EDP Sciences
Free Access
Issue
A&A
Volume 603, July 2017
Article Number A74
Number of page(s) 28
Section Stellar structure and evolution
DOI https://doi.org/10.1051/0004-6361/201527494
Published online 10 July 2017

© ESO, 2017

1. Introduction

The influence of the multiplicity on the evolution of the protoplanetary disc is an open question of astronomy (see e.g. Bouwman et al. 2006; Kraus et al. 2012). Multiple systems are present abundantly among young stars, as a large fraction of stars form in binary or multiple systems, and, for example, Lada (2006) has found that the single star fraction is only 40% for G-type stars, rising to 70% for late M-type stars. This shows that multiplicity might be a significant factor in the stellar evolution. The young multiple systems are of special interest, because the low-mass evolutionary models at the early phases of stellar evolution are currently only poorly constrained by observations, and observing such systems can help to refine and calibrate these models (Stassun et al. 2009).

There is evidence that the age at which the star still has a disc is shorter in multiple systems than around single stars (see e.g. Damjanov et al. 2007; Bouwman et al. 2006). In addition, Osterloh & Beckwith (1995) claimed that companions closer than 100 AU inhibit disc formation, based on a 1.3 mm continuum survey of 121 young stars. Andrews & Williams (2005) found that among the 150 young stars in Taurus (including 62 multiple systems), the sub-millimeter flux densities (and thus the disc masses) are lower for binaries with a projected semi-major axes <100 AU. Cieza et al. (2009) also found that circumstellar disc lifetimes are reduced in binaries with separation smaller than 10–100 AU. Although the discs of the components are usually assumed to be coeval, they may evolve with different pace. This differential disc dispersion is in the focus of recent studies (Patience et al. 2008; Kraus & Hillenbrand 2009; Cieza et al. 2009; Kraus et al. 2012; Daemgen et al. 2012a,b, 2013). However, although these studies derived some correlations between the presence of a disc and binarity, they did not perform any detailed analysis of disc content and structure. Simulations combining viscous draining and X-ray photoevaporation suggest that in systems with separation below 100 AU the secondary disc disappears first, while for wider systems this trend cannot be seen (Reipurth et al. 2014, and references therein). The actual disc removal process that causes the relatively short disc lifespan and the differential disc dispersion is unclear. The disc removal processes may consist of tidal effects in a binary or multiple systems that destabilizes the disc structure, or the disc may deplete over time by accretion onto the star. It is also possible that planet formation is responsible for clearing the disc, or the disc may suffer photoevaporation from the star.

Which of these effects is the main driver for disc dissipation in multiple systems is still unclear. However, it is likely that the actual disc dissipation process depends on the binary parameters of the systems, such as the age, separation, mass ratios, and multiplicity of the systems. To determine which process plays the dominant role in the dissipation, we therefore need to witness the differences in disc clearing and determine binary parameters in a sample large enough to make statistical conclusions.

Obtaining orbital parameters for multiple T Tauri systems has been the focus of recent works (Schaefer et al. 2006, 2014; Köhler & Hiss 2015), and although some closer binaries have well-constrained orbital solutions, many wider systems still lack proper time coverage of their orbits (see e.g. Csépány et al. 2015, for the case of the T Tauri system). This can be alleviated by observing the same systems over a long period of time to obtain data that cover a significant fraction of the orbit.

In this paper we report high spatial resolution observations in the optical and infra-red of a sample of 18 multiple T Tauri systems. Most of the systems have been first resolved about 20 yr ago and therefore have a timeline on which we may be able to see orbital motion. One of our aims was to determine whether the systems are gravitationally bound, and if they are, to derive the orbital parameters and to ascertain a correlation between their binary configuration and disc states.

2. Sample

We studied a sample consisting of 18 T Tauri multiple systems (comprising 16 binary and 2 triple systems, which are treated as 2 binary pairs for each triplet, where the components B and C are both measured relative to A, the brightest star in the V or R bands, depending on available measurements; altogether 16 + 2 × 2 = 20 binary pairs) in the Taurus-Auriga star-forming region. Our sample is based on Leinert et al. (1993), who conducted a survey from September 1991 to October 1992 using speckle imaging at the 3.5 m telescope at Calar Alto. They found 44 multiple systems out of the 104 observed young low-mass stellar systems, and their measurements serve as an astrometric and photometric epoch for our targets. Our selection criteria considered observability using the employed telescope and instrumentation, together with the available observation time to ensure that we only include stars for which separation and relative brightness of the components allow a reliable detection. The selected sample covers spectral types from K1 to M5, with separations from 0.22′′ to 5.8′′. Most of the systems are well-studied stars, with multiple epochs available, but we also included a few systems that have not been extensively observed. Therefore our new observations present the first time at which these system have been spatially resolved since the pioneering work of Leinert et al. (1993). A detailed description of the systems is presented in Appendix A.

The proper motions of the main stars that we used for the analysis in Sect. 5 are obtained from the publicly available UCAC4 all-sky star catalogue (Zacharias et al. 2012) and for one star (FV Tau/c) from the USNO-B catalogue (Monet et al. 2003).

2.1. Naming scheme

The naming of the companions is not evident when we compile a database from observations that span several decades. Some companions may have been resolved only recently, while others were simply renamed. The renaming usually was minor, such as “HK Tau/c” to “HK Tau B”. However, sometimes we have to be very careful, as in the case of “FV Tau”, where “FV Tau” and “FV Tau/c” denotes two binary systems with a distance of ≈ 12″ from each other. We tried our best to match the different names with each other’ and to ensure that one name indeed refers to the same physical object. The naming scheme we employed is that the optically brightest star is A, the second brightest is B, and so on, with historical considerations to stay compatible with the most popular nomenclature in the literature. The final naming scheme is listed in Table 1.

