Off-nucleus dust concentrations: The observations impressively show that in this colliding galaxy pair the dust is accumulated not only in the two nuclei, but is also strongly concentrated in giant cloud complexes in the overlap region of the galaxy disks. At our spatial resolution the knots represent rather cloud complexes than single clouds (10" 1 kpc). Furthermore, besides the relatively warm dust at T 30 K which is a characteristic temperature for starbursts in luminous IR galaxies, much of the dust is probably cold at T 20 K which is more typical for quiet galaxies and extremely dense clouds. In the following we focus on knots in the overlap area only. Nature of the knots: The two knots K1 and K2 in the overlap area show an excellent spatial coincidence with peaks seen on maps at 15 µm, CO 1-0 and 6 cm. We argue now that both knots show starbursts as well as quiescent regions.
The brightest 15 µm peak (#2) is located at K1. Thus, K1 contains already clouds well heated by the strong starbursts. K2, however, could still be mainly composed of clouds in a pre-starburst phase. Firstly, the flux of the 15 µm peaks coincident with K2 is about a factor of three fainter than for K1. If the star formation rate were actually higher in K2 than in K1, then also the extinction in the MIR must be higher for K2 than for K1, but K2 will outshine K1 somewhere in the FIR and submm domain. This, however, is neither indicated by our FIR data simulations, nor is it observed in the submm domain. The possibility, that K2 suffers still from extinction at 450 µm, is very unlikely. K2's opacity can be derived from the dust mass where we adopt that a dust column density of 0.1 corresponds to an optical depth of 0.5. For example, distributing a dust mass of 107 homogeneously within a slab of 300 pc radius (3") yields 170, or following Mathis et al. (1983, their Table C1) 0.1, hence negligible. Secondly, the 450/850 µm flux ratios indicate for K2 rather a lower temperature and luminosity than for K1, suggesting that in K2 starbursts are spatially confined or moderate or too young to have yet heated the bulk of the dust like in K1.
Stanford et al. (1990) already suggested from the lack of emission in the area around K2 that, if star formation took place, then it would be totally obscured. Their 6" resolution CO maps show, besides the two nuclei, also very prominent emission in the overlap region which is resolved into four peaks. The brightest one in the south coincides with the warm knot K1, while the three much fainter ones (E,W and N in Stanford et al.'s notation) are located at the position of the cold knot K2. Thus, K2 appears to be clearly resolved into smaller units on the CO maps. Notably the CO emission of K2 is about 50 of that for K1 (Stanford et al. 1990), but new CO maps with the BIMA array show a more equal CO brightness for K1 and K2 (Gruendl et al. 1998, Gao et al. 1998). Nikola et al. (1998) derived a low flux ratio CII 158 µm line / CO for the whole galaxy and concluded that the active regions might be surrounded by quiescent ones. These could just be the cold regions like we see in K2 (and also in K1).
VLA 6 and 20 cm radio continuum maps also show - besides the two nuclei - a couple of prominent peaks in the overlap region (Hummel & van der Hulst 1986). Their peak #2 lies in K1, and their peaks #3 and #4 in K2. At 20 cm peak #3 in K2 is by far the brightest one and has the steepest spectral index ( -0.6 0.2) among the other peaks ( -0.4 0.2). While K2 is less luminous and shows less signatures for starburst activity compared with K1, the 20 cm radio continuum in K2 is brighter and has a steeper spectrum than that in K1. Therefore, the strong radio continuum could be due to small clusters of compact HII regions and supernova remnants which are hidden on the back side of the dust clouds. They are invisible at optical wavelengths and possibly suffer strong extinction in the mid-infrared, and they belong to the first generation of starbursts, so that the bulk of the dust has not yet been heated substantially. An alternative explanation is, that in K2 the strong radio continuum is the consequence of an enhanced magnetic field still compressed within the dense clouds. Then a modest star formation or shocks in the interstellar medium are sufficient to provide the electrons for the strong synchrotron emission. Note, that in a turbulent environment like K1 and K2 only little is known about the validity of the radio-FIR correlation and that magnetic field strength needs not to be in equipartition with the cosmic rays or star formation rate. If K1 is more evolved than K2, then any compression of the magnetic fields could already be relaxed in K1 and less radio continuum would be seen. Future polarimetric radio observations should provide clues to this puzzle.
In conclusion, besides the strong starburst activities in particular in K1, many of the dust clouds in K1 and K2 are cold. K2 shows the lower dust temperature and luminosity and also lower or younger starburst activity, but the strongest radio peaks at 6 and 20 cm. This suggests that, in addition to the starbursts, exceptional conditions are present. They are consistent with the picture of an extreme compression of the ISM in clouds, in which case these clouds could be in a pre-starburst phase housing many protostellar condensations. Genesis of the knots: From the morphology we sketch the following evolutionary scenario: The separation of K1 and K2 is about 2 kpc (20"). The overlap region contains about 107 dust which comprises a considerable fraction (10-30) of the typical dust content of a whole spiral galaxy. Such a concentration within a projected size of 1-3 kpc could be explained as a consequence of an initial inelastic collision of clouds of the two galaxy disks which serves as a kernel into which the other clouds of the rotating disks are running - like a traffic jam in the fog.
The life time of molecular clouds in the Milky Way is less then 107 years and probably much shorter in turbulent environments (A. Burkert, private communication). Taking a typical maximum sound speed of 10-100 km/s for the dust clouds, the dynamical time scale to cross the distance between K1 and K2 is about 108 to 109 years or longer. This clearly exceeds the 106 to 107 years lifetime of the clouds in K1 and K2 as well as that of the OB stars generated in K1. Thus the knot K2 can not be triggered by K1, rather it was independently created. Generalisation to mergers: While the bulk of the dust in Sb and Sc galaxies has a temperature of about 15-20 K (e.g. Chini et al. 1986), dust in mergers and active galaxies was believed to be warmer, with most of their FIR-submm luminosity coming from dust at T 30 K (Chini et al. 1989). The spatially resolved observations of Arp244, however, suggest that mergers of dusty galaxies in general contain dense concentrations of cold dust at T 20 K as well.
© European Southern Observatory (ESO) 2000
Online publication: April 17, 2000