4. Binarity of YSOs in the Hipparcos catalogue
A few known young binary systems (FK Ser, etc.) were included in the Hipparcos observation program, but most of the discoveries of close T Tauri systems result from recent progress in high-resolution interferometric and imaging techniques, after the Hipparcos launch. This fast moving research field opens the exciting prospect of understanding the molecular core fragmentation process and the relationship between the formation of binaries and disk (cf. Mathieu 1994).
The theoretical angular resolution of Hipparcos is 0.45 arcsec, but double stars with a moderate brightness ratio () separated by 0.10-0.13 arcsec already produce a measurable broadening of the diffraction peak (Lindegren 1997). Unresolved systems with separation below this limit (and below about 0.3 arcsec for larger brightness ratios) were seen by Hipparcos as single point sources located at the photocentre of the system. Both orbital motion and light variability of at least one of the system's components can lead to deviations of the photocentre's path from its expected uniform motion, giving rise to the suspicion that the star is an unresolved binary. As explained above, if an astrometric solution for a binary system could be derived by the Hipparcos data reduction teams, it is indicated in the Hipparcos Catalogue (Field H59, cf. Column 10 of Table 1). Table 1 shows that Hipparcos observations resulted in several detections of astrometric binaries in the small YSO sample that was observed.
We went back to the intermediate astrometric data of these objects and of other young stars for which the astrometric solutions derived by the Hipparcos teams did not take into account newly discovered information, notably as far as binarity is concerned. We tried on a star-by-star basis to explore alternative astrometric solutions, and we discuss below those systems where an improvement has been obtained, in order of increasing for each type of double star distinguished in Hipparcos astrometric solutions.
4.1. Variability-induced movers
Since TTSs are variable stars, it is not surprising to find that several of them are called Variability-Induced Movers (VIMs, Field H59=V), for which Wielen (1996) derived astrometric solutions based on the assumption that one star in the binary system is variable, while the other one is constant. Since the photocentre varied back and forth between the two components, both the position angle (with its standard deviation ) and a lower limit of the separation can be found in this way. The results for the 8 VIMs among our TTS sample are given in Table 7, and we discuss each of these objects briefly in turn.
4.1.1. V773 Tau
This is a triple system made up of a double line spectroscopic binary star (Welty 1995) and a third component which has been resolved at several optical and IR wavelengths (cf. Ghez et al. 1997a and references therein). The position angle of V773 Tau C ranges from in 1990 to in 1994, and the separation is about 115 mas until 1993, then decreasing to about 60 mas from 1993 to 1995. This is in good agreement with the VIM solution given in Table 4. Note that the position angle refers to the constant component of the binary (as assumed in the VIM formalism) with respect to the variable component, while it refers to the primary component in Ghez et al. (1997a). The phase shift of between the two position angle values thus confirms that V773 Tau C is more variable than V773 Tau AB, as observed by Ghez et al. (1997a). It is therefore likely that the two flares recorded by Hipparcos in its photometric database originate from V773 Tau C.
There is an unmodelled scatter in the astrometric data residuals of the Hipparcos solution, which could be due to the motion of the photocentre of the two-lined spectroscopic binary. However, if one uses the luminosities of the components from Ghez et al. (1997a) and the masses () given by Welty (1995), the semi-major axis for the motion of the photocentre is % of the semi-major axis of the relative orbit, i.e., around 0.15 mas. This is too small for detection by Hipparcos, and even if the secondary had a much smaller luminosity, the effect would be negligible compared to the VIM effect ( mas variation of the photocentre between the minimum and maximum luminosity of the system).
Recently, Lestrade et al. (1999) determined the astrometric parameters of V773 Tau using high precision VLBI astrometry, and found a parallax of mas. The derived parallax and proper motion agree within with Hipparcos results. The fact that the position angle of V773 Tau C changed during the mission has an influence on the astrometric parameters: constraining the position angle to vary linearly with time during the 3-year mission with gives a parallax of mas, closer to the VLBI value but with a large uncertainty.
4.1.2. RY Tau
This star has long been suspected of being a binary (Herbig & Bell 1988) based on apparent changes in the stellar radial velocity, and Hipparcos confirmed this suspicion.
A new VIM solution was computed, using all available intermediate data, and slightly improving the published solution. The position angle is and the lower limit on the separation is 23.6 mas. This is compatible with the lack of detection of this system in current high angular resolution observations. The projected minimum distance between the two components is 3.27 AU.
