Astron. Astrophys. 318, 472-484 (1997)

5. Results and discussion - individual sources

In this section we discuss individually the binaries known so far in our sample, with emphasis on the nature of the companions. In addition to the known binaries, we also consider here AB Aur. This star is a good example of an unresolved object, and it serves us to demonstrate typical upper limits for the near-infrared brightness of undetected companions. Finally we examine whether we can see a trend in the character of the companions with increasing luminosity of the Herbig Ae/Be star. The results on binary separation and brightness ratio are summarised in Table 4. The binary statistics of the sample are given in Sect. 6.

Table 3. Photometry of Elias 1 and its companion source

Table 4. Parameters of companions with projected separations 3600 AU

5.1. LkH  198

The Herbig Ae/Be stars LkH  198 and V376 Cas are the dominating visible stars in a small isolated molecular cloud at a distance of 600 pc (Chavarria 1985) in Cassiopeia. Both are known for circumstellar matter extended on the scale of 1000 AU which has been studied at visible and near infrared wavelengths (Piirola et al.  1992, Leinert et al.  1991). The geometry of the close environment of LkH  198 proved to be complicated and is not related to the conspicuous adjacent elliptical reflection nebula or to the elongated bipolar outflow (Levrault 1988) in an obvious way (Mundt and Ray 1994). In 1991 Lagage et al.  (1993) discovered a deeply embedded ( 35 mag) companion to LkH  198, north of the optical source. This new source, LkH  198 IR, is estimated to have a luminosity of 100 , not much less than the total luminosity of 160 estimated for the sytem (Chavarria 1985). On this basis LkH  198 IR appears to be the third Herbig Ae/Be star in this molecular cloud. It fits better into the outflow geometry than LkH  198 itself and therefore has been suspected to be the actual driving source. Contrary to the similar case of LkH  234 below, where the companion was only detected at 8.7 µm, LkH  198 IR was also seen in the near infrared from 1.2 µm to 2.2 µm (Li et al.  1994, Lenzen, private communication). The projected separation of 3300 AU is small enough for LkH  198 to be considered a binary within the definition of this paper. Because of the small size of the molecular cloud it is probable that the two sources lie physically close together. As with other similar cases, the actual physical association cannot be demonstrated but appears plausible.

5.2. Elias 1

Elias 1 has been taken as a B9 star with extinction =8.85 mag (Strom and Strom 1994) and alternatively as an A6 star with =3.9 mag (Zinnecker and Preibisch 1994). Its system luminosity is L 38 (Berrilli et al.  1992). The apparently stellar companion in the northeast is fainter in the near-infrared by 4-5 mag, but similar in the rising near-infrared spectral distribution (SED). Only at the longest measured wavelength (3.8µm, see Table 3), its SED, although still substantially high, is decreasing again. The SED of the system Elias 1, on the other hand, is rising with wavelength throughout the infrared range. Therefore it is improbable that the companion provides an important contribution to the far-infrared emission of the system. Measurements at additional wavelengths would help to determine the nature of the companion, which is not a priori guaranteed to be a young star. The available measurements at least allow tentative conclusions.

The near-infrared colours of Elias 1 NE, J-H 2.0, H-K 1.0, K- 0.7 are too extreme for being a cool foreground dwarf. We also discard the possibility that it is a Herbig-Haro object, as was found for an infrared source from HV Tau (Magazzu and Martin 1994). In this case we would not expect the still comparatively strong infrared emission at 3.8 µm and the substantial radio continuum flux at cm wavelengths.

A reddened background object is also a possibility to be considered: while a B star reddened with a standard extinction law would turn out to be too faint at J by half a magnitude, a M2III giant at 500 pc behind 10 mag of visual extinction would have just the measured near-infrared colour and brightness. However, the radio continuum flux of Elias 1 NE (0.2-0.6 mJy at 6 cm, Skinner et al.  1993) tends to be higher than measured by Drake and Linsky (1986) for the M giants µ Gem and  Per, which are at only 50 pc and 90 pc, respectively. The spectral index of Elias 1 NE between 3.6 cm and 6 cm has been measured to -0.75. This is also not expected for thermal emission in the wind of a red giant, and therefore we discard this possibility.

