EXCITATION AND TEMPERATURE OF EXTENDED GAS IN ACTIVE GALAXIES .2. PHOTOIONIZATION MODELS WITH MATTER-BOUNDED CLOUDS

Recent reliable measurements of the weak lines of [O III]lambda 4363, He II lambda 4686 and [Ne V]lambda 3426 seen in emission in extended a nd nuclear regions of AGN (Storchi-Bergmann et al. 1996) are used to d iscriminate between different types of photoionization models. The sam ple comprises Seyferts and Narrow Line Radio Galaxies. A well known pr oblem, the scatter and the extreme values of He II/H beta encountered in the nuclei and extended regions, is solved in a similar fashion to Viegas & Prieto (1992) by considering two populations of ionized cloud s: a matter-bounded (MB) component responsible for most of the He II e mission, and an ionization-bounded (IB) component emitting low-to-inte rmediate excitation lines. A new sequence of photoionization calculati ons is thereby obtained by varying A(M/I), which is defined as the sol id angle ratio occupied by the MB component relative to the IB compone nt. In various line ratio diagrams, the A(M/I)-sequence is compared to the traditional single component U-sequence in which one arbitrarily varies the relative intensity of the ionizing source. An original aspe ct of the new calculations is that the IB clouds are photoionized excl usively by the radiation which has filtered ('leaked') through the MB component. This situation could arise if, for example, the MB componen t is the low density 'halo' of a cloud, and the IB component the high density core. Such a filtered ionizing spectrum enlarges the range in He II/H beta accessible to models without requiring changes in the spe ctral index of the ionizing source, which we fix at alpha=-1.3. The ma in success of the model is that it provides a natural explanation for the newly found correlations between both the [O II]/[Ne V] and [O II] /[O III] ratio and the He II/H beta ratio. On the other hand, the stan dard U-sequence cannot account for such correlations between the gaseo us excitation and the He lI/H beta ratio. Furthermore, the so-called ' temperature problem' is solved with the A(M/I)-sequence provided the t hickness and the ionization parameter (U(MB)greater than or similar to 0.03) of the MB component are appropriately selected. Finally, the A( M/I)-sequence produces much stronger high excitation lines of [Ne V] a nd C IV lambda lambda 1549 than the U-sequence, in accordance with obs ervations. The A(M/I)-sequence predicts a temperature difference of 5 000 K between the T-OIII and T-NII temperatures while for the U-sequen ce this difference is less than 1 000 K. To the extent that 'cooling f lows' and LINERs share the same excitation mechanism as our IB compone nt (i.e. photoionization by an absorbed ionizing continuum), we predic t a value of He II/H beta<0.01 while the U-sequence is charaterized by a value >0.1.