Understanding the H2/HI Ratio in Galaxies
Total Page:16
File Type:pdf, Size:1020Kb
Mon. Not. R. Astron. Soc. 394, 1857–1874 (2009) Printed 15 July 2009 (MN LATEX style file v2.2) Understanding the H2/HI Ratio in Galaxies D. Obreschkow and S. Rawlings Astrophysics, Department of Physics, University of Oxford, Keble Road, Oxford, OX1 3RH, UK Accepted 2009 January 12 ABSTRACT galaxy We revisit the mass ratio Rmol between molecular hydrogen (H2) and atomic hydrogen (HI) in different galaxies from a phenomenological and theoretical viewpoint. First, the local H2- mass function (MF) is estimated from the local CO-luminosity function (LF) of the FCRAO Extragalactic CO-Survey, adopting a variable CO-to-H2 conversion fitted to nearby observa- −5 −1 tions. This implies an average H2-density ΩH2 = (6:9±2:7)·10 h and ΩH2 =ΩHI = 0:26±0:11 galaxy in the local Universe. Second, we investigate the correlations between Rmol and global galaxy properties in a sample of 245 local galaxies. Based on these correlations we intro- galaxy duce four phenomenological models for Rmol , which we apply to estimate H2-masses for galaxy each HI-galaxy in the HIPASS catalog. The resulting H2-MFs (one for each model for Rmol ) are compared to the reference H2-MF derived from the CO-LF, thus allowing us to determine the Bayesian evidence of each model and to identify a clear best model, in which, for spi- galaxy ral galaxies, Rmol negatively correlates with both galaxy Hubble type and total gas mass. galaxy Third, we derive a theoretical model for Rmol for regular galaxies based on an expression for their axially symmetric pressure profile dictating the degree of molecularization. This model is quantitatively similar to the best phenomenological one at redshift z = 0, and hence repre- sents a consistent generalization while providing a physical explanation for the dependence galaxy galaxy of Rmol on global galaxy properties. Applying the best phenomenological model for Rmol to the HIPASS sample, we derive the first integral cold gas-MF (HI+H2+helium) of the local Universe. Key words: ISM: atoms – ISM: molecules – ISM: clouds – radio lines: galaxies. 1 INTRODUCTION the local Universe (e.g. 0.14, 0.42, 1.1 stated respectively by Boselli et al. 2002, Keres et al. 2003, Fukugita et al. 1998). The Interstellar Medium (ISM) plays a vital role in galaxies as their Ultimately, the uncertainties of H2-measurements hinder the primordial baryonic component and as fuel or exhaust of stars. Hy- reconstruction of cold gas masses Mgas = (MHI + MH2 )/β, where drogen constitutes 74 per cent of the mass of the ISM. When it is β ≈ 0:74 is the standard fraction of hydrogen in neutral gas with cold and neutral it coexists in the atomic phase (HI) and molecular the rest consisting of helium (He) and a minor fraction of heav- phase (H ). While the former follows a smooth distribution across 2 ier elements. The limitations of comparing MHI to Mgas caused by large galactic substructures, the latter is found in dense molecular the measurement uncertainties of MH2 culminate in severe difficul- clouds (Drapatz & Zinnecker 1984) acting as the sole creches` of ties to compare statistically tight cold gas-mass functions (MFs) of newborn stars. The dissimilar but interlinked roles of HI and H2 in modern cosmological simulations with precise HI-MFs extracted substructure growth and star formation have caused a growing in- from HI-surveys, such as HIPASS. Both simulations and surveys terest in simultaneous observations of both phases and cosmologi- have reached statistical accuracies far better than any current model galaxy cal simulations that distinguish between HI and H2. for Rmol , and hence the comparison of observations with simula- galaxy Extragalactic observations of HI often use its prominent 21- tions is mainly limited by the uncertainty of Rmol . cm emission line, and currently comprise several thousand galaxies As an illustration, Fig. 1 displays the observed HI-MF from (HI Parkes All Sky Survey HIPASS, Barnes et al. 2001), and a max- the HIPASS sample (Zwaan et al. 2005) together with several sim- imum redshift of z = 0:2 (Verheijen et al. 2007). By contrast, most ulated HI-MFs. The latter are based on the cold gas masses of H2-estimates must rely on indirect tracers, such as CO-lines, with the simulated galaxies produced by two different galaxy forma- uncertain conversion factors. Consequently, the phase ratio of neu- tion models applied to the Millennium Simulation (Bower et al. tral hydrogen Rmol ≡ dMH2 =dMHI and its value for entire galaxies 2006; De Lucia & Blaizot 2007). We have converted