Astron. Astrophys. 347, 55-62 (1999)

5. Composite spectra

As it was found in the previous section, only 4 galaxies could be classified as LINERs and 1 as Seyfert 2 in at least one of the diagnostic-diagrams, but several galaxies are in the transition zone. These transition objects could have a composite spectrum due to the simultaneous presence on the slit of a Seyfert nucleus and an H II region (V97f). In order to test this hypothesis we have built composite spectra, combining emission-pure spectra of a typical Seyfert 1 (J 15.22), Seyfert 2 (NGC 1068) and LINER (AM 2054-433S) with the H II region spectra of solar abundance R1a, R5a and R10a, of the Gumm 38a complex. These galactic H II regions have been selected because they are as metallic as the circumnuclear H II regions of nearby galaxies (Storchi-Bergmann et al. 1996a), are purely photoionised by stars, and also show a wide range of excitation (G97). For J 15.22 and NGC 1068, suitable stellar population templates were also subtracted following the method described in Sect. 3.

The composite spectra were built adding different fractions of the H II region spectrum to that of the AGN . To do this, the stellar-population subtracted AGN and H II region spectra have been normalised at the peak of the emission line. This process is illustrated in Fig. 7, in which we combine 30% of the spectrum of the Seyfert 1 J 15.22 with 70% of the moderate excitation H II region R10a. As a result, the H II region spectrum masks that of the Seyfert 1, leaving only a slight broadening at the base of and .

Fig. 7. (A) The Classical Seyfert 1 J 15.22. (B) Combined spectrum: 30% of J 15.22 with 70% of R10a; fractions in flux at Å. Spectra are plotted in units.

The resulting emission line ratios of the composite spectra are plotted in the diagnostic-diagrams (long dashed lines) in Figs. 5 and 6, where the effect of the H II region contamination can be seen: for the composite model shown in Fig. 7, the 70% H II region contribution shifts the line ratios of the Seyfert 1 to the typical H II region zone. In terms of flux, this H II region fraction corresponds to 64% of the total, a reasonable amount for low luminosity nuclei. Indeed, in the moderate star-forming galaxy NGC 1097, the contribution of a few individual H II regions in the ring around the nucleus to the total nuclear flux is 50% (Storchi-Bergmann, Wilson & Baldwin 1996b). A more dramatic example is NGC 3310 in which the H II region contribution amounts to 90% (Pastoriza et al. 1993).

In order to check if the selection of H II regions with different properties would affect the results, we calculated using the CLOUDY code (Ferland 1993), a series of H II region models photoionised by black-body spectra with effective temperatures T=35 000K, 40 000K and 50 000K. Since we are dealing with circumnuclear H II regions, and the low spectral resolution prevents a direct calculation of the electron density () from the lines, we have adopted the average value obtained for the circumnuclear H II regions observed in NGC 3310 (Pastoriza et al. 1993), i.e. . The Hydrogen gas density was fixed at 500 , and models were calculated for solar ¹ and half-solar metallicity, and ionisation parameter in the range . The T=50 000K and solar metallicity models for the diagram are shown in Fig. 6. The use of models with the same range of U and lower abundances would shift the H II region loci further to the left, increasing the number of AGN candidates among the galaxies of our sample.

From Figs. 5 and 6, all the points to the right of the Gumm 38a H II regions (heavy-dashed line) can be reproduced by a combination of an AGN spectrum with that of an H II region. In this sense, only the points to the left of that line would be pure H II regions. Notice that for our sample galaxies, the and lines are more sensitive indicators of the presence of a composite spectrum than . Indeed, in the diagnostic-diagram (Fig. 5, bottom panel), 50% of the line ratios appear to be from composite spectra, whereas in the (Fig. 5, top panel) and (Fig. 6) diagrams, this fraction increases to 78%.

Taking into account the position of the observed line ratios with respect to the Gumm 38a H II regions zone in the 3 diagnostic-diagrams (Figs. 5 and 6), we give in column 6 of Table 2 a revised classification for each galaxy. The objects having a composite spectrum (CS) are those with line ratios lying to the right of the H II region line in all 3 diagnostic-diagrams. Objects with a possible composite spectrum (CS?) are those with line ratios satisfying the above criterion in at least one diagnostic-diagram.

From this approach, we conclude that 15 objects (28% of the sample) may be AGN contaminated by some fraction of H II region emission due to the simultaneous presence on the slit of a Seyfert nucleus and an H II region. Including the possible composite spectra (CS?) this number is increased to 20, which corresponds to 37% of the present galaxy sample, which is consistent with previous results (Ho, Filippenko & Sargent 1997; Liu & Kennicutt 1995).

© European Southern Observatory (ESO) 1999

Online publication: June 18, 1999
helpdesk.link@springer.de