The spectral index of Elias 1 NE, however, would be typical of the radio emission of galaxies; and for a galaxy with a spectral energy distribution similar to a K giant (J-H 0.6, H-K 0.15, K- 0.1), the colours would turn out approximately correct for 12 mag of visual extinction. Given the adopted range of extinctions to Elias 1 of 4-9 mag, which may be lower bounds because of expected local scattering, an extinction of 12 mag could occur in this region. On the other hand, the observed variation in radio flux at 2 cm by at least a factor of 20 within 2 days is very unusual for an extragalactic source, and the dereddened K brightness of 9 mag comparatively high. Also the image of Elias 1 NE does not look extended in our near-infrared frames. None of these arguments is fully conclusive by itself. Taken together, they lead us to come back to the interpretation of Skinner et al.  (1993) that Elias 1 NE is a T Tauri star and probably a true companion.

Radio emission in T Tauri stars is not unusual. In classical T Tauri stars it is thought to arise by free-free emission in the optically thick part of a stellar wind, in weak-lined T Tauri stars it is attributed to gyrosynchrotron radiation in which coronal electrons are accelerated in strong, ordered magnetic fields (Chiang et al.  1996). Flare-like variations of radio emission on time scales of  1 day, which can exceed the levels observed in Elias 1 NE, have been observed (Feigelson and Montmerle 1985).

The spectral index for this nonthermal emission is not bound to be positive, and the near-infrared colours of a weak-lined T Tauri star with spectral type M2 reddened by 8-12 mag of visual extinction are very close to those measured for Elias 1 NE. Five examples of M2 stars picked from the first part of the Herbig-Bell catalogue (1988) all agreed reasonably well, and the best fitting one, LkCa 5, is included in Table 3 for comparison. For these reasons we adopt Elias 1 NE as T Tau companion to Elias 1 of the weak-lined type. Based on the comparison to LkCa 5 its mass and luminosity can be roughly estimated to 0.4  and 0.3 . And the observed X ray luminosity of the Elias 1 system (1.9 0.8 10³⁰ erg/s, Zinnecker and Preibisch 1994) then could be fully or in part be explained by such a configuration (Wichmann et al.  1996).

In February 1990 the VLA observations of Skinner et al.  (1993) showed a second radio peak north of Elias 1 NE. We see no evidence for a companion to Elias 1 NE in our near-infrared data, neither in conventional imaging, nor by standard speckle reduction, nor by speckle holography with the primary of Elias 1 as reference source. Also speckle holography with Elias 1 NE as reference source showed no signs for a close companion or of significantly extended structure close to the primary of Elias 1. In the near-infrared we basically see a wide binary system.

5.3. HK Ori

The results are shown in Fig. 1 and summarised in Table 4. HK Ori, with spectral type A5 and = 1.2, has an observed spectral energy distribution (SED) rising from the optical region through the near-infrared to 20 µm. The strong wavelength dependence of the brightness ratio of its components in the near infrared (see Fig. 1) suggests that only one component contributes substantially to the emission longward of 5 µm. We tried to identify whether this emission is due to the main optical component or to an infrared companion. At 1.25 µm, the SEDs of components A (i.e. the brighter component at K) and B are almost crossing. The measurement at 917 nm is at the limits of the performance of our NICMOS camera for such a comparatively faint object (I=10.8). We still are able to determine the brightness ratio reasonably well to about 0.85. We also find that the northeast component is the fainter one at this wavelength, too, but this finding is less certain since the phases show distortion by noise. Therefore, in modelling the two components each by the simple star plus disk model mentioned above, we considered two extreme cases: a) component A is also the main optical component (as suggested by the turn down of component B between 1.25 µm and 0.9 µm in our measurements) and b) component A is an infrared companion and the optical radiation is dominated by component B. Both assumptions result in fits to the SEDs of similar quality longward of 1µm. Of course, shortward of 1µm a more realistic separation would also have to allow for a more even distribution of the optical radiation between the two components. But we think that such more definite modelling attempts should wait until the optical brightness ratio and the association of the main optical to one of the infrared components have been determined by optical high resolution observations.

Fig. 1a and b. Left: One-dimensional visibilities and phases for HK Ori. The observed values are shown with 1  error bars. The lines represent binary model fits. Note that at J the brightness ratio approaches 1, while at K it is about 0.1. The sign of the phase steps indicates that component B is to the northeast of component A. Right: Decomposition of the spectral energy distribution of HK Ori (filled circles) into spectral energy distributions for its components (A: squares, and B: triangles). The solid lines correspond to case a) in the text, where both the far-infrared emission and most of the optical radiation are assigned to component A.

In the more interesting case b), HK Ori has an infrared companion which dominates the luminosity. In this case, both the infrared companion with an inferred luminosity of 78 , and the main optial component (14 , = 8200 K) may qualify as intermediate mass stars. In case a), more probable according to our measurements and shown on right side of Fig. 1, 78 are due to the dominating component, and the luminosity of component B, fitted with = 3800 K, would be 5 , more typical of a T Tauri star. Optical resolution of HK Ori would allow to find a realistic description of the system between these two extreme cases.

