2.2. Integrated properties
The overriding result of all ISOPHOT studies of the integrated properties of normal galaxies in the FIR is that their SEDs in the 40-200 µm spectral range require both warm and cold dust emission components to be fitted. Although the concept of warm and cold emission components is as old as IRAS (de Jong et al. ), it only became possible to directly measure and spectrally separate these components using ISOPHOT's multi-filter coverage of the FIR regime out to 200 µm.
In order to investigate the integrated properties of local universe gas-rich galaxies, a number of statistical samples were constructed. All these projects were complementary in terms of selection and observational goals. In descending order of depth (measured in terms of a typical bolometric luminosity of the detected objects), the published surveys are:
The ISOPHOT Virgo Cluster Deep Survey (IVCDS; Tuffs et al. , , Popescu et al. ) represents the deepest survey (both in luminosity and surface brightness terms) of normal galaxies measured in the FIR with ISOPHOT. A complete volume- and luminosity-limited sample of 63 gas-rich Virgo Cluster galaxies selected from the Virgo Cluster Catalogue (Binggeli et al. ; see also Binggeli et al. ) with Hubble types later than S0 and brighter than BT 16.8 were mapped with ISOPHOT at 60, 100 and 170 µm. The IVCDS sample was (in part) also observed with the LWS (Leech et al. ) and with ISOCAM (Boselli et al. , see also Boselli et al. ).
The IVCDS provides a database for statistical investigations of the FIR SEDs of gas-rich galaxies in the local universe spanning a broad range in star-formation activity and morphological types, including dwarf systems and galaxies with rather quiescent star-formation activity.
The Coma/A1367 Survey Contursi et al.  consists of 6 spiral and 12 irregular galaxies having IRAS detections at 60 µm. The galaxies were selected to be located within 2 or 1 degrees of the X-ray centres of Coma and A1367 clusters, respectively, with emphasis on peculiar optical morphologies. Each galaxy was observed in a single pointing with ISOPHOT, at 120, 170 and 200 µm, as well as mapped with ISOCAM in the 6.75 and 15 µm broadband filters. The sample provides a database of integrated flux densities for a pure cluster sample of high luminosity spiral and irregular galaxies.
The ISO Bright Spiral Galaxies Survey (Bendo et al. , , ) consists of 77 spiral and S0 galaxies chosen from the Revised Shapley-Ames Catalog (RSA), with BT 12.0. Almost all are IRAS sources. Mainly an ISOCAM mapping survey with the 12 µm filter, the project also used ISOPHOT to take 60, 100 and 170 µm short stares towards the nucleus of the galaxies and towards background fields. The sample provides a database of MIR morphologies and FIR surface brightnesses of the central regions of bright spiral galaxies, including S0s.
The ISO Bright Spiral Galaxies Survey and the IVCDS represent the principle investigations of optically selected samples of normal galaxies. It should be emphasised that the main difference between them is primarily one of shallow versus deep, rather than field versus cluster, since by design the Virgo Sample predominantly consists of infalling galaxies from the field, and no cluster specific effects could be found (see also Contursi et al.  for Coma / A1367 Sample).
The ISO Key Project on Normal Galaxies Sample consists of 69 galaxies selected to span the whole range of the classical IRAS colour-colour diagram (Helou ). Since IRAS detected a vast number of galaxies in its four bands, the selection was also made to span the Hubble sequence evenly and provide a broad range of IR luminosities, dust temperatures (as determined by IRAS) and in star-formation activity. Single pointing toward the centres of the galaxies in the sample were made by ISOPHOT-S (Lu et al. ) to measure the 2.5-12 µm spectra and by ISOPHOT-C at 170 µm. The galaxies from this sample were also mapped with ISOCAM (Dale et al. ) and their main cooling lines in the FIR were measured with ISOLWS (Malhotra et al. ).
The ISOPHOT Serendipity Survey (Stickel et al. ) has initially catalogued 115 galaxies with S 2Jy at 170 µm and with morphological types predominantly S0/a-Scd. This sample provides a database of integrated 170 µm flux densities for relatively high luminosity spiral galaxies, all detected by IRAS at 60 & 100 µm. Recently a catalogue of 1900 galaxies was released (Stickel et al. ), of which a small fraction does not have IRAS detections. Most of the 1900 galaxies are spirals. The measured 170 µm flux densities range from just below 0.5 Jy up to ~ 600 Jy.
