Authors: The Fermi/LAT collaboration
Gamma Ray astronomy has helped unveil many of the Universe’s secrets, from the most energetic explosions to the possibility of probing dark matter directly. Another lesser-known but equally important facet of this subfield is the ability to study the distribution of interstellar gas, particularly in dense, star-forming molecular clouds. The Large Area Telescope (LAT) on the Fermi gamma ray observatory has been surveying the sky at high energies for the last four years (see this astrobite for a nice summary of recent Fermi science results), and the Fermi/LAT team recently released a study of three nearby star-forming regions. Their results are strong evidence that traditional methods of surveying molecular clouds are missing a sizable fraction of the gas. They also confirm through gamma ray observations that the so-called “X-factor” – the ubiquitously used conversion factor between Carbon Monoxide (CO) and total molecular gas – might vary within galaxies.
Gamma rays as ISM tracers
Tracing interstellar gas is a major challenge in astronomy; in molecular clouds, CO is used to infer the total amount of gas present using the “X-factor” (a.k.a. XCO; see this excellent astrobite for more). This number is often assumed to be constant, but there has been some evidence in previous studies that XCO may actually vary by a factor of a few with environment. Measuring XCO directly is difficult: the most straightforward method would involve simultaneous observations of CO and molecular hydrogen (H2), but the latter is unobservable in molecular clouds. The Fermi/LAT team present another, indirect method of inferring this important parameter, and it’s not something one would usually associate with cold interstellar gas: gamma rays!
The method used in this paper relies on the fact that cosmic rays – nuclei and other subatomic particles that have been accelerated to insanely high speeds – permeate the Universe. They are thought to be generated primarily in and shortly after supernova explosions and in Active Galactic Nuclei (AGN), the active supermassive black holes at the centers of galaxies (see these astrobites for more on cosmic rays). But the important fact for this Fermi study is that cosmic rays are simply particles that can interact with other particles in interstellar space. When a cosmic ray encounters an atom or molecule, the resulting collision is highly energetic and consequently produces the highest energy photons: gamma rays. If the flux of cosmic rays into the interstellar medium (ISM) is known (or assumed), observing at gamma ray energies can thus probe the distribution of interstellar gas. Furthermore, cosmic rays don’t really care how hot or cold the gas is, or what phase (atomic vs. molecular) it is in, for the most part; the “cross-section”, or probability of a collision that leads to a gamma ray being produced, is almost entirely independent of the chemical or thermodynamic state of the gas, and thus directly probes the overall distribution of matter. Gamma rays are also so energetic that they pass through basically everything unimpeded, so we don’t have to worry about optical depth effects (i.e. the possibility of photons being absorbed along their path to us) that plague observations from the x-ray to the radio. The paper’s analysis depends centrally on this important point: if the ISM is transparent to gamma rays, then the total gamma ray emission can be modeled as a sum of the emission from (1) atomic hydrogen (H I), (2) molecular gas (presumably traced by CO), and (3) non-ISM sources (e.g. Galactic background, point sources, and Inverse Compton scatterings). Fitting this model to observed gamma ray data allows the gamma ray emissivity “q” (i.e. the amount of gamma rays produced) for H I and CO to be separately calculated.
The study presented in this paper combines Fermi/LAT observations with measurements of H I, CO, and extinction (i.e. dust) in order to measure q in three nearby (less than 300 pc away) regions containing molecular clouds: Chamaeleon, R Coronae Australis, and the Cepheus and Polaris flare region. Using these auxiliary datasets allows the authors to (1) test the assumption that the cosmic ray flux is isotropic (result: it isn’t as isotropic as we assume), (2) estimate the amount of interstellar gas absent from the H I and CO data, and (3) directly compute the X-factor. The gamma ray maps for all three regions are shown in Figure 1 with CO contours from Dame et al. (2001) overlaid.
The missing gas
Assuming a constant gas-to-dust ratio (which is thought to be roughly accurate), the dust traced by an extinction map also corresponds to a particular distribution of gas, and since gas should be either atomic or molecular, H I and CO maps (with the latter converted to total gas through XCO) of the same region should add up to the same gas distribution. The authors tested this idea by converting the extinction map to a gas map using this gas-to-dust ratio, then subtracting the sum of the H I and CO maps. They call the resulting map “A_Vres”, for “visual extinction (A_V) residual”, and find that there seem to be concentrations of gas near the borders of the star-forming regions that are unaccounted for in the H I and CO surveys. Figure 2 shows the H I, CO, and A_Vres maps for the Chamaeleon region. The authors thus add an additional term to their gamma ray model to account for any gas that isn’t traced by either H I or CO, and the model fit improves greatly when they include this additional term, confirming the presence of this unseen gas. This isn’t the first time this “dark” gas has been “seen” at cloud interfaces; submillimeter dust continuum observations have also predicted that CO isn’t always telling the whole story.
Deriving the X-factor
Since the cross section for cosmic ray interaction doesn’t depend on whether the gas is atomic or molecular, the gamma ray emissivity for H2 is simply twice that of H I. Because H2 comprises the majority of the molecular gas by mass, the X-factor can thus be calculated as qCO/(2 qHI). The authors compute XCO by plotting qCO vs. qHI at several gamma ray energy bins (see Figure 3) and then using a maximum likelihood linear fitting technique (this statistical method is similar to the more commonly known linear least-squares fitting, but is does not assume that the quantities in question are independent). The slope of the linear fit will then be the X-factor simply due to the choice of units plotted.
The most commonly assumed value for XCO in ISM studies is about 2 x 1020 cm-2 (K km s-1)-1*; the Fermi/LAT team derives values more than a factor of two lower in all three regions studied. While a factor of two is not headline news in the ISM community, the systematically lower values inferred from gamma rays are sure to raise some eyebrows. Furthermore, XCO in the Cepheus and Polaris flare region is about 2/3 the value computed for the other two regions, which suggests that the X-factor can vary over scales of only about 300 pc, the distance between the regions. Often XCO is taken to be constant over entire galaxies, so the conclusions from this Fermi/LAT study will have repercussions for astronomers interested in studying the gas content of molecular clouds in galaxies near and far.
*the units for XCO may seem obtuse, but it is simply units of column density (molecules of H2 per cm2) divided by units of CO luminosity (in the radio astronomy convention of temperature multiplied by velocity). This radio convention comes from the fact that (1) intensities are measured in units of temperature in K, and (2) the data are three-dimensional (two dimensions on the sky and one in velocity), and so integrating along the velocity axis (which should correspond to all the gas along the line of sight, modulo optical depth effects) gives temperature times velocity in K km/s.
Latest posts by Chris Faesi (see all)
- Astrophysical Classics: The Observed Relation between Star Formation and Gas in Galaxies – November 5, 2013
- Dark in the Lightness – October 9, 2013
- Kepler’s view on the gravity of stars – August 29, 2013