The basic observed structure of a compact radio jet consists of a "core" at one end with a series of knots extending more or less in the direction of the extended, kiloparsec-scale jet (although not all sources possess detectable large-scale jets). The knots tend to become more diffuse the farther they are from the core. In most sources, they separate from the core with apparent speeds that are superluminal. (The derivation of the velocity assumes that the distance to the quasar can be established using the Hubble Law.) In some sources, the knots trace out an essentially linear jet, whereas in other objects the jets are twisted. Typically, the compact jets either point directly toward the extended structure or curve toward that structure. Figure 1 shows an example of a straight compact jet, while the jet in Figure 2 exhibits pronounced twists.
Figure 1. VLBI image of the core-jet quasar NRAO 140 at a wavelength of 6 cm at two epochs. Contours are 2, 3, 5, 10, 20, 30, 40, 50, 60, 70, 80, and 90 percent of the 1984.40 peak of 3.2 x 1010 K (= 0.65 Jy beam-1). Motion of the knots relative to the core (at the right-hand end of the jet), corresponding to an apparent speed of ~ 6.5c (H0 = 100, q0 = 0.5), can be seen. From observations obtained by the author.
Compact extragalactic radio sources do not always follow the core-jet morphology (Pearson & Readhead 1988). Of particular interest are the compact steep-spectrum and gigahertz-peaked spectrum sources, which typically have either complex compact structure or appear to be miniature versions of extended double-lobed radio sources. Some or all of the more complex sources could be either disturbed jets (for example, those that interact strongly with an external medium) or jets observed directly down the axis such that the full complexity of the microstructure of the jet is apparent. In this review I will discuss the more well-behaved core-jet sources, since these are more readily compared with current theoretical models.
Figure 2. VLBI image at = 6 cm of the twisted compact jet of the quasar 1156+295, observed in November 1988 (McHardy et al. 1990). The contour levels are 0.3, 0.5, 1, 2, 3, 5, 10, 20, ..., 90% of the peak brightness of 1.2 Jy beam-1. An arcsecond-scale jet is also present, extending along position angle -19°.
The parsec-scale core (not to be confused with the "core" seen on VLA images, which includes the entire compact emission region) is usually unresolved by Earth-based VLBI at centimeter wavelengths. Its spectrum is flat or inverted such that the core emission tends to dominate over that of the other components at frequencies higher than 30-100 GHz in most sources. It is not clear whether the core itself is highly variable, since during the early stages of a flare a new knot, if formed, is too close to the core to be distinguished from it using Earth-based VLBI. Hence, the core could be rather stable with variations occurring mainly in newly created knots. Space-based VLBI might be able to resolve the issue.
The most promising technique for studying the structure of the compact core is VLBI imaging at high frequencies. Recent progress has been made at 44 GHz, at which relatively high dynamic-range maps have been made of strong sources (e.g., Krichbaum & Witzel 1992). These images show "more of the same" in the sense that the structures are still core-jets, although in 3C 84 the position angles of the knots are quite time variable, indicating some interesting complexity. At 100 GHz the structures are still core-jets, with the suggestive result that a strong knot in 3C 273 detected in 1988 appeared to be significantly elongated perpendicular to the jet axis, suggestive of a thin shock wave (Bååth et al. 1992). It will be of great interest to extend these high-frequency VLBI observations to 230 GHz, which lies above the turnover frequency of the core in some strong sources. The images should therefore reveal the true structure of the core region, unimpeded by optical depth effects.
The knots in core-jet sources tend to become progressively more extended as they move away from the core. Observationally, it is the knots that define the compact jet, although Reid et al. (1989) find that the compact jet in M 87 is continuous when observed at very high dynamic range. In the quasar 3C 345, Biretta, Moore, & Cohen (1986) measure an opening half-angle (as projected on the sky) of 13°, while in the quasar NRAO 140 (see Fig. 1), Marscher (1988) measures 6°. If the jets are pointing in our direction, the true opening angles are considerably smaller (see Section 3 below). The morphology of individual knots does not seem to follow any pattern, although the resolution and dynamic range of VLBI images are typically insufficient to explore the structure of each knot in detail. The few good spectral dissections made with multifrequency VLBI (e.g., Marscher 1988; Unwin et al. 1985) reveal that individual knots have spectra which roughly correspond to that expected from a uniform, self-absorbed synchrotron source. The turnover frequencies tend to be lower the farther the components are down the jet. In contrast, the core has a flatter spectrum below the turnover, with a power-law slope of 0.5 to 1.
Compact jets are overwhelmingly one-sided. For example, the jet:counterjet brightness ratio is observed to exceed 1600 in the case of 3C 345 (Rantakyrö et al. 1992). Under the standard relativistic jet model (see Section 4 below), this upper limit provides a lower limit to the Doppler beaming factor and hence to the Lorentz factor and orientation angle of the jet. A very small number of (weak) compact jets appear to be two-sided (e.g., Venturi et al. 1993), although one must take care to observe at a sufficiently high frequency that a downstream knot is not mistaken for the core and the core for a counterjet knot.
