Spatially Resolved Velocity Structures in Jets of DF Tau and UY Aur A

Star formation occurs when large clouds of gas and dust collapse due to gravity. The clouds are inhomogeneous in density and they fragment into smaller structures, where the center of each collapsing subcloud may become a star or system of stars. Very few of the resulting stars are massive and hot. By far the largest number will evolve into late-type stars with spectral types in the M–F range. Most stars are members of a binary or multiple system (see, e.g., a review by Duchêne et al. 2007). The infalling envelope flattens to a circumstellar disk, making the stars visible in the optical. Low-mass stars in this stage are called classical T Tauri stars (CTTS). For a few megayears, planet formation can take place before the disk disperses. For binaries or higher-order multiple systems, the disk can belong to an individual star or surround a close binary pair depending on the mass and separation of the components.

Mass is accreted through these disks onto the stars. However, conservation of angular momentum demands that some mass is ejected and carries away the angular momentum accreted through the disk. Mass loss occurs through wide-angle disk winds, but in some systems we also see highly collimated jets (see Frank et al. 2014, for a review). Based on different theoretical ideas, models of stellar winds (Kwan & Tademaru 1988; Matt & Pudritz 2005), X-winds (Shu et al. 1994), and disk winds (Blandford & Payne 1982; Anderson et al. 2005; Stute et al. 2014) have been proposed that can launch mass into an outflow. Ultimately these mass flows must be powered from the gravitational energy released in the accretion process. The collimation of the outflow into a jet is hard to explain without toroidal magnetic fields generated by a magnetic field that is dragged inwards in the accretion process (see a review by Lovelace et al. 2014). However, we do not know in detail how energy and momentum are converted from an inflow to an outflow (e.g., Matt et al. 2010, and references therein). Observationally, we see that the outflow rate is roughly one-tenth of the accretion rate (Cabrit et al. 1990; Coffey et al. 2008), supporting a causal relation between outflows and energy released in the accretion process.

Full 3D magnetohydrodynamic (MHD) simulations of jet launching are numerically challenging and most work has been limited to ideal MHD and/or less than three full dimensions (e.g., Casse & Keppens 2002; Dyda et al. 2015). Observationally, jets seem to be layered like an onion, where fast and hot inner components are surrounded by successively cooler and slower outflow components (see, e.g., Agra-Amboage et al. 2014; Güdel et al. 2018, for the well-studied case of DG Tau). The hottest emission is seen in X-rays and if those X-rays are formed through shocks, e.g., diamond shocks (Bonito et al. 2010; Ustamujic et al. 2016) or recollimation shocks (Günther et al. 2014), then the flow velocities of these inner components must reach at least 400 km s⁻¹. Forbidden optical emission lines (FELs) are a good way to find and study such jets, since the stellar photosphere and the accretion shock are too dense to contribute significantly to the emission. FELs typically have several velocity components and provide the highest kinematically determined velocities (Giannini et al. 2019).

Fendt & Gaßmann (2018) simulate jets launched by dynamos in the disk, as opposed to just stellar fields connecting to the disk. The jets in their simulations reach velocities comparable to the Keplerian velocity at the inner disk edge and show an onion-like velocity profile. If true, all jet components can be launched from the disk and there is no need to invoke stellar winds or X-winds to explain the fastest and hottest components. Observed line profiles are consistent with this new picture (Giannini et al. 2019). Takasao et al. (2017) propose the existence of a magnetically heated disk corona capable of launching jets that are initially hot. They estimate that even a weak magnetic field in the disk can cause a hot (>10⁶ K) corona, and, if true, no shocks would be required to explain the X-ray emission. However, in that case, the kinematically observed velocities in FELs might well be the fastest outflow components.

