Astron. Astrophys. 323, 363-373 (1997)

2. Star formation activity

2.1. Star formation tracers

There does not seem to exist any observational quantitative estimator of active star formation devoid of ambiguity. The present situation is briefly discussed and summarized in the following four subsections, whereas our final choice is discussed in the fifth one.

2.1.1. H

One can expect that H emission leads to a good estimator of instantaneous star formation rate (SFR) insofar as it can be checked that they have been corrected for extinction, for instance by comparing with thermal radio luminosities (Sauvage & Thuan 1992). The reader is referred to Kennicutt (1983), Kennicutt & Kent (1983), Keel (1983), and Pogge (1989) for data on the H emission properties of normal galaxies. For a general discussion see Kennicutt (1989).

2.1.2. UV

UV fluxes in a bandpass of about 125Å centered at 2000Å have been used by Donas et al. (1987) to obtain quantitative estimates of the current SFR. Difficulties and uncertainties in the correction for extinction are similar to those present for the H data.

2.1.3. FIR

The connection between IRAS far-infrared (FIR) and H emissions is still controversial. Sauvage & Thuan (1992) outline the strong linearity in a correlation. They show that the decrease of the ratio along the Hubble sequence can be explained by a model of FIR-emitting ISM consisting of two components, i.e. star forming regions and quiescent cirrus-like regions as originally introduced by Lonsdale Persson & Helou (1987) and Rowan-Robinson & Crawford (1989).

Comparing these last approaches, we observe that in two-colour diagram versus , galaxies with and have a probable contribution to the flux from recent star formation at least of the order of 50%. Moreover, Sauvage & Thuan (1992) infer from the decreasing fraction of associated to the cirrus from Sa to Sdm that the high mass star formation efficiency increases toward late spiral types. This efficiency is defined as the fraction of FIR-emitting ISM directly associated with star formation.

Various authors directly used FIR colour indices as estimators of current star formation activity. According to Puxley et al. (1988) galaxies with are believed to contain regions of star formation. Eskridge & Pogge (1991) admit that is the value above which a major part of the FIR emission is expected to be due to star formation. Sekiguchi (1987) indicates that the fraction of 60 µm emission attributable to a warm component can be used as an indicator of star formation activity. Dultzin-Hacyan et al. (1990) consider that the best IRAS tracer of recent star formation is the ratio . In normal galaxies, the mean value of this ratio is -1.30, whereas for liners and starbursts it is respectively -1.15 and -0.75. Moreover, considering , the global to central H flux ratio taken from Kennicutt (1983) and Keel (1983), we observe a correlation between and . This clearly suggests that some information on star formation activity is contained in this indicator in spite of the caveats mentioned above.

From the analysis of these different suggestions, we can admit that objects for which most of the FIR emission is due to current star formation are separated from those with FIR colours indistinguishable from galactic cirrus by the alternative conditions: , , , or . The ratio of the respective contributions of starburst and cirrus in the observed spectrum of IRAS galaxies (Rowan-Robinson & Crawford 1989) could also be considered. The frontier in color indices defined above corresponds to .

Finally, in view of strengthening our statement, we can compare the global values of estimators for various galaxies with local values obtained in nearby galaxies. In fact, the FIR sources in M31 seem to coincide with giant HII regions complexes (Xu & Helou 1996). The same observation is reported by Rice et al. (1990) from IRAS maps of M33 or by Xu et al. (1992) for the LMC. Tomita et al. (1996) localize galaxies in different part of the versus diagram corresponding to HII regions, non-HII regions and central regions of M31. This approach suggests that could be a useful indicator of current versus recent star formation rates.

2.1.4. Radio

The origin of the tight correlation between the FIR and the radio continuum emission of late-type galaxies has been discussed by various authors (e.g. Helou 1991 for a review). The 60 µm to 20 cm IR-to-radio ratio seems to be a signature of star formation activity resulting from stars with . As shown by Puxley et al. (1988), the majority of SB galaxies with have central radio source emission arising from the center of a burst of star formation. However, such an emission could also be powered by a density enhancement or a darkened active nucleus.

2.1.5. Our choice

The necessity to have a sample of galaxies as large as possible with both data on bar morphology and star formation activity leads us to prefer FIR to H or radio data. We also have to take into account all the different caveats and problems listed in Sect. 2.1.3. Therefore, after having consulted various sources of data, we choose to use as indicator of star formation activity for the late-type SBs galaxies from Martin's catalogue (Martin 1995). The effects observed are qualitatively confirmed by using other color indices.

2.2. Star formation rates

For the sake of comparison with numerical simulations (see Sect. 5), we would like to have estimates of the SFRs. However, the SFR values calculated from data must be used with caution. The results are strongly affected by at least three factors: 1) The choice of the shape and the mass range for the IMF. 2) The IR wavelength range used in published data. 3) The cirrus contribution to as a function of Hubble type.

Concerning the IMF, it must be emphasized that in the general relation , the value of k will strongly vary depending on the IMF. For instance, for a Salpeter IMF with mass ranges 0.1 - 60 , 2 - 60 , 8 - 60 , k will respectively be , , and (Telesco 1988). The reality of a top-heavy IMF for starburst galaxies is still debated (see e.g. Sommer-Larsen 1996).

The main source of data for still come from IRAS fluxes which do not cover the whole IR spectrum. Some extrapolations have been tried to correct for missed fluxes beyond 120 µm and shortward 40 µm. This correction depends on the ratio (see e.g. Young et al. 1989).

Finally, the percentage coming from the cirrus contribution can be accounted for by using the two-component model introduced by Lonsdale Persson & Helou (1987). Following Sauvage & Thuan (1994), a contribution of 77% for Sbc's, 70% for Sc's, and 45% for Scd's is adopted.

The previous considerations mean that only relative SFRs are really relevant although access to absolute SFRs remains necessary. Assuming the same IMF for all galaxies in the sample and a linear relation between the FIR flux and the SFR, we use the following relation inferred from Telesco (1988):

where is taken on the 1 - 500 µm range from Young et al. (1989), corrected for the cirrus contribution according to Sauvage & Thuan (1994). Equation 1 concerns the mass range 0.1 - 60 . The numerical coefficient must be divided by two if the mass range is 2 - 60 as mentioned before. We introduce a normalizing factor obtained by inserting into Eq. 1 the mean FIR luminosity for normal Sc galaxies, i.e. (Becklin 1986). Then, we relate all the SFRs to ,

where C is a weighting factor to take into account the respective cirrus contribution ( for Sbc's; for Sc's; for Scd's). Both f and are given in Table 1.

Table 1. Galaxy sample

For some galaxies in our sample, the SFR deduced from the H emission, , has been given by Kennicutt (1983) and is also indicated in Table 1. As noted by the author, the individual entries probably possess uncertainties of the order due to variable extinction. The is not directly comparable to f. However, for our purpose we can be satisfied when observing a qualitative agreement on high and low values of the SFR inferred from FIR and H data. Using the same assumptions for the IMF, the distance, and the corrections for extinction as Kennicutt (1983), the (Donas et al. 1987) are on average larger by a factor 1.2 than the for common objects.

© European Southern Observatory (ESO) 1997

Online publication: June 5, 1998

helpdesk.link@springer.de