Astron. Astrophys. 363, 93-107 (2000)

3. Results

3.1. The distribution of the ¹²CO emission

3.1.1. Morphology and central kinematics

The ¹²CO distribution is displayed in Fig. 1a. It extends in one continuous structure over the whole length of the bar, with a number of secondary peaks along the bar axis. The total ¹²CO flux in the map is 380 Jy km s^-1. The distribution has a central concentration with a (deconvolved) FWHM size of ( pc at D=32 Mpc). This structure holds about 30% of the total CO flux of the bar. The peak position is at and , coincident to within 1 0 with the radiocontinuum peak.

Fig. 1. a The map of the 12CO (1 0) total integrated intensity. The contour levels range from 7.6 K km s-1 (1.6  Jy beam-1 km s- 1 or 1.5 for a velocity width of 400 km s-1) to 76 K km s-1 (16  Jy beam-1 km s- 1) in steps of 11.4 K km s-1 (2.4  Jy beam-1 km s- 1), 2) (black), then change to a spacing of 38 K km s-1 (8.0  Jy beam-1 km s- 1, 8) up to 380 K km s-1 (80  Jy beam-1 km s- 1) (white). b  Representative spectra from the outer regions of the maps, demonstrating that the lines in these regions become narrow, but are clearly detected. The spectra correspond to single beams centered on the positions indicated. c  Dispersion map. The contours range from 10 km s-1 to 100 km s-1 in steps of 10 km s-1. The peak dispersion is 110 km s-1.

In Fig. 1b, we display some representative spectra from the outer parts of our ¹²CO map, regions where the confidence level of the integrated intensity maps (which refers to the full velocity width in the center of 400 km s^-1) drops below . The lines are clearly detected in these parts of the bar, but become very narrow. The same effect is evident from the dispersion map (Fig. 1c): The dispersion (calculated along the line-of-sight, ) falls off sharply toward the bar edges, but remains at 40 km s^-1 to 60 km s^-1 close to the major axis.

To show that the change in dispersion is not a result of projected rotation, we display position-velocity (pv) diagrams along the bar major axis (position angle (PA) 5^o, Fig. 2a) and the disk major axis (PA 25^o, Fig. 2c). The velocity width is not systematically larger along the disk major axis, where rotation should have the largest effect. Also, the regions of narrow lines are not associated with the disk minor axis. The only possible effect of rotation is a tentative alignment of the 60 km s^-1 to 80 km s^-1 velocity contours in the ¹²CO dispersion map (Fig. 1c and Fig. 4a) with the disk major axis.

Fig. 2. a  Position velocity diagram along the major axis of the 12CO bar (position angle 5o (north-to-east)). Thus, the bar is oriented very close to the north-south direction. b  Position velocity diagram along the axis with position angle 50o (i.e. 45o relative to the bar major axis). This is the angle that best shows the central rapidly rotating structure. c  Position velocity diagram along the disk major axis, at a position angle of 25o (i.e. 20o relative to the bar major axis). For all panels, the contours are multiples of 2 (0.17 K or 36  mJy beam-1).

There is no sign of a `twin peaks' structure (as coined by Kenney et al. 1992) perpendicular to the bar, perhaps suggesting that there are no antibar orbits, occuring within an Inner Lindblad resonance (ILR). However, the question of the existence of an ILR in NGC 7479 is controversial and remains undecided: Quillen et al. (1995) derived the existence of an ILR close to the nucleus, while in the models of Sempere et al. (1995) no ILR exists. Laine et al. (1998) argue that the bar perturbation close to the center is so strong that the issue cannot be resolved without non-linear orbit analysis. In any case, the ILR would be located within 700 pc - 800 pc (4" - 5") from the nucleus; thus, we would barely resolve a `twin peak' structure.

Another possible signature of an ILR is a ring of material close to its positions, due to the expected pile-up of gas in the collision region where and orbits intersect. Such a structure may be indicated in our map. There certainly is a clear peak not just in intensity but also in velocity dispersion (Fig. 1c) in the center. This dispersion peak fragments into a high velocity system, consisting of a number of clumps, as is evident from the pv diagrams we show in Fig. 2. This system is confined to within of the center and distinctly different from the material within the `main' bar, where the velocity changes much more slowly, which is seen in panel (a), a cut along the major bar axis. The central rapidly rotating structure is most easily seen at a position angle close to (panel (b)). While there is a central peak (at  km s^-1), this peak does not dominate the emission: other features at velocities ranging from 2200 km s^-1 to 2500 km s^-1 are almost as strong and spatially separated by .