For binary stars, an added complication is the influence that one member of the binary has on the disk and outflow from the other star. The most important parameter to determine the strength of any gravitational torque is the binary separation. Simulations for binaries separated by no more than 15 au or so indicate that no jets are launched for high disk–orbit inclinations. For moderate inclination, the jet axis can precess in a cone of a few degrees opening angle (Sheikhnezami & Fendt 2018). We refer the reader to the introduction in Sheikhnezami & Fendt (2018) for a review of this topic. In contrast, in wide binaries, simulations and observations show random alignments between outflows from both members of the binary (Offner et al. 2016). At larger scales, the motion of the jet source on its binary orbit could cause observable wiggles along the jet (Moraghan et al. 2016).

If a jet is detected in several FELs, line ratios can be used to calculate density and the ionization fraction of the emission components in the jet with typical jet densities in the range 10³–10⁵ cm⁻³ (e.g., Solf & Boehm 1993; Bacciotti & Eislöffel 1999; Lavalley-Fouquet et al. 2000; Schneider et al. 2013). In turn the density and the velocity give mass loss rates. Different outflow components show different velocities. For example, in the well-studied CTTS DG Tau Schneider et al. (2013) find [O i] in a low-velocity component (LVC; about 60 km s⁻¹) which can be detected as close as 15 au from the star and a medium-velocity component (MVC; about 130 km s⁻¹) first detected at about 50 au from the source. The MVC in DG Tau slows down at larger distances. FELs can be seen up to at least 600 au from the star in DG Tau (Solf & Boehm 1993). FELs have also been detected in outflows from binaries, in some cases each star launches its own jet (e.g., in T Tau Solf & Böhm 1999), and in other systems it appears that the outflow is launched from a common circumbinary disk or at least that the individual jets from both components of the binary merge early (Mundt et al. 2010).

While a single spectrum is sufficient to detect the presence of a FEL, spatially resolved data is necessary to study how outflows accelerate and decelerate. In this work, we reanalyze archival data from the Hubble Space Telescope (HST) Program ID 7310 observed between 1998 and 2000 to search for FELs that are spatially resolved.

In Section 2 we describe the observations and the data reduction. Section 3 presents our immediate results. We discuss the resolved emission from DF Tau and UY Aur in Section 4 and end with a short summary in Section 5.

HST Program ID 7310 targeted binary T Tauri stars with long-slit spectroscopy using the Space Telescope Imaging Spectrograph (STIS). The long slit is always oriented such that both components of the binary are observed. Hartigan & Kenyon (2003) analyze the spectra of both stellar components to determine stellar properties and accretion diagnostics. In this work, we aim to spatially resolve the emission in FELs along the slit.

Table 1 lists the observations presented in this paper. The observations for each of the stars were taken using two different gratings, G750L and G750M, with central wavelengths of 7751 Å and 6252 Å with exposure times of 360 s and 1080 s respectively. For each target system the observations were consecutive in the same orbit. Position angles are measured east of north. Hartigan & Kenyon (2003) list properties (including binary separation and position angle, but also derived properties like stellar mass and accretion) in tables for all targets.

Our analysis begins from the pipeline reduced 2D sx2 files; these files have one spectral axis and one spatial axis for the coordinate along the slit. For each wavelength in sx2 we fit a single Gaussian to approximate the spatial flux distribution (dominated by the point-spread function (PSF) of the star), taking into account regions flagged for data quality by the pipeline. While this does not capture all features of the instrumental PSF, it describes the signal close to the peak of the emission well and allows a numerically stable fit of the position of the peak.

The measured peak position of the Gaussian changes with wavelength in a smooth manner, both due to minor misalignment between slit and CCD as well as due to the distortion corrections done in the data pipeline. The scatter in the fit results can be taken as an estimate of uncertainty of our fits. If both components of the binary system are within a few arcseconds of each other, a fit using two Gaussians is performed. All fits are visually inspected.

FELs are tracers of outflows because they can only be formed in low-density environments. We search for changes of the fitted position of the Gaussian, i.e., changes in the mean position of the emission caused by a resolved emission component offset from the central star around the wavelength of the FELs in Table 2. Limiting our search in this way reduces the rate of false positives.

