Although the CO molecule is arguably the best tracer of the H2 content of a galaxy, it is primordial to gather results from several tracers, to inter-compare them, and avoid some of the main biases of any given diagnostic. At high redshift, this is even more required, since the CO lines observed are from a high J-level, and several CO lines are needed to derive the gas excitation, and estimate the CO(1-0) intensity calibrated in terms of H2 mass. The main alternate tracer is the dust continuum emission in the Rayleigh Jeans domain (i.e. close to the CO(4-3) to CO(7-6) if excited), where the dust emission is linear in both temperature and column density. The assumed dust-to-gas ratio will account for metallicity effect. Other tracers at very high z are fine structure line emission such as the [CII] at 158 µm, or the [OIII] at 88µm, with some specificity as gas tracers, as will be developed in Section 6.
Casey et al (2014) have written a detailed review of far-infrared and sub-millimeter survey of high redshift galaxies with dust emission. Although the temperature of the dust heated by star formation in molecular clouds, is expected in average of the order of 20-40K, in some cases, nuclear starbursts or AGN, the dust can peak at 60K. This produces large uncertainties in the detection rates of continuum surveys, since their success rate depends in the SED distribution of the sources.
Scoville et al (2014), Scoville et al (2016) have proposed that the Rayleigh-Jeans tail of the dust emission spectrum, peaking around 100 µm, acts as a good tracer of the gas content of galaxies: the Rayleigh-Jeans regime means that the dust temperature is involved only linearly, and does not introduce too much uncertaintly. Of course the dust abundance is also proportional to metallicity, so the conversion factor between the dust emission and gas mass, through the dust-to-gas ratio, is as incertain as the CO method. However, the detection of the dust continuum might be easier than the line, and does not require specific tunings. In compensation, there is more confusion and no redshift or kinematical information on the detected objects. While it is relatively easy to detect actively star forming galaxies and starbursts, main sequence objects at relatively low redshifts are more difficult to detect in dust continuum than in the CO lines, given the non-linear LFIR - LCO scaling relation. For instance, in a sample of normal star-forming galaxies of the COSMOS field at z ∼ 3, about half of the galaxies are detected in continuum with ALMA, and the rest of the undetected sources have to be stacked (Schinnerer et al, 2016). Besides, the CO line observation provides more very useful information as the dynamical mass, and gas excitation.
Another difference between dust and CO tracers, is that the former traces both atomic and molecular gas. A calibration experiment has shown however that at low and high redshift the two main tracers of the molecular/interstellar gas agree very well, see Figure 2.
 |
Figure 2. Comparison between the dust continuum and CO line tracers of the interstellar gas. The left panel compares the 850 µm and CO luminosities for normal low-z star forming galaxies (SF gal), low-z ULIRGs, and z ∼ 2 submillimeter galaxies (SMG, many of them are lensed). At right, the ratio between these two luminosities (L’CO being converted to Mmol) shows in more detail an almost constant proportionality factor. The conversion factor used is Mmol = 6.5 L’CO [K km/s pc2]. Image reproduced with permission from Scoville et al (2016), copyright by AAS. |