. The material slams into the stellar photosphere, releasing energy as a shock front forms a hot spot on the surface of the star, heating both the shocked gas and the trailing column of infalling material. Ionization and recombination of hydrogen in the dense gas flows are thought to be responsible for the majority of flux produced in the H i emission lines present in the spectra of TTSs. For some T Tauri systems, recent observations have demonstrated that a small fraction of the hydrogen emission is spatially extended and associated with outflowing gas (Beck et al. 2010; Coffey et al. 2010; Davis et al. 2011). While the contribution of emission from the outflows to the integrated line flux is negligible for most T Tauri systems, it is potentially more important for less evolved, embedded Class I systems. Given the relationship between accretion and outflows and the fact that many observational studies have shown that emitting H i gas correlates well with other tracers of accreting gas, it is reasonable to conclude that H i line emission remains a trusted proxy for determining mass accretion rates in TTS systems. The wavelength coverage of CorMASS, TSpec, and SpeX and multiple epochs of observations provides the opportunity to study several hydrogen transitions simultaneously in the Paschen and Brackett series and to measure the variations in mass accretion rates as a function of orbital phase.

In addition to the H i features, the He i 1.083 μm feature is another well-studied spectral characteristic of many TTSs that falls within the wavelength coverage of CorMASS, TSpec, and SpeX (e.g., Beristain et al. 2001; Dupree et al. 2005; Edwards et al. 2006; Fischer et al. 2008). At moderate resolution, this feature may appear with a P Cygni and/or inverse P Cygni profile, possessing either a blueshifted or a redshifted absorption component, or both. Models of this feature describe the formation in either accreting and/or outflowing gas with moderate success (Kwan et al. 2007; Fischer et al. 2008). We monitor the strength of this feature and the appearance of absorption components as a potential indicator of differing modes of accretion activity in the DQ Tau system.

Lastly, the Ca ii infrared triplet near 0.85 μm has also been observed to correlate with accretion activity (Muzerolle et al. 1998; Azevedo et al. 2006) and falls within the wavelength coverage of CorMASS and SpeX. We will assume that the presence of the H i and Ca ii emission features directly indicate ongoing mass accretion in the DQ Tau system.

Figure 2 shows the variations in strength of both the He i and Paγ emission features. Notably, the He i feature remains strongly in emission throughout the entire orbit of the system with only moderate variations in strength. The moderate-resolution spectra available at ϕ = 0.402, 0.466, 0.528, 0.539, 0.546, and 0.666 show that the structure of the feature undergoes some interesting variations. For instance, the appearance of the blueshifted absorption feature at ϕ = 0.546 and 0.666 suggests the presence of a variable outflowing wind (v ∼ 200–300 km s⁻¹) in the system since the feature is much weaker or completely absent at the other orbital phases. The presence of the blueshifted absorption component in the He i feature observed in a few epochs of the moderate-resolution spectra suggest that some of the variation observed in the low-resolution spectra are correlated with the presence or absence of the blueshifted absorption component. The Paγ feature becomes undetectable in two epochs of the low-resolution data. As expected for orbitally modulated accretion, the Paγ feature appears strongest in epochs near periastron. The TSpec and SpeX spectra collected far enough from periastron allow us to confidently detect this feature during the quiescent phase of the orbit.

Zoom In Zoom Out Reset image size

Figures 3 and 4 similarly show Paβ and Brγ features simultaneously collected with the He i and Paγ features presented in Figure 2. Paβ has a larger flux than Paγ and, as a result, is more easily detected during the quiescent accretion phase of the orbit. The moderate-resolution spectra provide insight into the structure of the line and a higher signal-to-noise ratio with which to better estimate the mass accretion rate. In most epochs for which we have moderate-resolution spectra, the feature appears slightly asymmetric with a sharp edge on the red side of the line. Due to the presence of a notable photospheric absorption feature (unidentified by Rayner et al. 2009) in the spectra of late K and early M main-sequence stars, we conclude that the asymmetry of Paβ is most likely due to this photospheric feature. For all epochs, there is no clear evidence in any of the Paschen and Brackett series features for the redshifted absorption component frequently reported for hydrogen lines in the near-IR spectra of accreting TTSs.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Mass accretion rates in T Tauri systems are most reliably determined by measuring excess Balmer continuum emission at wavelengths shorter than the Balmer jump (Valenti et al. 1993; Gullbring et al. 1998a; Herczeg & Hillenbrand 2008). However, the observed correlations between mass accretion rates and infrared H i and the Ca ii infrared triplet emission line luminosities have led to empirical relationships relating infrared line strengths to mass accretion rates (Muzerolle et al. 1998). Using the following empirical relations,

$$\begin{equation} \log \left(\frac{L_{{\rm acc}}}{L_{\odot }} \right) = (1.14 \pm 0.16)\log \left(\frac{L_{{\rm Pa}\beta }}{L_\odot } \right) + (3.15 \pm 0.58) \end{equation} \tag{ 1 }$$

$$\begin{equation} \log \left(\frac{L_{{\rm acc}}}{L_{\odot }}\right) = (1.26 \pm 0.19)\log \left(\frac{L_{{\rm Br}\gamma }}{L_\odot } \right) + (4.43 \pm 0.79), \end{equation} \tag{ 2 }$$

we can calculate the accretion luminosities from the Paβ and Brγ line luminosities. We then use the formula

$$\begin{eqnarray} \log \left(\frac{\dot{M}}{M_\odot }\right) &=& \log \left(\frac{L_{{\rm acc}}}{L_\odot }\right) +\log \left(\frac{R_{\ast }}{R_\odot }\right) \nonumber\\ && -\log \left(\frac{M_{\ast }}{M_\odot }\right) - \log \left(-7.18\right) \end{eqnarray} \tag{ 3 }$$

to convert the accretion luminosity to a mass accretion rate. In Table 2, we present the equivalent widths and upper limits of the four strongest hydrogen emission lines (Paβ, Paγ, Paδ, and Brγ) and the 1.083 μm He i feature as well as the accretion luminosities and mass accretion rates. We note that the $\log ({\dot{M}}/{M_\odot })$ of the accretion luminosities presented were determined using the stronger, higher signal-to-noise ratio Paβ line luminosity. In Section 3.6, we discuss the determination of the spectral type of the DQ Tau system from which we base a mass and stellar radius (M_* = 1.3 M_☉ and R_* = 1.6 R_☉, respectively) for the determination of $\dot{M}$. In Figure 5, we plot the accretion luminosity versus orbital phase.

