4 Signatures favouring a hidden quasar in Arp 220

   
4.1 The amount of the PAH 7.7 $\mu$m extinction

Firstly, we argue in favour of the "mixed case'' extinction, in contrast to the simpler case by a dust screen.

The difference in the PAH 7.7 $\mu$m/100 $\mu$m distributions between ULIRGs and the normal galaxies can be entirely explained by the higher 100 $\mu$m/850 $\mu$m colour temperature of the dust in ULIRGs. Hence, the PAH emitters do not appear particularily related to the warm - presumably starburst heated - dust in ULIRGs. Rather, the similar PAH 7.7 $\mu$m/850 $\mu$m distributions for the typical ULIRGs and the normal galaxies suggests that the PAH emitters and the sub-mm emitting dust grains are intimately coupled and actually mixed along the line of sight through the galaxies. Therefore, in the following we will use the mixed case, unless otherwise stated.

Now we consider the minimum additional mixed case extinction required to shift Arp 220 towards the lower border of the PAH 7.7 $\mu$m/850 $\mu$m range for the other ULIRGs (Fig. 1). This depends on the extinction curves used. $A_V \approx 330$, when using the interstellar extinction curve for graphite and silicate grains in our Galaxy, with a grain size distribution cutoff at about 0.3 $\mu$m (MRN-model: Mathis et al. 1983, Table C1).

In ULIRGs the dust may have properties more like those found in protostellar cold dense clouds. Their observed extinction curves have been successfully modelled using flaky grains with a large size cutoff at about 30 $\mu$m (KS model: Krügel & Siebenmorgen 1994). Using even larger grains (>30 $\mu$m) would result in a low emissivity exponent of $\beta < 1.2$, contrary to the observed value of $\beta \approx 1.6$ (Lisenfeld et al. 2000; Klaas et al. 2001). In contrast to the MRN model, the KS extinction curve has no 5-25 $\mu$m features in the MIR, consistent with the smooth curve observed toward the Galactic centre (Lutz et al. 1996). Therefore we give the KS model a slightly higher preference. If the entire dust in Arp 220 has such properties, then the extinction required to provide the PAH 7.7 $\mu$m/850 $\mu$m offset to the other ULIRGs is at least $A_V \approx 110$, which we adopt as a conservative estimate. It is also consistent with the lower limit $A_V{\rm (mixed)} > 110$, which corresponds to $A_V{\rm (screen)} > 45$ derived from [SIII] 18.7 $\mu$m/33.5 $\mu$m line ratio (Genzel et al. 1998). Our value $A_V{\rm (mixed)} \approx 110$ is a lower limit, since it refers to the minimum extinction required to shift Arp 220 to the range of the other ULIRGs.

4.2 The extinction of the MIR continuum

So far, we have derived the average extinction of Arp 220 which, however, could vary between different lines of sight.

The central region of Arp 220 exhibits a complex geometry on NICMOS images (Scoville et al. 1999) with two nuclei, each intersected by dust lanes, which are coincident with the CO disks (Downes & Solomon 1998). At 2 $\mu$m the actual nuclear centres appear to be entirely hidden behind the dust lanes. On KECK high resolution 3-24.5 $\mu$m images the PAH emission, the silicate absorption and the 12.5 $\mu$m continuum show similar morphologies and only a moderate variation in the intensity ratios for various image areas (Soifer et al. 1999). Most of the MIR emission arises from an extremely compact area with a diameter of less than 100 pc centred on the western nucleus (Soifer et al. 1999). It contributes to the total flux by 66% at the 12.5 $\mu$m continuum and 55% at the 7.7 $\mu$m PAH line.

We argue now, that the extinction of the western nuclear MIR continuum is higher than previously inferred. If the extinction of the western nucleus were actually as low ( $A_V \approx 25$) as inferred from the 18 $\mu$m silicate absorption, then its dereddening does not increase the PAH strength significantly.

In this case the "rest'' (= total - western nucleus) of Arp 220 must contain all the hidden PAH emission necessary to reach the PAH 7.7 $\mu$m/850 $\mu$m and PAH 7.7 $\mu$m/100 $\mu$m levels of the other ULIRGs. Hence this "rest'' has to be much more extinguished ( $A_V{\rm (rest)} \approx 1.7 \cdot A_V$(average)). But this contradicts to the moderate variation in the intensity ratios for various image areas emphasized above. Furthermore, a detailed calculation shows that the dereddened position of the entire galaxy Arp 220 then lies in the same range as for the simple average dereddening (Fig. 1). Since the silicate absorption feature may probe only the shallow surface, the actual extinction of the western nucleus may well be much higher than $A_V \approx 25$.

