The torus simulations for NGC 2110, ESO 428-G14 and MCG-05-23-016 are still under- way. Therefore, I have to restrict myself to presenting the results of the observations themselves. The recorded spectra – as well as the archival spectrum of NGC 1068 – are shown in figs. 7.1 to 7.4. For comparison, we have also plotted archival Spitzer IRS spectra for NGC 2110 and ESO 428-G14 and the Gemini spectrum of NGC 1068 from Mason et al. (2006). For MCG-05-23-016, no comparison spectrum was available.
7.1 High resolution mid IR spectroscopy of Seyfert 2 nuclei 87
shaded parts of the spectrum, the S/N ratio drops and features seen there are probably not real.
The spectral shapes of NGC 1068, ESO 428-G14 and MCG-05-23-016 are quite similar. In all three objects we see a 9.7 µm silicate absorption feature of moderate (NGC 1068) or weak strength and a continuum that is rising toward longer wavelength. The continuum shape in NGC 2110 is flatter than that of the other objects. In addition, it does not exhibit the silicate feature. NGC 2110 is also the only object that does not have any emission lines in the MIR. The other three objects show [ArIII], [SIV] and [NeII] lines of
differing strengths. Interestingly, none of the objects show emission in the 11.3 µm PAH line. This is a good indication for the observed spectra to be largely free of contamination from SF regions. The atomic lines likely originate in the NLR of the AGN. Since they are relatively weak, the VISIR spectra are probably good representations of the pure torus spectra of these AGN.
In fig.s 7.2 and 7.3, we have also plotted archival IRS spectra of NGC 2110 and ESO 428-G14, respectively. It is obvious that these spectra look quite different than ours. First of all, the IRS spectra show very strong [SIV], PAH and [NeII] lines – and the
PAH line is a good indicator for active star formation. Secondly, the spectral slope also varies. In both cases, the continuum observed with Spitzer has a steeper rise toward longer wavelengths than the one measured with VISIR. This means that Spitzer sees an additional, cooler dust component that is likely associated with SF regions. In addition, in ESO 428-G14, the IRS spectrum exhibits a slightly stronger silicate feature than the VISIR one. The differences in continuum slope and silicate absorption strength between IRS and VISIR spectra adds to the doubt I put forward in section 6.1.1 whether detailed comparisons between AGN torus models and Spitzer spectra of Seyfert galaxies can produce meaningful results.
For the three objects with detected emission lines, we find evidence for the presence of the “Baldwin effect”. The Baldwin effect, an anti-correlation between continuum luminosity and emission line equivalent width, was first found for the broad Lyα and
CIVlines in QSOs (Baldwin, 1977). The physical origin of the effect is still unclear but
usually it is attributed to a change in the continuum slope with luminosity. According to a theoretical work by Korista et al. (1998), the continuum gets softer as the luminosity increases. This means that, relative to the total continuum power, the number of ionising and effectively excitating photons decreases. In turn, this results in reduced equivalent widths of the broad emission lines.
In the MIR, the Baldwin effect was first observed by Keremedjiev & Hao (2006) in a sample of 68 AGN, observed with the Spitzer Space Telescope. Their analysis, however, was obstructed by the low spatial resolution of their data, as they note themselves. In our sample we clearly find the effect for the three sources that show MIR emission lines. In table 7.2, we list the emission line equivalent widths Wλ and 12 µm continuum
luminosities λLλ of these objects. We then plot the anti-correlation between Wλ and
log λLλ in fig. 7.5. We clearly detect the Baldwin effect in the [ArIII], [SIV] and [NeII]
lines. The best fit slope to all emission lines is -0.6. It should be noted that our sample is small. Therefore, the value for the slope is of low statistical significance.