Zoom In Zoom Out Reset image size

Table 2. DQ Tau Line Measurements

Orbital	He i	Paδ	Paγ	Paβ	Brγ	$\log \left(\frac{L_{{\rm acc}}}{L_\odot }\right)$	$\log \left(\frac{\dot{M}}{M_\odot \, {\rm yr}^{-1}}\right)$
Phasea	Wλ(Å)	Wλ(Å)	Wλ(Å)	Wλ(Å)	Wλ(Å)
0.033±$^{0.031}_{0.093}$	−11.40 ± 0.52	−5.19 ± 0.52	−6.59 ± 0.52	−8.91 ± 0.62	−6.63 ± 1.03	−1.03 ± 0.05	−8.44 ± 0.05
0.037±$^{0.034}_{0.102}$	−28.67 ± 0.52	−6.19 ± 0.52	−11.05 ± 0.52	−16.53 ± 0.62	−9.6 ± 1.03	−0.72 ± 0.03	−8.13 ± 0.03
0.372±$^{0.035}_{0.107}$	−8.57 ± 0.52	−1.01 ± 0.52	−4.51 ± 0.52	−3.27 ± 0.62	⋅⋅⋅	−1.53 ± 0.16	−8.94 ± 0.16
0.402±$^{0.048}_{0.139}$	−11.40 ± 0.19	−0.70 ± 0.19	−1.77 ± 0.19	−2.30 ± 0.23	−1.462 ± 0.38	−1.70 ± 0.07	−9.11 ± 0.07
0.433±$^{0.034}_{0.102}$	−8.89 ± 0.52	−2.22 ± 0.52	−2.98 ± 0.52	−6.16 ± 0.62	−3.49 ± 1.03	−1.21 ± 0.08	−8.62 ± 0.08
0.437±$^{0.028}_{0.084}$	−8.84 ± 0.52	⋅⋅⋅	⋅⋅⋅	−1.78 ± 0.62	⋅⋅⋅	−1.83 ± 0.35	−9.24 ± 0.35
0.466±$^{0.048}_{0.139}$	−10.45 ± 0.19	−0.85 ± 0.19	−2.29 ± 0.19	−2.36 ± 0.23	−1.55 ± 0.38	−1.69 ± 0.07	−9.10 ± 0.07
0.479±$^{0.031}_{0.093}$	−6.77 ± 0.52	−2.70 ± 0.52	−3.26 ± 0.52	−2.40 ± 0.62	−2.43 ± 1.03	−1.68 ± 0.23	−9.09 ± 0.23
0.500±$^{0.028}_{0.084}$	−7.61 ± 0.52	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	−1.82 ± 0.34	−9.23 ± 0.34
0.528±$^{0.048}_{0.139}$	−10.99 ± 0.19	−0.79 ± 0.19	−1.89 ± 0.19	−2.95 ± 0.23	−1.74 ± 0.38	−1.58 ± 0.05	−8.99 ± 0.05
0.539±$^{0.045}_{0.135}$	0.94,-7.80 ± 0.09	−0.82 ± 0.09	−1.22 ± 0.09	−1.80 ± 0.11	−2.78 ± 0.18	−1.82 ± 0.04	−9.23 ± 0.04
0.546±$^{0.045}_{0.135}$	1.01,-7.00 ± 0.09	−0.97 ± 0.09	−1.16 ± 0.09	−1.41 ± 0.11	−1.41 ± 0.13	−1.94 ± 0.06	−9.35 ± 0.06
0.565±$^{0.028}_{0.084}$	−7.49 ± 0.52	⋅⋅⋅	−1.56 ± 0.52	−2.43 ± 0.62	⋅⋅⋅	−1.67 ± 0.23	−9.08 ± 0.23
0.627±$^{0.028}_{0.084}$	−7.66 ± 0.52	−2.06 ± 0.52	−3.04 ± 0.52	−2.07 ± 0.62	−3.01 ± 1.03	−1.80 ± 0.32	−9.21 ± 0.32
0.666±$^{0.045}_{0.135}$	1.02,-7.44 ± 0.09	⋅⋅⋅	−1.11 ± 0.09	−1.65 ± 0.11	−1.61 ± 0.18	−1.86 ± 0.05	−9.27 ± 0.05
0.713±$^{0.034}_{0.102}$	−10.90 ± 0.52	−1.90 ± 0.52	−5.97 ± 0.52	−5.59 ± 0.62	⋅⋅⋅	−1.26 ± 0.09	−8.67 ± 0.09
0.876±$^{0.027}_{0.082}$	−9.54 ± 0.52	−6.24 ± 0.52	−6.67 ± 0.52	−11.98 ± 0.62	−6.75 ± 1.03	−0.88 ± 0.04	−8.44 ± 0.04
0.897±$^{0.034}_{0.102}$	−22.85 ± 0.52	−3.86 ± 0.52	−4.92 ± 0.52	−9.18 ± 0.62	−5.71 ± 1.03	−1.02 ± 0.05	−8.43 ± 0.05
0.961±$^{0.051}_{0.148}$	1.56,-21.45 ± 0.19	−6.29 ± 0.19	−12.56 ± 0.19	−16.56 ± 0.23	−6.25 ± 0.38	−0.72 ± 0.01	−8.13 ± 0.01
0.962±$^{0.051}_{0.148}$	1.58,-21.62 ± 0.19	−7.90 ± 0.19	−12.65 ± 0.19	−16.05 ± 0.23	−6.17 ± 0.38	−0.74 ± 0.01	−8.15 ± 0.01
0.975±$^{0.051}_{0.148}$	1.21,-18.52 ± 0.19	−5.62 ± 0.19	−10.49 ± 0.19	−13.80 ± 0.23	−4.93 ± 0.38	−0.81 ± 0.01	−8.22 ± 0.01

Note. ^aOrbital phases were calculated beginning on Julian date 2449582.54 using the orbital period of 15.8016±$^{0.002}_{0.006}$ measured by Huerta et al. (2005).

