next previous
Up: Spectroscopic observations of southern galaxies


Subsections

4 Discussion

4.1 Optical emission line intensity ratios

The emission line intensity ratios at the nuclear region (Table 2) clearly indicate that it is a LINER (Osterbrock 1989), which is compatible with the suggestion of Shobbrock (1966) that this galaxy may be a Seyfert or other emission type. Véron-Cetty & Véron (1986) classified it as a Seyfert-like galaxy, a designation for either a Seyfert or a LINER. However, from their measurements, they concluded that NGC 2442 is a N galaxy, with H$\beta$ in absorption, because they found that H$\alpha$ < 1.2 $\times$ [N II] $\lambda$6584 but no other lines were present which would allow them to distinguish between a Seyfert 2 and a LINER. Being effectively a LINER, the dominant excitation mechanism should be photoionization by a continuous spectrum similar to but weaker than those present in Seyfert 2 galaxies (Osterbrock & De Robertis 1985).

The electron temperature $T_{\mathrm{e}}$ and the density $N_{\mathrm{e}}$ at the nuclear region were derived from the [N II] ($\lambda$6548 + $\lambda$6584)/$\lambda$5755 and [S II] ($\lambda$6717/$\lambda$6731) ratios, obtaining $T_{\mathrm{e}} \sim$ 14 000 K and $N_{\mathrm{e}}$ $\sim$ 530 cm-3, respectively, which are normal values for this type of objects. We might have here the situation mentioned by MB97 of seeing simultaneously the emission of a central point source (Seyfert) and of a star forming ring. Since the ring, in the H$\alpha$ maps, appears quite thick and the inclination is about 69$\hbox{$^\circ$}$(as estimated by MB97), an appreciable amount of its emission should be present in the nuclear spectrum. We shall come back later to this possibility since we find also the central ring and we believe that our double peaked spectra in the nuclear region are produced by the two H$\alpha$ sources detected by H98 and by the central ring.

  
\begin{figure}
\resizebox {8.8cm}{!}{\includegraphics{eb8111f3.eps}}
\\ end{figure} Figure 3: Spectrum of an emission region located on the major axis at 87$\hbox{$^{\prime\prime}$}$ NE from the nucleus along PA = 40$\hbox{$^\circ$}$. Dispersion 32 Å mm-1

In Fig. 3 a spectrum of the region at the NE, mentioned above, is shown. Its spectral characteristics are typical of an H II region. The internal reddening is much higher than in the nuclear region. For this region the N(O)/N(H) and N(N)/N(H) abundance ratios were obtained. The N(O)/N(H) abundance was derived from empirical calibrations from Edmunds & Pagel (1984). Because the [O II] $\lambda$3727 line lies in a noisy zone of the spectrum, the [O II] $\lambda$3727/H$\beta$ratio was calculated from the first predicting equation given by McCall et al. (1985). Assuming N(O)/N(H) = (N(O+) + N(O++))/N(H+) and N(N)/N(O) = N(N+)/N(O+), the nitrogen abundance is N(N)/N(H) = (N(O+) + N(O++))/N(O+) $\times$ (N(N+)/N(H+)). The corresponding electron temperature was deduced from the equation for N(O)/N(H) searching for the required value of $T_{\mathrm{e}}$ for the previously derived N(O)/N(H) abundance. The resulting oxygen and nitrogen abundances, N(O)/N(H) = 9.7 10-4 and N(N)/N(H) = 7.1 10-5, are 1.20 and 0.72 of the corresponding solar abundances. The ratio N(N)/N(O) = 0.07 reflects a relative deficiency of N with respect to O but it is very close to usual values in galactic emission regions (Shaver et al. 1983). Its electron temperature and density: $T_{\mathrm{e}}$ $\approx$ 6500 K and $N_{\mathrm{e}}$$\approx$ 10 cm-3, are slightly low but within the range of normal values.

4.2 Correlation with other frequencies

Our results consist of spectra at scattered points over NGC 2442, along the slit positions. They do not allow us to build contour maps for the comparison with the results of observations at other frequencies which did produce such maps. We can, however, make such a comparison along those slit positions. In particular, we were interested in the position at PA = 40$\hbox{$^\circ$}$ and through the nucleus, which we assumed to coincide with the line of nodes, because there are several important features along it. We shall start comparing the intensities.