4.1.3. DF Tau
The binarity of DF Tau was first detected in a lunar occultation (Chen et al. 1990), and was observed on several occasions since then. Available data are summarized by Ghez et al. (1997a), who find that the position angle decreases from in 1990 to in 1995, while the separation stays approximately constant over this time, at about 90 mas. The position angle from the Hipparcos VIM solution is consistent with these values and indicates that the primary is the variable component. This is also suggested by Ghez et al. (1997a), who note that both components display infrared excess but that the UV excess, which signals accretion activity, is much stronger in the primary than in the secondary. The VIM formalism (used to derive DF Tau's parallax in the Hipparcos Catalogue) assumes uniform orbital motion, which is obviously not the case here. We tried to improve the astrometric solution in two ways.
Obviously, a solution combining both orbital motion and VIM is necessary for this object, but this is impossible without more precise information, notably about the orbital parameters. None of the above derived parallaxes is significant, and the published parallax should obviously be considered with caution. Can we conclude that a location of DF Tau outside of the Taurus cloud is ruled out by our analysis? Not definitely. We have merely shown that the derived parallax in the Hipparcos Catalogue is probably not significant, and that its standard error is so large that DF Tau's weight in mean parallax derivations is very low. The only firm conclusion that we can draw is that, as a group including DF Tau, the TTSs associated with Taurus are indeed located at the cloud's distance; as for DF Tau, its location remains uncertain.
4.1.4. UX Tau A
Prior to Hipparcos launch, this object was a known triple system (1979). UX Tau A and B components are WTTSs, while UX Tau C is a low-mass object with H emission. The astrometric companion found by Hipparcos is most likely the B component. Table 8 summarizes the current position angles and separations of components B and C with respect to A.
4.1.5. UX Ori
Binarity of this star was first detected by Hipparcos. The small minimum separation found in this VIM solution may explain why this star has not been detected in IR with the shift-and-add technique (Pirzkal et al. 1997). It has been argued that the observed variability of this star and of BF Ori (also suspected of binarity by Hipparcos) is due to violent comet-like activity (Grinin et al. 1994, de Winter et al. 1999).
4.1.6. Z CMa
This star was already known as a binary with position angle around and separation 100 mas (Koresko et al. 1991, Leinert et al. 1997). The Hipparcos solution is in agreement with these solutions. In the optical range, the variable component is the primary.
4.1.7. IX Oph
The evolutionary status of this F star is not entirely clear, and it appears to have attracted little observational attention so far. The detection by Hipparcos of its binary nature is a new development. An improved VIM solution gives a position angle and minimum separation of 46.84 mas.
4.1.8. V1685 Cyg
Although the minimum separation, according to Hipparcos, is rather large, the binarity of this Herbig B2,3e star was detected neither in high angular resolution IR observations (Pirzkal et al. 1997), nor by speckle-interferometry (Leinert et al. 1997). However, the region around this star is a small stellar cluster with a molecular outflow oriented north-south (Palla et al. 1995), with V1318 Cyg and V1686 Cyg located south of BD +40 4124, which could explain why a VIM solution was found with a position angle .
4.2. Component solutions
4.2.1. XY Per
This binary has a arcsec separation, position angle with 0.88 magnitude difference. This is consistent with the results of Pirzkal et al. (1997), who find respectively 1.2 arcsec and . Hipparcos thus resolves the ambiguity noted in Pirzkal et al. (1997) and caused by the nearly equal brightness of the components.
4.2.2. NX Pup
This is a HAeBe star whose binary nature was discovered with the HST Fine Guidance Sensor giving arcsec, , with a magnitude difference (Bernacca et al. 1993). Hipparcos found consistent but less precise estimations, respectively arcsec, , and mag.
4.2.3. CV Cha
CV Cha and CW Cha are the two components of an optical binary T Tauri system with separation equal to 11.4 arcsec and p.a. (Reipurth & Zinnecker 1993). HIP 54744 is identified as CCDM J11125-7644A in SIMBAD, while HIP 54738 is identified as CCDM J11125-7644B. The component solution derived in the Hipparcos Catalogue gives a separation equal to 8.48 arcsec and a position angle of . The solution quality is given as poor (`C'), and the Hipparcos solution is not consistent with the images obtained by Reipurth & Zinnecker (1993). However, we note that the Hipparcos-derived position angle would be consistent with observations, if HIP 54738 were in fact CV Cha and if HIP 54744 were CW Cha. Given the weakness of CW Cha, the separation derived by Hipparcos is likely to be inaccurate. As discussed also in Sect. 2, we conclude that there is a misidentification in the Hipparcos Catalogue, and we believe that HIP 54738 = CCDM J11125 -7644A = CV Cha, while HIP 54744 = CCDM J11125 -7644B = CW Cha. A single star astrometric solution for CV Cha gives the following results: mas, mas/yr, and mas/yr (Falin, priv. comm.). Results given for HIP 54744 in the Hipparcos Catalogue should be discarded.