Download table as:  ASCIITypeset image

The randomly collected data sample the strengths of the emission features throughout the entire orbital phase, though they do not provide continuous coverage of any one full orbit of the system. Using H i and the Ca ii lines as indicators of accretion activity, we detect evidence of accretion flares in eight epochs of observations located near four different periastron passages (ϕ = 0.033, 0.037, 0.713, 0.876, 0.897, 0.961, 0.962, and 0.975; see Table 2). The flares are denoted by a strengthening of several Paschen and Brackett series H i emission features, as is shown in Figures 2–4. Upper limits are placed on the strengths of the emission features during the quiescent portion of the orbit when the line strengths fall below our sensitivity thresholds (see Figure 5). To determine a quiescent (non-flaring) accretion rate, we select epochs which do not indicate flaring activity (weak Hydrogen features, low intra-night variability) to estimate the mean quiescent mass accretion rate as 6.72±$^{3.09}_{2.17}$ × 10⁻¹⁰ M_☉ yr⁻¹ (this corresponds, via linear mapping, to a quiescent accretion luminosity of −1.76 ± 0.1 log (L_acc/L_☉)). We note that our value for the quiescent mass accretion rate compares favorably with the estimate of 6 × 10⁻¹⁰ M_☉ yr⁻¹ (Gullbring et al. 1998a), which was derived from the excess Balmer continuum emission.

Selecting the previously defined quiescent observations, we find (as noted above) a mean accretion luminosity of log (L_acc/L_☉) = −1.76 with a dispersion of σ = 0.1. Interpreted in this context, the first night of the proposed anomalous flare at ϕ = 0.372 is a 2.3σ outlier, while the second night of the flare at ϕ = 0.433 is a 5.51σ outlier. In fact, the line strengths and accretion luminosities on these two nights compare more favorably to the periastron accretion flares, which have a mean accretion luminosity of log (L_acc/L_☉) = −0.845 and a dispersion of σ = 0.125. The second night of the flare is more consistent with originating in the periastron flaring distribution (3.65σ) than the quiescent distribution (5.51σ). Note also that an observation at ϕ = 0.71 exhibits a 5.01σ flare. Based on the photometric monitoring of Mathieu et al. (1997), such a flare constrained to within Δϕ = ± 0.3 of periastron, would have been considered one of the periastron flares. Therefore, we did not consider it part of the quiescent portion of the orbit nor anomalous in any way. Unfortunately, additional observations from the same orbit were not acquired and we cannot comment on whether the system is simply experiencing an early beginning to a periastron accretion flare or if this is a completely random event. The data also do not posses temporal coverage within 0.8 ⩽ ϕ ⩽ 0.35 to definitively say that infrared periastron flares are similarly constrained in phase as the optical flares.

The fact that the anomalous apastron flare or outburst occurs when the stars are near their greatest separation strengthens the conclusions of Jensen et al. (2007) that the flaring events are not solely produced by the interaction of the stellar magnetospheres, a potential flaring mechanism originally proposed by Basri et al. (1997). Neither the hydrodynamic model of Artymowicz & Lubow (1996) nor that of Günther & Kley (2002) predict significant pulsed accretion activity occurring near apastron. Unfortunately, we only have two observations preceding the apastron passage. As a result, we do not know if an increase in the accretion activity continues as the binary approaches apastron and whether that behavior is then followed by a decay or if the increased activity continues through the rest of the orbit. Previous data collected at multiple wavelengths show relatively little variation in the quiescent emission levels of the H i features at both optical and infrared wavelengths.

The simulations of Günther & Kley (2002) and de Val-Borro et al. (2011) provide a framework for considering the nature of the accretion flare observed at apastron. Both simulations predict that perturbations in the circumbinary disk will lead to the formation of irregularly shaped, non-axisymmetric structures extending inward from the inner edge of the disk. The presence of such structures coinciding with the close passage of one or both stellar cores at apastron, may lead to an enhancement of the mass accretion rate such as that observed in the anomalous flare. If the size, shape, density, and position of these structures vary considerably over time, such events may be both rare and aperiodic. The viscosity of the disk is a key parameter to the size and extent of such structures. Therefore, additional detections of apastron flares, including durations and distances from the disk at which the flares begin should provide rough constraints on the viscosity of material in the circumbinary disk. We plan to continue the spectroscopic monitoring of DQ Tau and extend these observations to high resolution to compare kinematics of the accretion line diagnostics in both the quiescent and flare states.

In addition to being excellent proxies for measuring the mass accretion rates in TTSs, the multiple near-IR hydrogen lines collected simultaneously in the cross-dispersed data allow us to probe the temperature and density of the accreting gas. By comparing the values of multiple hydrogen line ratios to those predicted by temperature- and density-dependent hydrogen line excitation models, we can search for the models that produce the most statistically significant fits to the data and constrain the gas temperature and density. If the emitting gas is located in the accretion column, either at the point where it is loaded into the flow or where it resides in the column (or both), the physical conditions of this gas may be revealed by such a fitting procedure. Previously, Bary et al. (2008) used line ratios from over 70 spectra of actively accreting TTSs and the well-known case B line excitation model, which assumes a recombining hydrogen gas in which transitions above the Lyman series are optically thin, to solve for the average temperature and density of accreting gas in TTS systems. Bary et al. (2008) found a range of electron densities (10⁹ cm⁻³ ⩽ ρ ⩽ 10¹¹ cm⁻³) that were in good agreement with those predicted by magnetospheric accretion models. However, the temperatures constrained by the best-fit models (T_gas ⩽ 2000 K) were considerably lower than the typical 6000 K ⩽ T_gas ⩽ 12,000 K predicted for accreting gas (Muzerolle et al. 2001). A new line excitation model developed by Kwan & Fischer (2011), specified for infalling/outflowing gas (|v| = 150 km s⁻¹), includes line opacity as a free parameter, in addition to gas density, kinetic temperature, an ionizing flux, and a local velocity gradient. The Kwan–Fischer (KF) line excitation model places the emitting gas within a few stellar radii of the stellar surface and includes photoionization as an excitation mechanism. Using the data from Bary et al. (2008), the KF models found a best-fit density of N_H ∼ 10¹¹ cm⁻³ for temperatures in the range of 8750 K ⩽ T_gas ⩽ 3 × 10⁴ K and N_H ∼ 5 × 10¹¹ for temperatures T_gas ⩽ 7500 K. Temperatures returned by the KF models are in better agreement with the accretion models. This result suggests that the case B models are not the appropriate line excitation models for accreting gas and that the H i line emission arises from collisional excitation, rather than from a recombining gas.