2.2. Distances

Loinard et al. (2005, 2007), Torres et al. (2009, 2007, 2012) carried out a long VLBI campaign to measure the distance of the nearby star-forming regions, in which they measured three sections of the Taurus region. Taking into account all distances measured in this region and their spread (which probably reflects the three-dimensional extent of the Taurus cloud), we adopted 140 ± 21 pc as the distance of the stars in our sample.

Table 1

Spectral types, extinction magnitudes, classification, and naming scheme notes of the stars.

thumbnail Fig. 1

AstraLux images from the sample showing a pair of similar brightness, a pair of a bright and a faint component, a wide and a tight pair, and the two triple systems. The halos due to the lucky imaging addition are visible around the clearly resolved central cores of the stars. The intensity scaling is logarithmic.

Open with DEXTER

3. Observations

3.1. AstraLux Norte

We observed the selected 18 multiple systems between November 2006 and November 2007 at Calar Alto using the 2.2 m telescope and the Astralux Norte lucky imaging system (for a detailed description of Astralux Norte and the reduction pipeline see Hormuth et al. 2008). In the lucky imaging process we took 10 000 images per object and exposed for 30–50 ms for each image. Johnson I and SDSS i and z filters were used, since at that exposure time and telescope size, those are the optimal wavelengths (8001000 nm) to minimize the effect of the atmospheric turbulence (Fried 1965). Then we measured the Strehl ratios of the reference star in the images and selected the best 1–5% of the frames, which were composed into a final image using shift-and-add technique with the Drizzle algorithm (Fruchter & Hook 2002). This method selects the images in which the effect of the atmospheric turbulence was the lowest (as the turbulences in the air cells at the given wavelength operate at a longer time span than 30–50 ms), allowing us to reach the diffraction limit of the telescope. The typical FWHM of the primary stars in the AstraLux images range between 80 and 92 mas, depending on the filter.

The composite Fig. 1 shows six systems from the AstraLux observations, presenting a stellar pair of similar brightness, a pair of a bright and a faint component, a wide and a tight pair, and the two triple systems.

3.2. SPHERE

We also obtained observations of the V1000 Tau system during the Science Verification of SPHERE (Beuzit et al. 2008; Dohlen et al. 2008; Vigan et al. 2010; Claudi et al. 2008), the newly installed extreme adaptive optics facility at the VLT. These observations were acquired on 2014 December 10, when SPHERE was operated in the IRDIFS mode with a 155 milliarcsecond (mas) diameter apodized Lyot coronagraph, offering simultaneous Integral Field Spectrograph (IFS) observations from 0.95–1.65 μm, and imaging in the Ks (2.181 μm) filter. We also obtained observations with the star offset from the coronagraphic mask to be able to measure total fluxes. A point-spread function (PSF) reference star (TYC 1290-457-1) was observed immediately after the IRDIFS observation of V1000 Tau, at similar airmass and atmospheric conditions as the science target. The obtained data sets were reduced with the pre-release version 0.15.0-2 of the SPHERE pipeline. In addition to the usual steps of bias subtraction, flat-fielding, and wavelength calibration, the pipeline also includes a correction for geometric distortions. In the calculated positions of the stars, we also took into account the detector’s deviation from true north orientation in the position angle (− 1.788° ± 0.008°, detailed in the SPHERE manual1). Post-processing of the data was made using the FITSH software package (Pál 2012) to sum the data resulting from the two IRDIS channels.

3.3. NACO

We also took advantage of the extensive ESO archive and searched for publicly available data on the given systems. We found 19 previously unpublished adaptive-optics-assisted imaging observations of our targets captured by VLT/NACO (Lenzen et al. 2003; Rousset et al. 2003). We reduced the previously unpublished observations (which employed standard J, H, K and narrow-band 1.64 μm, 2.12 μm, 2.17 μm filters) with the ESO provided NACO pipeline2 (version 4.4.0), achieving ≈ 65 − 85 mas FWHM resolution in the K band. These observations are also included in our analysis.

3.4. Auxiliary data from the literature

A literature search for spatially resolved data of the systems in our sample resulted in including measurements from papers listed in Table C.1. The most extensive literature data sets come from White & Ghez (2001), McCabe et al. (2006), and Correia et al. (2006). McCabe et al. (2006) observed 65 T Tauri binaries using mid- and near-infrared adaptive optics instruments on the Keck 10 m telescopes between May 1998 and November 2002, and Correia et al. (2006) observed T Tauri triple and quadruple systems employing VLT/NACO. The observations of the objects are listed in Table C.1.

4. Data analysis

In the data analysis we obtained the positions of the stars in the AstraLux, SPHERE, and NACO images and calculated their separation and brightness ratios. We combined these results with the other surveys and list them in Table C.1.

4.1. Photometry

We performed PSF photometry on the AstraLux images and aperture photometry on the other images that we analysed to calculate the brightness ratios of the binary pairs.

In the aperture photometry calculations, we included the following term in the uncertainty of the brightness ratio (nph is the raw photon count):

  • due to the Poisson-process nature of observation;

  • due to the uncertainty caused by the sky background.

In the AstraLux PSF photometry, we used the PSF of lucky images that consists of an Airy disk convolved with a Gaussian and a Moffat function (described by Staley et al. 2010; and successfully used in AstraLux Norte images by Wöllert et al. 2014, 2015). This PSF can be expressed as where W is a weighting factor between the two components, σm is the width of the Moffat-profile, β is the Moffat power-law index, and σg is the width of the Gaussian. We constructed the theoretical AstraLux images as where Ai are the amplitude scaling factors, the PSFi(r) terms are the individual stars, and Csky is the sky background. The PSFi(r) terms share the common seeing parameters σm, β, and σg, as these parameters do not change in the field of view of the camera with images taken in 2050 ms. We used the modelling and fitting framework of the Astropy software package (Astropy Collaboration et al. 2013) to obtain the PSF fits.

