The infrared spectral energy distributions of star-forming galaxies

The spectral energy distribution (SED) is one of the key observational features of galaxies. It allows us both to gain insight into the physical properties of galaxies and compare populations selected at different wavelengths or redshifts. In this thesis we address all steps in the study of SEDs, including how to constrain them from the available data, how to derive the underlying properties of the galaxies, and how to use them to interpret and predict survey data. The first step is accomplished with a phenomenological SED model and a Markov Chain Monte Carlo fitting technique. General trends in the mid-IR spectra of a wide range of galaxy types are discussed as well as argument made for considering a continuum of SED shapes, as opposed to discrete templates. Based on this approach, predictions are made for the evolution of diagnostic mid-IR colours. We extend our model into the far-IR/sub-mm regime, and apply it to a sample of 22 170 /μm-selected, star-forming galaxies representing the brightest ~ 10% of the Cosmic Infrared Background (CIB) contributors. The sample spans the range of infrared luminosity log(L[sub IR]/L[sub Θ]) ~ 10-12 and redshift z ~0-l. From these model fits and standard relations we derive a host of underlying physical properties including: stellar mass, opacity, dust mass, dust temperature. These allow for direct comparisons with other samples of optical- or IR-selected star-forming galaxies, allowing us to explore in detail the nature of our sample. We find that, although not actually starbursts, these galaxies have enhanced SFRs compared with local quiescent spirals. Our highest luminosity sources are found to be colder than local templates, which is potentially explained by their having more extended star-formation activity than observed in local ULIGs. We conclude by addressing the implications of our study to other IR-selected samples. In particular we discuss the possible nature of the remaining ~ 90% of the CIB contributors. This thesis represents the first systematic fitting of SEDs of a sample of galaxies across the entire infrared wavelength range. This work is also the first use of Markov Chain Monte Carlo (MCMC) in infrared SED fitting, hence our error bars are much better defined and more believable than from earlier studies. The approach we take here can potentially be applied to larger samples from future multi-wavelength surveys extending to higher redshifts. This would allow for far greater precision in galaxy evolution studies than is currently possible.Science, Faculty ofPhysics and Astronomy, Department ofGraduat