The DQ Tau system with its periodic accretion flares provides a unique opportunity to search for changes in the physical conditions of the accreting gas with mass accretion rate. While we have 20 distinct epochs of observations, 10 epochs are low-resolution CorMASS spectra that were not sensitive enough to detect multiple H i features during the quiescent phases of the orbit, making the line ratio analysis described above impossible with this data set. Using data from one moderate-resolution spectrum during periastron and the two low-resolution spectra that caught the system during a flare, one during periastron and another near apastron, we measured the line ratios for five Paschen series lines. We compared the observed Pan/Paβ decrement for each of these epochs to the range of KF models (10⁹ cm⁻³ ⩽ N_H ⩽ 10¹² cm⁻³; Edwards et al. 2013). In Figure 6–8, we present plots of the Paschen decrement for the five lines with measurable fluxes for these three epochs. We find that our data constrain the densities to be between N_H equals 6.3 × 10¹⁰ cm⁻³ and 1.0 × 10¹² cm⁻³, while placing no meaningful constraint on the gas temperatures between 7500 K to 12,500 K. Figures 9–11 present three-dimensional surface plots of the reduced χ² values for the density and temperatures. The "trough" that runs from low temperature to high temperature over a small range of densities clearly illustrates the wide range of acceptable temperatures for the accreting gas, while simultaneously demonstrating the KF models ability to constrain the gas density. Given the large uncertainties in our measurements of the line ratios, we cannot constrain the range of temperatures and densities well enough over the three epochs to detect variations of the physical conditions of the accreting gas. In the future, with the addition of more high signal-to-noise resolution spectra during both quiescent and accretion flare epochs, we hope to search for variations in the physical conditions of the gas as a function of accretion activity.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The moderate-resolution near-IR spectra collected with SpeX and TSpec provide an opportunity to measure a spectral type for DQ Tau from a wavelength region inaccessible to previous spectroscopic studies that were centered only on visible wavelengths (e.g., Joy & Abt 1974; Herbig 1977; Basri et al. 1997). Using the IRTF SpeX Spectral Library (Rayner et al. 2009), we begin by following a similar spectral typing procedure outlined in Vacca & Sandell (2011), which involves comparisons of spectral shapes and the strengths of molecular and metallic features. We make additional comparisons to giant stars using spectral standards from the SpeX Library and model spectra from Coelho et al. (2005) to test surface gravity effects. After finding compelling evidence for a much cooler spectral type than previously reported for DQ Tau and demonstrating the inability of variations in surface gravity to simultaneously account for the observed strengths of the TiO and FeH bands, we investigate the potential effects of large cool spots on the surfaces of the stellar companions. We show that composite spectra of warm photospheres and large, cool spots produce reasonable fits to the TiO bands at 0.85 and 0.88 μm and the FeH feature⁴ at 0.99 μm. These fits also distinguish between the three different values of A_V published for DQ Tau.

3.6.1. IR Spectral Typing

In Figure 12, we present a sequence of spectral standards taken from the SpeX Library that spans the range of spectral types previously determined for the DQ Tau system (M0V, M1V, and K4–5V/M1–1.5V; Joy & Abt 1974; Herbig 1977; Basri et al. 1997). Bracketed by the main-sequence standards that share the same general spectral shape are three DQ Tau spectra generated from one SpeX observation collected on 2011 November 13 UT and corrected for reddening using three values of the visual extinction (A_V) taken from the literature: 0.97, 2.0, and 2.13 (Kenyon & Hartmann 1995; Mathieu et al. 1997; Strom et al. 1989) and the extinction law found in Martin & Whittet (1990). Figure 12 clearly shows that one magnitude difference in visual extinction (ΔA_V = 1) can account for considerable variation in the shape of the DQ Tau spectrum shortward of 1.5 μm, which would result in a spectral type mismatch. For example, M4V–M6V seems appropriate for DQ Tau with an A_V of 0.97 whereas M0V–M2V is best for an A_V of 2.0 or 2.13.

Zoom In Zoom Out Reset image size

The presence of several molecular absorption features provide another means for spectral typing. Broad water absorption bands at 1.4 and 1.9 μm strongly affect the overall spectral shape, producing the hump-like or triangular features in the H and K bands (see Figure 12). The magnitude of the hump in the H band is best approximated by the M5V and M6V standards over the entire range of A_V, while the feature in the K band is better fit by M3–3.5V. It is important to note that the H-band spectra of young stars has a more pronounced triangular shape as compared to the more smoothly rounded features in the spectra of evolved field stars (Lucas et al. 2001; Allers et al. 2007; Allers & Liu 2013). Thus the shape of the DQ Tau spectrum in the H band will not be perfectly matched by the spectral shape of the more evolved standards in the SpeX Library. Nonetheless, clear variations in the strength of the H-band feature are evident in Figure 12.⁵

In addition to the broad water absorption features in the DQ Tau spectrum, we note the presence of the following molecular features: (1) TiO absorption bands at 0.85 and 0.88 μm, (2) FeH at 0.99 μm, (3) a strong H₂O band at 1.33 μm, and (4) ¹²CO bandheads in the K band. The strengths of the TiO absorption bands at 0.85 and 0.88 μm, though slightly affected by the choice of A_V, are much deeper than those of the K5-M2V spectra for all values of A_V. Figure 13 provides a direct comparison between a dereddened DQ Tau spectrum (2011 November 14 UT;⁶ A_V = 0.97) and the spectral standards that bracket DQ Tau between 0.8 and 1.1 μm. Based on the strengths of the TiO bands and the FeH feature, fits between the main-sequence standards and DQ Tau constrain the spectral type to be between an M3 and M4V over the 0.85–1.0 μm wavelength range, a value 2–3 spectral classes cooler than quoted in the literature. Schiavon et al. (1997) clearly demonstrate that the FeH feature heavily depends on the surface gravity with cool dwarf stars possessing the strongest FeH features. Therefore, the strength of the FeH feature, in addition to suggesting a cooler spectral type, also indicates a dwarf-like surface gravity for these stars.

