Ultraviolet variability of the Seyfert 1 galaxy Fairall 9

The large amplitude of the variations of the Seyfert I Galaxy Fairall-9 between 1978 and 1991 make this Active Galaxy especially suitable for a combined study in terms of reverberation analysis and photoionization modeling over the velocity field of the broad lines of its ultraviolet spectrum. We have combined our ultraviolet data with those available at other wavelengths to derive the intrinsic ionizing continuum and to compare the predictions of the photoionization models with the observations. The UV continuum varies with a factor of 33+/-4 on a characteristic time scale of 182 days. The intrinsic spectral index UV-optical is alpha =-0.07+/-0.02 and the optical variations do not lag behind the UV variations at the mean sampling interval of 96 days. In the near IR, the J band flux presents a direct extension of the UV-optical continuum (alpha_{J-UV}=-0.07+/-0.05). Only at low flux levels is the 2-10 keV flux proportional to the UV continuum, at higher UV flux levels proportionality between the X-rays and the UV brightness breaks down. The continuum spectral energy distribution (SED) of the nucleus supports the previously reported FIR-NIR excess associated with 0.02 M_{\odot} dust at a distance of some 400+/-100 light-days, as well as a soft X-ray excess, possibly associated with the reprocessing on an accretion disk of the hard X-rays emitted from a region above the disk. The presence of strong Fe K alpha line in the GINGA spectra of this galaxy does lend support to this model. Line profile variability has been used to isolate four gaussian line components, which are sufficient to describe all lines at all levels of brightness in a consistent way: one narrow (i.e. unresolved at the IUE resolution), and three broad components: a central (velocity same as the narrow line), a red shifted (v=3300 km s^{-1}) and a blue shifted (v=-3600 km s^{-1}) one. The three broad components are strongly correlated with the UV continuum, indicating that photoionization is the dominant mechanism in the BLR. Correlation analysis shows different delays between the broad components and the UV continuum: respectively, 400+/-100, -4+/-70 and 230+/-95 days for the central, red and blue one. Only the red component of Ly alpha lambda 1216 and CIV lambda 1549 appear to vary linearly with the continuum and give significant transfer functions. The resulting transfer functions are peaked at zero days delay and are unresolved at the average time resolution (96 days) of our data. The photoionization models (CLOUDY) applied to the line ratios in these components, indicate that the Broad-Line-Region (BLR) is situated between 50-250 light-days from the ionizing source, with an hydrogen density of 10^{12-10} cm^{-3}, a column density of 10^{23-26} cm^{-2} and a covering factor of 12-2%, with an ionization parameter between 0.003 and 0.089. However, no optically thick model reproduces the (Ly alpha/CIV lambda 1549) and (Ly alpha/NV lambda 1240) ratios. From these results we propose a model for the structure and dynamics of the BLR: the mass of the central compact object is M \approx 2 X 10⁸ M_{\odot}. Around this exist two distinct gas zones within the BLR: the gas producing the central component at \approx 200 light-days and gas, emitting the red and the blue components, at \approx 80 light-days moving inward to a central source. These results require that both these gas zones be localized along the line-of-sight or, alternatively, that the continuum emission must be strongly anisotropic. Besides, the gas emitting the central component is most likely mixed with the dust and the central gas to dust mass ratio is Mass (central gas)/mass(dust) = 100 - 750. Tables 1, 10, 11, 12, 13 only available in electronic form at CDS via anonymous ftp to CDSARC.U-strasbg.fr (130.79.128.5) or via http://cdsweb.u-strasbg.fr/Abstract.html.