THE MOLECULAR GAS IN LUMINOUS INFRARED GALAXIES. II. EXTREME PHYSICAL CONDITIONS AND THEIR EFFECTS ON THE X-co FACTOR

In this work, we conclude the analysis of our CO line survey of luminous infrared galaxies (LIRGs: L-IR greater than or similar to 10(11) L-circle dot) in the local universe (Paper I) by focusing on the influence of their average interstellar medium (ISM) properties on the total molecular gas mass estimates via the so-called X-co = M(H-2)/L-co,L-1-0 factor. One-phase radiative transfer models of the global CO spectral line energy distributions (SLEDs) yield an X-co distribution with similar to (0.6 +/- 0.2) M-circle dot (K km s(-1) pc(2))(-1) over a significant range of average gas densities, temperatures, and dynamic states. The latter emerges as the most important parameter in determining X-co, with unbound states yielding low values and self-gravitating states yielding the highest ones. Nevertheless, in many (U)LIRGs where available higher-J CO lines (J = 3-2, 4-3, and/or J = 6-5) or HCN line data from the literature allow a separate assessment of the gas mass at high densities (>= 10(4) cm(-3)) rather than a simple one-phase analysis, we find that near-Galactic X-co similar to (3-6) M-circle dot (K km s(-1) pc(2))(-1) values become possible. We further show that in the highly turbulent molecular gas in ULIRGs, a high-density component will be common and can be massive enough for its high X-co to dominate the average value for the entire galaxy. Using solely low-J CO lines to constrain X-co in such environments (as has been the practice up until now) may have thus resulted in systematic underestimates of molecular gas mass in ULIRGs, as such lines are dominated by a warm, diffuse, and unbound gas phase with low X-co but very little mass. Only well-sampled high-J CO SLEDs (J = 3-2 and higher) and/or multi-J observations of heavy rotor molecules (e.g., HCN) can circumvent such a bias, and the latter type of observations may have actually provided early evidence of it in local ULIRGs. The only way that the global X-co of such systems could be significantly lower than Galactic is if the average dynamic state of the dense gas is strongly gravitationally unbound. This is an unlikely possibility that must nevertheless be examined, with lines of rare isotopologues of high gas density tracers (e.g., (HCN)-C-13, high-J (CO)-C-13 lines) being very valuable in yielding (along with the lines of the main isotopes) such constraints. For less IR-luminous, disk-dominated systems, we find that the galaxy-averaged X-co deduced by one-phase models of global SLEDs can also underestimate the total molecular gas mass when much of it lies in an star-formation-quiescent phase extending beyond a central star-forming region. This is because such a phase (and its large X-co) remains inconspicuous in global CO SLEDs. Finally, detailed studies of a subsample of galaxies find ULIRGs with large amounts (similar to 10(9) M-circle dot) of very warm (>= 100 K) and dense gas (>= 10(5) cm(-3)), which could represent a serious challenge to photon-dominated regions as the main energy portals in the molecular ISM of such systems.