Maps: The SCUBA and ISOPHOT maps are shown in Fig. 1 together with maps at other wavelengths for comparison.
On the SCUBA maps (Fig. 1a-c) the most striking features are the two equally bright 850 µm emission knots in the image centre (henceforth called K1 and K2). They are located in the overlap region of the two galaxy disks. At 850 µm the northern nucleus NGC 4038 is also prominent, while the southern nucleus NGC 4039 appears faint and shows a diffuse elongation towards southwest which is also found on the CO maps of Stanford et al. (1990). At 450 µm K1 and the northern nucleus are clearly seen, but the southern nucleus and K2 are dim. At the low flux level only an upper limit for K2 can be derived.
On the ISOPHOT maps two emission maxima show up at 100 µm (Fig. 1d), centred on the northern nucleus and the overlap region. The southern nucleus is not seen, nor does it produce a significant wing in the map. At 60 µm (Fig. 1e) only one broad peak located on the overlap region rises above the extended emission. Thus, the overlap region contains dust which is even warmer than in the northern nucleus.
Comparing the SCUBA and ISOPHOT maps, the location of the FIR maximum appears at a first glance closer to K2 than to K1 (Fig. 1d). But this is the consequence of the ISOPHOT observing and mapping procedure, where for the oversampled maps the 46" detector is scanning with steps of 15" 23", while the Airy profile for the ISO 60 cm telescope has about 21" and 32" FWHM at 60 and 100 µm, respectively. Note that no deconvolution was applied to the ISOPHOT maps. Instead, in order to investigate whether in the overlap area the FIR emission originates preferentially from K2 or also considerably from K1, we make use of known maps which trace the location of dust (and ISM) at high spatial resolution. These are the ISOCAM 15 µm, SCUBA 450/850 µm and VLA 6 cm maps. Note that in all these maps K1 is of similar or higher brightness as K2, and also that, in particular at 15 µm and 6 cm, the brightness profile is strongly asymmetric with a peak near the southeastern border and an extended emission towards northwest. We take them as input maps and simulate the ISOPHOT observing and mapping procedure. The input maps were (1) convolved to the corresponding Airy size of the ISO telescope at 60 and 100 µm, respectively, then (2) scanned as did the ISOPHOT detector along PA=33o with 15" 23" steps, then (3) from the scanned raster grid an output image with 15" 23" pixel size was created and (4) interpolated to 1" pixel size for better visualisation. The results of these simulations are shown in Fig. 1(f-i). Firstly, for all four input images the location of the maximum of the output image shifts about 10-20" towards northwest, with an example shown in Fig. 1f. This shift can be understood as a consequence of sampling an asymmetric input brightness distribution with a peak near the southeastern border. Thus, the true 60 and 100 µm maximum could lie towards southeast of the apparent location in the observed ISOPHOT images. Secondly, the observed images coincide well with the simulations (Fig. 1g-i). This suggests that the FIR emission, in fact, originates also considerably from K1, and not only from K2 (or northwest of it), as was the impression at the first glance (Fig. 1d-e). Furthermore, the 6 cm and 15 µm input maps result in simulated peaks encompassing the observed one at 60 µm (Fig. 1g-h). Thus, the brightness ratio R between K1 and K2 at 6 cm (R 0.7) and at 15 µm (R 2.3) may provide limits for a tentative estimate of the K1/K2 ratio at 60 µm; the mean of about R 1.5 argues in favor of K1 being slightly brighter at 60 µm than K2. This is also well consistent with K1 being brighter than K2 at 450 µm. At 100 µm (Fig. 1i) the observed broad maximum in the overlap region is excellently reproduced by the simulation with the 850 µm input map. Furthermore, compared to K1 + K2 the northern nucleus appears fainter at 850 µm than at 100 µm, indicating that the overlap region contains a stronger cold dust component (besides the warm one mentioned above).
The ISOPHOT maps are consistent with KAO maps presented by Evans et al. (1997) at 60 µm, and by Bushouse et al. (1998) at 100 and 160 µm. Similarly the KAO maps are oversampled using a detector pixel size of about 45", thus they likely suffer also from the convolution and show the observational shift of the flux maximum pointed out by our simulations.