Zoom In Zoom Out Reset image size

Figure 14 shows the DQ Tau spectrum between 1.1 and 1.35 μm, which includes the H₂O feature at 1.33 μm. The strength of the H₂O feature is best fit in the M2.5V to M3.5V spectral type range. Figure 15 contains the K band and several CO bandhead absorption features. The CO feature strengths were best fit between an M3.5V and M4.5V, in rough agreement with the spectral type suggested by the spectral shape in the K band. Any significant continuum veiling in the K band due to an infrared excess would result in a determination of spectral types corresponding to higher temperatures than the that of the actual photosphere. Therefore, M3.5V–M4.5V are upper limits to the spectral type determination from the K band. Regardless of the visual extinction value we choose to adopt, the strengths of these molecular features collectively and definitively suggest a spectral type cooler than those previously proposed for the DQ Tau system, with the earliest type being an M2.5V and other features suggesting a type as cool as an M6V.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

On the other hand, the strength of temperature sensitive metal lines such as K i at 1.16934, 1.17761, and 1.51725 μm (see Figures 12 and 14) and several Mg i features at 1.18314, 1.4881874, 1.50518, 1.57450, 1.57533, and 1.71133 μm (see Figures 14 and 16) indicate a better match with slightly early-M-type spectral standards, M2V and M0V, in agreement with those found in the literature. Joy & Abt (1974) used the strengths of TiO absorption bands at visible wavelengths to determine a spectral type of M0V. Herbig (1977) later classified DQ Tau as an M1V based on ratios of metallic lines in the optical (5850 < λ < 6700 Å). Most recently, Basri et al. (1997), using high-resolution echelle spectra, reported an M1–1.5V type based upon the strength of the TiO band at 7125 Å. However, in the same study, a comparison of the ratio of the temperature sensitive 6210.7 Å Sc i line to a pair of nearby temperature insensitive Fe i lines resulted in a far earlier spectral type of K4–5V. Continuum veiling was discussed as a potential explanation for the inconsistency in spectral types (Basri et al. 1997). However, the line ratios were formed from spectral features in a narrow region of the spectrum, which should have mitigated any effects of veiling on the spectral type. Basri et al. (1997) also suggested that the discrepancy may be due to cool star spots or strong chromospheric activity, but suggested that clear evidence in support of either scenario did not exist. Therefore, Basri et al. concluded that the lower surface gravity of the "puffed-up" TTS photospheres may lead to a strengthening of the TiO band, resulting in a cooler spectral type relative to that determined by the metallic lines.

Zoom In Zoom Out Reset image size

3.6.2. Testing Spectral-type Dependence on log g

The SpeX Spectral Library also includes spectral standards for giant stars over the spectral range presented in Figure 12; these standards enable us to test the suggestion that the "puffed-up" photospheres affect the spectral typing. In other words, do warm "puffed-up" or giant stars have TiO bands in the IR with strengths comparable to those observed in DQ Tau? In Figure 17, we compare two dereddened DQ Tau spectra to the spectra of giant stars in a spectral type range similar to those presented in Figure 12. By visual inspection, the DQ Tau TiO features at 0.85, 0.86, and 0.88 μm are clearly stronger than that of the giant stars as cool as an M4III. The low pressure photospheres of giant stars cannot account for the optical/IR spectral type discrepancy we observe for DQ Tau. In addition to the TiO feature, the giant star spectra also appear to lack the pronounced triangular spectral shape in the H band and have comparatively weak FeH and H₂O absorption features also suggesting that the photospheres in the DQ Tau system are more like that of a dwarf than a giant star.

Zoom In Zoom Out Reset image size

We investigated this further, using a library of synthetic spectra (0.3 μm ⩽ λ ⩽ 1.8 μm) calculated by Coelho et al. (2005) to probe the dependence of the TiO and FeH feature strengths on surface gravity. The library covers a broad range of effective temperatures (3500 K ⩽ T ⩽ 7000 K), metalliticies (−2.5 ⩽ [Fe/H] ⩽ +0.5), and surface gravities (0.0 ⩽ log g ⩽ 5.0). Figure 18 plots two model spectra over a wavelength region that includes a portion of the visible and very near-IR (0.65 μm ⩽ λ ⩽ 1.0 μm) for two stars with T = 4500 K (roughly a K4–5 per Basri et al. 1997), solar metallicity, and log  g values of 3.5 and 4.5. The lower log  g value of 3.5 represents a "puffed up" TTS,⁷ while 4.5 corresponds to a main-sequence star. The two model spectra were normalized over the same wavelength region and subtracted to produce the residuals plotted in the bottom panel. The magnitude of the residuals clearly shows that there is no discernible difference in the TiO features at 0.71 and 0.85 μm between the spectra with different log g values. A similar differencing of spectra between the target DQ Tau and the log g = 3.5 model spectrum ([DQTau − 3.5]) shows that the strength of the TiO and the FeH absorption features in DQ Tau far exceeds that of both model spectra. If the DQ Tau stars have surface temperatures on the order of a K4 or K5 spectral type, lower surface gravity is not a strong enough effect to account for the strength of the TiO features observed in this study and by Basri et al. (1997).

Zoom In Zoom Out Reset image size

Since most studies have determined DQ Tau to have an M0V to M1.5V spectral type, we also compare the strengths of the TiO bands in the synthetic spectra of two 3500 K stars (roughly the T_eff for an M2III–V), again with log g values of 3.5 and 4.5 in Figure 19. A comparison to model spectra in Figure 18 shows that the TiO features for the cooler stars are clearly much stronger and closer approximations of the TiO absorption in DQ Tau, as should be expected. The residuals left by differencing the synthetic spectra with different surface gravities and cooler temperatures show evidence for some dependence on surface gravity with the TiO features in the star with the lower log g value appearing stronger. Once again, we subtract the low surface gravity spectrum from DQ Tau ([DQ Tau–3.5]) and find that the TiO features are slightly stronger in the photospheres of the DQ Tau stars. However, what is most telling about the lack of sensitivity of the IR spectral type on log g is the strength of the FeH absorption feature at 0.99 μm. In contrast to the synthetic spectra of the 4500 K stars, the spectra of the 3500 K stars both possess FeH, a result of the temperature sensitivity of this feature. However, the two stars with different surface gravities have similarly strong FeH features both of which are clearly weaker than the FeH in the spectrum of DQ Tau. According to the study of Schiavon et al. (1997), FeH is strongest in the spectra of the coolest, dwarf stars (lowest temperature, highest surface gravity). According to the comparison made in Figure 18, the higher surface gravity of the log g = 4.5 is not enough to account for the strength of the FeH feature. Figure 18 also demonstrates that a photospheric temperature of 3500 K is not low enough to produce an FeH feature as strong as is observed in DQ Tau. In order to simultaneously fit the TiO features and the FeH feature in DQ Tau, a synthetic spectrum with a lower photospheric temperature (cooler than an M2) and/or a slightly higher surface gravity (greater than an M2III) is required. Therefore, we conclude that the strengths of both the FeH and the TiO features indicate that a low surface gravity associated with the "puffed up" envelopes of contracting TTSs cannot account for the discrepancy between the infrared and optical spectral types for such sources.

