6. Comparison with FIR bright galaxies
Since the IRAS survey, FIR bright galaxies have been the subject of numerous studies because these objects experiment an intense star formation activity. The extreme case is that of UltraLuminous Infrared Galaxies (ULIGs) with a bolometric luminosity larger than essentially emitted in FIR and star formation rates of several hundreds solar masses per year: they generally are violent mergers and may represent an important phase in the formation of large galaxies like ellipticals (Sanders & Mirabel 1996). Such objects are known to be rare at low z but they might be far more numerous at high z as suggested by the sub millimetric surveys with SCUBA (e.g. Sanders 1999).
With the launch of the ISO satellite, the sensitivity of the ISOCAM camera has allowed mid-infrared surveys at intermediate redshift (). In particular Flores et al. (1998) have observed one CFRS field, therefore UV (0.28 µm) and infrared data are available for these galaxies.
We now compare the FIR and UV properties of these galaxies (ULIG and ISOCAM/CFRS) to that of our IRAS/FOCA sample of nearby galaxies. The comparison is rather straightforward since all these objects are IR selected.
6.1. Ultraluminous Infrared Galaxies (ULIG)
6.1.1. Nearby ULIGs
Trentham et al. (1999) have obtained HST observations for three ultra luminous infrared galaxies: VII Zw031, IRAS F12112+0305, IRAS F22491-1808. These galaxies are selected to be cool in order to avoid a non thermal origin for the FIR emission. We can calculate directly their ratio using the data at 0.23 µm for the UV emission. The three objects are reported in Fig. 6 (similar to Fig. 2b) with empty stars for symbols. As expected for this type of objects they appear to be very luminous at 60 µm with a high FIR to UV flux ratio. Such objects are not represented in our FIR selected sample of nearby galaxies: this emphasizes how much these objects are rare in the local Universe and with extreme properties as often underlined (e.g. Sanders & Mirabel 1996). Since the three ULIG have also been detected at 100 µm we can estimate their UV extinction (we neglect the difference in the UV wavelengths i.e. 0.23 µm versus 0.2 µm). We find mag: more than 99 of the UV flux of these objects is emitted in the FIR.
6.1.2. High redshift galaxies detected by SCUBA: ULIG candidates
Hughes et al. (1998) have observed the HDF field at 850 µm with SCUBA. 5 objects detected by SCUBA in the HDF field have been tentatively associated to optical sources for which photometric redshift are available but such an identification is difficult because of the uncertainty on the 850 µm positions. Indeed, the identification of the most brightest source (HDF850.1) has not been confirmed (Sanders, 1999). Moreover the nature of these sources, starbursts or AGN, is not clear: at FIR luminosity larger than about half of the nearby ULIGs are predominantly powered by AGNs (e.g. Sanders 1999).
The 60 µm luminosity of the 4 remaining galaxies is obtained from their emission at 850 µm accounting for the redshifting and an assumed spectral energy distribution chosen to be that of M82. The optical data from the HDF lead to the estimate of the UV flux at a rest frame wavelength of 0.28 µm. All these estimates rely on the resemblance of all ULIGs with M82 and can lead to false results (e.g. Sanders 1999). In spite of these caveats, we have reported the 4 high redshift galaxies in Fig. 6 (filled triangles). They appear very extreme, being more luminous in FIR and probably more extincted that all the other galaxies studied in this paper. A tentative estimate of the extinction is obtained by using the instead of the one. We find values spanning from 8 to 11 mag. As a comparison M82, which belongs to our IRAS/FOCA sample exhibit "only" 5.4 mag of extinction at 0.2 µm. These high redshift ULIGs seem also to be much more extincted than the most luminous Lyman break galaxies of the HDF studied by Meurer et al. (1999) for which they derive an extinction not larger than 3.5 mag. Although their UV luminosity corrected for extinction are comparable (), these two classes of galaxies do not seem to exhibit the same properties in FIR and UV as suggested by Heckman (1999). Indeed we can try to roughly locate the most luminous galaxies of Meurer et al. in Fig. 6. The FIR luminosity can be estimated from their star formation rates and the ratio from their extinction using the Fig. 1. It gives and for an extinction of mag. Therefore it seems that the Lyman Break Galaxies detected in the HDF by their U-dropout do not follow the steep trend of Fig. 6 found for FIR bright galaxies but instead exhibit a lower increase of the extinction with the intrinsic luminosity of the galaxies. Such a difference may be due to the contribution of AGNs in ULIGs. Indeed the extrapolation of the mean trend found in the IRAS/FOCA sample (Fig. 3) reported as a full line in the Fig. 6 does not lead to the extreme case of ULIGs and seems more compatible with LBGs.
6.2. ISOCAM/CFRS galaxies
Flores et al. (1998) have obtained ISO/ISOCAM Mid Infrared images of one CFRS field, most of the detections are at 15 µm. The infrared 8-1000 µm luminosities have been deduced from MIR and/or radio measurements using templates of spectral energy distributions and are probably not very secure but an approximate value is sufficient for our comparison with local templates. We differentiate AGNs and starbursts as classified by Flores et al. Only the global IR (8-1000 µm) flux is available for these objects and not the flux at 60 µm. We adopt a mean value of for the ratio of the total dust emission to that intercepted by the 40-120 µm band (see Sect. 5.2). Then we estimate the ratio between the flux at 60 µm and the FIR (40-120 µm) for our IRAS/FOCA sample: . Therefore the total IR fluxes given by Flores et al. have been divided by a factor 2 in order to roughly represent the flux at 60 µm.
The UV emission is taken at 0.28 µm as given by Flores et al. It is difficult to estimate a correction factor to translate the UV data to 0.2 µm in the absence of observations of a large sample of galaxies at both wavelengths since the ratio depends on the star formation history and the dust extinction. We can try to use synthesis models for this estimate: assuming a constant star formation rate over 1 Gyr and using the models of Leitherer et al. (1999) for a solar metallicity we find the flux being defined as .The difference of extinction between 0.28 and 0.2 µm has been calculated using the extinction curves of the Milky Way and the LMC (Pei 1992) and that of Calzetti (1997). The ratio . Therefore the two effects (star formation and extinction) roughly compensate each other and we do not perform any correction between 0.28 and 0.2 µm. The galaxies are plotted in Fig. 6 as crosses for the true starbursts and empty circles for the Seyferts. They all fill the gap between the IRAS/FOCA sample and the ULIGs. Therefore, they are not as extreme as ULIGs but their extinction is larger than the nearby galaxies of the IRAS/FOCA sample. For the galaxies classified as starbursts, we have tentatively estimated this extinction from their FIR to UV flux ratio. The extinctions found span from 2 to 5.5 with a mean at 3.3 mag (and a median at 3 mag). This is much larger than that estimated by Flores et al. by matching the global star formation rates deduced from the total FIR and the UV luminosities in the observed field: they find extinctions around 2 mag at 0.28 µm. This discrepancy between the extinction occurring in individual galaxies selected in infrared and that deduced from the total FIR and UV luminosity of a selected field (i.e. the sum of the luminosity of all galaxies detected in the wavelength band (FIR or UV)) is well illustrated in the Table 5 of Flores et al. where the ratio calculated for individual objects observed at both 15 and 0.28 µm is times larger than the ratio of the global luminosities IR and UV luminosities in the CFRS field. This is in full agreement with our own results presented in Sect. 4.
© European Southern Observatory (ESO) 1999
Online publication: December 2, 1999