Performing photometry on the AstraLux images differs in a few points compared to the conventional CCD imaging, see for example Staley et al. (2010). The main issue affecting the PSF photometry is that the bias level of an electron multiplying CCD (EMCCD) camera can change during acquisition (“bias drift”). This bias drift can be as high as 0.6% of the pixel counts. The bias drift can be accounted for by stabilizing the temperature of the camera or by exploiting the overscan region of the camera (see e.g. Harpsøe et al. 2012). Since we did not have any of these options, we calculated the spread of the bias from each night using the sky area of the data cubes. The change in the sky area in the data cubes can be attributed to the bias drift, since no other change is foreseen in the images that were taken 2050 ms apart. The mean standard deviation of the sky photon count is 0.7%, which we took into account as a 1% error term as a safe overestimation of this error.

The final brightness ratios are shown as magnitude differences (with uncertainties) in Table C.1.

4.2. Astrometry

In the case of the SPHERE and NACO observations, the analysis of the data is based on the pipeline-produced images (including pixel scale and detector position angle), whereas in the case of AstraLux images we calculated the pixel scale and the position angle of the images by using images of the M15 globular cluster (van der Marel et al. 2002) and the Orion Trapezium cluster (Olivares et al. 2013) taken both by the AstraLux system and by the Wide-Field Planetary Camera 2 (WFPC2) on the Hubble Space Telescope (HST). We picked five to nine stars in the clusters that were used to align the AstraLux images and calculated the pixel scale to be 23.71 ± 0.01 mas per pixel (this is the pixel scale of the re-sampled and drizzled images that we used in the further analysis; the physical pixel scale is ≈ 47.4 mas/pixel). The error of this calculation is the root-mean-square deviation of the pixel scale and the position angle of the star pairs.

The position of each companion in the images was obtained by fitting a 2D Gaussian to the stellar profiles using the FITSH software package (Pál 2012) where the binaries were wide and bright enough to obtain a straightforward fit, and used the PSF fitting (as described in Sect. 4.1) to obtain the coordinates of the close or faint pairs.

The position of the stars relative to each other is also affected by the relative atmospheric diffraction, which we take into account by defining an error term. The possible maximum offset in the relative positions due to this effect is calculated by assuming the maximum relative shift in position of the components diffraction in the used filter (i.e. we take the extreme assumption that the spectral slope of one binary component is very blue while the other is very red, compared to each other). We took the wavelengths at the full width at half maximum (FWHM) value of the transmission function of the optical system, where the transmission function is composed of the transmission function of the filter (the AstraLux system used the RG8303, Johnson I4 and SDSS i, z5 filters) and the quantum efficiency6 of the CCD. Using the airmass at the observed star, we obtained the upper limit of the relative atmospheric diffraction when one or the other of the stars is shifted to the red or blue end of the transmission function.

The astrometric measurements for each epoch of each pair are listed in Table C.1. The errors listed there are the combined errors due to the uncertainty of the Gaussian fitting of the stellar positions (mean: 2.8 mas), the errors due to the relative atmospheric diffraction (mean: 57.7 mas), and the uncertainty from the pixel scale conversion (mean: 7.8 mas).

4.3. Orbital fit

We estimated the orbital elements for each companion by fitting models of circular orbits to the astrometric data. A detailed description of the procedure can be found in Köhler et al. (2008). In short, it works as follows: to find the period, we employed a grid search in the range 100 to 10 000 yr. For each period, the Thiele-Innes constants were determined using singular value decomposition. From the Thiele-Innes elements, the semi-major axis a, the position angle of the line of nodes Ω, and the inclination i were computed. We have restricted the eccentricity to be zero in all cases as the number of available epochs is moderate and only covers a small portion of the orbit. The obtained orbital parameters are listed in Table 2.

Table 2

Parameters of the fitted orbits.

4.4. Spectral energy distribution

To properly characterize the systems in question, we need to know the spectral energy distributions (SEDs) of the stars. The SED can tell us about the extra or extended emission in the systems, which can be signs of the dust in the discs. Since for our purposes the most interesting part of the spectrum is between the optical and millimeter wavelength, we looked for photometric measurement in this region. We employed many photometric surveys: the measurements from Kenyon & Hartmann (1995), Howard et al. (2013) and the IRAS, Spitzer, 2MASS, WISE all-sky surveys, a few ALMA and HST observations, and our lucky imaging data.

Table 3

SED fitting parameters.

The magnitude data were converted into λFλ using the conversion formulas from Johnson (1966), Bessell (1979), Glass et al. (1982), Berrilli et al. (1992), Gehrz et al. (1974), Helou & Walker (1988), while the data from the surveys were either already given in Jy or had their own conversion formulas to obtain λFλ (e.g. WISE).

We fitted an SED curve on all observations, using the NextGen2 atmospheric models from Hauschildt et al. (1999)7 and the extinction formulae from Cardelli et al. (1989). We adopted RV = 3.1 for the total-to-selective extinction ratio, which is the general value for standard interstellar matter. We note, however, that Vrba & Rydgren (1985) also found RV = 3.1 applicable to Taurus. The spectral types were collected from the literature as listed in Table 1, and we calculated the effective temperatures from the spectral types of each component (using the relations from Table 6 in Pecaut & Mamajek 2013). We used the calculated effective temperature to select the model of the correct temperature from the NextGen2 models. We fitted the SEDs up to the H band (except for DD Tau, where we had to use K-band observations because of missing resolved magnitudes at shorter wavelengths) to avoid the influence of the excess emissions from the circumstellar discs present in many systems. The fitted values were the AV extinction and a scaling value that we obtained using a grid search. We have derived the uncertainties of the extinction magnitude using a Monte Carlo approach by randomly varying the measured flux values within 1σ and recording the extinction magnitudes of the different runs, repeated a few hundred times. We employed a “combined flux” fitting procedure where we took into account both the individual flux of the components and the combined flux of the whole system at the same time.

thumbnail Fig. 2

SED plots. The triangles show upper limits, the filled markers designate the combined flux from the systems, and the empty markers show the measurements of the individual components. The dotted curves are fitted SED models.