4.2.1 Intensities along PA = 40$\hbox{$^\circ$}$

Figure 4 shows the flux densities of H$\alpha$ (our results), in erg cm-2 s-1, and of the 843 MHz (from Plate 2 of Harnett 1984) and 1415 MHz continuum (from Fig. 5.4 of H98), in mJy beam-1, and the areas of the 12CO(1-0) (from Fig. 6b of Bajaja et al. 1995) and 21 cm H I (from Fig. 5.7a of H98) velocity profiles, in K $\rm \, km\, s^{-1}$, along the line at PA = 40$\hbox{$^\circ$}$, as a function of the distance to the center. In spite of the different angular resolutions, which are evident in the widths of the central peaks, the correlation between the five components is clear. Except for H I, which shows a rather flat curve in the central region, all the curves show three peaks, one at the center, one at about 80$\hbox{$^{\prime\prime}$}$ towards the NE and a third one, much less defined, at about 125$\hbox{$^{\prime\prime}$}$ towards the SW. The differences between the distances to the NE and SW peaks are another evidence for the asymmetry of the galaxy. The smallest width for the central source is shown by H$\alpha$ with a FHMW of $\sim$16$\hbox{$^{\prime\prime}$}$, which indicates that it is not resolved by the other observations. Differences in the positions of the peak centers at the NE, and probably at the SW, may be due to the fact that the data were taken from the figures and these may have untraceable scale errors, but the intensities are supposed to be correct.

  
\begin{figure}
\resizebox {8.8cm}{!}{\includegraphics{eb8111f4.eps}}
\\ end{figure} Figure 4: Intensities as a function of the distance to the center, along a line at PA = 40$\hbox{$^\circ$}$, for the 12CO(1-0) (Bajaja et al. 1995), 21 cm H I (H98) and H$\alpha$ (this paper) lines and for the continuum at 843 MHz (Harnett 1984) and 1415 MHz (H98)

In Table 3 are listed the peak values for the three regions (NE, Center and SW) and for the five types of emission. In order to be able to compare the emissions in the CO and H I lines, the profile areas were converted to column densities using for the conversion factors the values 3 1020 and 1.823 1018 (cm2 K $\rm \, km\, s^{-1}$)-1, respectively. For CO the conversion factor is the same as the one used by Bajaja et al. (1995) although there is not a general agreement about the value that should be used. The continuum emissions are given in mJy beam-1 so, for the comparison of these intensities, observing extended sources, the values should be corrected for the beams. In our case, however, we have seen that at both frequencies, 843 MHz and 1415 MHz, the NE and central sources are not resolved. As for the southern part, the peaks in the curves are badly defined (values enclosed in brackets in Table 3) except for the H I whose peak has a FHMW similar to the one on the NE source.


  
Table 3: Peak values of intensities at different frequencies

\begin{tabular}
{\vert ccccc\vert}
\noalign{\smallskip}
\hline
\noalign{\smallsk...
 ...cm$^{-2}$\space seg$^{-1}$
& 22.5 & 60.5 & [7] \\ \noalign{}
\hline\end{tabular}


  
Table 4: Intensity ratios

\begin{tabular}
{\vert ccccc\vert}
\noalign{\smallskip}
\hline
\noalign{\smallsk...
 ...$\space s erg$^{-1}$\space & 2.1 & 0.4 & [3.7] \\ \noalign{}
\hline\end{tabular}

Some of the ratios between the emission intensities are listed in Table 4. We used the flux densities quoted in Table 3, for the continuum emissions at 843 and 1415 MHz, to estimate the spectral indices which are also quoted in Table 4. The ratio is larger at the center than at the NE and SW but all the spectral indices are indicating steep gradients in the flux densities for the three regions between both frequencies. As mentioned in Sect. 1, Harnett (1984) derived for the whole galaxy a spectral index value of 0.92 $\pm$ 0.08. All these values indicate a large pre-eminence of non-thermal radiation in the galaxy. The column density ratios for the lines also differ greatly in the three regions. The H2/H I column density ratio is higher at the center, as in the case of the continuum, but the ratios H2/H$\alpha$ and H I/H$\alpha$ are much lower at the center. Evidently, the conditions for the conversion of atomic to molecular Hydrogen and from this to stars at the center are quite different from the conditions at the other regions.