5.4. T Ori

Shevchenko and Vitrichenko (1994) found T Ori to be an eclipsing and spectroscopic binary with a period of . Hillenbrand (1995) finds at 2.2µm a faint companion away. Because of the enhanced stellar density in the Orion area, it is not unlikely that this second companion looks close by projection only. With a system luminosity of 130  , the companion, if associated to T Ori, probably would have a luminosity of less than 2  , which is in the range typical of T Tauri stars.

5.5. V380 Ori

The results are shown in Figs. 2 and 3 and summarised in Table 4. V 380 Ori presents a similar case like HK Ori in the sense that the near-infrared brightness ratio of the components is also strongly wavelength dependent (see the Figures). There could be a crossover of the components' SEDs at 0.9µm. Component A (i.e. the brighter one at K) therefore again could be an infrared companion. In the case of V380 Ori we also have the same difficulty to determine which of the components is the brighter one at 917 nm. However, this uncertainty is less critical than in the case of HK Ori, because the brightness ratio at this wavelength is close to 1. Nevertheless it is not obvious how the component SEDs will continue into the optical. We again consider two extreme cases, where again each component is described by a simple star plus disk model: a) the optical main component is responsible for the infrared emission, too. b) most of the infrared emission is due to an infrared companion, while component B is identical to the main optical component. Fig. 3 shows that in both cases satisfactory fits to the data result, such that we cannot prefer one model over the other one on the basis of our measurements. Certainly the measurements at 917 nm favour a model between the two extremes, but again we think that such more detailed modelling should be done only after the object has been optically resolved.

Fig. 2. One-dimensional visibilities and phases for V380 Ori measured in north-south direction. Note the wavelength dependence of the brightness ratio as evident from the varying depth of the minima in visibility and from the varying size of the steps in the phase curves.

Fig. 3a and b. Two alternative model fits to the spectral energy distribution of V380 Ori (filled circles) and its components (A: squares, and B: triangles). For 917 nm, components A and B have the same brightness. This is the value with the highest uncertainty, and the 1  error bar is shown. Left: case a), with the primary dominating the companion at all wavelengths. Right: case b) with an optically dominating star and an infrared companion.

However, the estimates for the luminosities of the components do not dramatically depend on which association with the optical source is correct. Because of the decrease in the infrared flux of component B at 3.5 µm it is suggestive to assign the far infrared flux to component A, which then both in case a) and in case b) has a luminosity of the order of 170-180 , typical of an intermediate-mass star. For component B the model fits shown result in luminosities of 30 to 70 , the higher value resulting when the optical radiation is assigned to component B. Based on the luminosities, then both of the components of V380 Ori qualify as intermediate mass stars, and at least one of them shows a SED strongly dominated by thermal emission of circumstellar material.

We add a cautionary remark. In a similar case, LkH 198, where the 100 µm emission is known to be extended by , Butner and Natta (1995) have shown, that such extended far-infrared emission is to be attributed to the optical star rather than to the deeply embedded companion. The extreme infrared companion modeled in case B in the right part of Fig. 3 therefore may not correspond to a physically acceptable solution.

5.6. LkH  208

LkH  208 was found to be a comparatively close binary, with a projected separation of and a brightness ratio at K of 0.54 (see Fig. 4). This moderately reddened B7 star has a luminosity of 270 (Hillenbrand et al.  1992). Based on the flux partition of 1:2 at 2.2 µm, both primary and companion are expected to be Herbig Ae/Be stars. For further conclusions, measurements at other wavelengths are needed.

5.7. Z CMa

Z CMa with a system luminosity of 3000 (Hartmann et al.  1989) is a FU Orionis star and one of the best observed young star binary systems. Leinert and Haas (1987) could only partially resolve this system with projected separation of and preferred a halo as explanation for the structure of this system. Koresko et al.  (1991) achieved better spatial resolution and showed that Z CMa had an infrared companion. Haas et al. (1993) noted the substantial variability of both components. Whitney et al.  (1993) proposed a model in which the infrared companion actually is the primary of the system, heavily obscured and seen in scattered light only, at least at the shorter wavelengths. The finding that the infrared companion is relatively bright at optical wavelengths (Barth et al.  1994) agrees with this model. Spatially resolved polarisation observations would make the case still stronger. The X-ray emission of 1.4 0.7 10³¹ erg/s is attributed by Zinnecker and Preibisch (1994) to the optical component. Based on their luminosities, both components could be intermediate mass stars. - The claim of having seen a circumbinary disk at 3-5 µm (Malbet et al.  1993) has been questioned (Tessier et al.  1994).