Zoom In Zoom Out Reset image size

3.6.3. Color Anomalies and Star Spots

Using infrared measurements, Gullbring et al. (1998b) and Fischer et al. (2011) demonstrated that weak-line or non-accreting TTSs appear brighter at longer wavelengths than main-sequence stars of the same spectral types. One result of this color anomaly is that A_V values determined from IR photometry will be larger than those determined from optical photometry. Gullbring et al. (1998a, 1998b) and Fischer et al. (2011) find systematically larger values of A_V for both non-accreting and accreting TTSs as compared to values determined from optical wavelengths. Gullbring et al. also demonstrate that spectral types derived for their sample of wTTSs will be later types when determined from longer wavelengths, similar to what we find for DQ Tau in the previous section (see Figures 4 and 5 from Gullbring et al. 1998b).

We used near-IR photometry from Kenyon & Hartmann (1995) to calculate A_V for DQ Tau (M1V) and found A_V = 3.6 and 6.6 from E(J − K) and E(H − K),⁸ respectively. Dereddening the DQ Tau spectra with such large values for A_V would drastically affect the shape of the infrared continuum, forcing it to appear as a much earlier type star. Gullbring et al. (1998b) suggest that very large axisymmetric spots may be responsible for the color anomalies and differences in values of A_V.

A defining characteristic of TTSs is their irregular photometric variability (Joy 1945, 1949), with one of the well-established sources of variability being the rotation and evolution of large, cool star spots on their surfaces (Bouvier & Bertout 1989; Herbst et al. 1994). The spot temperatures and fractional coverage of the stellar surfaces in TTSs are shown to be similar to those found on RS CVn stars (Hall 1972; Bouvier & Bertout 1989). Considerable effort has been committed to understanding the sizes and temperatures of spots on the surfaces of magnetically active stars, perhaps, more so than have been carried out for TTSs. Spot sizes, temperatures, and locations are deduced from photometric light-curve modeling and spectral line diagnostics (i.e., Doppler imaging; Strassmeier 2009). Relying on the fact that spots on active, low surface gravity stars are similar to those on TTSs, we discuss findings from a few studies of the TiO molecular bands in the photospheres of TTSs and RS CVn stars that are relevant for the analysis we perform below.

Neff et al. (1995), O'Neal et al. (1996), and O'Neal et al. (1998) used high-resolution spectroscopy of the TiO bands in the spectra of several RS CVn-type stars to constrain the spot and non-spot surface temperatures as well as the spot sizes. In these three papers, the authors developed a technique for fitting the RS CVn spectra using inactive G and K stars as spectral standards for the non-spotted photosphere and inactive M standards for the spots. For late-type active stars, O'Neal et al. (1998) found spot temperatures between 3400 and 3900 K and spot filling factors between 0.1 and 0.5 (10% and 50% fractional coverage, respectively), in good agreement with photometric light-curve modeling.

Herbst & Levreault (1990) and Berdyugina et al. (1992) first showed that variations in the strengths of the TiO bands near 7100 Å correlated with the existence, rotation, and evolution of cool star spots on the surfaces of TTSs. In light of this correlation, the anomalous strengths of both the TiO and FeH absorption features in the spectra of DQ Tau may strongly suggest the presence of cool spots on the surfaces of the DQ Tau companions (Schiavon et al. 1997). Inspired by the evidence pointing toward the contribution of cool spots to the spectral type discrepancy, the inability of surface gravity effects to reproduce the strengths of the TiO and FeH features, and the spectral fitting technique developed by O'Neal et al. (1998), we attempted to fit the TiO features at 0.85 and 0.88 μm and the FeH feature at 0.99 μm to determine if cool spots could be responsible for the spectral type discrepancy observed for DQ Tau. Modeling of the entire near-IR spectra would require a disk model to account for the infrared excess emission beyond 1 μm. Therefore, we focus on fitting the DQ Tau spectra in the spectral window from 0.8 to 1.0 μm and reserve the inclusion of disk models and associated parameters to future study.

Main-sequence spectral standards from the SpeX Library were used to construct composite spectra, representative of spotted stars. We chose M0V and M1V standard spectra to represent the photosphere based on optically derived spectral types. M6V and M7V standards were chosen to represent the spot temperatures of 3050 K and 2650 K, respectively. Composite spectra were calculated for six spot filling factors (f_sp ∼ 0.1, 0.2, 0.3, 0.4, 0.5, and 0.6). The spot filling factors weighted the contributions of the spectra produced by the spots to the final composite spectra. A goodness-of-fit to the observed DQ Tau spectrum was determined by finding the composite spectrum with the minimum χ² value for the two different photosphere and spot temperatures, six spot filling factors, and two of the published values of A_V: 0.97 and 2.13.

