The data of the 1991 outburst are different from previous VLBI observations of Cygnus X-3 in two ways: First, the total intensities at radio wavelengths are dominated by a superposition or blending of multiple flares, preventing a unique identification of single events. Second, the first high sensitivity VLBI data on this source show that considerable structure is contained in the low amplitude part of the visibility curves.
The investigation of the October 1985 flare (Schalinski et al. 1995) has shown evidence for "jet-like" radio emission on milliarcsecond scales, confined to a compact core and extended emission (30 mas) along a position angle of about . The analysis presented here contains evidence for the structure of Cygnus X-3 during the 1991 flare to be much more complex. Although a five component model is the least complex representation of the data, the real source structure is likely to be more complicated. This also applies to the analysis of the other epochs, especially to the single baseline observations of January 24 and 27. These modelfits, however, are all non-unique solutions, and not considered as reliable representations of the structure. On the other hand the well-defined minima of the visibility amplitudes, although low, allow the p.a. of the structure of the January 25 data to be constrained in the range of . The data show that the emission is confined to jets rather than a simple spherical expansion. A multiple component representation is best explained as a continuous outflow from Cygnus X-3. On the basis of these data it is however not possible to decide, whether this outflow is two sided or alternating.
Previous observations during outburst (e.g. Geldzahler et al. 1983; Spencer et al. 1986; Schalinski et al. 1995) and quiescent stages (Molnar et al. 1988) have shown evidence for outflow of the source with a velocity of 0.3c. Again the complexity of these data along with the sparse uv-coverage does not allow an unambiguous kinematic analysis. An estimate of the duration of the large flare, assuming that the lightcurve in later stages is a superposition of multiple outbursts, may be determined from inspection of Fig. 1 to be on the order of three days. This is similar to the timescales involved in previous flares (e.g. Johnston et al. 1986; Schalinski et al. 1995), and let us thus speculate that the kinematics involved may be similar. The five component modelfit of Table 3 is consistent with a total source size of at least 60 mas, and puts a lower limit to an expansion rate of 0.25c. The flux density constraints from the lightcurves (Fig. 1), requiring almost equal flux density contained in extended structure and "core", are also roughly consistent with the above model fit, showing about 3 Jy in a central region of about 30 mas, and about 2 Jy at core separations of 60 mas.
The low visibility amplitudes indicate sizes of components exceeding the beamsize of the array of about 5 mas. The modelfits show minimum sizes of about 16 mas. This is consistent with the compact components being dominated by image broadening most likely due to electron inhomogeneities along the line of sight towards the source. The Cygnus region has been shown to be subject to enhanced interstellar scattering on the basis of pulsar timing, interplanetary scintillation and VLBI measurements (e.g. Cordes et al. 1984; Fey et al. 1989). Schalinski et al. (1995) on the basis of two VLBI observations at 6 cm during the October 1985 flare derive compact component sizes of 16 mas. They also find a wavelength dependence of -2.02 on frequency consistent with refractive interstellar scattering with powerlaw index . This is slightly in excess of the Kolmogorov value for neutral turbulence, suggesting lengthscales of cm. Image broadening is most likely responsible for the strong resolution effects visible in the 1991 data. To obtain structural information on smaller scales and determine individual outbursts short wavelength VLBI observations with high sensitivity are required.
The September 1972 radio outburst (Gregory et al. 1972) has been modeled as particle injection into twin jets (Marti et al. 1992), extending the spherical solution of Marscher & Brown (1975) to confined jets. The data presented here give supporting evidence for jet emission, with emphasis on a continuous outflow. It is interesting to relate these radio flaring events to the X-Ray emission of the source.
Smale et al. (1993) report observations with the Broad Band X- Ray Telescope onboard the Space Shuttle Columbia showing Cygnus X-3 in an "ultrahigh", soft X-Ray state on December 5, 1990, prior to the large radio flare, with a luminosity of 2 x ergs (1-10 keV, at an assumed distance of 10 kpc). More general, from quasi-simultaneous radio and X-Ray measurements during a 4.5-year interval with Ginga and the GBI, Watanabe et al. (1994) find that during seven radio flaring events with flux densities exceeding 5 Jy Cygnus X-3 displayed X-ray flux densities corresponding to a high state with a luminosity of about ergs sec-1, but no correlation with X-Ray variability.
They argue that, although a high accretion rate, as suggested by the high X-Ray state, might cause jets, the absence of simultaneous X-Ray outbursts requires additional physical processes to be relevant for the radio flares, such as large mass transfer from the companion star, and a sudden relaxation of magnetic field wound up by accreting matter. As can be seen from Fig. 4 in Watanabe et al. (1994) compared to other epochs the softness ratio displays a steep gradient prior to the large flaring event of January 1991. A confirmation of a possible correlation of X-Ray softness ratio and radio flaring events from simultaneous X-Ray and VLB-interferometric monitoring during outburst should help to understand the underlying mechanisms in detail.
In order to obtain a reasonable analysis of flaring events of Cygnus X-3 the individual flares must be identified. This can be accomplished by observing the flaring event as soon to initial onset as possible at 1.3 cm wavelength using an array with a minimum resolution of 1 mas which is sensitive to structure on size scales 1 - 20 mas. Scattering at 1.3 cm is expected to be 0.5 mas. Continuous monitoring at 1.3 cm with this resolution could identify each individual flare and confirm that the Cygnus X-3 jet emission is ballistic in nature. Further with instruments such as the VLBA, the evolution of the flaring event and emergence of a jet could be confirmed by studies at 3.6 and 6 cm wavelength (with sensitivity to spatial structure on scales 5 - 60 mas), as the 1.3 cm emission will probably decay below detectable limits.
This study would require dedicated use of the VLBA and EVN networks for a continuous period of thirty to forty days. At the present time this is probably impossible due to the scheduling of these networks for other programs, such as the study of jets associated with extragalactic sources. However Cygnus X-3 allows us to determine the evolution of jet-like structure from the beginning to its disappearance, i.e. "birth to death". Such an opportunity with its associated contribution to understanding jet like radio emission is certainly worth the observing time when we consider all the observations of extragalactic sources displaying jets so the morphology of these sources may be understood.
© European Southern Observatory (ESO) 1998
Online publication: December 8, 1997