Open with DEXTER

The obtained extinction magnitudes are listed in Table 3, and plots of the fluxes and SED curves are shown in Fig. 2. The fitted extinctions agree with the literature data within uncertainties for many systems, but there are exceptions. In a few systems, only the extinction of the primary is in agreement with the literature (FV Tau, FV Tau/c, HK Tau, V710 Tau, GH Tau, and HV Tau); in these cases we anticipate that the discrepancy is attributed to the low number of resolved observations to fit (FV Tau/c, HK Tau, V710 Tau, GH Tau), that our optical measurement shows a higher extinction for the secondary (in the case of FV Tau), or that we know that the secondary probably has a high extinction magnitude due to an edge-on disc (in the case of HV Tau), where the companion may be seen through their discs. There are five systems where the obtained extinction magnitudes do not agree with the literature (LkCa 3, XZ Tau, HN Tau, V999 Tau, and RW Aur), where we know that two of them show significant variability in time (XZ Tau and RW Aur, see the notes in Appendix A), one system that is a known quadruple system consisting of two spectroscopic binary systems (LkCa 3, Torres et al. 2013), and there are two systems where the optical fluxes may be the cause of discrepancy (HN Tau and V999 Tau). The latter case, V999 Tau, simply lacks reliable optical measurements (although it was observed in the SDSS, it is flagged as a not clean observation), and HN Tau features a fairly flat SED in the optical that we cannot achieve a reliable fit using the NextGen2 models.

thumbnail Fig. 3

Motion of the companions. The coordinates are relative to the primary star, which is at (0, 0) in the plots (it is marked with a black star symbol when the (0, 0) coordinate is in the plotted area). The blue markers show the oldest and the brown markers the latest epochs. The maroon curve shows the fitted orbit with e = 0.

Open with DEXTER

thumbnail Fig. 4

Motion of the companions, plotted separately for the RA and Dec dimensions, time dependent. The red line is the average separation, i.e. if the companion is gravitationally bound and has no observable movement, it would line up on the red line. The blue line is the proper motion with the parallax added, the dashed blue line shows the uncertainty of the proper motion.

Open with DEXTER

Table 4

Proper motions, average movements per year, reduced χ2 of the orbital fits, and classification of the binary pairs.

The NextGen2 models only provide the photospheric SED curve up to 2.5 μm. To measure the infra-red excess at higher wavelengths, we therefore attached a black-body radiation curve to the K-band section of the model curve (with the same effective temperature that we used in the NextGen2 model), making the SED available in the whole wavelength range.

5. Results

5.1. Astrometry

The individual epochs of the system are shown in Fig. 3 in RA-Dec plots. The markers of different colours indicate different epochs, the blue markers show the oldest epochs and the brown markers the latest (the epochs are listed in Table C.1). The dark red curve is a circular fitted orbit (see details in Sect. 4.3).

We also show the individual epochs and the proper motion together in Fig. 4. Here we separated the RA and Dec coordinates to display the movement of the companion in a (time - spatial dimension) plot. The motions of the systems are summarized in Table 4.

If the companion were a background star, its epochs would line up with the proper motion of the primary star. However, we can see that in all systems the epochs deviate from the expected proper motion (see the deviation of the black markers from the blue line in Fig. 4). Since none of our binaries show motion comparable with the proper motion, we can conclude that each pair is either a common proper motion pair or a gravitationally bound pair.

5.2. Orbital fit

We studied the relative motion of the companion around the primary by obtaining orbital fits to decide whether the pair is a common proper motion pair or is gravitationally bound (see Sect. 4.3 for the description of the fitting procedure). We calculated the average relative motion of the companions and compiled Table 4 to convey the following classifications: relative motion over 3σ, which shows whether there is at least a 3σ difference between any two astronomical epochs, that is to say, there is detectable relative motion; and orbital motion, which is a classification of the binary pair based on the value of the orbital fit and the visual inspection in Fig. 5.

We based our visual inspection on the expectation that if the companion is orbiting the primary, then its coordinates must show a rising or decreasing trend in either RA or Dec (or both), in Fig. 5. If we do not see such a trend, then the companion is likely to move with the same proper motion as the primary star. The classification letters in Table 4 indicate the pairs orbiting each other with “Y” (which are also gravitationally bound); those that are probably orbiting each other, but the observational uncertainties are too high to draw certain conclusions with “C”; and those that do not show orbital motion with “N”. The last group most likely contains common proper motion pairs, because we do not see any pairs where the companion would be a background star (as mentioned in Sect. 5.1 and shown in Fig. 4). Since stars in the same star-forming region may have a similar evolutionary history, it is possible to find several binary pairs that have similar proper motions without being gravitationally bound. As for the statistics, there are ten pairs with detected orbital motion, five with possible orbital motion, and five that are most likely common proper motion pairs.

5.3. Stellar parameters and disc statistics

We calculated several properties from the SED fits. In Table 3 we list the parameters of the SEDs (spectral type, effective temperature, and extinction) and the derived parameters (luminosity, mass, and IR excess).

We calculated the stellar masses by using the 2 Myr isochrones from Baraffe et al. (2015), where we used the effective temperature and the luminosity as input parameters from Sect. 4.4. The derived masses of the components vary mainly between 0.08 and 1.32 M, which agrees with the general assumption that T Tauri stars are low-mass stars.

Infrared excesses were calculated at 10 μm, using either N-band, W3 (WISE), or IRAS 11.8 μm measurements. The difference is calculated between the fitted SED curve (which depends on the effective temperature and therefore on the spectral type) and the individual photometric data points, that is, we calculate the excess over the photospheric level. The IR excesses of the individual stars vary between 0.1 and 7.2 mag.

There are two resolved triple systems in our sample, UX Tau and UZ Tau. Unlike binary systems, triple systems can be unstable in some configurations. We studied two stability indicators, the separations and the masses.

