Time-monitoring Observations of Br$\gamma$ Emission from Young Stars

Young stellar objects have long been known as variable from UV to IR wavelengths (e.g., Joy 1942; Herbst et al. 1994; Skrutskie et al. 1996; Carpenter et al. 2001; Eiroa et al. 2002; Forbrich et al. 2007; Flaherty et al. 2012). In addition to continuum variability, changes in spectral features with time have been observed in some sources, with the first time-monitoring spectroscopy conducted soon after the discovery of T Tauri stars (Joy 1945). Subsequent spectroscopic monitoring of T Tauri stars, predominantly at optical wavelengths, has continued to the present day, providing constraints on gas kinematics and variability around young stars.

Variability in the line profile shapes of Balmer transitions, for example the appearance or disappearance of inverse P Cygni profiles, has been linked to changes in winds from, or infall onto, T Tauri stars (e.g., Bertout et al. 1977). The characteristic timescale of this variability is often observed to be hours to days, similar to expectations for magnetically mediated accretion (e.g., Bertout et al. 1988; Guenther & Hessman 1993). In particular, optical spectroscopic variability on timescales comparable to stellar rotation periods has been interpreted as evidence for rotationally-modulated accretion along stellar magnetic field lines (e.g., Giampapa et al. 1993; Johns & Basri 1995; Bouvier et al. 2007). Later studies extended similar monitoring, and results, to the higher-mass analogs of T Tauri stars, the Herbig Ae/Be stars (e.g., Mendigutía et al. 2011; Costigan et al. 2014).

Not all young stars exhibit spectroscopic variability consistent with changes in the magnetospheric accretion process. In a handful of objects, notably those characterized as UX Ori variables (e.g., Grinin et al. 1994), spectroscopic variability has been attributed to time-variable extinction from dusty clouds orbiting in the inner circumstellar disk (e.g., Rodgers et al. 2002; Schisano et al. 2009; Rush et al. 2011). Variability in accretion rate may also lead to changes in inner disk structure that cause time-variable extinction (e.g., Stone et al. 2014).

Observed variability of optical emission lines—notably H$\alpha$ and other Balmer series lines—is often interpreted in terms of accretion variability, which is reasonable given the empirical correlation of line luminosities with accretion rates (e.g., Gullbring et al. 1998; Rigliaco et al. 2012). However, optical line profiles for many young stars appear to trace a combination of infalling and outflowing matter (e.g., Reipurth et al. 1996; Kurosawa et al. 2006). This need not destroy the statistical correlation of optical line flux and accretion rate, since the outflow rate generally traces the accretion rate (with a multiplier of about 0.1; e.g., Hartigan et al. 1995). However for any given object, it may be hard to distinguish whether observed variability is due to a variable accretion flow, a variable outflow, or a combination of both.

One way to distinguish infall and outflow variability is to monitor Br$\gamma$ emission. Br$\gamma$ emission is also empirically correlated with accretion rate (Muzerolle et al. 1998; Mendigutía et al. 2011). While H$\alpha$ line profiles often show blue-shifted absorption components, suggestive of outflowing matter, this absorption is typically absent in Br$\gamma$ profiles. Br$\gamma$ lines also generally show blue-shifted emission centroids. Overall, the line profiles of Br$\gamma$ are more consistent with pure infall than H$\alpha$ and other optical lines (e.g., Najita et al. 1996; Folha & Emerson 2001; Muzerolle et al. 2001). Br$\gamma$ has the additional advantage that it suffers less extinction from foreground material; this can be particularly important for younger, more embedded sources.

A handful of young stars have multiple epochs of Br$\gamma$ emission in the literature. For some of these, the epochs were observed with the same instrument and analyzed in a consistent manner (Folha & Emerson 2001; Eisner et al. 2007, 2010a; Sitko et al. 2012; Pogodin et al. 2012; Mendigutía et al. 2013; Stone et al. 2014). Two of these works obtained simultaneous observations of optical and near-IR emission lines, clearly demonstrating the wind-like line profiles of H$\alpha$ along with the infall-like line profiles of Br$\gamma$ (Pogodin et al. 2012; Mendigutía et al. 2013). These studies show that the variability of the optical and near-IR lines are correlated, but suggest that the variability amplitude may be (marginally) smaller for Br$\gamma$ emission than for H$\alpha$ emission.