5.8. HR 5999

This A7III-IV star with a luminosity of 130 has long been known to have a nearby companion, Rossiter 3930, fainter by about 4.5 mag in V (Jeffers et al.  1963). The system has been studied recently by Stecklum et al.  (1995). The relative near-infrared brightnesses of HR 5999 and Rossiter 3930 appear to have changed by 0.1-0.2 mag between our 1993 measurements and their measurements one year later. Both sources could be variable on this scale. Stecklum et al.  argue that Rossiter 3930 probably is a T Tauri star companion to HR 5999 and might be responsible for a considerable fraction of the X ray flux of 3.1 0.7 10³⁰ erg/s found by Zinnecker and Preibisch (1994) for this system.

5.9. KK Oph

This A6 star with a luminosity of 35 (Hillenbrand et al.  1992) has a spectral energy distribution typical of Hillenbrand's group I: with the photospheric decrease out to 1 µm, followed by a rising spectral energy distribution through the near-infrared wavelength range. There is a possible association with a 3.6 cm radio source nominally to the NE. The companion appears to have a similar spectral energy distribution as the main component: in the visible it was estimated to be fainter than the primary by about 1 mag (Herbig and Bell 1988), while we found it in the near infrared fainter by about a factor of 6. Since the apparent luminosity of KK Oph is dominated by emission from 2 µm to 10 µm, the luminosity of the companion may be in the range of 5-10  . As far as luminosity and spectral energy distribution are known at present, they would qualify the companion as a classical T Tauri star.

5.10. LkH  234

LkH  234, one of the brightest objects in the star forming region NGC 7129, is a B3 or B5e-B7e star with a luminosity of 2300  at 1000 pc (Hillenbrand et al.  1992). It is associated with a CO outflow (Edwards and Snell 1983) and an optical jet (Ray et al.  1990). Within of LkH  234 there are several faint infrared sources, many of which appear to be field stars (Li et al.  1994, Weintraub et al.  1994). The latter group of authors conclude from their 2 µm polarisation map that besides LkH  234 itself there must be a deeply embedded companion at the center of the observed elliptical polarisation pattern, approximately NW of LkH 234. They also proposed that this companion actually be the source driving the CO outflow around LkH  234. It coincides with the infrared companion IRS 1 found NW of LkH 234 by Cabrit et al.  (1994b) at 10 µm, and which has a steeply rising spectrum in this wavelength range. The detection of radio continuum only at the position of LkH  234 IRS 1 (Skinner et al.  1993) is also an indicator that this source is related to wind and outflow. With a projected separation of 2500-3000 AU between LkH  234 and the companion source IRS 1, they can be considered as a binary system within the definition of this paper. The luminosity is difficult to estimate from the presently available data. Cabrit et al.  (1994b) mention that it has about 1/4 of the flux of the infrared companion to LkH  198 at 10µm. Both infrared companions are considered to be sources driving outflows or jets. Since the polarization pattern around LkH  234 at 2µm is oriented towards the embedded source IRS 1, the intrinsic K band luminosity also should not be very low. Based on this scattered information we assume that LkH  IRS 1 is similar to the infrared companion of LkH  198 also in luminosity, and probably an intermediate mass star.

5.11. MWC 1080

This high luminosity object (spectral type B0, L 6500 ) has a strong, fast wind with velocities of up to 1100 km/s. It was found to be an eclipsing binary with a period of 2.89 days (Shevchenko et al.  1994). Zinnecker and Preibisch (1994) consider this close system as an example where the observed X ray luminosities of 9.4 4.9 10³¹ erg/s could be due to colliding winds from the components. Our 1D speckle observations showed an additional companion west (see Fig. 5). Based on the observed brightness ratios and published magnitudes, its broad band spectrum is increasing from 0.9 µm to 5 µm. This means that most of the observed luminosity of the companion is due to emission by circumstellar dust, as is the case for the main component. To obtain approximate information on the distribution of circumstellar dust around the components, we fitted the spectral energy distribution by the models of stars surrounded by geometrically thin disks mentioned above. We stress again that, since the actual spatial distribution of dust is not known, such a fit only provides a qualitative description of the the general distribution of the material with temperature (rather flat for these spectra, q 0.4-0.5). For the primary, Fig. 5 shows that our simple two-component model is not completely adequate, but certainly a large extent and large mass of the circumstellar material are needed to account for the long-wavelength measurements at 100µm to 1.3 mm (see also Hillenbrand et al. 1992). For the companion, we found the assumption most convincing that it is fainter than the primary at all wavelengths. The adopted fit predicts a luminosity for he companion of 250 L , which would classify it as a Herbig Ae/Be star, too.

