IR (1-1000 μm) emission due to the re-processing of stellar light by dust is responsible for ~40% of the total bolometric energy density within the local universe (Calzetti et al. 2000; Driver et al. 2008). The contribution of the radio band to the cosmic energy budget is about 100 times smaller. Despite the vastly different energy scales, IR and radio emission are strongly correlated in star forming galaxies (SFGs; e.g. Helou et al. 1985; Condon 1992; Yun et al. 2001). In a summary sense, the birth and demise of massive (>5 M☉), dust-enshrouded stars has long been recognized to form the link between the two (e.g. Harwit & Pacini 1975); their strong UV emission heats the surrounding interstellar medium (ISM) which re-radiates in the IR, and at the end of their short life span they produce supernova remnants which accelerate the radio-emitting cosmic ray (CR) electrons.
Following first indications of the correlation in ground-based observations at 10 μm and 1.4 GHz (van der Kruit 1973; Condon et al. 1982) the ubiquity and tightness of the IR-radio relation became fully appreciated during the analysis of the combination of data from the Very Large Array (VLA) and the Infrared Astronomical Satellite (IRAS) which measured the far-IR (FIR) properties of ~20,000 galaxies at z < 0.15 (e.g. Dickey & Salpeter 1984; de Jong et al. 1985; Helou et al. 1985; Yun et al. 2001). IR observations of sources at higher redshifts became available with the advent of the Infrared Space Observatory (ISO) and, in recent years, with the Spitzer Space Telecope. They have provided increasing evidence that the locally observed correlation likely holds until z ~ 1 (Garrett 2002; Appleton et al. 2004; Frayer et al. 2006) and that the linearity of the correlation is maintained as far back as z ~ 3 although the slope may change, especially for sub-mm galaxies (Kovacs et al. 2006; Vlahakis et al. 2007; Sajina et al. 2008; Murphy et al. 2009, but see also Beelen et al. 2006; Ibar et al. 2008). The statistical significance of these high-z studies, however, is still low because so far the number of sources detected at z > 0.5 is limited.
Thanks to its large area and multi-wavelength coverage, the data base of the COSMOS survey provides sufficient information to chart the evolution of the IR-radio relation in the strongest IR- and radio-emitters out to high redshift. Our work on the radio-IR relation is strongly based on the VLA-COSMOS 'Joint' catalog which contains ~2500 galaxies detected at 1.4 GHz in the redshift range z < 1.5, and ~3000 sources (down to 5σ) in total. The catalog was compiled following a detailed analysis of the sources lists obtained from the images of the VLA-COSMOS Large and Deep projects. Its reliability was extensively checked by comparison to both the previous VLA-COSMOS radio catalog as well as to optical COSMOS data sets.
In Sargent et al. 2010a we presented the largest sample of band-matched radio, IR and optical sources that has been used so far to study the evolution of the IR-radio relation. Another important improvement over previous work lies in the consistent application of the statistical tools of 'survival analysis' (e.g. Feigelson & Nelson 1985) to the entire data set; this ensures that the information implicitly carried by sources lying under the detection threshold at either IR or radio wavelengths is retained. By measuring the shift — which agrees with theoretical expectations (e.g. Kellermann 1964; Condon 1984) — between the average IR/radio flux ratio in IR- and radio-selected populations, it was possible to reconcile discrepant claims in the literature and to show that studies relying on only one of the two selection bands run the risk of inferring spurious evolutionary trends.