The knot:inter-knot ratio is not so easily determined, yet is an important parameter since it measures the enhancement to the emissivity caused by the knot, which is directly related to the physics of knot formation. In the 1984 image of the quasar NRAO 140 at = 6 cm shown in Figure 1, this ratio appears to be as large as 40 downstream of the brightest knot. The brightness of the ambient jet is, however, difficult to determine from this and most other existing VLBI images, since the ambient emission is more extended than the knots and hence can be resolved out by observations that do not include a sufficient number of short baselines. The VLBA will improve this situation significantly.
The flux density per unit length along the jet is observed to decline monotonically with distance from the core, save for an occasional abnormally bright downstream knot. In the cases of 3C 120 (Walker, Benson, & Unwin 1987a) and 3C 345 (Unwin & Wehrle 1992), the fall-off is described well by a power law of slope -1.3. In the case of 3C 120, the decrease in brightness follows this law over five orders of magnitude in distance from the core, and the spectral index of the jet emission remains roughly constant at a value = -0.65 (flux density F ).
The superluminal proper motions (measured relative to the core) of the knots are typically less than about 10c (H0 = 100 km s-1 Mpc-1 and q0 = 0 assumed) (Wehrle et al. 1992), although higher speeds are occasionally reported. Within a single source, the proper motions from one component to the next (referred to the same distance from the core) are similar in some documented cases (3C 273: Unwin 1990); 3C 120: Walker, Benson, & Unwin 1987b; NRAO 140: see Fig. 1), although in other cases (one component in 3C 120: Walker et al. 1987b; 3C 345: Biretta, Moore, & Cohen 1986, Tang et al. 1990) substantial differences have been measured. It is interesting in the two cases of 3C 120 and 3C 345 that several recently appearing components are measured to have similar velocities, whereas older components have significantly different speeds. The superluminal speed in a jet may therefore be stable over a number of years before changing. In the case of 3C 345, components very close to the core (as seen on high-frequency VLBI maps) are observed to have curved trajectories with accelerating proper motions (Biretta, Moore, & Cohen 1986). In addition, the knots do not seem to follow identical trajectories as they separate from the core (Zensus 1990). Farther than ~ 0.5 millarcseconds from the core, however, the speed of a given knot is usually constant within the observational error. In some sources, such as 4C 39.25 (Shaffer et al. 1987), one or more components are superluminal while others are stationary relative to the core. In other sources, extremely rapid changes in the structure indicate that the time sampling is too sparse to determine systematic trends (e.g., 2134+004; Pauliny-Toth et al. 1990). In the case of 3C 345, Bartel et al. (1986) have verified that the core is stationary and the knots moving. (Normally, VLBI observations do not measure absolute phases and therefore do not yield absolute positions on the sky.)
As mentioned above, there is substantial curvature observed both within the compact jets and between the parsec-scale and kiloparsec-scale jets. In the case of 3C 120, the jet is found to curve continuously between the two scales (Walker, Benson, & Unwin 1987a, b). In their survey of strong sources, Pearson & Readhead (1988) found that, at a moderate significance level, the difference between small- and large-scale position angle tends to cluster near 0° and 90°. Conway (1993) finds that this can be explained within the context of the relativistic jet model if the jet is presumed to execute a helical twist. The differential Doppler boosting caused by the changing orientation of the jet then causes the preference toward 90° if the jet lies nearly along the line of sight, while 0° is favored for other orientations or if the intrinsic curvature is low.
The polarization structure of compact jets has recently been explored by the group at Brandeis University (Roberts et al. 1990; Gabuzda et al. 1992). In general, they find that the cores tend to have very low polarization, perhaps indicating that differential Faraday rotation (Faraday depolarization) occurs. The knots, on the other hand, tend to be strongly polarized. Curiously, the polarization position angle tends to be nearly parallel to the jet in BL Lac objects (magnetic field direction perpendicular to the jet axis), and perpendicular in quasars. This strong dichotomy stands in contrast to the basic similarities in morphology of the compact jets in the two classes. On the other hand, the compact jets in BL Lac objects tend to be shorter (i.e., have steeper intensity gradients) and radio-bright BL Lac objects tend to have stronger X-ray emission than do quasars. The overall polarization characteristics of compact jets - substantial linear polarization and very low circular polarization - plus the tendency of observed brightness temperatures to obey the 1012 K inverse Compton limit, indicate that the emission process in compact jets is incoherent synchrotron radiation.
Although most VLBI observations have insufficient dynamic range to detect anything but the basic structure, some high dynamic-range images have shown considerable complexity, similar to that found on arcsecond scales. For example, the compact jet of M 87 displays filamentary structure (Biretta, these proceedings), while 3C 345 shows wiggles at the jet boundary, edge-brightening over a portion of the jet, and a sharp edge on the inner side of a broad component oriented perpendicular to the jet axis, suggestive of a reverse shock wave (Unwin & Wehrle 1992). These all correspond to features expected in fluid jets (e.g., Norman, these proceedings).