4 Global properties and discussion

4.1 Near-IR morphologies

The structural parameters of the 8 galaxies with embedded structures are summarized in Table 3. Obviously no reliable statistics could be performed on such a restricted and biased sample, but it enriches previous studies. Note that morphological parameters are difficult to compare with those of the literature since each author uses various definitions and computation methods. Thus we will make a comparison only with Paper II.

For the B+B category (NGC3393, ESO215-G031, ESO320-G030, ESO443-G017), ranges of bar lengths and luminosity ratios ( $2.7\!\leq\!\beta_{12}\!\leq\!7.1$, $1.5\!\leq\!\gamma_{12}\!\leq\!2.4$ ) are shifted toward lower value than in Paper II. ( $4.0\!\leq\!\beta_{12}\!\leq\!13.4$, $1.8\!\leq\!\gamma_{12}\!\leq\!7.5$). Thus our new data tend to decrease the mean $\beta_{12}$ and $\gamma_{12}$ from 7.2 and 3.6down to 6.5 and 3.0 respectively. Those four double-barred galaxies confirm that no preferential angle are found between the two bars, so that nested bars are dynamically decoupled. Moreover while there is a lack of known objects with low $\theta_{12}$, two galaxies have nearly aligned bars ( $\vert\theta_{12}\vert\!\approx\!6^{\circ}$ and $\vert\theta_{12}\vert\!\approx\!14^{\circ}$, in ESO215-G031 and NGC3393 respectively).

Some galaxies (ESO264-G036, NGC3393, NGC4939, ESO323-G077, NGC5135, NGC 6221) exhibits significant differences in their isophote shapes (PA and e) between the two NIR bands. Quillen et al. (1996) have suggested that non-circular stellar motion and radial colour gradient (resulting from the stellar population gradient), could lead to such ellipticity differences between the two bands. But, in the present case, the ellipticity deviation occurs in the central region where some of the Quillen et al. assumptions are not respected; principally the colour gradient could be strongly affected by dust lanes and/or star formation. Moreover, as we have seen in the case of NGC3393, isophote shape deviations on small-scale could also result from seeing discrepancy between the two bands.

4.2 Behaviour of $\mu _{J}-\mu _{K'}$ profiles

Figure 3: Central $\mu _{J}-\mu _{K'}$ profile behaviour versus star formation intensity (see text). The whole sample is presented in Table 4. Seyfert stands for both Seyfert and/or Liner objects

Whereas all $\mu _{J}-\mu _{K'}$ profiles are roughly constant outside the inner region ( $\mu_{J}-\mu_{K'}\!\approx\!1\,{\rm mag}\hbox{$^{\prime\prime}$ }^{-2}$ at ), all of them increases toward the center or at least are indicative of redder central J-K' colour (if we exclude NGC3393, see Sect. 3.1.3). These $\mu _{J}-\mu _{K'}$profiles are qualitatively and quantitatively very similar to those published in Hunt et al. (1997).

Could the central activities give rise to such central NIR colour profile features? For studying this issue, we have reduced the $\mu _{J}-\mu _{K'}$ profile behaviour to one single parameter $\Delta$(J-K'). It is the difference between the inner and the outer J-K' colour. This differential colour has the definite advantage of being independent of the photometric calibration and corrections; in particular it is independent of the airmass, the Galactic extinction or the K correction. The inner J-K' colour is defined to be the difference between $\mu_{J}$ and $\mu_{K'}$, both integrated within the fitted ellipse $a_{0}\,=\,$1.5 $\hbox{$^{\prime\prime}$ }$ semi-major axis. The outer J-K' colour is computed between the semi-major axis a₁and $1.2\,a_{1}$, where $a_{1}\,=\,0.1D_{25}$ and D₂₅ is taken from de Vaucouleurs et al. (1993). Thus this outer aperture has a linear spatial extent roughly equivalent for all bright galaxies, regardless to its distance. Concerning the inner NIR colour, a₀has to be as short as possible in order to probe a region affected as little as possible by activities other than the nuclear ones (e.g. rings of star forming regions). Ideally a₀ should also be scaled to D₂₅ to make it consistent for all our sample, but we consider that the smallest significant aperture must at least have two pixels width ( $\sim a_{0}$ = 1pixel). $a_{0}\,=\,$1.5 $\hbox{$^{\prime\prime}$ }$respects this requirement and avoids, for our sample, the problem of contamination by nuclear rings. However, it must be used carefully for more distant objects.

If the central profile behaviour is linked to the Seyfert activity, $\Delta$(J-K') should be different for Seyfert and non-Seyfert population. The hypothetic link with starburst activity is less straightforward to enlighten since all disc galaxies are forming stars (and not only in the central region). As the starburst classification is almost arbitrary, we prefer to study the behaviour of $\Delta$(J-K') as a function of the star formation rate (SFR). The choice of a SFR estimator is dicussed below.