4.2.2 The rotation curve

The velocities derived from our observations are indicated in Fig. 5 superposed on the CO velocity field (Fig. 6c in Bajaja et al. 1995). There is, in general, a good agreement between both if we take into account the errors. There are, however, differences which are due to the much better angular resolution of the optical observations and to the different types of emission sources. This is particularly evident at the three regions mentioned in the previous section.

  
\begin{figure}
\includegraphics [width=8.8cm,clip=]{eb8111f5.eps}
\\ end{figure} Figure 5: Optical heliocentric radial velocities (small numbers at the positions indicated with dots) in $\rm \, km\, s^{-1}$ and CO velocity field with contour levels in $\rm \, km\, s^{-1}$ (large numbers), from Bajaja et al. (1995)

In Fig. 6 the optical velocities along the PA = 40$\hbox{$^\circ$}$ are shown superposed on the position-velocity diagram obtained with the CO spectra (Fig. 6d in Bajaja et al. 1995). The optical and the radio rotation curves, as derived from this figure, show similar velocity gradients in the central part in spite of the very different angular and velocity resolutions. The optical gradient, within 12$\hbox{$.\!\!^{\prime\prime}$}$5 from the center, is 17 $\rm \, km\, s^{-1}$ per arcsec and it is approximately constant along its length. The velocities at the ends of this part of the velocity curve are 1262 and 1694 $\rm \, km\, s^{-1}$ and, from its symmetry, we adopt for the systemic velocity the value of 1478 $\rm \, km\, s^{-1}$ (the errors of these velocities are $\pm$ 4 $\rm \, km\, s^{-1}$). This feature may be associated either to a disk in solid rotation or to a ring rotating at 216/sin(i) $\rm \, km\, s^{-1}$. Our spectra, along a line, do not allow us to have a two-dimensional picture of the object. MB97 derived similar parameters for a ring for which they obtained a radius of 8$\hbox{$^{\prime\prime}$}$ and a rotation velocity of 220/sin(i), and for the systemic velocity a value of 1475 $\rm \, km\, s^{-1}$. All these values are practically the same as ours since the differences are within the errors. H98, however, derived for the systemic velocity, from the H I velocity field, the value of 1431 $\rm \, km\, s^{-1}$ which is much lower than the velocities mentioned above.

  
\begin{figure}
\includegraphics [width=8.8cm]{8111f6.eps}
\\ end{figure} Figure 6: Optical heliocentric radial velocities (dots) and CO velocity position diagram (from Bajaja et al. 1995) along the PA = 40$\hbox{$^\circ$}$.Contour levels in the latter are 12 to 108 mK, at intervals of 12 mK, of corrected antenna temperature

Our velocity for the galaxy, with respect to the centroid of the Local Group, is 1208 $\rm \, km\, s^{-1}$ so, using a Hubble constant H0 = 75 $\rm \, km\, s^{-1}$ Mpc-1, the distance to NGC 2442 would be 16.1 Mpc and the radius of the central disk or ring would be $\sim$ 1 kpc. MB97 estimated, from the geometry of the central ring, an inclination angle of 69$\hbox{$^\circ$}$. Using this value the rotation velocity would be 231 $\rm \, km\, s^{-1}$ and the mass within the ring $\sim$ 1.2 1010 $M_{\hbox{$\odot$}}$.

The optical rotation curve, as determined from our observations, extends up to 137$\hbox{$^{\prime\prime}$}$ (10.7 kpc at 16.1 Mpc) where the observed velocity, with respect to the center, is 275 $\rm \, km\, s^{-1}$.From the geometry of the optical image of NGC 2442 we may not assume a value as high as 69$\hbox{$^\circ$}$ for the inclination angle of the outer parts of the galaxy. Although, most probably, it is not 24$\hbox{$^\circ$}$ (Bajaja & Martin 1985; Baumgart & Peterson 1986) either, in the absence of a better value we adopt this one for the mass estimation. In consequence, the mass within 10.7 kpc would be about 11 1011 $M_{\hbox{$\odot$}}$. From these values we may estimate then that, roughly, the mass within the ring is about 1% of the total mass, i.e. approximately proportional to the areas. The use of very different values of i for difeerent distances from the center is a consequence of the large distortions present in NGC 2442.

4.2.3 Velocities in the nuclear region

Our optical lines in the nuclear region show, as mentioned in the previous section, double velocity components. The spectra at 3$\hbox{$^{\prime\prime}$}$ NE and 5$\hbox{$.\!\!^{\prime\prime}$}$1 SW also show two velocity components, with about the same separation as in the nucleus, but with some differences: a) the mean velocities of the spectra agree with the central rotation curve; b) the intensities of the velocity components at the NE are lower than at the nucleus but higher than at the SW, and c) while the intensities of both velocity components are approximately the same at the nucleus, they are different in the spectra at both sides of it and their intensity ratios are inverse respect to each other, i.e. in the spectrum at the NE the component with the shortest wavelength has the largest intensity and the opposite happens in the SW spectrum.

