2 Observations, data reduction and analysis

The sample of galaxies is presented in Table 1. All Seyfert 2 galaxies have been selected from the Véron's catalog Véron-Cetty & Véron (1993) with the following criteria: V magnitude less than 15, not or weakly interacting, and not too inclined. Galaxies with an IRAS colour are considered as starbursts (see Sect. 3.2.2) and have been picked out from Rowan-Robinson & Crawford (1989). Despite its "non-activity'', ESO264-G036 has also been observed because it has $\log(S_{60}/S_{100}) \!\approx\! -0.37$, and its starburst component defined in Rowan-Robinson & Crawford (1989) suggests a transition case between starburst and non-starburst galaxies.

A visual inspection of these objects was performed to obtain the final sample. We do not make any attempt to make our selection criteria free of bias. For instance, our sample is obviously biased towards the search for nuclear structure.

2.1 Observations and reductions

The observations were carried out on April 7-9, 1997, with the ESO/MPI 2.2 m telescope at La Silla; the conditions were photometric with a FWHM seeing ranging from 1.2 to 1.6 $\hbox{$^{\prime\prime}$ }$. The infrared imaging camera IRAC-2b was equipped with a 256$\times$256 NICMOS-3 array with a pixel size of 0 $\hbox{$.\!\!^{\prime\prime}$ }$507pixel^-1, giving a field of view of 129 $\hbox{$^{\prime\prime}$ }$$\times$129 $\hbox{$^{\prime\prime}$ }$. For each object, detector saturation was avoided by taking a series of exposures on the object interspersed with sky exposures. This procedure of alternating between the source and the sky was repeating until the total integration time in Table 1 was reached.

Each frame was cleaned from cosmic rays as well as from cold and hot pixels. Flat-field frames were obtained from exposures taken each night on a uniform illuminated blank screen (dome flat-field). After stars were removed from sky frames, the sky background was computed using the mean of the two sky frames gathered before and after a science frame. Galaxy frames were then sky subtracted and divided by the normalized dome flat-field.

As they follow the rapid variation of the sky structures, mean sky frames were preferred to median sky frames (median of all sky frames taken over the observation of one galaxy). As they are not affected by the intrinsic time-dependent unflatness of the sky, dome flat-fields were preferred rather than median sky flat-fields (median of all sky frames taken over the night). These choices have been confirmed by the flatness of the images, the quality of the background substraction and the photometric calibration.

Finally, all frames of one galaxy were co-added to create a time-cumulated science frame. As the telescope was moved between two science expositions, galaxy frames had to be shifted with regard to reference points; they were defined as the peak of a Gaussian adjusted to the intensity profile of stars present in all frames of a same object.

2.2 Photometric calibrations

   

Table 2: Aperture photometry comparison between our study and published magnitudes $\Delta J$ and $\Delta K'$

Galaxies	Aperture	$J_{\rm our}$	$K'_{\rm our}$	$\Delta J$	$\Delta K'$
	[ $\hbox{$^{\prime\prime}$ }$]	[mag]
NGC3393 (1)	15	11.3	10.3	0.31	0.20
	30	10.8	9.8	0.31	0.16
NGC4941 (2)	14	11.2	10.2	-0.21	-0.13
NGC5135 (3)	12	11.3	10.0	-0.07	0.06
	34	10.5	9.4	-0.28	-0.08
NGC5643 (4)	34	10.2	9.1	0.03	0.07
	51	9.9	8.8	0.11	0.20
NGC6221 (5)	34	9.8	8.7	-0.09	-0.03
	51	9.4	8.3	-0.12	-0.04
Mean algebraic difference				0.00	+0.05
Mean absolute difference				0.17	0.11
Root mean square deviation (rms)				0.21	0.12

(1): from Alonso-Herrero et al. (1998);
(2): from Dultzin-Hacyan & Benitez (1994);
(3), (4), (5): from Glass & Moorwood (1985).

Frame calibration was achieved by observing infrared standards from Carter & Meadows (1995) during each night (2 on April 7th and 9th, and 3 on April 8th). The zero points of each night are all compatible within their rms error. We have used the zero point obtained from the 3 nights put together; that way the rms error in the determination of the photometric zero points was $\approx\!0.04$ (J) and $\approx\!0.03$ (K'). In order to check our calibration, we have compared our results with aperture photometry taken from literature (see Table 2 and Fig. 1). Since the difference between K' and K is weak and subject to uncertainties Wainscoat & Cowie (1992), we have directly compared K'with K photometry. The weak mean algebraic difference (see Table 2) reveals very little systematic deviation between our measurements and the published ones. The K' case may reflect the slight difference between K' and K; Wainscoat & Cowie (1992) found that, for a sample of 16 A stars or M dwarfs, the K' filter has a zero point which is 0.03 - 0.04 mag fainter than the K one. But they also emphasize that this zero point departure may differ for a wider population of stars. Since there is very little systematic deviation and with regard to the mean absolute differences and rms, our calibrations are accurate enough in the context of this morphological study.