To investigate resolved emission in more detail we subtract the contribution of the stellar continuum similar to the method of Schneider et al. (2013). For every row of in the image (in the dispersion direction) we select a region of interest centered on the position of the FEL (15 pixels wide for the G750M data and 2 pixels wide in the G750L data, where the FELS are not spectrally resolved). Assuming that the PSF is only slowly changing with wavelength and that the stellar emission around the FELs is continuum dominated, we fit a second-degree polynomial to the data to the left and right of the FEL. The region used for the fit is 10 pixels wide in the G750M data and 5 pixels wide in the G750L data. We subtract the value of the polynomial for every row of data in the FEL region. The remaining images should display signatures of jets if there are any, and can be used to examine jet profiles and intensities.

3.1. Significantly Extended FELs

We detected significantly extended emission only in 2 out of 21 objects, DF Tau and UY Aur. In both cases, the extension is seen in the [O i] lines, but is not significantly detected in any other FEL. Hartigan & Kenyon (2003) already noted that the on-source [O i] line profile shows different kinematic components for both components of the DF Tau binary and for the primary of UY Aur, while all other targets have simple [O i] line profiles. Hartigan & Kenyon (2003) also discuss the FS Tau primary where they note an [O i] line that appears spatially extended. FS Tau shows an extension in our processing, however, this is due entirely to the signal from a single pixel, which is flagged by the data processing pipeline as having a high dark current rate. No spatial extension is seen after removing that spurious signal.

We show position–velocity diagrams (PVD) for DF Tau in Figure 1 and for UY Aur in Figure 3. In the PVDs the origin of the spatial coordinate is set on the position of the primary star. Resolved FEL emission is seen as a bulge in the vicinity of the rest wavelength in the emission line. Where resolved FEL emission is seen, we quantify the peak velocity and width of the emission by fitting Gaussians to the spectrum at each position (row of data in the PVD). See Figure 2 for examples of those fits at different positions.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

3.1.1. DF Tau

The [O i] PVD for DF Tau is shown in Figure 1 corrected for the stellar radial velocity from Gontcharov (2006). In order to trace how outflow components accelerate or decelerate with increasing distance from the source, we fit one or more Gaussian emission components to each row of data (each row represents a spectrum at a constant distance from the source). Red lines in the figure mark the velocities of the peak of the fitted Gaussians. Examples of these fits are shown in Figure 2. The statistical uncertainty from the fit is less than the line thickness. We sum the fluxes and average the FWHM for all fitted Gaussians further than three pixels from the star center for each outflow component (Table 3). Uncertainties on the fluxes are dominated by the subtraction procedure and can be estimated by comparing the fluxes seen in the [O i] 6300 Å and [O i] 6363 Å lines, whose theoretical ratio should be 3 (Storey & Zeippen 2000). Based on the observed ratios, fluxes are accurate to approximately 50%.

Table 3.  Measured Fluxes

Source	Velocity	FWHM	6300 Å	6363 Å	$\dot{M}$
	(km s−1)	(km s−1)	a	a	b
DF Tau	−150	40	3	1	14
DF Tau	−50	60	2	.4	3
DF Tau	+150	40	3	5	14
UY Aur A	−150	30	0.6	2	5
UY Aur A	+50	40	0.3	1	0.8

Notes.

^aIn units of 10⁻¹⁵ erg s⁻¹ cm^{−2 b}In units of 10⁻¹⁰ M_⊙ yr⁻¹ assuming an outflow density of 10³ cm⁻³, see test for scaling to other densities.

Download table as:  ASCIITypeset image

Both sides of the jet can be traced to a similar distance. The observed velocity of the high-velocity component (HVC) of the approaching jet is about −150 to −200 km s⁻¹. Since we do not know the inclination angle with respect to the line -of-sight observed velocities are lower boundaries to the true jet speed. We also identify an MVC around −50 km s⁻¹, but the emission is weaker than the HVC. There is a redshifted counterjet at a velocity around +150 km s⁻¹ and a single emission feature at 50 km s⁻¹, located about 06 from the star (72 au projected on the plane of the sky). The HVC of the jet and counterjet is seen to about 05 (60 au). All those features are seen in both [O i] lines, which gives us confidence that even the weaker features are real detections and not just background fluctuations. We do not see significant changes in the velocity, thus all outflows must be accelerated in the inner 01 (12 au) where we do not resolve the jet emission.