The explanation for these features has been searched for in the objects already mentioned: the two small H$\alpha$ regions detected by H98 in the nuclear region and the central ring found by MB97 and by us. The H$\alpha$ regions are about 2$\hbox{$^{\prime\prime}$}$ in diameter and their peaks are separated by about 4$\hbox{$.\!\!^{\prime\prime}$}$5. The regions are, apparently, symmetrically positioned at both sides of the nucleus along a line at a PA of about 97$\hbox{$^\circ$}$ so the angle between this line and our slit (at PA = 40$\hbox{$^\circ$}$) is 57$\hbox{$^\circ$}$. The extension of the line seen by the 3$\hbox{$.\!\!^{\prime\prime}$}$3 aperture is 4$\hbox{$^{\prime\prime}$}$ so a large part of the emission of the H$\alpha$ regions should have been detected and be present in the spectra on the nuclear region. The regions are displaced by 2$\hbox{$.\!\!^{\prime\prime}$}$2 along the slit. Our H$\alpha$ spectrum at the nucleus of the galaxy encompasses both H$\alpha$ regions symmetrically positioned but the two spectra at each side, to the NE and to the SW, cross them at different distances, i.e. with different weights.

About the central ring, the H$\alpha$ maps of MB97 and of H98 show emission distributed over an oval region, centered in the nucleus and with the major axis at a PA of about 40$\hbox{$^\circ$}$. If it is a ring then it is quite thick. The H$\alpha$ velocity field of H98 shows clearly the rotation of this central feature and indicates that the PA of the line of nodes is the same as the PA of the major axis of the oval region. From the axis ratio of this region, MB97 derived an inclination angle of 69$\hbox{$^\circ$}$.The two H$\alpha$ regions and the points corresponding to our spectra in the nuclear region are inside this central fast rotating thick ring so, besides encompassing the two H$\alpha$ regions, our three spectra should cover also the emission from the ring.

Therefore, if the intensities of both H$\alpha$ regions are approximately the same, we would expect the following effects on our H$\alpha$ spectra: a) the two velocity components at the nuclear position should be similar in intensity and displaced in velocity by the difference in the projected radial velocities of the two H$\alpha$ regions; b) since the spectrum at the NE is closer to the two H$\alpha$ regions its integrated line intensity should be larger than in the spectrum at the SW; c) in each spectrum, the velocity component corresponding to the closest H$\alpha$ region, should be the strongest; d) the velocity difference for the two components should be approximately the same for the three spectra, and e) in each of our spectra, the material of the ring which is closer to its position will be seen with higher weight, pushing the mean velocity accordingly.

Qualitatively, the five effects are present in our spectra. Even quantitatively we have been able to reproduce the peak intensities on each of the three spectra assuming, for the sensitivity along the slit, a gaussian with a standard deviation $\sigma$ = 3$\hbox{$.\!\!^{\prime\prime}$}$1 $\pm$ 1$\hbox{$^{\prime\prime}$}$.We might say then that the interpretation made above could be acceptable. There is a problem, however, with the central velocities given by H98 for the two H$\alpha$ regions: 1367 and 1458 $\rm \, km\, s^{-1}$,with an average of 1412.5 $\rm \, km\, s^{-1}$ and a difference of 91 $\rm \, km\, s^{-1}$. We have, for the velocity components in the spectrum at the nuclear region, 1373 and 1518 $\rm \, km\, s^{-1}$, with an average of 1445.5 $\rm \, km\, s^{-1}$ and a difference of 145 $\rm \, km\, s^{-1}$. The separations between the component velocities differ by 54 $\rm \, km\, s^{-1}$ and the averages by 48 $\rm \, km\, s^{-1}$, differences which are much larger than the errors in our velocities. H98 also obtained, for the systemic velocity of the galaxy, 1431 $\rm \, km\, s^{-1}$ which is 47 $\rm \, km\, s^{-1}$ and 44 $\rm \, km\, s^{-1}$ lower than the values derived by us and by MB97, respectively. We do not know the reasons for these differences. We are inclined to think that our values are correct but the errors involved are much too high and strongly dependent on the procedure used for deriving the velocities.