Using a sample jointly selected at IR and radio wavelengths to reduce selection biases, our work firmly supports previous findings (e.g. Garrett 2002; Appleton et al. 2004; Murphy et al. 2009, but see also Seymour et al. 2009, Ivison et al. 2009) of an unchanged IR-radio relation out to at least z ~ 1.4 (cf. Fig. 1). Moreover, based on the analysis of ~150 sources with an unprecedentedly large fraction of direct radio detections, we concluded that the local relation likely still holds at 2.5 < z < 5 (cf. Fig. 2). This is strong evidence that ISM physics and, e.g., the configuration of magnetic fields in SFGs were very similar more than 10 Gyr ago; IR and radio emission are apparently equally good tracers of star formation (SF) activity over much of the history of the universe. This has been a central assumption underlying both measurements of the cosmic SF history (SFH) with deep radio surveys (e.g. Haarsma et al. 2000; Smolčić et al. 2009), and distance estimates for optically unidentified sub-mm galaxies (e.g. Carilli & Yun 1999; Dunne et al. 2000).
The similarity of the IR/radio properties of high-z galaxies and local SFGs is remarkable because the former are extremely IR-luminous systems
(LTIR > 1013 L☉) that possess intra-galactic radiation fields a thousand times as strong as those in typical
local late-type galaxies (e.g. Thompson 2006). The size of the COSMOS data set makes it possible for the first time to not only study the evolution of the IR/radio
properties for all detectable starbursts in general, but to do so selectively for a luminosity range that is consistently sampled at all z < 2. Our
measurements (see Sargent et al. 2010b) on volume-limited samples of both ultra luminous IR galaxies (ULIRGs), and of galaxies that populate the bright end of the evolving IR luminosity function (e.g. Magnelli et al. 2009), have revealed no change of IR/radio properties of these brightest IR-luminous galaxies (cf. Fig. 3).
SFGs and galaxies with different modes of nuclear activity often have similar average IR/radio ratios (e.g. Sanders et al. 1989; Sopp & Alexander 1991; Roy et al. 1998; Murphy et al. 2009), although the dispersion in the latter class is usually larger (Obrić et al. 2006; Mauch & Sadler 2007). In Sargent et al. 2010a we mapped the relative abundance of optically selected SFGs and AGN as a function of the logarithmic IR/radio flux ratio q and found that AGN populate the locus of the correlation for starbursts in almost equal proportions as the SFGs themselves out to at least z ~ 1 (cf. Fig. 4). A possible explanation for this is that both the IR and radio emission are predominantly powered by SF rather than AGN activity (e.g. Barthel 2006). It is also conceivable, however, that AGN come to lie on the IR-radio relation owing to a 'conspiracy' (i.e. combinations) of other astrophysical processes (e.g. Sanders et al. 1989; Colina & Perez-Olea 1995).
More information on the IR-radio relation (and the VLA-COSMOS catalogs):
- 'The VLA-COSMOS Perspective on the IR-Radio Relation. I. New Constraints on Selection Biases and the Non-Evolution of the IR/Radio Properties of Star Forming and AGN Galaxies at Intermediate and High Redshift'
Sargent, M. T., et al. 2010, ApJS, 186, 341
- 'No Evolution in the IR-Radio Relation for IR-Luminous Galaxies at z < 2 in the COSMOS Field'
Sargent, M. T., et al. 2010, ApJL, 714, 190
- Publically released VLA-COSMOS catalogs (available from the IRSA/IPAC COSMOS archive)
- 'The VLA-COSMOS Survey. IV. Deep Data and Joint Catalog'
Schinnerer, E., Sargent, M. T., et al. 2010, ApJS, 188, 384
Related Links & Literature:
- Collaborators: Eva Schinnerer, Eric Murphy, Vernesa Smolčić, George Helou, Chris Carillli, and the VLA-COSMOS, S-COSMOS and z-COSMOS teams.
- 'Thermal infrared and nonthermal radio - Remarkable correlation in disks of galaxies'
Helou et al. 1985, ApJL, 298, L7 - 'Radio emission from normal galaxies'
Condon, J. 1992, ARA&A, 30, 575 - 'The FIR/Radio correlation of high redshift galaxies in the region of the HDF-N'
Garrett, M. A. 2002, A&A, 384, L19 - 'The Far-Infrared-Radio Correlation at High Redshifts: Physical Considerations and Prospects for the Square Kilometer Array'
Murphy 2009, ApJ, 706, 482