To better constrain the physical mechanism behind near-IR spectroscopic variability, we seek monitoring observations over a range of timescales, for a large, diverse sample of young stars. Here we present multi-epoch observations of Br$\gamma$ emission from young stars spanning a wide range of both stellar and circumstellar properties. This paper presents a larger sample, with better temporal coverage, than existing work. We also present the first observations of Br$\gamma$ emission from a number of sources, expanding significantly the existing sample of early-type young stars.

We selected a sample of T Tauri and Herbig Ae/Be stars that were visible from Tucson during the summer months³. The sample includes: 5 T Tauri stars, pre-main-sequence analogs of solar-type stars; 17 Herbig Ae/Be stars, which are young, intermediate mass stars surrounded by circumstellar material; 2 FU Orionis stars, thought to be young stars surrounded by particularly active accretion disks; and one heavily-veiled object whose spectral type is uncertain. We provide some basic information about our sample below.

The T Tauri stars in our sample are AS 205 N, V1002 Sco, V2508 Oph, AS 209, and V521 Cyg. The first four are in the Ophiucus or Upper Sco regions, all at an assumed distance of 160 pc (e.g., Chini 1981; Walter et al. 1994). V521 Cyg is located in the “Gulf of Mexico” region (Herbig 1958; Armond et al. 2011), at an estimated distance of $\sim 520$ pc (Laugalys et al. 2006). These sources have spectral types between K6 and K0 (Eisner et al. 2005; Walter et al. 1994; Torres et al. 2006; Herbig 1958), corresponding to stellar masses of $\sim 0.9$–1.5 M${}_{\odot }$ (e.g., Eisner et al. 2005; Hartigan et al. 1995).

The Herbig Ae/Be stars in our sample include 6 Herbig Ae stars: HD 141569, MWC 863, 51 Oph, MWC 275, VV Ser, and V2020 Cyg; and 11 Herbig Be stars: EU Ser, MWC 297, V1685 Cyg, V1972 Cyg, AS 442, V2019 Cyg, LkH$\alpha$ 169, V645 Cyg, V380 Cep, V361 Cep, and MWC 1080. These sources have spectral types between A2 and O7 (e.g., Hillenbrand et al. 1992; Mora et al. 2001; Cohen 1977), corresponding to a stellar mass range of $\sim 2$–10 M${}_{\odot }$ (e.g., Palla & Stahler 1993). These targets are located in several different regions, at distances between $\sim 100$ and 3500 pc (Herbig 1958; Strom et al. 1972; Chavarria-K. 1981; Goodrich 1986; de Lara et al. 1991; Hillenbrand et al. 1992; Perryman et al. 1997; Miroshnichenko & Corporon 1999; Laugalys & Straižys 2002; Straižys et al. 2014).

We included two FU Orionis objects in our sample: V1057 Cyg and V1515 Cyg. These are two of the prototypical FU Orionis objects (Herbig 1966), and are thought to represent circumstellar disks in the midst of enhanced accretion events (e.g., Hartmann & Kenyon 1996). V1057 Cyg is in the North American nebula at a distance of 600 pc (Laugalys & Straižys 2002), while V1515 is found in the Cyg R1 region at $\sim 1000$ pc (Racine 1968).

V1331 Cyg is a heavily veiled object lacking clear stellar photospheric absorption features (e.g., Eisner et al. 2007), and hence its spectral type is unknown. However it is known to be an actively accreting object, as traced by strong Br$\gamma$ emission. V1331 Cyg is one of the few young stars known to exhibit strong CO overtone emission (e.g., Carr 1989), perhaps indicating a particularly active accretion flow for this source (e.g., Eisner et al. 2014). V1331 Cyg is found in the NGC 7000 region, at a distance of $\sim 700$ pc (e.g., Herbig 1958; Chavarria-K. 1981).

