S.J. Wagner and M. Dietrich
Landessternwarte, Königstuhl, 69117 Heidelberg, Germany
partly based on data collected at the European Southern Observatory partly based on data retrieved from the HST archive supported by the Deutsche Forschunggemeinschaft through SFB 328
We have taken 2D-spectra of NGC 1068 with high spectral resolution ( ). The slit covers the inner 6 arcsec of the center of NGC 1068. Due to the high resolution the blend of clouds is decomposed in velocity space. The deblended components can be located in the l-v diagrams with an accuracy of 0 along one spatial coordinate. Orthogonal slit orientations allow the precise location of knots and the identification with the clouds resolved in the HST narrow band images.
We derive the velocities of the 15 most luminous clouds within the central 2 arcsec. The components cover a velocity range of 1000 . The widths of the individual clouds are an order of magnitude larger than the thermal line width ( ). No single regular velocity field (rotation, inflow, outflow) fits the distribution of components in phase space completely. We find evidence for two different kinematic systems and high degree of turbulence and very high internal dispersions within individual clouds.
Keywords: NLR, Seyfert Galaxies, NGC 1068, Outflows
NGC 1068 (3C 71), one of the best studied Seyfert galaxies is among the nearest AGN. Lying at a distance of 23 Mpc, 1'' corresponds to 110 pc. Subarcsec resolution studies hence allow spatially resolved studies of the NLR. Although being regarded for a long time as the prototypical Seyfert 2 object, the discovery of polarized emission from a broad-line region suggests the presence of a hidden Seyfert 1 nucleus, which can be observed in scattered light. The position of the hidden nucleus has been determined from polarimetric HST observations (Capetti et al. 1995). The light from the nucleus is blocked along our line of sight, but nuclear radiation may escape more freely along other directions. Interferometric observations show a non-relativistic radio outflow (Wilson & Ulvestad 1987). The Narrow Line Region is extended along the same direction (Walker 1968) and shows indication of clumpy structure (Anderson 1971). If it is photoionized by a nuclear continuum this morphology suggests the NLR to be a radiation bounded ionization `cone' which owes its shape to an obscuring torus. Alternatively, the emission may partly be caused by interactions of the circumnuclear matter and the outflowing plasma.
In the former case, the widths of emission-lines reflect the strength of the gravitational potential at the luminosity-weighted radius where the corresponding line is emitted. In such a picture, non-gravitational forces are not assumed to contribute substantially to the line-widths of the NLR. This is independent of the details of the velocity field (chaotic motion or rotation) but contrasts those models which attribute the line-shapes to inflow or outflow dominated scenarios.
Emission-line widths of forbidden transitions are often found to be comparable to the widths of absorption features of the integrated light of the stellar population of the bulge of the host galaxy (Nelson & Whittle 1995). There are a few notable exception to this trend. Several AGN show forbidden lines whose widths exceed 1000 . The largest widths in an AGN have been recorded in NGC 1068 (FWHM 1200 ). This extreme value does not seem to be related to the presence of a hidden Seyfert 1 nucleus (with other sources of comparable polarization characteristics having normal line widths of forbidden transitions).
The possibility to resolve the NLR spatially now permits investigations of the velocity field on parsec scales. This enables us to address the question whether the large FWHM is due to large-scale velocity gradients, turbulence or due to scattering. High resolution imaging with HST has demonstrated that the NLR is composed of a large set of clouds with angular separations much smaller than the limited resolution of ground based telescopes. Measuring their bulk velocities can be used to discriminate between rotation and outflow which in turn can be used to derive masses or orientation and acceleration mechanisms. Line widths can be used to determine turbulence and study length scales and shock formation.
