Hydroxyl as a Probe of the Interstellar Medium
The molecular gas of the interstellar medium provides the raw material for star formation, yet its principle component – molecular hydrogen – is essentially invisible at radio wavelengths. Instead we must infer its presence and properties through the study of other tracer molecules, such as hydroxyl (OH). OH, with its four ground-rotational transitions at 1612, 1665, 1667 and 1720MHz is a challenging molecule to observe due both to the weakness of its lines and the complexity of their excitation. However, as I endeavor to demonstrate in this thesis, the complex excitation of OH holds valuable information about the host molecular gas. The satellite lines at 1612 and 1720MHz in particular, with their ubiquitous anomalous excitation, are much more sensitive to local conditions than the main lines which rarely diverge from their expected excitation at local thermodynamic equilibrium (LTE).
I demonstrate the wealth of information carried by the satellite line excitation by solving the decades-old mystery of the ‘satellite-line flip’: an intriguing profile pattern where the lines flip their relative excitation – one from absorption to stimulated emission and the other the reverse – across a closely blended double feature (Petzler et al., 2020). In Petzler et al. (2020) I show that 27 of the 30 flips found in the literature show the same orientation with respect to velocity, with the 1720MHz line in stimulated emission and the 1612MHz line in absorption in the blue-shifted component. All 27 of these examples also had a close on-sky and velocity association with Hii regions, leading us to propose that the flip originates in molecular gas both inside and outside a shock approaching the observer from an expanding Hii region, a claim we then support with extensive molecular excitation modelling.
In the second part of my thesis I develop a novel automated Gaussian decomposition algorithm specifically for the analysis of OH spectra (Amoeba Petzler et al., 2021a) that simultaneously fits features in all four lines. Amoeba utilises Bayesian model selection to take full advantage of not only the available data, but also prior expectations of its fitting parameters. Fitting all four lines simultaneously allows us to use the relatively strong signals in the main lines at 1665 and 1667MHz to infer and constrain the relatively weak but more informative signals in the satellite lines at 1612 and 1720 MHz.
Finally, I demonstrate the utility of Amoeba by applying it to OH spectra obtained as part of the Galactic Neutral Opacity and Molecular Excitation Survey (GNOMES) collaboration. This data set included 83 sets of observations from the Arecibo telescope and 15 from the Australia Telescope Compact Array (ATCA). The Arecibo data included off-source observations, allowing Amoeba to determine excitation temperature and optical depth in all four lines as well as the total OH column density. The ATCA data allowed Amoeba to find all four optical depths. I then investigate the relationship between our identified OH features (109 in total, 58 from our Arecibo data) and the Hi cold neutral medium (CNM) components where our sightlines matched. A preliminary analysis finds no clear correlations between OH and CNM parameters on a component-by-component basis, but there are some hints of differences in the populations of OH-associated CNM clouds. The lack of clear correlations itself may imply a decoupling of the molecular gas from the CNM gas in which it formed.