4.2.4. FK Ser
This visual binary was found by Herbig to be a possible post-T Tauri system (Herbig 1973). Hipparcos measured a separation of arcsec, and the position angle of the secondary is These values can be compared to those given by Herbig for the date 1972.5: separation of 1.32 arcsec, p.a. .
4.3. Acceleration solutions
An acceleration solution, using either a quadratic or cubic motion with respect to time, was applied to all stars not having a `component solution', and only the stars with significant non-linear terms were retained. The acceleration effect may be interpreted as the signature of binaries with an intermediate period (more than about 10 years).
4.3.1. CO Ori
This star has been detected as a binary by Reipurth & Zinnecker (1993), who mention a position angle and 2.0 arcsec separation. Given the distance of Orion and the very long period, it is unlikely that the acceleration term may be significant, so that the detected variation of the photocentre with time is probably an artifact due to the configuration of the system and the scanning law of the satellite.
If a stochastic solution (see below) is applied instead of an acceleration solution, the cosmic error is mas, i.e., with the same significance as a Gaussian level. From this cosmic error a magnitude difference mag is estimated, consistent with the 0.07 flux ratio in the Gunn z band. If a VIM solution is computed, the astrometric elements of the VIM motion are not significant at more than a level, but it should be noticed that the position angle found, , is also consistent with the Reipurth & Zinnecker 1993 observation.
4.3.2. AB Dor
The combination of Hipparcos and VLBI data allowed the determination of a dynamical mass of about 0.09 solar mass for the low-mass companion of this ZAMS star (Guirado et al. 1997). The Hipparcos data cover only a small fraction of the period, but the curvature of the photocentre motion was nevertheless clearly detected.
4.4. Stochastic solutions
These solutions were applied as a last resort during the Hipparcos data reduction, when all other solutions failed to give an adequate astrometric fit. A so-called cosmic error was added to the abscissae standard error, representing the unmodelled photocentre variations. Although the photocentre displacement may be due to short-period astrometric binaries (e.g. HIP 39903, Arenou 1998), it may also be caused by undetected long-period binaries, with separation of a few arcseconds. It may also be that a stochastic solution simply reflects bad abscissae measurements, without any binarity indication.
For resolved binary systems, there is a correlation - depending weakly on separation - between the magnitude difference of the two components and the cosmic error that would result if a stochastic solution was computed instead of a component solution (Arenou 1997). Using all Hipparcos component solutions with separation, e.g., between 1.3 and 1.5 arcsec, the magnitude difference can be calibrated against the cosmic error , leading to
This relationship allows us to estimate the magnitude difference between components when Hipparcos does not resolve a binary system.
4.4.1. RW Aur
No significant VIM solution can be found, but it is sufficient to reject the bad abscissae to obtain a better astrometric solution mas, and mas/y. This justifies the use of this star for computing a mean distance of the Taurus-Auriga region. Estimated magnitude difference between BC and A is mag, not far from the bolometric magnitude difference given by Ghez et al. (1997a).
4.4.2. DI Cha
Although the binarity of this object is not detected, e.g., in the infrared DENIS survey (Cambrésy et al. 1998), it is clear from Reipurth & Zinnecker (1993), Chelli et al. (1995), and Ghez et al. (1997b) that this is a binary of separation arcsec and position angle .
However, the binarity has probably not perturbed the Hipparcos astrometry too much. Indeed, if this star was reduced as a single star, its parallax would be mas, close to the stochastic solution, although with a 1.21 normalized .
Using the same method as above, the calibrated relation between cosmic error and magnitude difference is also valid for separation between 4 and 6 arcsec, so that the magnitude difference for DI Cha components is mag. Note that the primary is redder than the secondary with a difference of about 4 mag in the K band (Chelli et al. 1995).
4.4.3. R CrA
The Corona Australis molecular complex is at a distance of about 130 pc (Marraco & Rydgren 1981), which is corroborated by the parallax mas of HD 176386 (HIP 93425). R CrA is surrounded by several other YSOs (Wilking et al. 1997), which probably explains why the Hipparcos observations have been perturbed, leading to a stochastic solution. Although the error bar on the parallax prevents any safe distance to be derived, it is clear that the Hipparcos astrometric solution for this star should be completely discarded.