As the dust absorption peaks in the UV, the dust (mainly the big grains component of Désert et al. 1990 model) in thermal equilibrium with the radiation field reaches a temperature which mostly depends on the UV-flux (in fact regardless to the nature of the heating sources). Thus the dust temperature is well suited to study the intensity of the activity which dominates the UV-emission. It also has the advantage of probing these source intensities without requiring spectroscopic data which are quite sparse in the archives. In active star-forming galaxies, where UV-emission is dominated by massive stars, this thermal cold dust ( $\approx\!30$K, Siebenmorgen et al. 1999) mostly peaks around the $50-100~\mu$m spectral region 2000and references therein. As a consequence, the 60/100$\mu \rm m$ fluxes ratio is very sensitive to the cold dust temperature. Another reason why $\log(S_{60}/S_{100})$ may be chosen as tracer of star formation is that, contrary to the mid-IR ( $\approx\!10-40\,\mu$m), the transiently heated very small grains of dust do not contaminate the $60-100~\mu$m cold dust emission Cesarsky & Sauvage (2000). Unfortunately the low angular resolution of IRAS data ( $2\hbox{$^\prime$ }$ at $100~\mu$m) only allows a global estimation of the SFR intensity of a galaxy, regardless to its spacial distribution.

In view of the previous considerations, a link between Seyfert and/or starburst activity and NIR colour profile can be probed in the diagram $\Delta$(J-K') versus $\log(S_{60}/S_{100})$ plotted in Fig. 3. As already mentioned in Sect. 2.4, data from the literature were added in order to obtain more reliable statistics. The whole sample has Hubble Type ranging from $\approx\!-3$ up to $\approx\!7$ (SAB0-SBc), and has its IR properties summarized in Table 4.

   

Table 4: Main properties of the sample plotted in Fig. 3

Galaxies	${\rm log}(\frac{S_{60}}{S_{100}})$	$\Delta$(J-K')	J-K'(1.5 $\hbox{$^{\prime\prime}$ }$)	Activity	Nested
		[mag]	[mag]		Struct.
(1)	(2)	(3)	(4)	(5)	(6)
NGC573a	-0.30	0.30	1.56	Sy2	B+B
NGC1068a	-0.11	1.42	2.65	STB/Sy2	B+nS?
NGC2110a	-0.15	0.62	2.03	STB/Sy2
NGC2992a	-0.32	0.36	1.94	STB/Sy2
NGC3393a	-0.22	0.14	1.09	STB/Sy2	B+B
NGC4253a	-0.02	1.22	2.89	STB/Sy1.5
NGC4388a	-0.21	0.76	2.27	STB/Sy2
NGC470b	-0.28	1.41	1.47	STB	B+T
NGC4314b	-0.30	0.20	1.20	STB/LIN	B+B
NGC6951b	-0.44	0.14	1.23	Sy2	B+T
NGC7098b	-0.67	-0.09	1.18		B+T+B
NGC7479b	-0.31	0.50	1.72	STB/Sy2	B+T
ESO215c	-0.24	0.34	1.26	STB	B+B
ESO264c	-0.37	0.60	1.47
ESO320c	-0.11	0.55	1.34	STB	B+B
ESO323c	-0.21	0.64	1.85	STB/Sy1	B+T
ESO374c	0.02	0.63	1.58	STB	int.
ESO443c	-0.13	0.38	1.34	STB	B+B
ESO508c	-0.40	0.20	1.13	Sy2	B+dB?
NGC4903c	-0.39	0.23	1.17	Sy2
NGC4939c	-0.59	0.25	1.42	Sy2
NGC4941c	-0.49	0.24	1.18	Sy2
NGC5135c	-0.28	0.15	1.22	STB/Sy2	B+nS
NGC5643c	-0.37	0.04	1.35	Sy2
NGC6221c	-0.37	0.64	1.49	Sy2	T+B
NGC6300c	-0.46	0.81	1.79	Sy2

^a Alonso-Herrero et al. (1998).
^b Paper II.
^c Our sample.
(2): S₆₀, S₁₀₀ are non-colour corrected fluxes in the 60$\mu$m and 100$\mu$m IRAS-bands, from Catalogued Galaxies + QSOs observed in IRAS Survey, Vers.2 (IPAC 1989),
(3): defined in Sect. 3.2.2,
(4): J-K' integrated within the fitted ellipse with a 1.5 $\hbox{$^{\prime\prime}$ }$ semi-major axis,
(5): type of activity; (Sy, LIN)$\,=\,$(Seyferts, LINER) from respective sources/STB$\,=\,$starbursts ,
(6): nested structures; symbols are defined in Table 3.