The best fits to the TiO features and the spectral shape between 0.8 and 1.0 μm was given by an M1V photosphere (M1V_ph) and a spot temperature of 3050 K (M6V_sp), with spot filling factors, f_sp = 0.4 and 0.5, and A_V = 0.97 (see Figures 20 and (b)). The χ² values, which are calculated from the residuals for all of the data points in the spectral window, presented in Table 3 indicate that the M0V_ph+M6V_sp model with f_sp = 0.6 is the best fit. While the fit of the spectral shape is slightly better, the fit to the FeH feature is extremely poor. Therefore, we dismiss this model as the best overall fit to the DQ Tau spectrum. We also take the M1V_ph+M6V_sp model with f_sp = 0.4 to be a slightly better fit than the f_s = 0.5 model based on the fitting of the FeH feature. The FeH feature is best fit by the smaller spot size, f_sp = 0.3 for the M1V_ph+M6V_sp model. Based on the work of Schiavon et al. (1997), we can interpret this discrepancy between fits to the TiO and the FeH as a combination of the strict low temperature sensitivity of the FeH feature combined with its dependence on log g. In other words, both a higher spot temperature and/or a smaller log g for the model star could account for the spot size discrepancy. This is observed in the models with the hotter M0V photosphere as the larger spot size f_sp = 0.4 produces the best fit to the FeH feature.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

View figure set (24 panels)

Tarball of figure set images

Table 3. χ² Values for Spotted Star Model Fits

Spot Filling Factor	M0Vph+M6Vsp	M1Vph+M6Vsp	M0Vph+M7Vsp	M1Vph+M7Vsp
(fs)	(AV = 0.97, 2.13)	(AV = 0.97, 2.13)	(AV = 0.97, 2.13)	(AV = 0.97, 2.13)
0.1	5.82, 1.17	2.48, 1.02	6.42, 1.32	2.86, 0.95
0.2	4.32, 0.92	1.58, 1.37	5.42, 1.09	2.22, 1.10
0.3	2.88, 0.94	0.84, 2.07	4.33, 0.93	1.58, 1.40
0.4	1.59, 1.37	0.38, 3.27	3.16, 0.92	0.97, 1.95
0.5	0.65, 2.46	0.37, 5.22	1.97, 1.21	0.51, 2.95
0.6	0.35, 4.63	1.18, 8.27	0.92, 2.11	0.40, 4.75

Download table as:  ASCIITypeset image

The atomic absorption features of Na ii and Ca ii infrared triplet are much weaker in the best-fit models as demonstrated by the residuals in Figures 20(a) and (b). The Ca ii absorption lines in the spectrum of DQ Tau are likely filled in partially by Ca ii emission related to the accretion activity in the system as well as continuum veiling. The Na ii features though less affected by accretion activity, will also be affected by any veiling at these shorter wavelengths. We discuss difficulties associated with measuring veiling in this region of the spectrum in the following section.

In determining the significance of the fits, it is also important to note that for the larger value of A_V = 2.13, the TiO bands and overall spectral shape in the 0.8–1.0 μm window were fit reasonably well by models with the smallest spot filling factor of 0.1 (see Figure 20(c)). However, the poor fit at the shortest wavelengths in this spectral region as well as the poor approximation of the FeH band by the smaller spot size result in the comparatively large χ² values. Even though the χ² values steadily decrease with spot size for both M1V_ph models, the poor approximation of the FeH feature leads us to conclude that spots with f_sp < 0.1 are not viable models for DQ Tau. Table 3 presents the χ² values for the 24 different combinations of spot sizes and photospheric and spot temperatures.

Based on the quality of the fits of the models possessing the larger spots and the smaller A_V values of 0.97, we conclude that large, cool star spots make significant contributions to the near-infrared spectra of the DQ Tau system and may partially account for the color anomaly and optical/infrared spectral type discrepancies previously observed for many TTSs.

Such a conclusion is relevant for at least two recent studies that address the near-IR spectral types of TTSs. First, Vacca & Sandell (2011) present a detailed study of TW Hya, an accreting TTS in the TW Hydra Association (Kastner et al. 1997). Based on the comparison of multiple atomic and molecular features and the strengths of the water absorption features at 1.4, 1.9, and 2.7 μm to spectral standards in the SpeX Library, Vacca & Sandell revised the spectral type for TW Hya from a K7V to an M2.5V. The results of our study suggest that large star spots may account for the spectral mismatch for this source as well, explaining why the molecular components and the near-IR spectral shape of TW Hya would indicate a significantly later spectral type for the source. A second study, McClure et al. (2013), presented a detailed investigation of SpeX spectra of 10 accreting TTSs in the Taurus-Auriga star forming region. McClure et al. (2013) used optical spectra to determine the spectral types and then went about fitting simultaneously for A_V, veiling (r_λ), and the IR continuum excess. For stars with small spots, the results from this study should accurately describe both the stellar photospheres and IR continuum excesses associated with these sources. However, should any of these stars have large star spots, we find that the determination of IR continuum excesses measured for these sources will be impacted, with the value of A_V being most strongly affected. Therefore, one must address the strength of the molecular features and the spectral shape of the IR continuum when solving for A_V, veiling, and continuum excess.

Having established both a spectral type and visual extinction, we proceed to measure the veiling as a function of wavelength. We follow a standard measurement procedure based on equivalent widths of several temperature-sensitive metallic lines in the dereddened spectrum and derive the veiling without taking spots into account (Fischer et al. 2011; Vacca & Sandell 2011; McClure et al. 2013). Note that we do not simultaneously solve for veiling, visual extinction, and infrared excess as Fischer et al. (2011) and McClure et al. (2013). Instead, we assume an M1V spectral type, A_V = 0.97, and no infrared excess. The veiling measurements for all epochs with moderate spectral resolution are presented in Table 4 and Figure 21. Based on the strengths of the accretion indicators, we know that the magnitudes of the periastron accretion flares vary from passage to passage. Therefore, it is not surprising to see in Figure 21 that the veiling is not always the highest at periastron for all lines measured. The same can be said for the veiling outside periastron passage where the values also fluctuate and indicate that, for some lines, the veiling can remain high through much of the orbit. Despite the range of values measured for veiling at periastron, a general trend suggests that veiling is consistently higher at periastron than the rest of the orbit.