A high velocity system like this is a signature feature of many barred galaxies. It has been used by Binney et al. (1991) to argue the presence of a central bar in our Galaxy. The orbits of the gas that can cause this feature have been modelled by e.g. García-Burillo & Guelin (1995) for the case of the weakly barred edge-on galaxy NGC 891. The structure found in NGC 7479 (already noted by Laine et al. 1999) also greatly resembles the system seen in UGC 2855, a strongly barred galaxy where, in contrast to NGC 7479, quiescent conditions are found for the gas along the bar (Hüttemeister et al. 1999). As in UGC 2855, the central high velocity system in NGC 7479 can be interpreted as a clumpy tilted ring or torus close to the ILR (which we would then place at from the nucleus), as a fragmented, tilted, rotating nuclear disk that is freely fed from material inflowing through the bar without being stopped at an ILR, or even as a dynamically decoupled inner bar, as tentatively suggested by Baker (2000).

3.1.2. Comparison with prior ¹²CO maps

The general morphological structure and velocity field of our map are in excellent agreement with the results reported by Laine et al. (1999). However, the ¹²CO emission in our map is a continuous bar structure, while the structure seen by Laine et al. breaks up into separate clumps away from the central condensation, a large part of the body of the bar being devoid of emission. As can be expected if the two maps are consistent, the position of these clumps is in exact agreement with the peaks we see embedded in more extended emission.

Comparing the total fluxes contained in the maps, we find that our flux is higher by a factor of 1.45 than that seen by Laine et al. This discrepancy is caused by our map being more sensitive by a factor of , due to a larger beam size combined with a slightly longer integration time. Thus, the larger East-West diameter of the bar seen in our map and its continuity are not artifacts of smearing by lower resolution, but real effects: Our map picks up an additional, more smoothly distributed gas component traced by ¹²CO. We can compare our total flux to the flux seen in the single dish map of Sempere et al. (1995), which covered a larger area than our map. However, % of the flux is contained the region of the OVRO primary beam. We find that we recover more than 90% of this flux, which is consistent with Laine et al., who report recovering 60% of the single dish flux in their less sensitive map. Thus, there is very little, if any, missing flux in our ¹²CO map and we can analyse the complete molecular distribution as traced by ¹²CO at high resolution. This complete recovery of flux is the main distinguishing feature between our ¹²CO map and the interferometric maps published previously , both by Laine et al. and earlier by Quillen et al. (1995).

3.2. Distribution of ¹³CO and HCN

Since ¹³CO and HCN usually originate from regions that are more compact and confined than ¹²CO emitting gas, we consider it very likely that we also recover the entire flux for those molecules, being limited only by sensitivity, not by missing zero-spacings in the interferometer map.

The distribution of the ¹³CO emission is presented in Fig. 3. While Fig. 3a gives the moment 0 map for the full central velocity of 400 km s^-1 (see Fig. 5), panel b is integrated over only 200 km s^-1, so that very narrow lines are not lost in the noise. Now, a connection between the central distribution and the southern feature becomes apparent. The example spectra shown in Fig. 6 demonstrate the narrowness of these lines. The total ¹³CO flux we detect is  Jy km s^-1.

Fig. 3a and b. The distribution of the 13CO emission (contours) superposed on the 12CO intensity map (greyscale). a  Intensity integrated over the full velocity width in the center, i.e. 400 km s-1. The contours start at 1.4 K km s-1 (0.3  Jy beam-1 km s- 1 or for a width of 80 km s-1) and are spaced by 1.85 K km s-1. b  The inner region of the bar, integrated over 200 km s-1 only to bring out emission from narrow lines. The contours start at 0.9 K km s-1 (0.2  Jy beam-1 km s- 1, 1.5 for a velocity width of 40 km s-1) with a spacing of 1.35 K km s-1. The crosses mark the positions for which spectra are shown in Fig. 5. Thin gray contours refer to 12CO. The arrows mark the positions of the intensity cuts in Fig. 9.

Fig. 4a-c. Maps of the one-dimensional velocity dispersion in the inner region where continuous 13CO emission is detected. a  12CO. The contours range from 10 km s-1 to 100 km s-1 in steps of 10 km s-1. b  13CO. The contour levels are at 10 km s-1, 15 km s-1 and then range from 20 km s-1 to 80 km s-1 in steps of 10 km s-1. c  Ratio of the 12CO velocity dispersion to the 13CO velocity dispersion. The contours label ratios of 1.2, 1.4, 1.6, 2, 3 and 5.

