Astron. Astrophys. 333, 459-465 (1998)

3. The kinematical properties

Figs. 2 - 3 - 4 show the velocity fields observed for a total of 14 long slit spectra in the [OIII] and H wavelength ranges along nine different PAs. The velocities are given relatively to the systemic velocity and plotted as a function of distance to the nucleus for slits crossing the nucleus. For offset slits, the zero value in abscissa corresponds to the minimum distance to the nucleus. The velocities plotted in Figs. 2 - 3 - 4 reveal a rather complex velocity field which obviously cannot simply be due to rotation of a disc with a major axis (the photometric major axis, RC3). Notice that our data are in very good agreement with those of Heckman et al. (1981) along PA= and of Keel (1996) along PA= .

Notice also that the high and low excitation gases do not always have the same velocity structure.

3.1. Line profiles

The complexity of the velocity field is illustrated in two 2D plots of the high resolution spectra (Figs. 5 and 6). A blue asymmetry of the lines is detected in the central 3 arcsec (see Figs. 5 and 6), as already noted by Colina et al. (1987). On the west side, the velocity dispersion can be seen to be very large e.g. along PA= (see Fig. 5) giving a "mushroom" shaped spectrum with a typical FWHM of 300 km s^-1.

Fig. 5. Two-dimensional spectrum in the H -[NII] region along PA= , showing complex line structure. East is to the bottom and west to the top. The contours corresponding to the innermost regions have been superimposed.

Fig. 6. Two-dimensional spectrum in the H -[NII] region along PA= showing clear line splitting. Northeast is to the bottom and south-west to the top. The contours corresponding to the innermost regions have been superimposed.

Although in the [OIII] image almost no emission is seen to the north and east of the nucleus out to 5 arcsec, this is probably due to an oversubtraction of the continuum. However, weak emission is detected in our spectra (see for example Figs. 5 and 6).

The high spectral resolution H +[NII] data reveal the existence of double peaked profiles in the east quadrant, along PA= in the south, PA= in the northeast and PA= in the southeast. These regions extend over 10 arcsec. The most striking example is given Fig. 6 along PA= . In order to estimate the velocities of these two components, we performed a synthesis analysis using multiple gaussian fitting along this PA; the respective typical FWHMs of the two components are 90 and 250 km s^-1 on the northeast side, at a distance of about 5 arcsec from the nucleus. Both components (estimated at about 5 arcsec from the nucleus) are blueshifted by -135 km s^-1 and -10 km s^-1 with respect to the systemic velocity. Examples of one-dimensional spectra are shown in Fig. 7. Along PA= and at 5 arcsec from the nucleus, the two components are redshifted by 135 and 30 km s^-1. The fact that line splitting is observed only for the low excitation gas does not mean that the gas producing such features is less ionized, but may merely be due to the lower spectral resolution of our [OIII] data.

Fig. 7. Spectra in the H -[NII] region of several cross-sections along PA= showing complex line profile structure. The top spectrum is 2.3 arcsec northeast of the nucleus, and the following spectra from top to bottom are drawn every 1.8 arcsec towards the northeast.

Such line splitting probably represents radial motions (outflow or inflow) of the gas in the NLR, a rather common feature of extended high excitation gas in Seyferts (see Christopoulou et al., 1997; Lindblad et al. 1996; Morris et al., 1985; Storchi-Bergmann et al., 1992; Wilson et al., 1985). In fact, the morphology of the emission region is reminiscent of a cone (Fig. 1, and Fig. 4 in Wehrle & Morris, 1988) although the possibility that the arc is a ring or spiral structure around the nucleus cannot be discarded.

3.2. Velocity field

In order to disentangle overall rotation from peculiar motion, we first apply a simple galactic rotation model. The gas is considered in circular motion and lies in a disc with an inclination and a position of the major axis (RC3).

The velocity law, which assumes that the gas is in a spherical gravitational potential and follows circular orbits in a plane (de Zeeuw & Lynden-Bell 1988) is given by the following equation:

where r is the distance to the center in the plane of the galaxy, the maximum amplitude and the distance at which this velocity is reached; p is a parameter controlling the slope of the rotation curve in its "flat" outer regions.

We assumed that the dynamical center coincides with the peak of the continuum light and constrained the above parameters using the velocity field observed along the photometric major axis, i.e. . The resulting rotation curve computation can be seen as a dashed line in Figs. 2 - 3 - 4. We tried different sets of parameters and obtain a fairly good match for =350 km s^-1, =5 arcsec and p =1.3. Indeed, as mentioned above, p cannot be much different, since it is constrained by the shape of the rotation curve beyond 5 arcsec; could vary between 5 arcsec and 8 arcsec, and by no more than km s^-1.

However it is apparent on Figs. 3c and 3d, that is not the best estimate for the kinematic major axis. Indeed, the parameters lead to velocity fields which are systematically shifted with respect to the measurements along PA= (Figs. 3c-d). Along PA= (Fig. 3a), the velocity field which becomes flat at distances over 5 arcsec is badly represented. The steep central gradient and the amplitude of flat rotation regions further out indeed suggest that the kinematical major axis of the ionized gas is larger than = .