For a divergent flow with constant velocity, the divergence of the jet streamlines is related to the FWHM for the observed FEL. Using the simplifying assumption of a homogeneously emitting cone, we can use equation (5) from Mundt et al. (1990) and the values in our Table 3. The inclination of the DF Tau outflow cannot be directly pole-on, since it is spatially resolved in our data. Assuming an inclination >20°, we derive outflow opening angles <1° for the fast components and <3° for the slower components. As discussed in Section 1, jets from CTTS are typically not homogeneous, but this nevertheless shows that we are observing highly collimated jets and not wide-angle outflows. Note that the measured values for the FWHM are a convolution of the line profile of the FEL and the instrumental resolution and thus the true opening angles would be even smaller.

3.1.2. UY Aur

We use the radial velocities from Nguyen et al. (2012) for UY Aur and show a PVD in Figure 3. Like in DF Tau, the position of the peak flux for resolved FEL emission is determined by fitting Gaussians to the spectrum at each position and the resulting statistical uncertainties are smaller than the width of the red line in the figure. There is an HVC around −150 to −200 km s⁻¹ visible to about 025 (40 au). This outflow component seems to originate on the primary and move away from the secondary. There is a second component, also originating from the primary (the origin on the y-axis in the figure), that points toward the secondary with a velocity around +150 km s⁻¹. In [O i] 6366 Å, the same components are seen but the signal is lower. Both plots in the figures also show apparent emission close to 0 km s⁻¹, but this is an artifact of our subtraction procedure. The emission is centered on the star. Since our procedure subtracts the stellar continuum only, the wings of the PSF of the [O i] emission at the stellar location remain in the images. Fluxes are given in Table 3. We integrate only the resolved signal further than 3 pixels away from the star to account for that. The low FWHM sets strict limits on the opening angle of the UY Aur outflow, showing that we observe highly collimated jets and not wide-angle outflows.

3.2. Upper Limits

3.3. Mass Loss Rates

With some assumptions on the physical conditions in the jet, we can estimate the total mass-loss rate in the resolved outflow components. Following Hartigan et al. (1994), we take a plasma with solar abundances emitting close to the peak formation temperature of the [O i] lines. Any jet components that are located outside the aperture, so hot that oxygen is ionized or so cool that that [O i] 6300 Å is not excited, or with densities above the critical density for [O i] (10⁶ cm⁻³), are not accounted for in this estimate. Thus, the estimate is a lower limit in the total mass flux in the jet. We use

$$\begin{eqnarray}\begin{array}{rcl}\dot{M} & = & 5.95\times {10}^{-8}{\left(\displaystyle \frac{{n}_{e}}{{10}^{3}{{\rm{cm}}}^{-3}}\right)}^{-1}\left(\displaystyle \frac{{L}_{6300}}{{10}^{-4}{L}_{\odot }}\right)\\ & & \times \left(\displaystyle \frac{{v}_{\mathrm{sky}}}{100\,{{\rm{km}}{\rm{s}}}^{-1}}\right){\left(\displaystyle \frac{{l}_{\mathrm{sky}}}{{10}^{16}{\rm{cm}}}\right)}^{-1}{M}_{\odot }{{\rm{yr}}}^{-1},\end{array}\end{eqnarray} \tag{ 1 }$$