Reipurth & Mikkola (2015) have examined the stability of triple systems based on the masses of the companions. Their triple diagnostic diagram is based on the Ma/Mb and Mc/ (Ma + Mb + Mc) ratios of the systems. Based on our derived stellar masses, the ratios in the UX Tau system are Mb/Ma = 0.97 and Mc/ (Ma + Mb + Mc) = 0.04. In the UZ Tau system, they are Mb/Ma = 0.67 and Mc/ (Ma + Mb + Mc) = 0.26. These ratios place both the UX Tau and the UZ Tau system in the disrupted regime of the triple diagnostic diagram.

thumbnail Fig. 5

Motion of the companions, plotted separately for the RA and Dec dimensions, time dependent. The red line is the orbit from the orbital fit. The black markers are literature data points, the blue markers are data points from our observations.

Open with DEXTER

thumbnail Fig. 6

Scatter correlation plots. The filled circles show stars with discs observed at long wavelengths, the star markers stand for the stars without a disc detected at long wavelengths (however, we stress that a disc could still be present at detection levels lower than what was used by Harris et al. 2012 or Cabrit et al. 2006) and the diamonds are the stars where we have no high spatial resolution long-wavelength observation of the systems. The blue markers are primary, the green markers are secondary or tertiary stars.

Open with DEXTER

It is well known that triple systems, in which the third body is closer than ~ 10 times the separation of the other two bodies, are inherently unstable (Reipurth et al. 2014). In UX Tau, the semi-major axes are 383 AU and 1541 AU, while in the other triple system, UZ Tau, the semi-major axes are 110 AU and 797 AU. These values would indicate an unstable configuration in both cases. Moreover, we found that UX Tau AB and UZ Tau AC are very likely gravitationally bound, but UX Tau AC and UZ Tau AB do not show orbital motion. Therefore it is possible that these systems are not physical triple systems, but only binary pairs coupled with common proper motion stars. However, since the fraction of the orbits covered by the observations so far is small, the determined semi-major axes may be significantly affected by the inclination of the orbits to our line of sight, thus we cannot confirm the classification of either UX Tau or UZ Tau as a stable or unstable triple system.

6. Discussion

We plot the derived stellar and binary parameters as a function of the IR excesses in Fig. 6. The star markers show stars around which the long-wavelength observations did not resolve a disc, the filled circles show the stars where these observations did resolve a disc, and the diamonds are the stars that do not have long-wavelength measurements. The long-wavelength observations probe the emission from the outer disc (Beuther et al. 2014, and references therein), while the IR excess at 10 μm shows the emission from the inner disc.

In Fig. 6 (left panel, IR excess – AV) there are stars in all quadrants in this projection of the parameter space, and we do not see a correlation. The middle and right panels in Fig. 6 (IR excess – Teff and IR excess – semi-major axis) also shows no correlations, which also applies to the mass and luminosity as a function of IR excess.

In the IR excess plots we can see that all but one star with discs detected at long wavelengths have 10 μm IR excess over 3.9 mag. From the opposite direction, we can also see that all but one star with IR excesses over 3.9 mag have a disc detected at longer wavelengths. In the first case the exception is HV Tau C, which is a star with an edge-on disc, and its T Tauri classification falls between Class I and II because of its flat SED around 10 μm. Its IR excess is only moderate although it definitely harbours a disc, but the edge-on disc may decrease the detectable IR excess. In the second case the exception is GH Tau A, which has an IR excess of 4.5 mag, but we note that the uncertainties of its 10 μm measurements are high, therefore it may be labelled as an outlier. We also note that the intrinsic variability of T Tauri stars can be significant, and the outliers might be attributed to the fact that many of the photometric measurements were taken apart in time.

In a binary pair of young stars it is possible that one star has an influence on the disc of the other star, and such influence may be visible in the emission originating from the disc. Therefore we examined the distribution of IR excess in single T Tauri stars to determine whether we can see any difference between the distribution of the IR excess of single stars and stars in multiple systems. We selected 40 single stars that have full UVBRI photometry, have no sign of multiplicity, and present a similar spectral type distribution as the binary sample. We selected the single stars from the Kenyon & Hartmann (1995), Sicilia-Aguilar et al. (2004) and Luhman et al. (2017) papers, among six other stars found in SIMBAD by querying the Taurus-Auriga region for T Tauri objects of specific spectral types (listed in Table B.1). To ensure that the selected stars are single, we checked the stars in SIMBAD, Vizier, and the Washington Double Star Catalog (WDS), looking for any sign of multiplicity, such as dual SED curves, dual spectral types, notes on duality, or components present in the WDS with a separation smaller than 5.8′′, which is the upper limit of separations in our sample. The resulting sample was analysed in a way similar to our multiple system: we added 10 μm photometry data from the WISE survey, fitted the SED curves, and derived the IR excess.

The 10 μm IR excess is measured in the Rayleigh-Jeans region of the SED, which is not as dependent on the effective temperature as the K-band magnitudes, for instance, but it is still affected by it.

The results of the comparison of the 10 μm excesses is plotted in Fig. 7; the IR excesses of the single stars are between 1.7 and 9.1 mag. The main difference between the two distributions is that the multiple sample has slightly more stars with high IR excesses, while the single sample has more stars with low IR excesses. We performed a Kolmogorov-Smirnov test to compare the two distributions of the IR excesses. The resulting DKS = 0.27 difference and pKS = 0.20 probability indicates that the null hypothesis (in which the underlying distributions would be identical) cannot be rejected at a value of p< 0.05 (and a confidence of > 0.95). Therefore, although one could expect that the presence of a companion could affect the emission coming from the inner disc, the apparent difference between the IR excess of the single and multiple objects may be due to statistical fluctuations.

7. Conclusions

We have carried out a survey of 18 multiple T Tauri systems with the goal to detect orbital motion, determine orbital parameters, and also study correlations between the binary configuration, the disc state, and stellar parameters. Our sample covers binary separations from 0.22′′ to 5.8′′ and spectral types K1 to M5 (corresponding to masses between 0.08 and 1.32 M).