these cold gas galaxy galaxy Rmol ≡ MH2 =MHI remain debated, and estimates of the universal masses into HI-masses using four models for Rmol from the lit- universe density ratio Rmol ≡ ΩH2 =ΩHI vary by an order of magnitude in erature (Young & Knezek 1989; Keres et al. 2003; Boselli et al. c 2009 RAS 2 D. Obreschkow and S. Rawlings 2 THE VARIABLE CO-TO-H2 CONVERSION 2.1 Background: basic mass measurement of HI and H2 HI emits rest-frame 1.42 GHz radiation (λ = 0:21 m) originating from the hyperfine spin-spin relaxation. Especially cold HI (T ∼ 50 − 100 K, see Ferriere` 2001) also appears in absorption against background continuum sources or other HI-regions, but makes up a negligible fraction in most galaxies. Within this assumption, HI can be considered as optically thin on galactic scales, and hence the HI-line intensity is a proportional mass tracer, !2 MHI 5 S HI Dl = 2:36 · 10 · −1 · ; (1) M Jy km s Mpc where S HI is the integrated HI-line flux density and Dl is the lumi- nosity distance to the source. Unlike HI-detections, direct detections of H2 in emission rely Figure 1. Dots represent the observed HI-MF by Zwaan et al. (2005); on weak lines in the infrared and ultraviolet bands (Dalgarno 2000) lines represent simulated HI-MFs derived from the semi-analytic models and have so far been limited to the Milky Way and a few nearby by Bower et al. (2006, red lines) and De Lucia & Blaizot (2007, blue lines). galaxies (e.g. Valentijn & van der Werf 1999). Occasionally, H2 has galaxy also been detected at high redshift (z ≈2–4) through absorptions The four models of Rmol were adopted or derived from Young & Knezek (1989, solid lines), Keres et al. (2003, dashed lines), Boselli et al. (2002, lines associated with damped Lyman α systems (Ledoux et al. dotted lines), and Sauty et al. (2003, dash-dotted lines). 2003; Noterdaeme et al. 2008). All other H2-mass estimates use indirect tracers, mostly rotational emission lines of carbon monox- ide (CO) – the second most abundant molecule in the Universe. The most frequently used CO-emission line stems from the re- laxation of the J = 1 rotational state of the predominant iso- topomer 12C16O. Radiation from this transition is referred to as CO(1–0)-radiation and has a rest-frame frequency of 115 GHz 2002; Sauty et al. 2003). The figure adopts the Hubble constant (λ = 2:6 · 10−3 m), detectable with millimeter telescopes. The con- of the Millennium Simulation, i.e. h = 0:73, where h is defined version between CO(1–0)-radiation and H -masses is very subtle −1 −1 2 by H0 = 100 h km s Mpc with H0 being the present-day Hub- and generally expressed by the X-factor, ble constant. The differential gas density φHI is defined as φHI ≡ −2 dρHI=d log MHI, where ρHI(MHI) is the space density (i.e. number NH2 =cm −20 X ≡ −1 · 10 ; (2) per volume) of HI-sources of mass MHI. In Fig. 1 different models ICO=(K km s ) for galaxy formation are distinguished by colour, while the models where N is the column density of molecules and I is the inte- of Rgalaxy are distinguished by line type. Clearly, any conclusion re- H2 CO mol grated CO(1–0)-line intensity per unit surface area defined via the garding the two galaxy formation models based on their HI-MFs is surface brightness temperature T in the Rayleigh-Jeans approxi- affected by the choice of the model for Rgalaxy. R ν R mol mation. Explicitly, I ≡ T dV = λ T dν, where V is the radial This paper presents a state-of-the-art analysis of the galaxy- CO ν ν velocity, ν is the frequency, and λ = jdV=dνj is the wavelength. This dependent phase ratio Rgalaxy, the H -MF and the integral cold gas- mol 2 definition of the X-factor implies a mass-luminosity relation analo- MF (HI+H +He), utilizing various observational constraints. In 2 gous to Eq. (1) (see review by Young & Scoville 1991), Section 2, the determination of H2-masses via CO-lines is revisited 2 !2 and an empirical, galaxy-dependent model for the CO-to-H2 con- MH2 λ S CO Dl = 580 · X · · −1 · ; (3) version factor (X-factor) is derived from direct measurements of M mm Jy km s Mpc a few nearby galaxies (Boselli et al. 2002 and references therein). R In Section 3, this model is applied to recover an H2-MF from the where S CO ≡ S CO,νdV denotes the integrated CO(1–0)-line flux CO-luminosity function (LF) by Keres et al. (2003). The resulting and Dl the luminosity distance. S CO,ν is the flux density per unit H2-MF significantly differs from the one obtained by Keres et al. frequency, for example expressed in Jy, and thus S CO has units like −1 (2003) using a constant X-factor. Section 4 presents an indepen- Jy km s .