As mentionned above, $\Delta$(J-K') is independent of the photometric calibrations. Thus its uncertainties mainly come from the readout noise of the detector and are only marginal ($\leq 1\%$) compared to the typical conservative error on $\log(S_{60}/S_{100})$ ( $\approx 10-30\%$). This later uncertainty is deduce from the errors on the individual 60 and 100 $\mu$m IRAS fluxes given in the IRAS Point Source Catalogue (see e.g. Young et al. 1986) which are generally $5-15\%$.

Independently of the presence of a Seyfert nucleus, the range of $\Delta$(J-K') tends to increase from -0.1-0.9 up to 0.3-1.5 as $\log(S_{60}/S_{100})$ increases. Despite the efforts we made to extend the sample, it is still not large enough to unambiguously point out a possible correlation in such a plot. Using NIR colour profile tables given in Peletier et al. (1999), the same trend is observed for 29 objects among their Seyfert sample. As we do not have the original NIR data, these new points are not reported in Fig. 3 and cannot be directly compared to ours. But as they could only amplify the scatter of the $\Delta$(J-K') vs. $\log(S_{60}/S_{100})$ plot, they could be used to quantify the robustness of the observed trend: while the slope and the zero point of the regression remain nearly the same in both cases, the correlation coefficient is found to be 0.56 for the sample plot in Fig. 3, and 0.55 if the 29 Peletier et al. (1999) objets are added. This low but nearly constant value of the correlation coefficient, suggests that the link between $\Delta$(J-K') and $\log(S_{60}/S_{100})$ is marginally linear but real. Figure 3 looks very similar if one plots either the earliest-types ($T\!<\!3$) or the latest-types ($T\!\ge\!3$), so that it is not a Hubble sequence effect. Thus the integrated FIR colour is related to the nuclear NIR colour (scaled to the disc colour): significant starburst galaxies have central J-K'0.3-1.5 mag redder than the disc. Hunt et al. (1999) found that Seyfert 1 and nuclear starburst galaxies have the bulge J-K colour 0.1 mag redder than the disc, whereas Seyfert 2 bulges have the same colour as the disc. Of course these results could not directly be compared to ours but both are compatible: as the bulge scale-length is always larger than the inner photometric aperture we used, the contrast between their inner and outer colour naturally tends to be lower than ours.

4.3 Explanation attempt of the $\Delta$(J-K') excess

The interpretation of the $\Delta$(J-K') red excess observed in starburst galaxies is far from being straightforward. But the strong correlation between $\log(S_{60}/S_{100})$ and the nuclear H$\alpha$ luminosity density found among Markarian galaxies in Mazzarella et al. (1991) might be a guideline and suggests that, for their sample, the bulk of FIR luminosity arises from a nuclear starburst (young enough to still ionize the surrounding material).

Could the red nuclear NIR colour we observed result from a recent nuclear starburst? Apart from the usual contribution of the old stellar population, mainly giants, the J-K' colour depends at least on three factors^d: 1) the amount of dust (extinction is higher in J than in K'), 2) the gas and/or dust temperature (James & Seigar 1999 claimed that hot dust contributes to K'), and 3) the respective contribution of various types of young stars in star forming regions (K' luminosity increases much more than J one if K supergiants dominate; J increases slightly more than K' if OB stars dominate, James & Seigar 1999). Thus, in luminous and young star-forming regions J-K' naturally tends to increase, so that the trend observed in Fig. 3 may indicate that FIR luminosity is essentially produced by a nuclear starburst.

None of the three factors having an effect on J-K' could alone be responsible for the whole variations of $\Delta$(J-K') through our sample. Indeed, using Leitherer et al. (1999) starburst model, we are unable to explain J-K' differences larger than $\approx\!0.5$ mag between an old stellar population and a recent starburst. The maximum $\Delta$(J-K') we can obtain is $\approx\!0.5-0.6$ mag for a continuous SFR of 10Myr old with a Salpeter IMF and a solar metallicity, and assuming a J-K' $\!\approx\!0.5\,$mag for the old stellar population.

So additional factors that could also locally change J-K' have to be considered, for instance: the 150-170 K gas component associated with AGN and starburst founded with ISO Sturm et al. (1996) tends to increase J-K'. The non-thermal continuum of PAHs or VSG likely contributes more to the K'-band than the thermal emission of the 150-170 K gas Rouan et al. (1996). The stellar and interstellar metallicity gradient is another parameter, which effect on $\Delta$(J-K') could not accurately be estimated without having previously disentangled the age-dust degeneracy.

Thus, a general explanation about the trend observed in Fig. 3 cannot be stated. It requires a careful study of the gas and dust properties and of the stellar population in individual objects, for which additional data are essential. For that purpose, H2 and Br$\gamma$ NIR narrow-band imaging of ESO215-G031 and NGC3081, two double-barred galaxies, have already been performed and will be used to carry on our investigations in a forthcoming paper.