which is Equation (10) from Hartigan et al. (1994). Since no other FELS are detected, the density can only be constrained to be below the critical density for [O i]. For the following discussion, we will use ${n}_{e}={10}^{3}$ cm⁻³ as the fiducial density because this number was observed in the jet of DG Tau (Lavalley-Fouquet et al. 2000), a well-studied collimated outflow in the same star-forming region, launched from a star of similar age and only slightly earlier spectral type (K6 versus M0-M2). Several FELs are observed in DG Tau, thus the density can be calculated from FEL ratios, unlike in our data from DF Tau and UY Aur. In practice, the density likely varies with distance to the source in all jets (as it does in the jet from DG Tau) and might also differ between DF Tau and UY Aur. All values for the outflow mass flux scale with 1/n_e. Even if the density was known in some way (e.g., a future observation), the uncertainty in the remaining factors in the equation would still make the estimated mass loss uncertain by a factor of a few. As it stands now, the mass accretion rates given in the following could be up to three order of magnitude smaller (if the density matches the critical density) or also much higher (if the density is much lower than the fiducial density chosen here). However, we regard such extreme densities as highly unlikely, given that our fiducial value is based on the observed density in a similar jet and the derived mass-loss rates are compatible with the ratios of accretion to outflow rate observed in other objects (see the discussion in Sections 4.1 and 4.2).

We detect spatially extended emission in about 10% of observations. The two components of the DF Tau binary are the strongest accretors in the sample (Hartigan & Kenyon 2003), while UY Aur is close to the median. If mass flux of the outflow is roughly proportional to the accretion rate as suggested by the observations (Cabrit et al. 1990; Coffey et al. 2008), then we would not expect to detect outflows from any source with accretion weaker than UY Aur. There are six binaries in the sample with accretion rates between UY Aur and DF Tau, where a jet could be detected, if it scales with the accretion rate and happens to be roughly aligned with the slit in the observations. Assuming that we can detect spatially resolved emission for jets misaligned with the slit up to about 30°, actually detecting jets in 3 out of 16 stars (eight binaries) with mass accretion rates of UY Aur or higher is statistically consistent with all stars having jet emission in a random orientation. However, the fact that the two binaries where extended emission is seen are the same binaries where at least one component shows multiple kinematic components in the [O i] emission lines (Hartigan & Kenyon 2003) suggests that the other objects may not have as pronounced outflows.

We compare our detections to other observations of the same jets in the literature to put the properties of these jets into context.

4.1. DF Tau

4.2. UY Aur

4.3. Comparison to Other Jets

We present a search for resolved emission in FELs in a sample of 21 long-slit HST/STIS observations observed between 1998 and 2000. We detected resolved [O i] emission in the binaries of DF Tau and UY Aur, the only two sources in the sample with complex [O i] lines profiles at the stellar position confirming the [O i] line profile as outflow indicator. In DF Tau, we see an HVC and a MVC in both the jet and the counterjet. Both are detected as far out as possible, given that our slit position angle is misaligned compared to the jet direction. The HVCs are accelerated within about 15 au and retain a constant velocity after that. The LVC likely has a different origin, as it must be part of a larger and possibly less collimated structure. In UY Aur, a complex outflow geometry with components originating on both members of the binary was known before. We detect all components that overlap with the spatial position of our slit, indicating that the outflow pattern is stable for at least one decade. Although both DF Tau and UY Aur are binaries, our observations show velocity components and profiles similar to single stars, indicating that the FEL emitting components are launched by the same mechanism as in single stars.

Support for this work was provided for A.V.U. and H.M.G. by NASA through grant GO-13766.010 from the Space Telescope Science Institute. This research made use of Astropy,³ a community-developed core Python package for Astronomy (Astropy Collaboration et al. 2013, 2018). This work has made use of data from the European Space Agency (ESA) mission Gaia (https://www.cosmos.esa.int/gaia), processed by the Gaia Data Processing and Analysis Consortium (DPAC, https://www.cosmos.esa.int/web/gaia/dpac/consortium). Funding for the DPAC has been provided by national institutions, in particular the institutions participating in the Gaia Multilateral Agreement.

Facilities: HST (STIS), Gaia.

Software: Astropy (Astropy Collaboration et al. 2013, 2018).