Fig. 5. Spectra of the central position (taken at the position with the highest 12CO intensity, i.e. and ). The dashed line is 12CO divided by 10 for better comparability, the solid line is 13CO and the dotted line HCN. All three spectra above correspond to a region with the 7 2 5  2 size of the HCN beam; the 12CO and 13CO data were smoothed to the same beamsize as the HCN data. 20  mJy beam-1 correspond to 73 mK (HCN) or 44 mK (12CO).

Fig. 6. Offset spectra for 12CO divided by 10 (dashed) and 13CO (solid). The positions are marked in Fig. 3 and correspond to regions with the size of the beam of the CO observations (, see Table 2). For the 12CO beam used here, 20  mJy beam-1 correspond to 97 mK.

The ¹³CO emission is distributed very differently from the ¹²CO. The central ¹³CO peak is significantly offset to the north-west from the ¹²CO peak (by 2" or 320 pc in linear size).

While the ¹²CO distribution in the central concentration and south of the center along the bar runs almost exactly north-south, the central ¹³CO distribution is best described as extending along a position angle of , with a curvature toward the south-east (Fig. 3). The deviation between the ¹²CO and ¹³CO morphology is even clearer in the southern ¹³CO extension at offsets to to the south along the bar.

No ¹³CO is seen at corresponding offset along in the northern part of the the bar.

The weak feature offset to the north, close to the bar end, has no southern counterpart, but is coincident with a peak in the ¹²CO distribution and thus likely to be real. Further peaks at the bar ends are not detected with at a sufficient confidence level, since these regions are close to the edges of our primary beam, and any emission may be affected by the fall-off in sensitivity as well as sidelobes of the central peak. One might expect to recover ¹³CO again close to the bar edges, due to the crowding of elliptical () streamlines, resulting in increased cloud collisions and the formation of dense shocked gas. Further high sensitivity studies are needed to clarify this point.

Still, very dramatic changes in the ¹²CO to ¹³CO line intensity ratio are evident, both in the central region and along the bar. These changes, which are indicators of changing gas properties, will be discussed in detail in the subsequent section.

HCN is clearly detected only in the very center of the ¹²CO distribution (Fig. 7). It is unresolved and its position is consistent with the ¹²CO peak, i.e. offset from the maximum ¹³CO intensity. The total HCN flux in this central structure is 1.6 Jy kms.

Fig. 7. The total integrated intensity of HCN, superposed on 12CO (grayscale and thin grey contours). The HCN contour levels start at 1 K km s-1 (0.42  Jy beam-1 km s- 1 or 1.5) and increase in 2 steps. The ellipse in the lower left corner denotes the HCN beam and demonstrates that the HCN emission is not resolved.

3.3. Line intensity ratios

We present line intensity ratios taken between ¹²CO and ¹³CO () in different positions and over different areas in Table 3. In this table, we have computed and its errors individually for the peak intensity, the integrated intensity as measured within the velocity window where ¹³CO is detected and the velocity range within which we find ¹²CO emission. The respective velocity ranges are given along with the values of . Again, it is obvious that the ¹³CO line usually is narrower than the ¹²CO line, especially at and close to the ¹²CO peak region along the bar. Moving away from this peak, the ¹²CO linewidth drops and approaches that of ¹³CO (see the dispersion maps for ¹²CO and ¹³CO (Fig. 4) and the spectra displayed in Fig. 6). Especially from Fig. 4c), which displays the ratio of the ¹²CO and the ¹³CO velocity dispersion, it is clear that the velocity dispersion of ¹³CO always is smaller than the velocity dispersion of ¹³CO. The dispersions are similar (ratio ) in the central , where the ¹³CO line is fairly wide ( km s^-1) and in the southern ¹³CO concentration, where the ¹²CO transition becomes narrow ( km s^-1). The ratio of the dispersion is high (3-6) close to the ¹²CO peak region along the bar, where the ¹²CO line is wide, while the ¹³CO line is narrow.

Table 3. ¹²CO/¹³CO line ratios () for NGC 7479 and typical galactic ratios. The `cuts' refer to Fig. 9. Individual random errors are given in parentheses. The error in the scaling of all ratios due to calibration uncertainties is estimated to be 20%.
Notes:
a) The velocity range gives the region where the ¹²CO or ¹³CO line is detected above the noise.
b) Under extreme assumptions, this can be brought down to 20 (see text)
c) Range encountered in the central , both in peak and integrated ratios
d) 3 limit calculated over the range where ¹²CO is visible.
e) Polk et al. 1988;
f) Aalto et al. 1995

The global ¹²CO/¹³CO intensity ratio, i.e. the ratio we find when comparing the total fluxes for the entire map, is quite high at , certainly much higher than what is typical for gas in galactic disks or even `normal' galactic nuclei (see Table 3 for typical values). It is very unlikely that this ratio is an artifact of missing flux, since almost all flux in ¹²CO is seen and we do not expect to miss more flux in the the more compactly distributed ¹³CO. However, lack of sensitivity to ¹³CO is a concern.