We found that 10 pairs out of 20 are orbiting each other, 5 pairs may show orbital motion, and 5 are very likely common proper motion pairs. We found no obvious correlation between the stellar parameters and binary configuration. The 10 μm IR excess of the multiple systems varies between 0.1 and 7.2 mag, while it is between 1.7 and 9.1 in the sample of single stars. The distribution of the IR excesses in the two samples provides no statistical evidence for being different distributions, therefore the presence of the companion does not affect the emission coming from the inner disc.

We note that we did not detect any signs of circumbinary discs, which are usually also present in young multiple stellar system. However, they are more abundant around very close binaries with a separation of a few AU (Reipurth et al. 2014), which is not the case for our sample. Moreover, the circumbinary discs have to have larger inner holes because of the central stars (in our sample the holes would be larger than 30 AU), therefore their infrared excess can easily be below the sensitivity of our observations.

The obtained orbital periods vary from 138 year to over 10 000 yr. This suggests that even for the shortest orbital periods, the observation from the past ≈ 20 years may only cover ≈ 15% of the orbit, therefore obtaining precise orbital fits is not yet viable. We may be able to obtain a meaningful astrometric orbit by re-observing seven of the systems included in our study in 15 yr.

thumbnail Fig. 7

10 μm IR excesses by multiplicity. Blue: single stars, Red: multiple systems (individual flux measurements).

Open with DEXTER


6

The data sheet of the camera used in the AstraLux system is available at: http://www.andor.com/pdfs/specifications/Andor_iXon_897_Specifications.pdf

7

We used the SEDs available at http://www.am.ub.edu/~carrasco/models/NextGen2/, with the parameters M/H = 0.0, α/ Fe = 0.0 and log g = 4.0.

Acknowledgments

The authors would like to thank the SPHERE Science Verification team for their support in obtaining the observations. This work was supported by the Hungarian Scholarship Board Office. This work was supported by the Hungarian Academy of Sciences via the grant LP2012-31. This work was supported by the Momentum grant of the MTA CSFK Lendület Disk Research Group, and the Hungarian Research Fund OTKA grant K101393. This research has made use of the Washington Double Star Catalog maintained at the US Naval Observatory. Based on observations made at the La Silla Paranal Observatory under programme ID 60.A-9364(A), 70.C-0565(A), 70.C-0701(A), 72.C-0022(A), and 78.C-0386(A).

References

Appendix A: Description of the individual targets

LkCa 3:

This system was long known as a visual binary (since it was resolved by Woitas et al. 2001), but recent observations from Torres et al. (2013) have shown that it is a hierarchical quadruple system of M-type stars. With the assumption of coevality, Torres et al. found an age of 1.4 Myr and a distance of 133 pc for the system, which is consistent with previous estimates for the region of LkCa 3 and suggests that the system is on the near side of the Taurus complex.

DD Tau:

It was resolved as a binary by Bouvier et al. (1992), who have obtained high-resolution images of the T Tauri star DD Tau at optical and near-IR wavelengths, between 0.5 and 3.9 μm. The system was resolved into two intensity peaks, with a separation of 0.56 ± 0.01′′ and position angle (PA) of 184 ± 2°. The photometry of the two components suggested that DD Tau is a binary system consisting of two active T Tauri stars of similar luminosity. The authors also found evidence for an extended nebulosity, which may be a reflection nebula illuminated by the two stars. Gomez de Castro & Pudritz (1992) presented high-resolution and narrow-band images of DD Tau, obtained with the High-Resolution Camera of CFHT, in which they also resolved the forbidden line emission region. They found that the bulk of the continuum emission is concentrated in two knots that are separated by 0.55′′ and oriented in a PA of 6.4°. The forbidden-line morphology was found to be distinctly different from that of the continuum. The forbidden N ii emission appeared to be a “jetlike” extension connecting the two continuum knots, while the forbidden S ii emission seems to be confined to the northern knot. The southern knot (DD Tau B) is extended in the of PA ≈ 130°. The orientation of the disc as defined by IR polarization measurements is PA ≈ 125°.

LkCa 7:

The high-resolution survey by Leinert et al. (1993) showed that it is a binary system. The primary is known the be a weak-line T Tauri star, as observed by Grankin (1998), for example. It was part of a long-term variability survey (Grankin et al. 2008), where they measured the stellar rotation period.

FV Tau and FV Tau/c:

These two binary systems are separated by 12.29″ (Kraus & Hillenbrand 2009). They are usually included in the T Tauri binary surveys and are referred to as a wide binary pair, probably forming a quadruple system (see e.g. Ghez et al. 1993; Hartigan & Kenyon 2003). However, there is no evidence that they are gravitationally bound, thus we handle them as two separate systems. Since their separation is too large for most of the high spatial resolution instruments that we used, combining the separations from two different pointings would lead to higher astrometric uncertainties. Therefore our data do not allow us to test whether FV Tau and FV Tau/c indeed form a common proper motion quadruple system.

FV Tau:

This system was first resolved by Chen et al. (1990) using lunar occultation measurements, but Leinert et al. (1991) measured the binary parameters first. FV Tau (and FV Tau/c) was included in a recent survey conducted by Akeson & Jensen (2014), who used ALMA at two wavelengths (850 μm and 1.3 mm). They derived the stellar and disc masses of the primary, and (6.3 ± 1.4) × 10-4M, respectively.

FV Tau/c (also as HBC 387):

Simon et al. (1992) resolved this binary. As its name suggests, it is very close to FV Tau (12.3′′) and was also included in the survey by Akeson & Jensen (2014) (using ALMA at 850 μm and 1.3 mm). However, the measured separation does not agree with the previous observations (the discrepancy is ≈ 3σ in the separation, suggesting another body that emits the radio), therefore we did not include this epoch in our analysis.

UX Tau:

The two brightest components (A and B) in UX Tau have been resolved in 1944 by Joy & van Biesbroeck (1944). The third component (C) was seen by Herbig in 1975 (Jones & Herbig 1979), but Della Prugna et al. (1992) were the first to measure the actual separation. The fourth component (D) was discovered by Duchêne (1999) at CFHT, using adaptive optics.

