3.1. Line fluxes and ratios
The quasi-complete spectra of the nucleus and of knot C are shown in Figs. 2, 3 respectively, and the derived line fluxes are summarized in Table 1. Dilution by a stellar continuum is particularly strong in the nuclear spectrum where the equivalent width of [OIII]5007 is only 50 Å (cf. Table 2) and a factor of 10 lower than found in typical Seyfert 2's. The stellar contribution is normally estimated and subtracted using either off-nuclear spectra extracted from the same 2D long slit frames, or a suitable combination of spectra of non-active galaxies used as templates (e.g. Ho 1996, Ho et al. 1997). However, neither of the methods proved particularly useful because line emission contaminates the stellar emission all along the slit, and we could not find any template which accurately reproduces the prominent stellar absorption features typical of quite young stellar populations. The fluxes of weak lines (5% of the continuum) in the nucleus are therefore uncertain and, in a few cases, quite different than those reported in O94, the largest discrepancy being for [NI] which is a factor of 2 fainter here.
Table 1. Observed and dereddened line fluxes Observed line flux, relative to H=100 Dereddened flux (H=100), a colon denotes uncertain values. Blank entries are undetected lines with non-significant upper limits From Savage & Mathis (1979) and Mathis (1990) Units of erg cm-2 s-1
Table 2. Fluxes of significant lines in the various knots Dereddened fluxes relative to H=100, extinctions are computed imposing H=290 and the adopted visual extinctions are given at the top of each column. Blank entries are undetected lines with non-significant upper limits H is weak and the derived extinction is therefore uncertain. Possibly overestimated due to contamination by foreground gas with lower extinction. Dereddened flux, units of erg cm-2 s- 1 Equivalent widths in Å of nebular emission lines.
The spectrum of knot C has a much more favourable line/continuum ratio and shows many faint lines which are particularly useful for the modelling described in Sect. 4.
3.2. Spatial distribution of emission lines
The spatial variation of the most important lines is visualized in Figs. 4, 5 which show contour plots of the continuum subtracted long slit spectra and selected spectral sections of the various knots respectively. The fluxes are summarized in Table 2 together with the extinctions which were derived from hydrogen recombination lines assuming standard case-B ratios (Hummer & Storey 1987). A remarkable result is the large variations of the typical line diagnostic ratios [OIII]/H, [OI]/H, [NII]/H and [SII]/H which are plotted in Fig. 6 and range from values typical of high excitation Seyferts (nucleus, knots A, B, C, D), to low excitation LINERs (knots H, I) and normal HII regions (knots E, L). Another interesting result is the steep extinction gradient between the regions outside (knots C, D, H, I) and those close to the galactic disk (nucleus and knots A, E, L). However, a comparison between the Br map (Moorwood & Oliva 1994), the H images (M94) and the observed Br flux from the whole galaxy (M96), do not show evidence of more obscured () ionized regions such as those observed in NGC4945 and other starburst galaxies (e.g. Moorwood & Oliva 1988). Nevertheless, these data cannot exclude the presence of deeply embedded ionized gas which is obscured even at 4µm (i.e. mag).
Particularly interesting is the variation of the line ratios between the adjacent knots C and D. The ratio [FeVII]/H is a factor of 2 larger in knot D than in C, but this most probably reflects variations of the iron gas phase abundance (see also Sect. 4.5). Much more puzzling is the spatial variation of the low excitation lines [OI], [SII], [NII] which drop by a factor 1.8, while the high excitation lines HeII, [OIII], [NeIII], [ArIII] together with the [SII] density sensitive ratio and [OIII]/[OII] vary by much smaller amounts (cf. Table 2). This cannot therefore be explained by variations of the ionization parameters, which should first of all affect the [OIII]/[OII] ratio. A possible explanation for this is discussed in Sect. 4.8.
3.3. Diagnostic line ratios
Temperature and density sensitive line ratios are summarized in Table 3. Note that only a few values are available on the nucleus because of the strong stellar continuum which prevents the measurement of faint lines.
Table 3. Temperature and density sensitive line ratios. [NeV] FIR lines from M96
3.3.1. Electron temperature
The relatively large (OIII), slightly lower (SIII) and much cooler [NII], [SII] temperatures are typical of gas photoionized by a "typical AGN", i.e. a spectrum characterized by a power law continuum with a super-imposed UV bump peaked at 50-100 eV (e.g. Mathews & Ferland 1987). As (OIII) mainly depends on the average energy of eV ionizing photons, "bumpy" spectra, which are quite flat between 13 and 54 eV, yield hot [OIII]. The lower ionization species, such as [NII] and [SII], mostly form in the partially ionized region, heated by soft X-rays, whose temperature cannot exceed 104 K due to the powerful cooling by collisionally excited Ly and 2-photon emission. The contrast between the OIII and NII temperatures can be further increased if sub-solar metallicities are adopted, because [OIII] is a major coolant while [NII] only plays a secondary role in the cooling of the partially ionized region.
An alternative explanation for the OIII, NII etc. temperature differences is to assume that part of the line emission arises from density bounded clouds. In this case there is no need to adopt a "bumpy" AGN spectrum and detailed models, assuming a pure power law ionizing continuum, were developed by Binette et al. (1996, hereafter B96). However, it should be kept in mind that (OIII)(NII) does not necessarily indicate the presence of density bounded clouds.
3.3.2. Electron density
There is a clear trend between and excitation of the species used to determine the density.
The [SII] red doublet yields densities lower than [OII] which is compatible with a single-density cloud for the following reason. If the flux of soft X-rays (200-500 eV) is strong enough, then [SII] lines are mostly produced in the X-ray heated region where the average hydrogen ionization fraction is quite low (0.1). The lines of [OII] on the contrary can only be produced in the transition region, where the ionization degree is close to unity, because of the very rapid O-H charge exchange reactions. Hence (SII)(OII) most probably indicates the presence of a strong soft-X flux, as one indeed expects to be the case for an AGN spectrum. This is also confirmed by the detailed modelling described below.
The higher densities in the fully ionized region are new results because the blue [ArIV] doublet is usually too weak in AGN spectra and the FIR [NeV] lines are only accessible with the ISO-SWS spectrometer (M96). The [ArIV] density of knot C is equal, within the errors, to that derived by [OII] thus indicating that no large variations of densities are present in this cloud.
© European Southern Observatory (ESO) 1999
Online publication: December 22, 1998