To evaluate the possible magnitude of this effect, we assume that ¹³CO emission in a velocity range of 200 km s^-1 is present at the 2 level in all places where ¹²CO is detected and we do not find ¹³CO. This (unlikely) scenario would result in an additional flux of 10.8 Jy km s^-1, i.e. roughly double our ¹³CO flux and bring the global ratio down from 42 to 19, which we thus regard as a firm lower limit. We have also estimated the global value of using the global spectra of ¹²CO and ¹³CO, i.e. spectra constructed for the entire spatial extent of ¹²CO emission. Here, ¹³CO is only tentatively detected. Using a 3 limit and a velocity range of 450 km s^-1, we obtain a ratio of   30, in full agreement with the ratio determined using integrated fluxes.

In the central condensation, the variation in goes along with the morphological shift described above. ranges from close to the ¹³CO `ridge' to at distances of 5" or more from it. The average ratio over the central condensation is fully consistent with the ratio determined from the single dish spectra taken toward the central position (Fig. 8). Since we expect the central condensation to dominate the single dish flux in the central 33", this result confirms our prior conclusion that there is no missing flux in the interferometry maps.

Fig. 8. Single dish observations of 12CO and 13CO taken with the OSO 20 m telescope. The beamsize is 33", the intensity scale is (main beam efficiency ).

Fig. 9a-e. East-West cuts in integrated intensity through the bar at the positions marked in Fig. 3. The north-south width of the cuts correspond to the beamwidth of the CO observations. Solid lines are 13CO, dashed lines are 12CO divided by 20. The integrated intensities refer to a velocity range of 450 km s-1, i.e. the full range where 12CO emission is found. The thin dashed line is the 2 limit for 13CO and a velocity width (FWHM) of 40 km s-1, i.e. the typical width of off-center 13CO lines. Thus lower 13CO intensities have no meaning. All offsets are relative to the right ascension of the 12CO peak position. 1  Jy beam-1 km s-1 corresponds to 4.8 K km s-1.

The even more drastic variation of in the bar is evident from both Table 3 and Fig. 9. The latter presents another easy way of visualizing the changing line ratios. It shows cuts in integrated intensities (over a range of 450 km s^-1, chosen to include all ¹²CO emission) along the bar minor axis at a number of offsets. Here, we focus on the southern ¹³CO extension. As the ¹³CO emission shifts eastward from ¹²CO when we move south along the bar, the line ratios change from values around 25 in the central condensation to (integrated intensity) at the ¹²CO maximum along the bar. The ratios calculated using the peak flux and the velocity range of the ¹³CO emission are often smaller than the referring to the velocity range of the ¹²CO line since in many places the ¹³CO lines are narrower than the ¹²CO lines. At the declination of the maximum of the southern ¹³CO distribution, drops from 7337 to 3.50.6 (integrated) or 2410 to 3.51.4 (peak) within ! Interestingly, in this position, off the ¹²CO peak, the widths of the two transitions are identical. These values of are put into perspective by noting that of 6-7 is typical for global ratios in galactic spiral arms (Polk et al. 1988), which might already include the contribution of some diffuse gas, while Galactic GMCs typically show values of 3-5. An exceeding 30 is only seen in a few luminous mergers (Casoli et al. 1992, Aalto et al. 1991).

The central gas surface density of NGC 7479 ( pc^-2, using the `standard' CO to N(H₂) conversion factor for comparison purposes only, see discussion below) is high enough to place the nuclear region in the range of IR-bright starburst galaxies (Scoville 1991), even though ultraluminous IR galaxies (ULIRGs) can have core gas surface densities up to an order of magnitude higher (e.g. Bryant & Scoville 1999, who obtained exceeding 20000  pc^-2 for NGC 2623, Mrk 231 and NGC 6240, assuming a standard conversion factor). However, the gas surface density in the bar of NGC 7479, where the most extreme values of are reached, is far smaller than in luminous starbursts, pointing to a different dominant mechanism accounting for the high value of .

The large changes of have to indicate a very significant change in the properties of the emitting gas, scenarios for which will be discussed below

HCN  emission is detected only in the nucleus, with a ¹²CO/HCN line ratio .

© European Southern Observatory (ESO) 2000

Online publication: December 5, 2000
helpdesk.link@springer.de