There are further questions related to the two H$\alpha$ regions detected by H98. Being inside the central ring, it is natural to assume that they participate of its rotation and that the velocity difference is due to the difference in the projected rotational velocities. The question is then: why both regions are along a line with a PA of 97$\hbox{$^\circ$}$ and not at the edges of the ring, along the line of nodes? Furthermore, a difference of 145 $\rm \, km\, s^{-1}$ and a separation of 4$\hbox{$.\!\!^{\prime\prime}$}$5 on a plane inclined 69$\hbox{$^\circ$}$, would imply that the H$\alpha$ regions are rotating with a velocity of about 345 $\rm \, km\, s^{-1}$.This velocity is 50% larger than the velocity of the ring. With a velocity separation of 91 $\rm \, km\, s^{-1}$ (as given by H98) that velocity would become 216 $\rm \, km\, s^{-1}$ but the distance to the center would be then twice the value measured in the H$\alpha$ map of H98. The possibility that the two H$\alpha$ regions are on the edges of a ring on a different plane, with the line of nodes at 97$\hbox{$^\circ$}$ and with an unknown inclination angle, is not supported by the emission distribution in the H$\alpha$ maps. There might be still another possibility: two H$\alpha$ regions with expanding movements. It should be recalled that Sersic & Donzelli (1993) specified for the bar also a PA of 97$\hbox{$^\circ$}$. Might be there a connection? The questions are open.

We have to consider how the results described above could affect the previous conclusion, derived from the intensity ratios at the nuclear region, that it is a LINER. As was said then, this would imply that the dominant excitation mechanism is photoionization by a power-law continuum similar to but weaker than in Seyfert 2 galaxies. However, from the fact that our nuclear region spectrum sees the H$\alpha$ regions, and that H$\beta$ is also detected in absorption, it would be possible that this spectrum is the result of the integration of the emissions from a central point Seyfert source and from the star formation regions. This combination, which weakens the effect of a pure Seyfert source, would give as a result the characteristics of a LINER.

4.2.4 Velocities in the disc

The CO Velocity-Position diagram in Fig. 6 shows a tendency for double velocity components at both ends of the line at PA = 40$\hbox{$^\circ$}$. The optical velocities along this line, in those regions, seem to follow the CO components with the highest absolute velocities and the lowest intensities. It is not clear what this double velocity would mean. A warp could easily produce this kind of effect along the minor axis but it would have to be very strong to be visible along the major axis. More data are necessary for modelling it.

Figure 5 shows the velocities obtained for ten points along the external northern arm and it may be found that the velocities, at the far end, are between 1360 and 1370 $\rm \, km\, s^{-1}$. Since the errors are of the order of 3 to 10 $\rm \, km\, s^{-1}$, the conclusion is that these velocities are not compatible with a normally rotating inclined disc, with a PA for the line of nodes of 40$\hbox{$^\circ$}$ and a systemic velocity of 1478 $\rm \, km\, s^{-1}$, because the points are then close to the minor axis. There might be several possible explanations for the abnormal velocities: a) a smaller PA; b) a velocity component normal to the galactic plane (warp?), c) a radial component along the plane (expansion?), etc, some of which might be ocurrying simultaneously. This is another evidence for the distortion of the galaxy. Almost certainly there are no single values for the PA of the line of nodes and for the inclination that may be applied to the whole galaxy.

Interaction with other galaxies might be playing a fundamental role in determining the matter distribution and the velocities on this galaxy. There are in fact several galaxies around that might be interacting with NGC 2442, but none of them is so close and with such a clear evidence of interaction as to be defined as the candidate. The consequence is that different authors suggest different galaxies: NGC 2434 (Elmegreen et al. 1992), AM 0738-692 (MB97). H98 has considered several of the galaxies in the neighbourhood of NGC 2442 without pointing to any one in particular as the most probable candidate. The number of free parameters involved in the modelling of this galaxy requires many more data than those available for this work. The model, however, is largely necessary before any discussion about the dynamics of this galaxy may be attempted. MB97 considered the encounter with a nearby galaxy modelling the evolution of the stellar and gaseous components. H98 used two programs within the AIPS package, GAL and ROCUR, for deriving some basic parameters and then to make a three dimensional model of the galaxy using a constant inclination of 42$\hbox{$^\circ$}$ and fitting the PA. Both models are important but still partial contributions to the knowledge of NGC 2442.


next previous
Up: Spectroscopic observations of southern galaxies

Copyright The European Southern Observatory (ESO)