2.2.1. The FIR spectral energy distribution: Dust temperatures, masses and luminosities
The presence of a cold dust emission component peaking longwards of 120 µm was inferred from studies of the integrated SEDs of individual galaxies (see Sect. 2.1.1), from statistical studies of small samples (Krügel et al. ; Siebenmorgen et al. ) and was confirmed and generalised by studies of the larger statistical samples mentioned above. The latter studies also demonstrated the universality of the cold dust component, showing it to be present within all types of spirals (Tuffs & Popescu ). The cold emission component predominantly arises from dust heated by the general diffuse interstellar medium and the warm component from locally heated dust in HII regions, an interpretation consistent with what has been seen in the ISOPHOT maps of nearby galaxies (see Sect. 2.1.1) and with self-consistent modelling of the UV-FIR SEDs (see Sect. 2.3 and Popescu & Tuffs ).
The cold dust component is most prominent in the most "quiescent" galaxies, like those contained in the IVCD sample, where the cold dust temperatures were found to be broadly distributed, with a median of 18 K (Popescu et al. ), some 8 - 10 K lower than would have been predicted by IRAS. The corresponding dust masses were correspondingly found to be increased by factors of typically 2 - 10 (Stickel et al. ) for the Serendipity Sample and by factors 6 - 13 (Popescu et al. ) for the IVCD sample, with respect to previous IRAS determinations. As a consequence, the derived gas-to-dust ratios are much closer to the canonical value of ~ 160 for the Milky Way (Stickel et al. , Contursi et al. ; see also Haas et al.  for M 31), but with a broad distribution of values (Popescu et al. ).
It was found that the cold dust component provides not only the bulk of the dust masses, but even the bulk of the FIR luminosity, in particular for the case of the most quiescent spirals, like those in the IVCD sample. In contrast to the SEDs found by the other ISOPHOT studies, which typically peaked at around 170 µm, Bendo et al.  derived spatially integrated SEDs typically peaking at around 100 µm. The result of Bendo et al. may reflect the fact that these observations were single pointings, made towards the nucleus of resolved galaxies extending (in the main) beyond the field of view of ISOPHOT, and were therefore biased towards nuclear emission, which is warmer than the extended cold dust emission missed (or only partially covered) by these measurements. Nevertheless, the measurements of Bendo et al. constitute a useful probe of the FIR emission of the inner disks. This emission (normalised to K band emission) was found to increase along the Hubble sequence (Bendo et al. ).
Since the FIR carries most of the dust luminosity, it is interesting to re-evaluate the question of the fraction of stellar photons converted via grains into IR photons, taking into account the comprehensive measurements of the cold dust emission component made available by ISOPHOT. This was done by Popescu & Tuffs , who showed that the mean percentage of stellar light reradiated by dust is ~ 30% for the Virgo Cluster spirals contained in the IVCD sample. This study also included the dust emission radiated in the NIR-MIR range. The fact that the mean value of ~ 30% found for the Virgo Cluster spirals is the same as the canonical value obtained for the IRAS Bright Galaxy Sample (BGS; Soifer & Neugebauer ) is at first sight strange, since IRAS was not sensitive to the cold dust component. However the BGS sample is an IR selected sample and biased towards galaxies with higher dust luminosities, while the Virgo sample is optically selected and contain a full representation of quiescent systems. So the deficit in FIR emission caused by sample selection criteria for the Virgo sample is compensated for by the inclusion of the cold dust component. Popescu & Tuffs  also found evidence for an increase of the ratio of the dust emission to the total stellar emitted output along the Hubble sequence, ranging from typical values of ~ 15% for early spirals to up to ~ 50% for some late spirals. This trend was confirmed by Boselli et al.  who further utilised the new ISO data on dust emission to constrain the corresponding absorption of starlight and thus improve extinction corrections (using the technique pioneered by Xu & Buat  for the IRAS data).
2.2.2. Two outstanding questions of the IRAS era
The radio-FIR correlation
One of the most surprising discoveries of the IRAS all-sky survey was the very tight and universal correlation between the spatially integrated FIR and radio continuum emissions (de Jong et al. ; Helou et al. ; Wunderlich et al. ; see Völk & Xu  for a review). However all the pre-ISO studies of the FIR/radio correlation were based on FIR luminosities derived from the IRAS 60 and 100 µm flux densities, and thus were missing the bulk of the cold dust luminosity. The ISOPHOT measurements at 60, 100 and 170 µm were used to redefined the FIR/radio correlation (Pierini et al. ) for a statistical sample of spiral galaxies. The inclusion of the cold dust component was found to produce a tendency for the total FIR/radio correlation to become more non-linear than inferred from the IRAS 60 and 100 µm observations. The use of the three FIR wavelengths also meant that, for the first time, the correlation could be directly derived for the warm and cold dust emission components.