Most objects in our sample have existing Br$\gamma$ spectra in the literature. We compare these existing data with our measurements below in Section 4. We also present the first observations of Br$\gamma$ emission in 10 sources: 2 T Tauri stars (AS 209 and V521 Cyg); one Herbig Ae star (V2020 Cyg); and 7 Herbig Be stars (EU Ser, V1972 Cyg, V2019 Cyg, LkH$\alpha$ 169, V645 Cyg, V380 Cep, and V361 Cep). These Br$\gamma$ spectra of Herbig Be stars result in a sample of early-type young stars approximately twice the size of previous work (e.g., Garcia Lopez et al. 2006).

We used the near-IR spectrograph FSPEC at the Bok 90-inch telescope to monitor the Br$\gamma$ emission around our sample of young stars (Section 2). We employed a grating with 600 grooves per mm, providing a resolving power of $\lambda /\Delta \lambda =7400$ (per pixel). FSPEC provides a ${2.4}^{\prime \prime }\times {96}^{\prime \prime }$ slit, with ${1.2}^{\prime \prime }$ pixels. The slit is $\sim 2$ pixels wide, and the actual spectral resolution is closer to 3500.

In our observations we nodded along the slit, observing each object at 5–6 distinct slit positions. Total integration times (including all slit pointings) for each observation are listed in Table 1. Observations of target objects were interleaved with observations of telluric calibrators. We selected dwarf calibrators with spectral types later than G6V; in fact all calibrators except HD 174719—used to calibrate targets in Serpens—have spectral types later than G8V. Such late-type calibrator stars do not show photospheric Br$\gamma$ absorption, so no spurious signals are introduced into the target spectra.

We developed an IDL-based data reduction pipeline to produce calibrated spectra (described in detail in Eisner et al. 2013). The pipeline works on an entire night of data as a block, producing telluric-corrected, wavelength-calibrated spectra for each object observed during the night. Because we work with the entire night of data at once, we can use a weighted sum of all the data to produce higher signal-to-noise in certain calibrations.

The data reduction procedure includes: creation of median flat and dark images; generation of a bad pixel mask; correction of bad pixels in the data; sky subtraction; shifting-and-adding nod sets; spectral extraction; telluric calibration; and wavelength calibration. Our observations were not taken under photometric conditions, so we do not attempt to associate real continuum fluxes with our reduced spectra. Rather, we divide each spectrum by its continuum level so that all spectra for a given object share a common normalization. We thus ignore any potential variability in the continuum flux of our targets. Such variations would not be surprising given the known infrared variability of many young stars (e.g., Skrutskie et al. 1996; Eiroa et al. 2002).

Spectra of our targets are plotted in Figures 1–5. Nearly all targets show Br$\gamma$ emission. A few targets show absorption features, while two show no discernible Br$\gamma$ features. A number of objects show differences in spectra from epoch-to-epoch.

The random uncertainties in each spectrum can be judged from the channel-to-channel variations in continuum regions. While the uncertainties are different from epoch to epoch, typical error bars are $<1\%$ of the continuum level. Even for the faintest objects in our sample (EU Ser and V521 Cyg), the uncertainties are $<5\%$ of the continuum level.

Because we do not calibrate the continuum flux level of our observations, we can not calculate Br$\gamma$ line fluxes for our targets. Rather we determine equivalent widths (EWs), which provide a measurement of line flux relative to the continuum level. Although some fraction of inner disk continuum emission may trace gas, perhaps free-free emission from H or H${}^{-}$ (e.g., Tannirkulam et al. 2008; Eisner et al. 2009), to first approximation, the continuum emission in the near-IR traces dust. Within this approximation, one can view EW as a measure of Br$\gamma$ gas to dust emission.