NGC 2110 which is omitted in both table 7.2 and fig. 7.5 has a 12 µm luminosity of 0.96· 1043 erg s−1. Thus, from the Wλ – λLλ anti-correlation among the other three objects,
Figure 7.1: N-Band VISIR spectrum of NGC 1068. The red line is the VISIR spectrum, error bars are in yellow. The blue dashed line is a Gemini spectrum (Mason et al., 2006) that we plot for comparison. The grey shaded areas mark regions with strong and variable sky lines. Vertical dashed lines denote the position of common MIR emission lines with names given at the top.
Figure 7.2: N-band VISIR spectrum of NGC 2110. The red line is the VISIR spectrum, error bars are in yellow. Purple crosses mark VISIR photometry that was used for the flux calibration of the spectrum, the horizontal error bars depict the respective filter band width. Here, we compare the VISIR spectrum to an archival Spitzer IRS spectrum (blue dashed line). Observations in spectroscopic setting that covers the [NeII] line are missing.
7.1 High resolution mid IR spectroscopy of Seyfert 2 nuclei 89
Figure 7.3: N-Band VISIR spectrum of ESO 428-G14. All symbols are as in fig. 7.2.
Figure 7.4: N-Band VISIR spectrum of MCG-05-23-016. All symbols are as in fig. 7.1. A comparison spectrum was not available.
Table 7.2: Emission line equivalent widths Wλ for our VISIR sample. In order to illustrate the
presence of the Baldwin effect, we also state the 12 µm continuum luminosities λLλ. NGC 2110
is omitted in this table since we did not detect any MIR emission lines in this object.
Object λLλ Wλ [ArIII] Wλ [SIV] Wλ [NeII]
[1043 erg/2] [µm] [µm] [µm]
ESO 428-G14 0.33 0.0095 ± 0.0008 0.0259 ± 0.0014 0.0181 ± 0.0011
NGC 1068 6.81 0.0023 ± 0.0006 0.0034 ± 0.0005 0.0024 ± 0.0003
MCG-05-16-023 2.10 0.0030 ± 0.0003 0.0047 ± 0.0001 0.0098 ± 0.0011
the observation in the spectroscopic setting that covers the [NeII] line was not executed
for this object due to scheduling problems. Thus, we do not know whether the line is emitted in the nucleus or in the more extended region observed by Spitzer. The [ArIII]
and [SIV] lines, on the other hand, are clearly absent from the spectrum.
The MIR lines we detect are narrow and, thus, likely originate in the NLR. As the NLR can be very extended, the physical origin of broad lines and narrow lines may be quite different (Shields, 2007, and references therein). On the other hand, the high angular resolution of our spectroscopy confines the recorded spectra to those parts of the NLR closest to the central engine. Therefore, it seems reasonable to assume that the excitation of the ions we see in emission is caused by the radiation from the accretion disc. Since the accretion disc also heats the torus, we expect the MIR continuum to scale with the accretion disc emission. The MIR – hard X-ray correlation actually demonstrates that this is, indeed, the case. Therefore, the Baldwin effect in the MIR may be caused by the same mechanism as for the optical broad lines: The ionising continuum softens with luminosity. A problem for this interpretation is the slope. In broad lines, the typical slope for the Baldwin effect is ∼ −0.2 instead of −0.6 which we find among our sample. This discrepancy might be caused by the low number statistics of our sample.
We summarise the first results that we draw from this spectroscopic campaign: The nuclear spectra of type II AGN probably show less variation than expected from the analysis of Spitzer spectra of lower angular resolution. The strength of the 9.7 µm silicate feature as well as the slope of the continuum vary among the objects in our sample; the overall shape of the MIR SED, however, is rather similar. Weak [ArIII], [NeII] and [SIV]
lines are found in three of four observed objects. They probably originate in the NLR and show a significant Baldwin effect. The presence of the Baldwin effect in the MIR indicates a close physical connection between accretion disc and torus. Taking into account the main result of this thesis – the MIR – hard X-ray correlation – it seems natural that MIR emission lines show a similar Baldwin effect as broad lines.