Espaillat et al. (2007) have analysed the Spitzer IRS spectra for UX Tau A, and their SED fittings suggested the existence of a disc gap of ≈ 56 AU. A few years later, Espaillat et al. (2010) measured the gap to be 71 AU wide, using near-infrared spectral measurements. Tanii et al. (2012) also examined the disc around UX Tau A, using near-infrared observations from the Subaru telescope, who found a strongly polarized circumstellar disc surrounding UX Tau A and extending to 120 AU, at a spatial resolution of 0.1′′ (14 AU). The disc is inclined by 46° ± 2°, with the west side being the nearest. They did not detect the gap that was suggested by SED models at the limit of their inner working angle (23 AU) at the near-infrared wavelength.

UX Tau C was observed by White & Basri (2003) using the Keck I telescope to obtain high-resolution spectra. They re-determined the spectral type to be M5 and calculated the mass to be 0.166 ± 0.047 M. Andrews et al. (2011) found no evidence for remnant disc material and failed to detect 880 μm emission. This is in agreement with McCabe et al. (2006), who determined UX Tau A to be a classical T Tauri star, while UX Tau B and C are weak-line T Tauris.

FX Tau:

This system was also resolved by Leinert et al. (1993). Akeson & Jensen (2014) observed this system with ALMA, but did not detect the companion.

DK Tau:

The first observer of this binary system was Weintraub (1989), who used speckle imaging. Simon et al. (1992) also resolved the binary using occultation and optical measurements, but the position they reported deviates by more than 5σ from the other astrometric observations. However, their numbers strongly suggest a coordinate conversion error (a sign error in the right ascension H:M:S degree conversion), thus we revisited their data and recalculated the astrometric position. Recently, the system was resolved by Akeson & Jensen (2014) using ALMA, but since the separation does not agree with the positions therein, we recalculated the separation based the RA and Dec coordinates reported there.

XZ Tau:

Haas et al. (1990) resolved XZ Tau as a binary system using near-infrared speckle observations. Close et al. (1997) resolved the system and performed astrometric and photometric measurements. However, since their focus was HL Tau, they only provided coarse photometry for XZ Tau, which includes outflows around the system. Krist et al. (2008) have monitored the system and its bipolar outflow over ten years using the HST. They found traces of shocked emission as far as 20′′ south of the binary. VLA observations carried out by Carrasco-González et al. (2009) resolved the southern component into a double system with a separation of 90 mas, making the XZ Tau system a triple. The third component is likely deeply embedded and not visible at optical and IR wavelengths. Forgan et al. (2014) revisited the possibility of a third component by observing the system again with VLA, but they found no trace of that third component. Dodin et al. (2016) recently observed the system using a 6 m class telescope with speckle imaging and also derived preliminary orbital parameters.

HK Tau:

It was first resolved by Moneti & Zinnecker (1991) using classical infra-red imaging. It was also included in the ALMA survey by Akeson & Jensen (2014). It is known to be the first young binary system where an edge-on disc was found using HST (Stapelfeldt et al. 1998; McCabe et al. 2011). The secondary star is heavily extincted due to its disc, and, in optical and IR, only the scattered light is detectable from that companion. Later observations revealed that the primary also harbours a disc, making this system a young pre-main-sequence binary with two circumstellar discs (Harris et al. 2012).

V710 Tau:

This system was first resolved by Cohen & Kuhi (1979) using infra-red spectroscopy, and the binary parameters were determined later by Leinert et al. (1993) and Hartigan et al. (1994). It is a classic and weak-line T Tauri star pair, the primary is slightly more massive than the secondary (White & Ghez 2001; Jensen & Akeson 2003). Shukla et al. (2008) spatially resolved the system using Chandra and they found that both components emit X-rays. V710 Tau was observed by Akeson & Jensen (2014) as well using ALMA, but they could only detect the primary star.

UZ Tau:

The two brightest star of this multiple system have been noted by Joy & van Biesbroeck (1944), and the system was later resolved as a triple system by Simon et al. (1992). The UZ Tau W binary was extensively examined by Jensen et al. (1996), and in the same year, Mathieu et al. (1996) obtained spectroscopic measurements of UZ Tau E that showed it to be a spectroscopic binary, making the system quadruple. The UZ Tau E binary was closely examined by Jensen et al. (2007), who found that the brightness varies with a period of 19.16 ± 0.04 days, which is consistent with the previous spectroscopic binary period of 19.13 days. Their best orbital fit resulted in a separation of 0.124 ± 0.003 AU, which converts into a spatial separation smaller than 1 mas, making it an as yet unresolvable binary.

GH Tau:

The GH Tau binary was resolved first by Leinert et al. (1993). This binary was included in the surveys later by McCabe et al. (2006) and Hartigan & Kenyon (2003).

HN Tau:

HN Tau was resolved as a binary system by Moneti & Zinnecker (1991). It was also observed in the ALMA survey by Akeson & Jensen (2014), but they did not detect the secondary.

HV Tau:

This is a triple system, resolved first by Simon et al. (1992). HV Tau A-B is a close binary (74 mas, measured in 1996) with similar brightnesses, therefore in many surveys only the HV Tau AB-C pair is resolved. The HV Tau A-B pair was recently re-observed by Kraus et al. (2011) using the Keck telescope, and they reported 36.0 ± 0.2 mas for the separation and 326.6 ± 0.3 degree for the position angle. The system was closely examined by Duchêne et al. (2010) using NACO at the VLT. They found that HV Tau AB-C is a common proper motion pair and that the orbital motion within the close HV Tau AB system is slow, suggesting a highly eccentric orbit or a large de-projected physical separation. Previous spectroscopic and photometric measurements also showed that the AB subsystem does not experience accretion, nor does it show infra-red excess (White & Ghez 2001). HV Tau C star has an almost edge-on disc, which has a mass in the range of Mdisc ~ 10-3M, and Rout = 50 AU size (Duchêne et al. 2010).