After subtracting the continuum level, EWs are measured by simply integrating the flux across the Br$\gamma$ feature, from 2.159 to 2.174 $\mu$m. We do not attempt to correct the spectra for any photospheric absorption, and so measured EWs may include both stellar absorption and circumstellar emission (or absorption). To estimate errors in measured EWs, we simulate 1000 noise realizations for each spectrum. Each simulated spectrum consists of the observed spectrum plus Gaussian noise, where the $\sigma$ of the noise is determined from line-free regions of the observed spectrum. In addition, we include the effects of bad pixels at random locations, multiplying two pixel values in each synthetic spectrum by 5. The uncertainty in the EW measurement is estimated as the 1$\sigma$ confidence level of the EWs determined for the 1000 synthetic spectra. EWs and uncertainties for each target and observed epoch are listed in Table 1.

Two objects in the sample show no evidence of Br$\gamma$ emission (or absorption) in any observed epoch: V1002 Sco and V361 Cep. Another source, V1057 Cyg, shows only hints of Br$\gamma$ features in its spectra, with most epochs consistent with an EW=0. Br$\gamma$ emission or absorption features are detected in the remaining 22 targets in our sample. For 8 objects, these are the first reported detections of Br$\gamma$ emission: AS 209, EU Ser, V1972 Cyg, V2019 Cyg, V2020 Cyg, V521 Cyg, V645 Cyg, V380 Cep. Br$\gamma$ absorption is reported for the first time in LkH$\alpha$ 169, although the origin of this spectral feature is presumably the stellar photosphere rather than circumstellar gas.

About half of the objects with detected Br$\gamma$ emission show variability, as traced by measured EWs, across the observed epochs (Table 1). The sub-sample exhibiting significant Br$\gamma$ variability includes AS 205 N, V2508 Oph, MWC 863, MWC 297, VV Ser, V1695 Cyg, V1972 Cyg, AS 442, V2019 Cyg, and V1331 Cyg. Several objects also show some variability in observed Br$\gamma$ line profiles from epoch to epoch. A notable example is V1331 Cyg, where inverse P Cygni absorption features appear in some epochs (Figure 5).

In the following subsections, we describe the line profiles and measured EWs for each target in the sample individually. We also compare with previous Br$\gamma$ measurements in the literature, for sources where such data exist.

We presented spectra of a sample of 25 young stars, covering the Br$\gamma$ feature at a resolution of $\approx 3500$. For 10 of these targets, the data presented here are the first observations of the Br$\gamma$ spectral region, and emission was detected from 8 for the first time. For the other targets in our sample, which had been previously observed, we presented multiple epochs of new data, which we used to constrain the time-variability of Br$\gamma$ emission from our sample.

For most targets, observations were taken over a span of $>$1 year, sampling time-spacings of roughly one day, one month, and one year. We detected Br$\gamma$ emission from 21 sources, and saw significant variability in 10 of them (approximately 50%). This variability is constrained quantitatively through measurements of EW. In some objects, we also observed changes in line profiles with time.

The observed variability seems to occur over a range of timescales. Night-to-night variations were detected in some objects, and a periodogram analysis suggested that some variability occurs on such timescales. These timescales are similar to stellar rotation periods, and are compatible with those expected for rotationally-modulated accretion onto the central stars. However the line profiles—which typically lack redshifted absorption across all observed epochs—are inconsistent with rotationally-modulated accretion models. Furthermore, some targets in our sample show evidence for substantial variations in Br$\gamma$ EW and line profiles on timescales of months or years, much longer than stellar rotation periods.

An alternative explanation for observed Br$\gamma$ variability is time-variable extinction or changes in the continuum flux level. Such variability might occur on inner disk dynamical timescales, which are typically days to weeks, or on longer timescales if there are orbiting clouds of dust at larger radii. While continuum changes or time-variable extinction could lead to variability in Br$\gamma$ EW, line profiles would remain approximately constant. This scenario appears consistent with the observed variability for many of our targets.