Figure 1: Comparison with published data of aperture photometry for the objects listed in Table 2. Squares are for the J-band and triangles for the K'-band

As the airmass correction obtained from mean atmospheric extinction coefficient for this site ( $a_{J}\!=\!0.08$, a_K' = 0.11 mag/airmass) does not improve our calibration check, it was not applied. No attempt was made to take the Galactic extinction and the K-correction into account.

2.3 Surface photometry analysis

Ellipses were fitted to isophotes of the whole sample in both J and K'bands. As shown by various authors (e.g. Paper I), this technique provides good qualitative and quantitative estimations of the shape of embedded structures. As we want to observe the mean behaviour of isophotes, field stars and regions of intense emission (e.g. giant star forming regions) were flagged on galaxy frames before ellipse fitting. The program used is described in Paper I (and references therein). Each ellipse is characterized by:

• its semi-major axis length, a [ $\hbox{$^{\prime\prime}$ }$];
• its ellipticity e, $e\!=\!1-b/a$ where b is the semi-minor axis;
• its intensity converted into surface brightness $\mu$, [mag $\hbox{$^{\prime\prime}$ }$^-2];
• its position angle, PA [$^{\circ}$], which is the angle between the semi-major axis and the North. This angle was defined to match the conventional notation: North is up, East at the left. Angles are measured from North and are positive in the anti-clockwise direction modulo 180$^{\circ}$.

Thus, in each band, $\mu (a)$, e(a), and PA(a) profiles are derived, with a increasing by a factor 1.01 between each point. In addition, "differential profiles'' ( $\mu _{J}-\mu _{K'}$, e_J-e_K', and PA_J-PA_K') was computed as the difference between profiles in each band at the same semi-major axis. For that purpose individual profiles were linearly interpolate between each point.

2.4 HST and literature data

Because of their high spatial resolution, HST frames are useful to visually detect and/or confirm the presence of potential central asymmetries. Thus, when available, WFPC2 and/or NICMOS calibrated frames have been obtained from the Hubble Data Archive. The planetary camera CCD of WFPC2 instrument generally gave images of the galaxy centers in the F606W filter, with a pixel size 0 $\hbox{$.\!\!^{\prime\prime}$ }$0455 and a field of view of 37 $\hbox{$^{\prime\prime}$ }\times 37 \hbox{$^{\prime\prime}$ }$. NICMOS frames were gathered with the F160W filter ($\!\approx$H band) and have a pixel size of 0 $\hbox{$.\!\!^{\prime\prime}$ }$075/pixel and a field of view of 19 $\hbox{$.\!\!^{\prime\prime}$ }$2 $\times$ 19 $\hbox{$.\!\!^{\prime\prime}$ }$2.

In Sect. 3.2.2, where the behaviour of the NIR colour profile versus IRAS colour is studied, data from the literature have enriched our sample. Galaxies for which photometric J, K' frames and IRAS data were available were found in Alonso-Herrero et al. (1998) and in Paper II. Hence J and K' frames (with different pixel sizes 0 $\hbox{$.\!\!^{\prime\prime}$ }$286pixel^-1 or 0 $\hbox{$.\!\!^{\prime\prime}$ }$143pixel^-1) of 7 galaxies were obtained from the former study, as well as J and Kframes (with different pixel sizes 0 $\hbox{$.\!\!^{\prime\prime}$ }$49 pixel^-1, 0 $\hbox{$.\!\!^{\prime\prime}$ }$6 pixel^-1, or 0 $\hbox{$.\!\!^{\prime\prime}$ }$9 pixel^-1) of 5 galaxies from the latter. Ellipse fitting was performed on those galaxies in exactly the same way as for our sample.

2.5 Terminology

We use in this paper the following terminology:

• Primary bar: a bar with a scale-length of several kpc;
• Secondary bar: a bar with a scale-length around one kpc or less, and nested within a primary bar;
• Nuclear bar: the same feature as the secondary bar but with no evidence for a primary bar;
• $a_{\rm p}$ ($a_{\rm s}$), $e^{\rm max}_{\rm p}$ ( $e^{\rm max}_{\rm s}$), PA$_{\rm p}$ (PA$_{\rm s}$) are respectively the length, the maximum ellipticity, and the position angle of the primary (secondary) bar;
• $\theta_{12}$ is the angle between the two bars. Positive values are for leading secondary bars whereas negative values are for trailing secondary bars (with respect to the primary bar);
• $\beta_{12} \equiv a_{\rm p}/a_{\rm s}$;
• $\gamma_{12} \equiv L_{\rm p}/L_{\rm s}$ where $L_{\rm p} {\rm\ and\ } L_{\rm s}$ are the luminosity of the primary and secondary bars.

All the definitions above but the nuclear bar follow those of Papers I and II, and all the parameters have been computed in the same way as in Paper I.