The ground-based spectroscopic data were acquired with the CASPEC echelle spectrograph of the European Southern Observatory attached to the 3.6m telescope of ESO during two runs. In November 1993 data were taken in the red wavelength regime, including H and [NII], in December 1994, we could use the blue sensitive cross-disperser to get data of [OIII] and H . The spectral resolution was measured to be 15 , the seeing varied throughout both campaigns with an average effective spatial FWHM of 1. Details of the reduction procedure are described elsewhere. Following standard reduction, the individual orders were rectified, maintaining the full spatial information and rebinned into velocity with respect to to a systemic velocity of 1145 (Allen et al. 1971). The slit was oriented along PA 32^o and 122^o (along and perpendicular to the radio feature and the ENLR). Three sets of data in each of the two position angles were taken in both wavelength ranges. The effective exposure on NGC 1068 amounted to 12 hours.
The HST images were taken from the archive (frame x24e0101r). We used the F501N filter post-COSTAR observations with FOC (f/96) by Macchetto et al. (1994) which have a scale of 0/pix and a resolution of 0.
The velocity field of NGC 1068 was constructed by identifying all individual components in each of the long slit spectrograms. The integrated flux, velocity, and the position along the slit of each knot was determined by fitting elliptical Gaussians to the profiles. The identification was checked by treating all three frames taken with the same slit orientation separately. The uncertainty of the model fitting was determined by comparing the parameters derived for each of the knots in the three spectrograms. After locating individual knots in the HST images, we cross-identified the knots by requiring identical spatial coordinates along the slit and comparable intensities in the long-slit spectrograms and the [OIII] image. While the spatial coordinate is a unique but insufficient criterion, the fluxes (or flux ratios) cannot be expected to be identical. The F501N filter has a FWHM of 7.4 nm and contains a significant fraction of continuum emission. The equivalent width of the spatially integrated [OIII] line within the 6 1.5 arcsec slit is about 43 nm, but may vary for individual knots. In a few cases, this unknown contamination of continuum emission prohibits a cross identification of knots with identical locations along the slit. In these cases, we used the spectrograms taken with a PA of 122^o, perpendicular to the long axis to break the degeneracy. This led to a unique location of the 15 best defined clumps that can be seen the central 6 2 arcsec of the HST image in the longslit spectrograms. The comparison of the FOC image and the CASPEC spectrograms is shown in Figure 1. Peak velocities and line widths of each of the clouds were determined to derive the velocity field and the internal kinematic state of the clouds. The luminosity function of these clouds is fairly steep with more then 90 % of the line flux being emitted by four components.
Figure: Left panel: [OIII] FOC image (F501N) of the NLR of NGC 1068 within a centered aperture of 1.5 6 arcsec (equivalent to the CASPEC slit used to obtain the long-slit spectrogram on the right. All individual components visible in the direct image can be identified (and separated) in the velocity map. This allows a complete mapping of the velocity field of the clouds seen in the HST image.
Within the central 2 (275 pc), a large-scale gradient of V 1350 can be observed. While the range of velocities of the large-scale gradient is not centered on the systemic velocity (-850 to +500 ), the systemic velocity is observed at the `hidden' nucleus. Both of these results suggest that emission is due to outflow rather than to rotation. Since the linear bulk motion is oriented along the radio structure, it is tempting to identify it with the outflow. The blue shift towards the north-east then indicates that the outflow on this side is approaching the observer. The string of five of the brighter clouds which form a ridge at PA 10^o follow a common trend in velocity space with much less scatter than the overall field.
In addition to the coherent large scale gradient, a large velocity spread is observed at different separations from the nucleus. A few components along a common line of sight (l.o.s.) are found which have velocities that differ by up to v 1700 . While large spreads in velocities might be expected in case of outflows within cones of very wide opening angles, we attribute the difference to random motion (i.e., large scale turbulence). First, the morphology of the NLR in general suggests an opening angle of less than 90^o; second, the observed spread along a single l.o.s. v > V, the large scale gradient. Furthermore, l.o.s. integration through an axisymmetric cone does not explain complexity of large scale spread.
