The last few years have seen remarkable progress in the study of the cool, molecular gas content of galaxies, using centimeter and (sub–)millimeter telescopes. The cool gas content is a critical parameter in galaxy evolution, serving as the immediate fuel for star formation in galaxies. The state of the field in 2005 was reviewed by Solomon & vanden Bout (2005; see also Omont 2007). At that point, only a few handful of extreme starburst galaxies and luminous active galactic nuclei (AGN) host galaxies had been detected in molecular gas emission at significant lookback times, hardly anything was known about the gas excitation, and there were no detections of atomic fine structure lines. Research in recent years has resulted in an explosion in the number and type of galaxies detected in molecular line emission, as well as in atomic fine structure line emission, in the distant Universe. Detailed multi–transition, multi–species follow–up has been performed to determine the physical conditions of the gas in some of the brightest high–redshift systems known. The results are proving extremely telling for our understanding of galaxy formation and evolution and suggest that the molecular gas content of galaxies increases significantly with look–back time.
It is a good time to review the field of molecular line observations of high redshift galaxies for two reasons. First is the dramatic advance that has been made over the last decade, both in the number of galaxies detected, as well as the characterization of the molecular properties of these galaxies through followup observations. Second is the imminent full operation of revolutionary telescopes, the Atacama Large Millimeter/Submillimeter Array, ALMA (Wootten & Thompson 2009; Andreani 2010), and the Karl J. Jansky Very Large Array, JVLA (Perley et al. 2011), both of which promise to explore this evolution of the universal molecular gas content to an order of magnitude greater level of detail and sensitivity than previously possible. This review of the field, at this temporal cusp of knowledge, captures the current state of the field, and frames the fundamental questions that will be addressed with the next generation instruments.
1.2. Galaxy formation and the need for cool gas observations
Galaxy evolution has been the subject of many reviews in recent years (e.g. Shapley 2011, Renzini 2006, Giavalisco 2002, Silk & Mamon 2012) and the last decade has seen dramatic advances in our understanding of cosmic structure formation. Cosmic geometry, the mass-energy content of the Universe, and the initial density fluctuation spectrum, are now known to better than 10% (Spergel et al. 2007, Komatsu et al. 2011). Structure formation through gravitational instabilities has been calculated in exquisite detail through numerical studies (e.g. Springel et al. 2005, Klypin et al. 2011), and observationally verified through studies of galaxy distributions (e.g., Peacock et al. 2001, Reid et al. 2010). And the cosmic star formation rate density (the ‘star formation history of the Universe', SFHU), and stellar mass build-up, have been quantified back to first light and cosmic reionization (e.g. Bouwens et al. 2011a, Coe et al. 2013), within 1 Gyr of the Big Bang. Studies of galaxy formation are now turning attention to the evolution of the cool gas, the fuel for star formation in galaxies. In this section, we briefly summarize some of the general conclusions on galaxy formation that are relevant to our subsequent review of the gaseous evolution of galaxies.
Three main epochs have been identified in the SFHU , starting with a steady rise during cosmic reionization from z ∼ 10 to 6 (e.g. Bouwens et al. 2011b; Bouwens et al. 2012, Coe et al. 2013), corresponding to the epoch when light from the first galaxies reionizes the neutral intergalactic medium (IGM) that pervaded the Universe (Fan et al. 2006, Finkelstein et al. 2012). The comoving cosmic star formation rate density then peaks at z ∼ 1 to 3. This range is known as the ‘epoch of galaxy assembly,' during which about half the stars in the present day Universe form (Shapley 2011; Marchesini et al. 2009; Reddy et al. 2008). Last comes the order of magnitude decline in the comoving cosmic star formation rate density from z ∼ 1 to the present (e.g. Lilly et al. 1996, Madau et al. 1996).
The study of galaxy formation takes on the challenge to explain this observed star formation history of the universe in the context of ΛCDM, the hierarchical dark matter halo model. To understand galaxy formation we must investigate how stars and star formation are distributed over dark matter halos with different masses as a function of time. The most important feature of our current understanding of the field is that star and galaxy formation is inefficient: only ∼ 5% of all baryons (i.e. atoms of all kind) are in stars and dark stellar remnants at redshift zero (e.g., Fukugita & Peebles 2004).
Galaxies with a baryonic mass of slightly more than that of the Milky Way ( ∼ 5 × 1010 M⊙) are most efficient in converting the available baryons into stars ( ∼ 15−20%, e.g. Moster et al. 2010). One key observational result is that this ‘typical' galaxy mass has not greatly changed since z ∼ 3 (e.g., Marchesini et al. 2009, Ilbert et al. 2010). Dark matter halos and their baryon content, on the other hand, have grown by two orders of magnitude over that time span (e.g. Springel et al. 2005). It appears that dark matter halos with a total mass of ∼ 1012 M⊙ are, at all cosmic times, the most efficient star formation factories. For such halos, the star formation rate at different epochs is observed to roughly follow the cosmological accretion rate as predicted by ΛCDM: star–formation rates are observed to increase systematically with redshift, in a regular fashion such that a galaxy ‘main sequence' (defined below) is established, with a relatively small scatter in star formation rate for a given galaxy stellar mass (e.g. Noeske et al. 2007). At z ∼ 0, the cosmic star formation rate density is dominated by galaxies with star formation rates ≤ 10 M⊙ yr−1 (FIR luminosities ≤ 1011 L⊙). By z ∼ 2, the dominant contribution shifts to galaxies forming stars at ∼ 100 M⊙ yr−1 (Murphy et al. 2011; Magnelli et al. 2011). Once the halo and the galaxy grow beyond this mass, further growth through star formation is marginal (e.g. Peng et al. 2010) and accretion of other galaxies (merging) becomes the dominant evolutionary channel (van der Wel et al. 2009).