The T Tauri classification of HV Tau C is ambiguous, it is usually classified as “I?” because the SED is rising in the NIR regime, but it levels out after ≈ 20 μm. Owing to its almost edge-on and optically thick disc that totally block out the star as a point source, Stapelfeldt et al. (2003) suggested that an extinction magnitude of AV> 50 should be adopted for this system. The edge-on disc also prevented us from precisely locating the position of the star, which can be seen in the positional plot in Fig. 3, where the astrometric epochs overlap each other at the assumed position of the star. This makes the possible orbital fit difficult, and we would either need to measure the exact position of the star more accurately or wait for the companion to orbit around the primary for a few more decades before deriving precise orbital parameters for this system.

V955 Tau, V999 Tau, V1000 Tau:

These stars are also known as LkHα 332, LkHα 332 G2, LkHα 332 G1, respectively. They are three close binary pairs that are ≈ 11′′ and ≈ 26′′ away from each other, probably forming a wide triple system (Kraus & Hillenbrand 2009). All three stars were resolved to be binaries by Woitas et al. (2001) and V1000 Tau was resolved slightly earlier by Ghez et al. (1995). Hartigan & Kenyon (2003) have determined V955 Tau to be a classical T Tauri pair, while V999 Tau is a weak-line T Tauri pair.

RW Aur:

This star was resolved by Joy & van Biesbroeck (1944) as a binary, and it is also shown as a binary pair in the third Herbig-Bell catalogue (Herbig & Bell 1988). Cabrit et al. (2006) resolved an optically thick disc around the primary that has a radius of 40–57 AU. The primary also features one of the highest known accretion rates for a T Tauri star (Hartigan et al. 1995). The primary star features an optically visible asymmetric bipolar outflow and was recently resolved by Skinner & Güdel (2014) using Chandra. They found that both components are visible in X-rays, and the luminosity of the less-massive secondary is at least twice that of the primary and is variable. This binary also features a peculiar dynamical configuration: the disc around RW Aur A is counter-rotating with respect to the binary orbital motion (Bisikalo et al. 2012). It is a significantly variable system. Rodriguez et al. (2013) observed a dimming that had a depth of ≈ 2 mag and a duration of ≈ 180 days between 2010 September and 2011 March. They speculated that the dimming may be attributed to a one-time occultation by the tidally disrupted disc around the primary.

Antipin et al. (2015) obtained resolved UBVRI photometry of the system in 2015. However, they found that the primary has suffered a 3 mag grey extinction, while the secondary became 0.7 mag brighter in all bands, compared to the measurements of White & Ghez (2001). Therefore, since this is a high difference in magnitudes, we refrained from including this photometry data in our analysis. The variability was also observed in the IR regime: Shenavrin et al. (2015) noted ≈ 1.5 mag variability in the J band over four years.

RW Aur B had two astrometric epochs that deviate significantly from the hypothetical orbit and the other data points. One epoch is a Keck observation in a narrow-band N-band

observation (1999-11-16, IHW18 filter), which may have captured some other feature, such as scattered light from the jets, disc, or nebula around RW Aur. The other deviating epoch comes from the survey of Leinert et al. (1993).

Appendix B: Single star sample

In the following we list the single T Tauri stars that are used in Sect. 6 to compare the distribution of the 10 μm IR excess of single stars and stars in multiple systems.

Table B.1

Single star sample and their 10 μm IR excesses.

Appendix C: Additional table

Table C.1

Astrometric epochs of each pair in our sample.

All Tables

Table 1

Spectral types, extinction magnitudes, classification, and naming scheme notes of the stars.

Table 2

Parameters of the fitted orbits.

Table 3

SED fitting parameters.

Table 4

Proper motions, average movements per year, reduced χ2 of the orbital fits, and classification of the binary pairs.

Table B.1

Single star sample and their 10 μm IR excesses.

Table C.1

Astrometric epochs of each pair in our sample.

All Figures

thumbnail Fig. 1

AstraLux images from the sample showing a pair of similar brightness, a pair of a bright and a faint component, a wide and a tight pair, and the two triple systems. The halos due to the lucky imaging addition are visible around the clearly resolved central cores of the stars. The intensity scaling is logarithmic.

Open with DEXTER
In the text
thumbnail Fig. 2

SED plots. The triangles show upper limits, the filled markers designate the combined flux from the systems, and the empty markers show the measurements of the individual components. The dotted curves are fitted SED models.

Open with DEXTER
In the text
thumbnail Fig. 3

Motion of the companions. The coordinates are relative to the primary star, which is at (0, 0) in the plots (it is marked with a black star symbol when the (0, 0) coordinate is in the plotted area). The blue markers show the oldest and the brown markers the latest epochs. The maroon curve shows the fitted orbit with e = 0.

Open with DEXTER
In the text
thumbnail Fig. 4

Motion of the companions, plotted separately for the RA and Dec dimensions, time dependent. The red line is the average separation, i.e. if the companion is gravitationally bound and has no observable movement, it would line up on the red line. The blue line is the proper motion with the parallax added, the dashed blue line shows the uncertainty of the proper motion.

Open with DEXTER
In the text
thumbnail Fig. 5

Motion of the companions, plotted separately for the RA and Dec dimensions, time dependent. The red line is the orbit from the orbital fit. The black markers are literature data points, the blue markers are data points from our observations.

Open with DEXTER
In the text
thumbnail Fig. 6

Scatter correlation plots. The filled circles show stars with discs observed at long wavelengths, the star markers stand for the stars without a disc detected at long wavelengths (however, we stress that a disc could still be present at detection levels lower than what was used by Harris et al. 2012 or Cabrit et al. 2006) and the diamonds are the stars where we have no high spatial resolution long-wavelength observation of the systems. The blue markers are primary, the green markers are secondary or tertiary stars.

Open with DEXTER
In the text
thumbnail Fig. 7

10 μm IR excesses by multiplicity. Blue: single stars, Red: multiple systems (individual flux measurements).

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.