Once a VIM solution was attempted, the parallax shifted from to mas. Even constraining the parallax to the expected parallax of Corona Australis does not give a satisfactory solution (in the sense of obtaining significant values of astrometric or orbital parameters). We conclude that the astrometric intermediate data are clearly useless for this star.
4.5. Other astrometric solutions
4.5.1. V410 Tau
This is a triple system (Table 10, Ghez et al. 1997a), undetected by Hipparcos, apart from the `suspected non-single' flag in the Hipparcos Catalogue H61 field. The AB pair is not resolved by the HST either (Ghez et al. 1997a). There is no evidence in the Hipparcos astrometric intermediate data that a double solution can be obtained, and the published solution cannot be improved.
4.5.2. BP Tau
This is one of the few objects in Table 1 that Hipparcos did not flag as a suspected binary star. That it is single down to 0.01 arcsec is further confirmed by HST observations (Bernacca et al. 1995). Closer binarity would obviously not be detected by Hipparcos, and we can rule out variability-induced motions of the photocentre to explain the parallax found in the standard data reduction. An undetected orbital motion cannot be an explanation either, although a one-year orbital period would obviously result in a confusion between the orbital and the parallactic motion. We checked that this assumption would imply unreasonably large masses (on the order of ) for the two components.
Because the star is apparently single, we cannot dismiss BP Tau's apparent parallax easily. We do not believe, however, that it should be taken at face value for the following reason. The solution published in the Hipparcos Catalogue represents the best astrometric fit with normalized equal to 1.1. If we now compute a solution where we constrain the parallax to be that of the Taurus stars, we get a fit with normalized equal to 1.2. In other words, the solution is only marginally worse than the published solution, and a true parallax at from the published parallax is not unlikely. Also, one should note that the star, although it would be located in front of the Taurus group if the Hipparcos parallax were correct, has the same proper motion as confirmed members of the Taurus SFR. This casts additional doubt on the published parallax value.
Assuming for the moment that the Hipparcos parallax is correct, is it plausible that the current location of BP Tau could be explained by its relatively large heliocentric radial velocity, km s-1 (Barbier-Brossat & Figon 1999)? This velocity translates to a LSR velocity of about +5 km s-1. Given the LSR radial velocity of the local molecular cloud of +7.1 km s-1 (Herbig 1977), the radial velocity of BP Tau with respect to the cloud is about -2 km s-1. The radial displacement of the star in its estimated lifetime of 1 Myr (Siess et al. 1999) is thus about 2 pc. This is obviously not consistent with location of BP Tau in the molecular cloud at birth.
There is no obvious reason to dismiss the Hipparcos parallax for
this star, but conversely, there is no strong reason to believe it
either; the large statistical error on the result precludes a firm
conclusion. The reason for this large error is not obvious either.
Most likely, the culprits are the faintness of BP Tau and its
intrinsic variability. As in the case of DF Tau, the conclusion is
somewhat frustrating, as no clear-cut answer can be given to the
question of these stars' distance. A major conclusion that was drawn,
however, is worth being repeated here: whereas distances of individual
TTSs must be viewed with caution, the distance of Taurus TTSs as a
group is entirely consistent with the molecular cloud
4.5.3. GW Ori
One of the brightest CTTSs, GW Ori was found to be a single-lined spectroscopic binary by Mathieu et al. (1991), who give its orbital parameters. Because of the large distance of this star, the astrometric perturbation due to orbital motion (reflex motion mas) is too small for Hipparcos to detect it.
4.5.4. AK Sco
This is a SB2 system with well-defined orbital parameters (cf. Andersen et al. 1989). Unfortunately, the two components have nearly equal masses and luminosities, so that the photocentre orbital motion is not significant, precluding a computation of the other orbital elements from Hipparcos data.
Astrometry turned out to be a powerful tool for detecting the binarity of variable stars using the variability-induced motion of the photocentre (Wielen 1996). In spite of the limitations of the method, which must assume that only one component is variable and neglects all other causes of photocentre motions, the binary parameters derived for VIM systems are in remarkable agreement with other observations. VIM detection of the RY Tau binarity is a long awaited, but somewhat unexpected result of the Hipparcos mission.
© European Southern Observatory (ESO) 1999
Online publication: December 2, 1999