3. Results and Discussion
For the 9 sources observed in this experiment the projected baseline ranged from 600-800 M (at 215 GHz), which is equivalent to fringe spacings of 0.26 - 0.34 mas. In Table 2 we summarize the signal-to-noise ratios and the flux densities. Column 1 & 2 of Table 2 give the source names; column 3 the redshift; column 4 & 5 the total flux density at 215 and 86 GHz, respectively; column 6 the number of VLBI scans with significant detections; column 7 & 8 the SNR as obtained after integration over the full scan from standard fringe search (FRINGE) (col. 7) and from incoherent fringe search (SEARCH) (col. 8).
The difference of the SNR -values in columns 7 & 8 is the following: for the standard fringe search (FRINGE) without search windows the typical detection threshold is which, however, can be reduced to by setting smaller search windows. For the incoherent fringe search (SEARCH) the SNR depends on the number of coherent segments and the search area of rate and multi-band delay. In this experiment we averaged for each 6.5 min scan over about segments and searched over a rate-delay grid of 10 mHz µs. Under these conditions significant detections have a .
In column 9 we give for each source the correlated flux density at 215 GHz. For comparison we give also in col. 10 the correlated flux densities obtained at 86 GHz for 7 sources in another (longer) VLBI-observation performed on March 7-8, 1995 on the same baseline. The range of correlated flux densities does not only reflect the uncertainties of the amplitude calibration, which is of order of at 215 GHz ( at 86 GHz), but also includes variations most probably caused by source structure. We note, however, that because of the small number of VLBI scans and the remaining calibration uncertainties it is difficult to discriminate between systematic effects from the telescope (eg. focus, pointing, gaincurve), the atmosphere (eg. anomalous refraction, see Altenhoff et al. 1987), and intrinsic variations of the visibilities.
In the experiment we observed 9 sources and clearly detected 6 with signal-to-noise ratios of up to a factor of 4 higher than in the previous 1.4 mm experiment (Greve et al. 1995). For 3C 273 and 3C 279 we confirm the previous detections. In addition to the 6 clearly detected sources, we marginally detected the sources 3C 345 and SGR A*. The observations of SGR A* at 86 and 215 GHz will be discussed in a separate paper. The source 4C39.25 is not detected, while the source 2145+067 detected by Greve et al. (1995) was not observed again. The detection limit at 215 GHz is determined by the lowest correlated flux density in col.9 of Table 2, i.e. Jy, which is in good agreement with the theoretical expectation based on the antenna parameters of Table 1.
In the following we define the degree of compactness of a source by the ratio of correlated flux to total flux (). For the investigated sources we derive from Table 2 a compactness, which ranges between % at 215 GHz and % at 86 GHz. At 215 GHz, the most compact sources are 3C 279, 1334-127, and 1749+096 with C . Since most of the sources exhibit complex structures on sub-mas scales (eg. jets and multiple compact components), the variations of C can be partially attributed to variations of the visibility functions. However, it appears that on average the compactness of the sources at 215 GHz is lower than the compactness at 86 GHz and lower than that typically seen at cm-wavelengths (). In view of the overall calibration uncertainties and the limited uv-coverage (changes of the visibility function with hour angle could lead to underestimates of C in snap shot type observations) this result must be regarded with caution.
The snap shot type 1.4 mm VLBI observations of 3C 279 of December 1994 and March 1995 were performed at similar () interferometer hour angles (I.H.A.). This allows a direct comparison of the correlated flux densities, which show an increase by nearly a factor of 2 (at I.H.A.= 4.5 h), from 2.2 Jy to 3.8 Jy between both epochs. Such an increase is also seen in the compactness, which increased from C=0.21 to C=0.35, although the total 215 GHz flux density changed only from 10.5 Jy to 11.0 Jy. The 86 GHz VLBI map obtained in March 1995 showed a secondary jet component at mas core separation (Krichbaum, unpublished data). Although the observed variation of the visibility amplitudes and of the compactness are not completely outside the range set by the calibration uncertainties, motion of a jet component (at a speed of mas/yr typical for 3C 279 (eg. Carrara et al. 1993)) could easily explain the observed changes of the correlated flux densities.
A similar behaviour is also seen in 3C 273. In contrast to 3C 279, however, the compactness of 3C 273 decreased from C=0.14 in December 1994 to C=0.08 in March 1995, while the total flux density decreased from 13.5 Jy to 9.2 Jy. Within a mutual hour angle interval I.H.A.= 2 - 3 h, where data from both epochs are available, the correlated flux density decreased from 1.0 Jy in December 1994 to 0.4-0.7 Jy in March 1995. On sub-mas scales 3C 273 shows a prominent jet with components separating from the core at a typical speed of mas/yr (eg. Krichbaum et al. 1990). Again, variations of the correlated flux density and the compactness over timescales of a few months must be expected. In fact, recent 3 mm VLBI monitoring of 3C 273 shows the ejection of a new jet component between January 1994 and March 1995 (Krichbaum et al. 1996). It is worthwhile to note that the relative strength of the changes of the visibility amplitude and the compactness in 3C 279 seem to be larger than in 3C 273, although 3C 273 showed more pronounced variations in total flux density and exhibits a higher angular expansion rate. One possible reason for this may be a more complex sub-mas structure in 3C 273, which on mas-scales shows a more pronounced jet than 3C 279.
In the standard model of AGN, the radio emission originates from a continuous synchrotron self-absorbed relativistic jet and embedded compact structures, eg. shocks. Determination of the physical parameters of the innermost and most compact jet component (the jet base) yields some constraints for current jet models. If we assume that the observed compact emission results from a homogeneous synchrotron self-absorbed component with a maximum brightness temperature of limited by inverse Compton cooling, and that this component radiates predominantely near GHz, then a lower limit to the magnetic field strength is obtained: from we find , where D is the Doppler boosting factor. An upper limit for B follows from the observed high Gamma-ray luminosity of AGN (most of the objects of Table 2 are detected with EGRET), which indicates that the electron Lorentz-factor in blazar jets is considerably higher than previously assumed, eg. (eg. Maraschi et al. 1992). Maximum synchrotron radiation occurs near . This gives . With an average redshift and a typical Doppler-factor for the sources of Table 2, we thus obtain G.
The size of a homogeneous synchrotron self-absorbed component observed at 215 GHz is . Single component Gaussian modelfits to the visibility data of the detected sources (tab. 1) result in component flux densities of Jy and sizes in the range of as, corresponding to brightness temperatures of K. With respect to the relation above, these sizes must be regarded as upper limits to the true sizes. Correspondingly the brightness temperatures are minimal brightness temperatures. The fact, however, that only a fraction of the total flux is seen in our VLBI observations () indicates the existence of more complex and more extended emission, which must be partly resolved by the interferometer beam. For reasonable magnetic field strengths, the turnover frequency of the extended emission must be located at GHz (otherwise B gets too high). It is therefore likely that the brightness distribution of the objects is not pointlike but consists of several distinct components with sizes ranging from completely unresolved to largely resolved.
© European Southern Observatory (ESO) 1997
Online publication: June 5, 1998