These arguments support the picture that in Arp 220 both the regions emitting the PAH 7.7 $\mu$m and those emitting the 12.5 $\mu$m continuum are affected by a high MIR extinction of similar order of . Thus the extinction is so high that it might be difficult to discover a hidden AGN with common tracers.

4.3 Dereddening of the MIR continuum

The effects of reddening depend on the geometry of the absorbing/reemitting dust and we consider here the extreme cases "ideal sphere'' and "axis-symmetric geometry''.

If the dust were distributed in an ideal sphere, then the absorbed MIR continuum could only escape via reemission at longer wavelengths. In this case "dereddening'' will not increase the total luminosity, but only the $L_{\rm MIR}/L_{\rm FIR}$ ratio via a shift of $L_{\rm FIR}$ towards $L_{\rm MIR}$. But this case can be rejected as follows:

The presence of CO disks already indicates a non-spherical axis-symmetric geometry tilted about 45$\hbox{$^\circ$ }$ with respect to our line of sight (Downes & Solomon 1998). Therefore, a significant portion of the MIR continuum might escape along the polar directions, but along the line of sight a reduced portion of the luminosity is seen. In order to derive the true MIR luminosity, dereddening has to be applied. It will actually increase $L_{\rm MIR}$, and to a less extent $L_{\rm FIR}$.

If the emitters of the MIR continuum are mixed with the absorbing dust, then the same A_V as derived from the PAH diagnostics can be used for dereddening. The results are listed in Table 1 and shown in Fig. 2.

Dereddening with the MRN model shifts Arp 220 from the cool ULIRGs to the warm ones, each of which houses a strong AGN. With the KS model such a shift is present, but less pronounced. Nevertheless, in both cases $L_{\rm MIR}$ reaches high quasar-like values ($\approx$10$^{\rm 12}~L_{\odot}$).

The compactness of the MIR emitting region and the presence of the dusty CO disks suggests, that the MIR continuum emitters are not as well mixed with the absorbing dust as the PAHs. This does not conflict with the low variation of the observed MIR colours across the KECK images mentioned above. In this case the (hidden) MIR continuum originates more in the centre, which is surrounded by the absorbing material. Such a situation is typically seen in Seyfert2s (Clavel et al. 2000). Then half of the dust column according to $A_V{\rm (KS\ mixed)} \approx 110$ will work effectively on the central MIR continuum as screen absorber with an amount of $A_V{\rm (KS\ screen)} \approx 90$, placing Arp 220 clearly in the $L_{\rm MIR}/L_{\rm FIR}$ range of quasars (Table 1). With $A_V{\rm (MRN\ mixed)} \approx 330$ the screen dereddening of the centre becomes even more extreme.

4.4 Compact starbursts versus powerful AGN

With conservative dereddening of $L_{\rm MIR}$ by a factor of 4 (Table 1, KS: $A_V \approx 110$), the MIR luminosity density of the western nucleus reaches 7.310$^{\rm 7}~L_{\odot}$/pc$^{\rm 2}$
($L_{\rm MIR} \approx 0.56\,10^{\rm 12}~L_{\odot}$, $r \approx 50$ pc), about a factor of 1000 higher than in the prototype nearby starburst galaxy M 82 ( $L_{\rm MIR} \approx 9\,10^{\rm 9}~L_{\odot}$, $r \approx 200$ pc). Since the dust (and the gas) is more dissipative than the stars, is tends to be distributed in a more compact area than the stars. Bearing this in mind, it is difficult to imagine, how starbursts alone can create such a high luminosity density, and simultaneously how the dust hides them entirely.

Therefore, a more natural explanation would be that - in addition to the prominent starbursts mainly responsible for the FIR luminosity - one (or perhaps both) of the nuclei of Arp 220 contains a powerful AGN providing the quasar-like MIR luminosity which is hidden to us.

To summarize, Arp 220 shows an extraordinarily high - presumably mixed case - extinction, which also works on the MIR continuum of the western nuclear region. The presence of tilted non-spherical geometry justifies that after dereddening the true MIR luminosity reaches quasar-like values. Moreover, the MIR luminosity density of the western nucleus exceeds that of known circumnuclear starbursts, thus increasing the evidence for a hidden quasar in Arp 220.