2.1. ISOPHOT photometry
We present ISOPHOT data of the FIR spectral energy distribution of a sample of active and inactive galaxies. We performed observations in OFF followed by ON source measurements using the standard multi-filter observing template PHT22 (Klaas et al. 1994). For each filter, the exposure time per position was 32 s. The OFF positions were chosen from clean fields of the IRAS 60 µm and 100 µm plates. Typically, the OFF position is 10´ away from the source and well outside its optical diameter. The ISOPHOT sample is introduced in Table 1 with positions from IRAS.
Table 1. The two samples of active and inactive galaxies observed with ISOPHOT.
The raw data were extracted from the ISO processing system (OLPv7) and reduced by the ISOPHOT Interactive Analysis system (PIAv7.2) 1. Starting from edited raw data, the signals were corrected for non-linearities of the integration ramps, glitches and responsivity drifts (Laureijs et al. 1998). The background was subtracted by taking the difference between ON and OFF beams.
Photometry was done by coadding all pixels of the C100 and C200 arrays, respectively. The correction for single losses due to beam effects and inter-pixel gaps between neighboring detector elements is usually done by a simple model which calculates the intensity fraction of a point source () falling on the center of a single pixel. They are published in the ISOPHOT Observer's Manual. However, this model is not applicable for C200 since in most cases the source is falling on the center of the four pixel array. For this case, empirical correction factors have been obtained (Laureijs et al. 1998). They are close to unity for very extended objects. For our sample, the FIR sizes are comparable to the optical diameters of the galaxies so that we are dealing with the intermediate case of slightly extended targets. We therefore consider correction factors (intensity fraction) as given by the mean of the two extremes (Table 2).
Table 2. Intensity fraction of a slightly extended source falling on the center of the C100 or C200 detector array. The reference wavelengths [µm] of the C100 and C200 filters are indicated.
2.2. The color correction
The derived flux densities, , as they are obtained with PIA presuppose that the sources have an energy distribution such that const. However, the thermal dust emission should be described by a modified black-body spectrum. Therefore, the sources have an intrinsic spectral shape of the form . This will lead to flux densities that differ from . The difference is expressed by a color correction factor , which is defined as the ratio at the reference wavelength, , of the ISOPHOT broad band filters. Previously, we applied a color correction by assuming but simply prescribed to 30 K (KSZC). This time, we first determine the temperature and then perform the color correction.
The temperature can be found by applying an inverse color corection, e.g. requiring that the product best fits the derived flux densities . As initial guess we take a temperature of K and so derive a first estimate of the flux density . Then we calculate at each reference wavelength the color correction factor
where is the relative system response of the bandpasses. We solve the inverse color correction by minimizing the expression
Here denotes the uncertainty of and k is the running number of data points. We assume that for the active galaxies, the FIR has one dust component:
and for the inactive spirals two:
The column densities, N, and mean dust temperatures, T, are free parameters.
In our observing mode, the C100 detector has an absolute accuracy of and a relative filter-to-filter uncertainty of 10% (Laureijs et al. 1998). We see residual detector drifts at a level . On top of the statistical noise, we therefore apply a photometric error, , of 30%. For C200, we end up with very similar error estimates. Within the uncertainties, the influence of gas emission lines to the broad band photometry is negligible (Lord et al. 1996). We find typically a color correction of , in some cases it goes up to 40%.
To further constrain our fits we include FIR photometry by IRAS. As the C60 and IRAS 60 µm filters range down to m, we consider an additional warm dust component of K which fits the 12 and 25 µm IRAS fluxes. This gives us a boundary in the inversion algorithm at short wavelengths. A boundary at long wavelengths is given by the 1.3 mm photometry discussed in the following section.
2.3. Integrated millimeter wave photometry
With respect to the spatial distribution of cold interstellar material in galaxies, Chini et al. (1992a) found that in the active galaxies the bulk of the molecular gas, as traced by its CO emission, is concentrated within the central region. In contrast, both the CO and the 1.3 mm dust emission is more extended in inactive spirals (Chini et al. 1996). In the absence of high resolution FIR images, we are faced with the problem to spatially match the 1.3 mm and FIR data of the inactive spirals. The ISOPHOT detectors cover a large area of and in the C100 and C200 array, respectively.
Except for NGC5719, we take the 1.3 mm flux densities from Table 2 of Chini et al. (1995) which refer to a surface area across. If a galaxy is optically extended beyond 70", we assume that its dust emission also extends to the optical edge. We then linearly scale up the 1.3 mm flux to the optical diameter, but not beyond , the field of view of ISO. The linear scaling implies that we assume the surface brightness to decline towards the edge. In this way, we find, for example, that the integrated 1.3 mm flux of NGC6156 is mJy (optical diameter , flux within area 163 mJy). For NGC5719, which has an optical diameter of , there is only a 1.3 mm measurement for a beam (32.5 mJy, Chini et al. 1996). Here we estimate the total flux from Fig. 4 of Chini et al. (1995), which gives the statistical dust distribution in spiral galaxies. This figure suggests that the 1.3 mm flux of the whole galaxy is a factor 4 larger.
It has been stressed that the 1.3 mm fluxes in Table 2 of Chini et al. (1995) are lower limits because the maps from which they were derived are heavily undersampled and only those positions were considered where measurements had been performed that yielded a 3 detection. These maps obviously miss some flux. For example, in the annulus ranging from 24" to 70" it is quite likely that half of the flux is not included because this region is observed by four beams which only cover about half of the surface area. If one interpolates between the observed points of the maps to account for the missing flux, one obtains in the case of NGC6156, after Fig. 2 of Chini et al. (1995), a total flux within 70" of mJy, i.e. 283.5 mJy instead of the 163 mJy value which we use. Therefore, our 1.3 mm fluxes are conservative low estimates. The values that follow from interpolation would result in some 30% larger dust masses and an average dust temperature more than 1 K lower than what we will cite below.
© European Southern Observatory (ESO) 1999
Online publication: November 3, 1999