Fig. 4. Visibility and phase for LkH  208. The figure shows the projection of the two-dimensional data onto the radius at PA . The fit curves similarly represent the projection of a fit to the two dimensional visibility and phases.

Fig. 5a and b. Left: One-dimensional visibilities and phases for MWC 1080. Right: Observed spectral energy distribution for the primary of MWC 1080 (dots) and its visual companion (triangles) compared to models comprising reddened stars and thermal radiation from thin disks with a power-law temperature profile. In each case, the thin lines show the direct stellar contribution (with = 5 mag and = 27000 K and 12000 K, respectively). The bold lines give the system flux predicted by the models. For the primary component, two models are shown, one emphasising the mid-infrared measurements (solid line) with disk mass, outer radius of the disk and system luminosity of 0.001 , 50 AU and , and 3500 , and one emphasising the far-infrared and millimeter fluxes (broken line, 1.5 , 330 AU and 6200 ). The corresponding parameters for the companion are 0.00001 , 20 AU and 240 .

5.12. AB Aurigae

The spectroscopically very well studied B9/A0 star AB Aur has a both a fast wind, as seen from its P Cyg profiles (see, e.g. Finkenzeller and Mundt 1984), a strong near-infrared excess similar to a classical T Tauri star and cool dust as detected in the IRAS bands and in the continuum at 1.3 mm (see Hillenbrand et al.  1992). The object is unresolved in our one-dimensional speckle observations at K (Leinert et al.  1994). The data lead to the following limits on the extent of the source or the presence of a companion: an upper limit for the FWHM of , or alternatively an upper limit of for the FWHM of a compact halo contributing 10% to the object brightness. The brightness contribution of a possible halo larger than 50 AU () is limited to a fraction of at most 8% of the system brightness, which corresponds to an upper limit on the mass of about 10^-5 under standard assumptions on dust scattering properties; but this estimated mass increases with the square of the halo radius (see Leinert et al.  1991). We do not expect to resolve the source of the near-infrared excess radiation which, be it hot dust (the colour temperature of the near-infrared excess is 1800 K) or free-free emission in the stellar wind, in any case is expected to be within 1 AU from the star.

AB Aur also was found to be an X-ray source with an X ray luminosity (0.1-2.4 keV) of 3.3 0.9 10²⁹ erg/s (Zinnecker and Preibisch 1994). A possible explanation of the X ray emission would be an unresolved T Tauri star companion. In this case our observations give an upper limit for the brightness contribution at 2.2 µm due to such a companion of 0.04 at separations , rising for smaller separations to 0.2 at (4 AU). The luminosity for AB Aur has been given as 80 (Hillenbrand et al.  1992). Our upper limits therefore do not exclude the existence of a moderately faint or moderately close T Tauri star companion but show that, if it exists, it would be difficult to detect.

5.13. Classification of the companions

In Table 5 we summarise the binaries in our sample by luminosity and try to assign a stellar type to the companions on the basis of the inferred luminosity: we assume that the companion should be a T Tauri star for luminosities less than 10 and a Herbig Ae/Be star for luminosities L 40 . There are no companions in Table 5 with estimated luminosities in the intermediate range.Admittedly, this assignment is a coarse approximation which also neglects the difference between stellar luminosity and accretion luminosity, which may be an important effect. All of the classifications given below therefore are somewhat uncertain, and those qualified by a question mark are uncertain indeed. On the other hand, if we take the six most luminous T Tauri stars in Taurus from the extensive list of Kenyon and Hartmann (1995), we have the triple star RW Aur ( 54 ), the class I sources L1551 IRS5 (22 ) and Haro 6-10 (7 ) and the prominent class II sources T Tau N (16 ), RY Tau (17 ) and HL Tau (7 ), where the classes are assigned according to Lada (1987). Our tentative classification given in Table 5 therefore appears acceptable, and it is a convenient and informative way to summarise the individual data on binarity.

Table 5. Proposed stellar type of the companions

It appears from Table 5 that the higher luminosity sources also have higher luminosity companions, and that this correlation looks more pronounced than expected e.g. for random pairing of the components from an initial stellar mass function. Part of this relation must be a selection effect, since it is difficult to detect faint companions close to bright sources. Insofar we are back to the point that we may have underestimated the multiplicity in our sample. On the other hand, this effect could also be a natural outcome of intermediate mass star formation. This question should be followed up in future observations of Herbig Ae/Be stars.

© European Southern Observatory (ESO) 1997

Online publication: July 8, 1998
helpdesk.link@springer.de