The cold FIR/radio correlation was found to be slightly non-linear, whereas the warm FIR/radio correlation is linear. Because the effect of disk opacity in galaxies would introduce a non-linearity in the cold-FIR/radio correlation, in the opposite sense to that observed, it was argued that both the radio and the FIR emissions are likely to have a non-linear dependence on SFR. For the radio emission an enhancement of the small free-free component with SFR can account for this effect. For the cold FIR emission a detailed analysis of the dependence of local absorption and opacity of the diffuse medium on SFR is required to understand the non-linear trend of the correlation (see Pierini et al. ).
The improved angular resolution of ISO compared with IRAS also allowed a more detailed examination of the local FIR-radio correlation on sub-kpc size galactic substructures. Hippelein et al.  established the correlation for the star-forming regions in M 33. This correlation is shown in Fig. 7, overplotted on the correlation for integrated emission from galaxies. It is apparent that the local correlation has a shallower slope (of the order of 0.9) than for the global correlation. It was argued that the local correlation is attributable to the increase with SFR of dust absorption in increased dust densities, and to local synchrotron emission from within supernova remnants, still confining their accelerated electrons. Both emission components play only a minor role in the well known global radio-FIR correlation, that depends on the dominant large-scale absorption/re-emission properties of galaxies.
Figure 7. Plot of the monochromatic radio luminosity versus the FIR luminosities for M 33 (Hippelein et al. ) and its star-forming regions (filled circles) together with the data for the Effelsberg 100-m galaxy sample (Wunderlich et al. , open circle). The dotted line has a slope of 1.10 (Wunderlich & Klein ).
Far-infrared emission as a star-formation tracer
In the FIR, the correlation with the most widely-used star-formation tracer, H, is extremely non-linear, which has long led authors to suspect that the FIR emission is the result of more than one component (see e.g. Lonsdale-Persson & Helou ). With ISOPHOT the correlation was separately established for the warm and cold dust emission components. A good linear correlation was found between the warm FIR luminosities (normalised to the K band luminosity) derived for the IVCD sample and their H EW (Popescu et al. ). This is in agreement with the assumption that the warm dust component is mainly associated with dust locally heated within star-formation complexes. The scatter in the correlation was attributed to a small component of warm emission from the diffuse disk (produced either by transiently heated grains or by grains heated by the old stellar population), as well as to the likely variation in HII region dust temperatures within and between galaxies. A good but non-linear correlation was found between the cold FIR luminosities of the galaxies from the IVCD sample and their H EW, in the sense that FIR increases more slowly than H. Since the bulk of the cold FIR emission arises from the diffuse disk, the existence of this correlation implies that the grains in the diffuse disk are mainly powered by the UV photons (see also Sect. 2.3 and Popescu & Tuffs ). The non-linearity of the correlation is consistent with there being a higher contribution from optical photons in heating the grains in more quiescent galaxies.
For the late-type galaxies in the Coma and A1367 clusters, Contursi et al.  derived the relationships between the IR flux densities at 200, 170, 120, 100 and 60 µm, normalised to the H band flux, as a function of the H EW. It was found that the poorer correlation is in the 200 µm band and that the values of the fitted slopes decrease as the FIR wavelength increases. These results should be interpreted in terms of the increasing contribution of the diffuse component with increasing FIR wavelength.
Using the ISOPHOT observations from Tuffs et al. ,  and Stickel et al.  in combination with UV and K band photometry, Pierini & Möller  have tried to quantify the effects of optical heating and disk opacity on the derivation of SFR from FIR luminosities. For this they investigated trends in the ratio of the far-IR luminosity to the intrinsic UV luminosity, Ldust / LUV, with both disk opacity and disk mass (as measured by the intrinsic K-band luminosity). Using a separate relation between disk opacity and K band luminosity they were able re-express Ldust / LUV in terms of a single variable, the galaxy mass. In this way they found evidence for the relative importance of optical photons in heating dust to increase with increasing galaxy mass.