In spite of the blending of the individual clouds in ground-based images, the components can be separated well enough in the high spectral resolution spectrograms to investigate the line profiles of the individual knots. A few of the individual clouds have large line-widths of up to FWHM 300 . These features do not show any substructure in the line profile. The corresponding clouds in the HST image are not blended with any significant contribution from other clouds. We, hence, rule out that the large line widths are due to superpositions of two distinct components along the line of sight. The clouds are very compact in the HST images with diameters of about 100 mas, i.e., 11 pc, and separated from the hidden nucleus. Their peak velocities are different from the systemic velocity. Both of these characteristics clearly indicate that these clouds do not lie in the central potential well. They are not confined by the gravitational field of the nucleus. Irrespective of whether these clouds are photoionized, shock heated or scatter nuclear line emission, confinement by gas pressure of magnetic fields would require very extreme conditions. We conclude that the individual clouds are transitory. They dissolve on a time-scale of several 10 years.
A few of the components can be spatially resolved in the FOC image, indicating substructure on the scale of a few parsecs. Pronounced substructure within 0 is seen in clouds A, B, C, D, and F of Evans et al. (1991). In some of these cases, substructure is seen in the line profiles of these knots. The line widths of these compact clumps reaches up to 600 , but the role of l.o.s. integration of spatially separated cannot be estimated.
In spite of the overall trend in the large scale velocity field, many clouds are found to show substantial differences in their velocity. Small complexes of clouds spread over as much as 600 within about (30 pc). Individual clouds of about 10 pc diameter show line widths of up to 300 . We find evidence for turbulent motion with high velocities on all scales investigated in this study. The large velocities with respect to adjacent clouds and intercloud matter induce strong shocks which are likely to contribute significantly to the ionization of the NLR plasma. Although there does not seem to be a close relation between bright knots of the radio outflow and bright, line-emitting clouds, the former propagates through the turbulent NLR. Shock induced excitation has been shown to be an important contributor in the more powerful jets of radio sources (e.g., Sutherland et al. 1993), but is likely to play an important role in NGC 1068 as well. Quantitative analyses require additional information on the temperature and density structure within the NLR. Such information is currently derived by determining diagnostic line ratios for individual clouds.
We have combined the high spatial resolution of post-COSTAR FOC observations with high spectral resolution of ground-based longslit spectroscopy to determine the velocity field of the clumpy Narrow Line Region of NGC 1068. We find that the unusually large width of the entire forbidden line region is due to the superposition of the individual clouds. We find that a large fraction of the clouds cluster around a linear trend in the l-v diagram derived from the longslit spectrograms. This regular velocity field crosses the location of the hidden nucleus at the systemic velocity. The asymmetry of the maximum velocities and the large systemic velocities suggest that this regular velocity field is not caused by rotation. We favor an outflow hypothesis. Several individual clouds differ from this general velocity field by up to 1700 . Some of the compact, unblended, off-nuclear clouds exhibit large line widths. It is highly unlikely that the widths are due to blending along the l.o.s.. Since the clouds cannot be confined by gravity, we favor the hypothesis that the clouds are shock heated and dissolve fast.
We are grateful to Dr. A. Jüttner, who carried out most of the observations.
Allen, R.J., Darchy, B.F., & Lauqué, R. 1971, A&A, 10, 198
Anderson, K.S. 1971, ApJ, 162, 743
Capetti, A., Axon, D.J., Macchetto, F., Sparks, W.B. & Boksenberg, A. 1995, ApJ, 446, 155
Macchetto, F., Capetti, A., Sparks, W.B., Axon, D.J., & Boksenberg, A. 1994, ApJ 435, L15
Nelson, C.E., Whittle, M. 1995, ApJS, 99, 67
Sutherland, R.S., Bicknell, G.V., & Dopita, M.A. 1993, ApJ 414, 510
Walker, M.F. 1968, ApJ, 151, 71
Wilson, A.S. & Ulvestad J.S. 1987, ApJ, 319, 105
Evans, I.N., Ford., H.C., Kinney, A.L., Antonucci, R.R.J.,
Armus, L., & Caganoff, S. 1991, ApJ, 369, L27