Star Formation in Dwarf Galaxies: Using the Radio Continuum as an Extinction-Free Probe
Kitchener, Ben Gerald
To eliminate uncertainties introduced by extinction by dust in the optical, we examine to what extent the radio continuum (RC) can probe star formation in dwarf galaxies. Star formation (SF) drives galaxy formation and evolution; acquiring accurate measurements of SF thus becomes crucial in order to understand galaxies. As radio technology improves further, RC surveys will probe the fainter, more quiescent regime of the radio sky. Having a robust manner by which to convert RC luminosities to star formation rate (SFR) has the potential to provide millions of independent SFR measurements out to intermediate redshifts. In order to calibrate the RC to infer SFR, the 40 dwarf galaxies that make up the LITTLE THINGS sample were chosen as the bedrock of the thesis due to the large range of galactic parameters that they cover. RC observations of these galaxies were taken with the VLA between L- and Ka-band (1–33GHz) using the B-, C-, and Darrays, yielding images with 3–10′′ resolution and rms noise levels between 3 and 15 μJy beam−1. On a global scale, 27 out of the 40 dwarf galaxies exhibited RC emission above the detection threshold, 17 of which were new RC detections. The general picture is an interstellar medium (ISM) largely void of RC emission, interspersed by isolated pockets of RC associated with SF regions; this general picture agreed with what was expected given current models of dwarf galaxies—weaker magnetic fields in the ISM leading to a higher escape of CRe (and resulting reduction in RCNTh emission). This was also backed-up by the relatively low RCNTh fraction—61 ± 7% at C-band. The observed RC–SFR relation was calibrated to allow the observed RC luminosity of a gas rich dwarf galaxy to be used to infer the SFR; the calibration takes the form SFR [M⊙ yr−1] = 5 × 10−18(RC [WHz−1])0.85. On a resolved basis, only the RCNTh was examined—this is because whether scales of 1 pc, or 1 kpc are investigated, the relationship between the Hα (current SF) and RCTh was not expected to change. Calibrating the resolved RCNTh–SFR relation was best done when using discrete SF regions which varied from 10s up to 100s of pc in size. On these scales, the calibration allows the SFR to be inferred from an observed RCNTh luminosity, and takes the form SFR [M⊙ yr−1] = 1.36 × 10−23(LNTh [WHz−1])1.15. This calibration, however, is only valid for resolved regions forming stars at a rate & 2 × 10−4M⊙ yr−1. Despite the low flux densities of RCNTh measured from these discrete SF regions, the RCNTh still works well as a SFR tracer whereas Hα, which is largely dependent on stars with mass & 18M⊙, and is thus dependent on the high mass tail of the stellar IMF, will suffer from stochasticity. In a few dwarfs, the equipartition magnetic field strength reaches as high as 30 μG in multiple 100 pc regions, and in one case, 70 μG. However, generally, the weaker magnetic fields in the ISM give the CRe longer lifetimes, and thus more time to be advected out of the galaxy with the magnetic fields frozen into the gas in outflows, or diffuse. This explains in part the lack of RCNTh emission observed in the ISM of dwarf galaxies. Through implementing a simple galactic CRemodel, itwas found that the RCNTh emission associated with the CRe can be used as a SF tracer from approximately 5 up to 70Myr following a burst of SF, while RCTh can be used in its absence prior to 5Myr. The RCNTh luminosity reaches its peak approximately 55Myr after the SF episode, but actually remains nearly constant over the 60Myr following the SF episode, highlighting its potential to be used to infer SFR. The CRemodel also tracked the evolution of the RCNTh spectral index with time. Between values of about −0.4 and −0.7, the RCNTh spectral index can be calibrated to infer the time elapsed since a burst of SF through t[Myr] = −25αNTh. RCNTh spectral indices of −0.8 are consistent with ages between 20 and 55Myr, suggesting that the oft observed spectral index of −0.8 in galaxies may come from the fact that C-band RCNTh emission is dominated by the steep spectral indices of −0.8 from these older SF regions (20–55Myr). For the galaxies that displayed RCNTh emission that was bright enough and sufficiently well resolved, a spectral decomposition of the RC spectrum was performed to infer Hα-independent RCTh, RCNTh, and RCNTh spectral index maps. The spectral decomposition showed DDO50 and NGC1569 to have a low thermal fraction of 23% and 10%, respectively, at C-band, while NGC2366 and NGC4214 were shown to have higher thermal fractions of 48% and 66%, respectively. In summary, dwarf galaxies are not only faint in the RC due to their lower SF activity, but they are also fainter than expected due to CRe escape. Nonetheless, the RC can be used to probe SF in dwarf galaxies not only on a global scale, but also within discrete SF complexes 10s to 100s of pc in size. Theoretically, the RC can be used right from the onset of a burst of SF, where RCTh will dominate, up to ∼ 70Myr, at which point RCNTh will dominate. Calibrated by the RC observations in this thesis, both resolved and global SFRs of gas rich, low mass galaxies can be inferred with an uncertainty of ±0.2 dex; the relations allow SFRs of between approximately 2×10−4 and 0.1M⊙ yr−1 to be inferred.