Spectral analysis: At 15 µm and 6 cm the knots K1 + K2 together comprise about 30-35 of the total flux (within 2´ diameter). Assuming for the knots a similar contribution also at 60, 100 and 200 µm and taking the FIR flux values from Klaas et al. (1997), the FIR to submm spectral energy distribution (SED) can be derived for K1 + K2 (Fig. 2). A single modified blackbody fit of the 100 to 450 µm data with T = 25 K (which is unusually cool for an active starburst galaxy) yields a remaining flux excess at both 60 and 850 µm. This suggests the presence of dust components which are both warmer and colder than 25 K, respectively. A better fit to the SED is obtained with two modified blackbodies, one with T = 31 K (which is more typical for starburst galaxies) and one with T = 18 K (which is again rather cold). The presence of the 18 K component in the overall spectrum of K1 + K2 is consistent with the 450/850 µm colour temperatures derived for the single knots. Table 1 lists for all four knots the submm fluxes and derived parameters. Depending on the choice of the 450 µm upper limits for K2, the temperature reaches extremely cold values down to about 11 K, like found for protostellar condensations. The 450 µm data, however, are rather noisy and the difference between K1 and K2 should be considered with some care - we discuss it further below together with other observational results.
Table 1. Properties of the submm knots: Fluxes are within 30" apertures, with 1- errors about 0.3 Jy at 450 and 0.02 Jy at 850 µm. The formal colour temperature and L are derived from modified blackbodies with emissivity -2. The temperature errors were determined by fitting the steepened and flattened extreme constellations with 0.5 sigma added/subtracted to the measured fluxes. M is calculated according to Hildebrand (1983).
The radio 6 and 20 cm continuum (Hummel & van der Hulst 1986) extrapolated to 850 µm is very faint ( 10 mJy) and its contribution in the submm range can be neglected. On the other hand, high excitation CO lines are found in warm regions of starburst galaxies (e.g. Wild et al. 1992, Devereux et al. 1994). Thus, in our Arp244 maps the broad band 850 µm filter could contain also flux from the CO 3-2 line at 870 µm with an excitation temperature T 30 K. In this case the 450/850 µm colour temperature would be warmer and the dust mass lower than listed in Table 1. But as seen in Fig. 2 (right), the observed 850 µm flux lies clearly above the blackbody curves, allowing for a CO 3-2 line flux of 120 mJy, i.e. 30 of the 850 µm flux. This is well consistent with the spectral line contributions of 28 to the SCUBA 850 µm fluxes derived by Johnstone & Bally (1999) for the Orion A cloud which contains the nearest site of ongoing high mass star formation. Currently, for Arp244 no published CO 3-2 line fluxes are available, so we try some further estimates: is also compatible with the estimates ranging from 40 to 56 mJy derived for the CO 3-2 flux of K1+K2 by scaling the measured flux F (CO 1-0) with typical ratios R = CO 3-2 / CO 1-0 for starburst galaxies. We used = 225 Jy km s- 1 measured by Stanford et al. (1990, their Table 1, with a spectral resolution r = 13 km s-1 and a line width / bandpass ratio CO 3-2 / 850 µm filter 3.6 10-3) and the average ratio R = 0.64 0.06 obtained for nuclear starburst regions by Devereux et al. (1994, their Table 3) and R = 0.9 0.2 for M 82 by Wild et al. (1992), respectively. is even yet compatible for a three times higher CO 1-0 flux which was suggested by Nikola et al. (1998). Similarly, a possible CO 6-5 line contribution at 435 µm to the SCUBA 450 µm flux requires even higher excitation temperatures and might be too small to affect our dust temperature estimates (F (CO 6-5) 100 mJy or 5 , derived with values from Stanford et al. 1990 and Wild et al. 1992). Thus, the 450/850 µm data are well consistent with a cold dust component at temperature of about 18 K or less.
© European Southern Observatory (ESO) 2000
Online publication: April 17, 2000