Zoom In Zoom Out Reset image size

Table 4. Veiling Measurements by Orbital Phase

Orbital	rFeH	$r_{{\rm K}\,\scriptsize{I}}$	$r_{{\rm Mg}\,\scriptsize{I}}$	$r_{{\rm Fe}\,\scriptsize{I}}$	$r_{{\rm Mg} \,\scriptsize{I}}$	$r_{{\rm Fe} \,\scriptsize{I}}$	$r_{{\rm Mg} \,\scriptsize{I}}$	$r_{{\rm Al} \,\scriptsize{I}}$	$r_{{\rm Al} \,\scriptsize{I}}$	$r_{ {\rm Al} \,\scriptsize{I}}$	$r_{ {\rm Na} \,\scriptsize{I}}$
Phase	(0.99 μm)	(1.169 μm)	(1.183 μm)	(1.189 μm)	(1.244 μm)	(1.264 μm)	(1.588 μm)	(1.673 μm)	(1.676 μm)	(2.117 μm)	(2.206 μm)
0.402	0.52	0.73	1.25	0.68	0.64	0.80	0.66	1.26	1.01	1.73	1.58
0.466	0.45	0.76	1.19	0.78	0.71	1.26	0.70	⋅⋅⋅	⋅⋅⋅	1.81	0.04
0.539	0.46	0.85	1.56	1.02	0.75	0.88	0.91	1.76	1.78	2.03	1.75
0.546	0.48	0.83	1.47	0.98	0.77	0.95	0.98	1.72	1.95	2.33	1.63
0.666	0.49	0.81	1.85	1.03	0.76	0.93	1.10	1.58	1.97	⋅⋅⋅	1.95
0.961	0.56	1.09	1.09	1.84	1.10	1.17	1.01	1.59	2.23	⋅⋅⋅	2.43
0.962	0.52	1.19	1.00	1.90	1.09	1.14	1.08	1.72	2.17	⋅⋅⋅	2.42
0.963	0.56	1.05	1.45	1.25	0.89	1.01	0.86	1.91	2.36	2.40	2.13

Download table as:  ASCIITypeset image

The formation of the molecular absorption features within cool spots indicates the potential for variability in the strengths of the 0.85 and 0.88 μm TiO and 0.99 μm FeH bands in DQ Tau due to the rotation of the stars and evolution of the spots (Herbst & Levreault 1990; Berdyugina et al. 1992), in addition to veiling of the features and the potential for heating during accretion events. In Figure 22, we present 15 epochs of DQ Tau spectra for which we have data on both the TiO and FeH bands, plotting the percent difference between a DQ Tau reference spectrum taken at apastron (ϕ = 0.500) during an apparent quiescent accretion phase and each of the spectra that show clear evidence for variability in both the continuum shape and strength of spectral features. Visual inspection of these spectra reveal significant variations in the strength of the TiO features, but less so for the FeH feature. The shape and depth of the features are roughly consistent for observations outside of periastron passes, with the exception of ϕ = 0.372 and 0.433. In general, the largest variations in the strengths of both the atomic and molecular features coincide with epochs when the star is experiencing an increase in accretion activity (Figure 23). The TiO bands weaken significantly during epochs in which the accretion indicators are in or near a flare state, including the two exceptional flare epochs taken near apastron. However, one epoch very near periastron (ϕ = 0.897) shows little evidence of variation from the reference spectrum and thus retains a strong TiO band. Interestingly, the increase in strength of the Ca ii and H i features should correlate with a "filling" in of the TiO features due to veiling associated with the accretion activity.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The majority of the data that includes both the TiO and FeH bands are low-resolution CorMASS spectra, making difficult a simultaneous measurement of the veiling in both the TiO and FeH features. Establishing a continuum level in order to measure the strength of the TiO features also increases the uncertainty in such a measurement. In addition, a measurement of the strength of the TiO band is complicated by the contribution of the Ca ii emission features residing within the TiO band, which increases the uncertainty in the measurement of the TiO feature and the estimate of its variability. Assuming that the variation in the strength of the feature is due solely to veiling and after smoothing over the calcium features, we estimate r_TiO to be ∼0.4 during epochs in the quiescent phase of the orbit (ϕ = 0.404, 0.460, and 0.530) for which we have SpeX observations, and conservatively estimate a lower limit of r_TiO ∼ 0.6 and an upper limit of ∼1.0 for the heightened accretion activity occurring in the periastron passages observed with CorMASS.⁹ This is a significantly larger range of veiling than those measured for FeH. If both features, as suggested by the spot models, originate in large cool spots, we would expect them to vary in concert with one another. However, we do expect the FeH to be most strongly associated with the spots given its substantially lower dissociation energy of 1.6 eV as compared to the 6.8 eV for the TiO molecule. Unfortunately, the limited spectral resolution of our current data set limits our ability to draw a meaningful conclusion about the variability of these features. Future data sets with greater resolving power and high signal to noise will be important for probing the sources of variability such as veiling, spot evolution, spot heating, and stellar rotation.

Hartigan et al. (1995) and Basri et al. (1997) observe forbidden line emission with features possessing both blueshifted and redshifted velocity components characteristic of a disk wind. However, we find no evidence in the literature for a molecular outflow associated with DQ Tau binary. In a recent high-spatial resolution spectral imaging study of GG Tau Aa and Ab, another binary system that possesses standard indicators of outflowing gas (Hartigan & Kenyon 2003), but no evidence for a molecular outflow, Beck et al. (2012) find evidence for shocked molecular hydrogen emission (v = 1–0 S(1) at 2.1218 μm) residing inside the inner truncation radius of the circumbinary disk in this system. While this study lacks the spectral resolution necessary to determine if the gas is confined to an accretion streamer or an outflowing wind, previous high-resolution long-slit spectroscopy of the 2.1218 μm feature constrains the bulk motion of the gas to within 1 km s⁻¹ of systemic velocity for GG Tau A (Bary et al. 2003). With a full width at half maximum of ⩽14 km s⁻¹, it is not likely that the molecular gas is entrained in a fast moving bipolar outflow. Therefore, Beck et al. conclude that the H₂ gas most likely traces shocked molecular gas resulting from accretion streamers falling inward from the circumbinary disk.

If this novel stimulation mechanism is responsible for the emitting H₂ gas in GG Tau A, it may also be capable of producing similarly excited H₂ emission in a system like DQ Tau. Uniquely, in the case of DQ Tau, the dynamics of the system provides predictable accretion flares that may result in variable emission from shocked H₂ confined to the accretion streams. Given the strength of the H₂ emission detected in the GG Tau A system, we searched the only moderate-resolution spectrum taken of DQ Tau at periastron on 2012 October 2 UT for the 2.12 μm H₂ emission. Not surprisingly, without a high-resolution spectrum, the feature was not detected and an upper limit of 6.0 × 10⁻¹⁴ erg s⁻¹ cm⁻² was measured. Indeed, at this sensitivity, the GG Tau A H₂ emission feature (F_2.12 = 6.9 × 10⁻¹⁵ erg s⁻¹ cm⁻²) would not have been detected. We plan to observe DQ Tau at high spectral resolution during a periastron passage in the future to further test the idea of shocked gas as part of accretion streamers in binary systems.