Massive galaxies with low star formation activity, if any, are observed at all redshifts z ≤ 3 (e.g. Franx et al. 2003, Kriek et al. 2006) and their existence remains a puzzle as there is no trivial mechanism that prevents gas from cooling and forming stars in more massive halos. Many ideas abound: shock heating of in–falling gas (Kereš et al. 2005, Dekel & Birnboim 2006), feedback from active galactic nuclei (AGN, Croton et al. 2006; Bell et al. 2012), motivated by ubiquitously observed super–massive black holes in massive galaxies and the coincidence of the peak of QSO and star–formation activity at z ∼ 2 (Hopkins et al. 2006), and stabilization of gaseous disks as a result of bulge formation (Martig et al. 2009). Feedback by AGN–driven winds appears to be required to explain the evolution of young star forming galaxies into red, bulge–dominated galaxies at intermediate redshift (e.g., Feruglio et al. 2010), while powerful radio jets from AGN may be needed to heat the intercluster gas around massive galaxies at late cosmic epochs, thereby inhibiting further late–time growth of massive galaxies (Fabian 2012; McNamara & Nulsen 2007).
We briefly clarify our use of the terms ‘starburst' and ‘main sequence' (MS) for star forming galaxies in this review. These classifications have arisen in two, parallel situations. First, studies find that the majority of starforming galaxies at both high and low redshifts define a ‘main sequence', in which there is a relatively tight distribution (dispersion < 0.3dex) in specific star formation rate (sSFR = SFR/stellar mass) versus stellar mass. The sSFR is typically a slowly decreasing function of increasing stellar mass, and, at a given stellar mass, the sSFR increases by a factor 20 from z = 0 to 2 for these MS galaxies (Sargent et al. 2012; Rodighiero et al. 2011). However, at all redshifts the distribution function in sSFR requires a second component at a factor 4 to 10 times higher SFR than the nominal main sequence. This ‘starburst' component constitutes a few percent of the distribution by number, and about 10% in terms of the contribution to the cosmic star formation rate density.
Second, as presented in Sec. 4.5, there also appears to be a parallel dual-sequence in the Far-IR to CO luminosity ratio, with factor few higher ratios for starburst versus main sequence galaxies. The implied gas consumption timescales may be an order of magnitude, or more, shorter in starburst systems than in main sequence galaxies (modulo the conversion factor, see Sec. 4.2). There is some evidence to support the notion that starbursts are associated with major gas rich mergers, although this remains an area of open investigation. These two sequences of star forming galaxies are in addition to early-type galaxies discussed above, which are typically higher stellar mass, and show an order of magnitude, or more, lower sSFR.
In order to understand why star formation efficiency peaks at a certain halo mass and then declines for more massive systems, the interplay between gas accretion, cooling, star formation and feedback must be understood. Most of our current understanding of galaxy formation, as briefly summarized above, is based on studies of the stars, star formation, and ionized gas. There remains a major gap in our knowledge, namely, observations of the cool gas: the fuel for star formation in galaxies. Put simply, current studies probe the products of the process of galaxy formation, but miss the source. If we can trace the presence of cold gas and its distribution in different galaxies and halos over cosmic time, the puzzle of the efficiency of star and galaxy formation can be unraveled. Numerous observational and theoretical papers have pointed out this crucial need for observations of the cool molecular gas feeding star formation in galaxies (Dressler et al. 2009; Genzel et al. 2008, Obreschkow et al. 2009a; Bauermeister et al. 2010).
We review the current status of observations of the cool gas content of galaxies, as measured via the rotational transitions of common interstellar molecules, and via the atomic fine structure line transitions, predominantly [C II]. Our focus is on results since the reviews of Solomon & vanden Bout (2005) and Omont (2007). Our review is primarily observational. We present tools and concepts of studying the interstellar medium in distant galaxies (Sec. 2), then summarize observational results for different galaxy types at high redshift (Sec. 3). We then discuss implications of the recent observations (Sec. 4) and what they tell us about conditions in early galaxies and galaxy formation in general (Sec. 5). We end by raising some of the key questions that can be addressed with new facilities: the JVLA and ALMA (Sec. 6 and Sec. 7).
We only consider molecular and fine structure line emission in this review. For a review of the few rotational molecular absorption line systems seen at high redshift to date, see Combes (2008) and Carilli & Menten (2002).
Over the last two decades, the star formation history of the universe, and the build up of stellar mass, has been well quantified as a function of galaxy environment and luminosity, back to cosmic reionization (z ∼ 10). It is clear that massive galaxies form most of their stars early, and that the majority of star formation occurs at z > 1. The dominant contribution to the cosmic star formation rate density shifts to higher star formation rate galaxies with redshift. The next major step in the study of galaxy formation is the delineation of the cool gas content of galaxies, and, in particular, the molecular interstellar medium (ISM) out of which stars forms, as a function of cosmic time.