This value is strongly constrained by the PA= offset data (Figs. 3c-d), in order to shift the minimum model velocity to the centre, and can only vary between 25 and . The set of parameters is somewhat different in this case: =250 km s^-1 and p =1.1, the maximum velocity being reached at the same radius, =5 arcsec. The resulting rotation velocity field is plotted in Figs. 2 - 3- 4 as a dotted line.

Better matches to the data in most of the slit positions are obviously obtained. The fact that the photometric major axis derived from the kinematics does not agree with that derived from broad band imaging suggests that the gas disk is in a different plane than the stellar disk (for instance because of a warping due to the interaction with NGC 2993) and/or that a radial component is present in addition to the galactic rotation. It is worth noticing that PA= is perpendicular to the PA= of the elongated radio structure as measured by Hummel et al. (1983) from VLA observations. This model accounts for most of the low excitation gas structure along non offset slits in the central few arcseconds (excepting PA= ); however, further from the nucleus, the low and high excitation gases do not always follow the same kinematics.

As discussed in Sect. 3.1, outflow of the gas within a conical envelope or on the surface of a hollow cone is a possible picture to account for the double peaks and blueshift. The morphology of the emitting gas (see the [OIII] image in Fig. 1 and Fig. 4 in Wehrle & Morris 1988) is suggestive of a cone structure along an axis at PA= and with a projected full opening angle of about . Notice that the cone axis is almost aligned with that of the elongated radio structure observed by Hummel et al. (1983) with the VLA, and its linear extent is comparable to that of the radio emission, namely 20 arcsec on the east side and 10 arcsec on the west side.

Although the line splitting is ignored when measuring the H lines, two components are clearly present in the low excitation gas along PA= and (see Fig. 7 and Sect. 3.1). One can notice in Fig. 2c that +30 km s^-1, the velocity of one component along PA , is close to the velocity expected from a pure disk rotation (dashed curve). In Fig 2d, the value of -10 km s^-1 is not far from the disk rotation velocity. It is thus clear that the plotted weighted mean velocities are dominated by outflow in the east. We have therefore used a very simple model in which the outflowing velocities are considered as radial motions in a plane very close to that of the gas disk. A more sophisticated modeling of a true cone, such as that of Wilson et al. (1985), Hjelm & Lindblad (1996) or Christopoulou et al. (1997) is not possible due to the lack of detailed information on the line splitting and of photometrical information.

We have considered that the west side of the galaxy is the near one; this would explain why the emission cone is seen mainly to the east. We add to the general rotation pattern described above a constant outflow, modeled along a triangular region of axis PA= , with an opening angle of and a constant velocity of 150 km s^-1 along the outflow region (measured 5 arcsec from the nucleus). This is a simple addition of outflowing radial velocity in a region determined by the [OIII] image morphology and taking place in a plane very close to the gas disk. The projected size of the zone on which this region extends its influence is taken to be 9 arcsec and 4 arcsec in the east and west regions respectively; these values are strongly constrained by the high excitation gas data along PAs between and (centered and offcentered). The results are shown as full lines in Figs. 2 - 3- 4. Although this model is very simple, it accounts notably better than the previous one for the high excitation gas, as can be seen in Figs. 2c, 3a, 3b and 3c. Nevertheless, close inspection of Figs. 2 - 3 - 4 raises some comments.

The low excitation gas represented by the lines in the H domain follows more closely the pattern of normal rotation with a kinematical major axis = . Notice that along PA= the H data follows a rotation pattern with = ; the discrepant points in the SW might correspond to the annulus that is conspicuous at such distances on broad band images; it could correspond to part of a spiral arm or ring, in which non-circular motions are expected.

For the whole set of PAs, the data for both the high and low excitation gas at distances larger than 10 arcsec to the northwest show the largest differences with respect to the model. This could suggest that we are seeing kinematically distinct regions, i.e. line of sight gas which is not physically associated with the main emission structure. Indeed, the gas kinematics could be perturbed by the interaction of NGC 2992 with its close companion NGC 2993; notice that the existence of a warp in the gaseous disk is suggested by the difference between the photometrical and kinematical major axes.

Another feature that can perturb the measured velocity field is dust. Indeed, if the line profiles are affected by dust, the cross-correlation method used to infer velocity points will result in velocities which are not fully characteristic of the sampled region. This could explain the humps observed to the SE along PA= and , and also account for the differences observed between the [OIII] and H ranges in the NW along PA= .

Along PA= offset by 5 arcsec (Fig. 3c), the hump observed in the velocity field just southeast of the nucleus corresponds to the annulus of ionized gas already mentioned above, while the large values northwest of the nucleus correspond to the "finger" observed by Wehrle & Morris (1988). It is probable therefore that these two regions do not follow the general rotation pattern. Notice that, here again, H roughly follows the disc rotation while the [OIII] velocities are better accounted for by a model including outflow.

© European Southern Observatory (ESO) 1998

Online publication: April 20, 1998
helpdesk.link@springer.de