Multi-Wavelength Properties of Submillimeter-Selected Galaxies

Citation

Hainline, Laura Jeannine (2008) Multi-Wavelength Properties of Submillimeter-Selected Galaxies. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/Q4DB-K461. https://resolver.caltech.edu/CaltechETD:etd-01242008-093655

Abstract

The resolution of the cosmic far-infrared background at long wavelengths has uncovered a population of high-redshift, highly infrared-luminous galaxies that indicate that a significant amount of highly obscured star formation occurred in the early history of the Universe, which has not been taken into account in studies of the cosmic star formation and mass assembly histories. Since such studies are important tests of hierarchical galaxy formation models in a cold dark matter- dominated universe, it is vital that the contribution of these luminous submillimeter-selected galaxies to the buildup of stellar mass in the Universe be understood and included when testing galaxy formation models. Progress in understanding the nature of submillimeter-selected galaxies has been slow, however, due to the faintness of the population outside of the submillimeter bands and the coarse resolution of the single-dish submillimeter surveys in which the galaxies were discovered. The slow progress has at least two important consequences: (1) there is still much that is unknown about submillimeterselected galaxies at all wavelengths, so the properties of and evolutionary predictions for the population are often inferred by making analogies to local galaxies of similar luminosity and assuming the high-redshift galaxies represent a uniform population of galaxies; and (2) the relationships of the submillimeter-selected galaxies to other high-redshift galaxy populations selected at other wavelengths remain poorly understood.

In this work we capitalize on recently-available observing resources of the Spitzer Space Telescope, at near- and mid-infrared wavelengths, and the Green Bank Telescope, at 1 cm, to improve the characterization of the largest representative sample of submillimeter-selected galaxies with spectroscopic redshifts. We combine our new data with data from the literature at a variety of wavelengths to place new constraints on the fundamental properties of gas mass, stellar mass, infrared luminosity, and spectral energy distribution, and use our new constraints to test a variety of assumptions which have been used in the past to predict and infer characteristics of submillimeter-selected galaxies. We also use our new data to compare different populations of high-redshift galaxies selected at different wavelengths in an effort to understand the relationships between the different types of galaxies.

First, we use observations of CO rotational line emission with the Green Bank Telescope to constrain the cold gas mass and gas conditions in several submillimeter-selected galaxies. We obtain the first detection of CO(1→0) emission from a submillimeter-selected galaxy, finding that the CO(4→3)/CO(1→0) brightness temperature ratio of ~ 0.26 suggests n(H₂) > 3 − 10 × 10² cm⁻³ and the presence of sub-thermally excited gas. The integrated line flux implies a cold molecular gas mass 4 times larger than the mass predicted from the CO(4→3) line, assuming a brightness temperature ratio of 1.0, suggesting that extrapolating molecular gas masses from J _upper ≥ 3 transitions of CO, which is the primary method of estimating molecular gas masses of high-z galaxies in the literature, leads to considerable uncertainties.

Next, we use deep imaging with the Multiband Imaging Photometer for Spitzer (MIPS) on the Spitzer Space Telescope of the spectroscopic sample of radio-detected submillimeter-selected galaxies of Chapman et al. (2005) to derive new estimates of the infrared luminosity for these objects. Our Spitzer data constrain the Wien side of the infrared spectral energy distribution peak of high-redshift submillimeter-selected galaxies, and thus are extremely important to determine the contribution of hot dust emission to the total infrared luminosity. We find that most submillimeter-selected galaxies do not have dominant contributions from hot dust at rest-frame mid-infrared wavelengths. We also find that the spectral energy distribution of the nearest ultraluminous infrared galaxy, Arp 220, is significantly different on average from most high-z submillimeter-selected galaxies and is thus a poor template with which to predict properties of submillimeter-selected galaxies even though it has been often used in the past. Using our new infrared luminosity estimates constrained by multiple infrared and submillimeter data points, we show that submillimeter-selected galaxies display a relatively tight, almost linear correlation between total infrared luminosity and radio luminosity, which is not largely different from the far-infrared–radio correlation of local galaxies selected by the Infrared Astronomical Satellite (IRAS).

We examine the rest-frame ultraviolet through near-infrared spectral energy distributions of the same sample of radio-detected submillimeter-selected galaxies with spectroscopic redshifts, obtained using measurements from imaging with the Infrared Array Camera (IRAC) on Spitzer in combination with observed-frame optical and near-infrared data from the literature. We find from these spectral energy distributions, which trace the stellar light from a galaxy in the absence of an active nucleus, that submillimeter-selected galaxies suffer significant extinction at rest-frame optical wavelengths, and that both stars and dust emission contribute to the near-infrared luminosity in many z > 2 submillimeter-selected galaxies. We estimate stellar masses for the individual galaxies in the sample using restframe H-band luminosities interpolated from the observed spectral energy distributions, obtaining a median stellar mass for the sample of 6 − 7 × 10¹⁰ Msun. By comparing our stellar mass estimates to molecular gas and dynamical mass estimates for 13 individual submillimeter-selected galaxies in our sample observed in CO emission lines, we determine that the molecular gas fraction in submillimeter-selected galaxies declines with increasing stellar mass, which is suggestive of an evolutionary trend. If the molecular gas masses for the 13 galaxies for which gas mass estimates are available are typical of the entire radiodetected submillimeter-selected galaxy population, then a typical lower limit to the total baryonic mass of submillimeter-selected galaxies is ~ 10¹¹ Msun, and these galaxies are unlikely to significantly increase their stellar mass in the current epoch of activity which is the source of their enormous infrared luminosity.

Lastly, we compare the IRAC and MIPS properties of submillimeter-selected galaxies and their stellar masses to those of high-redshift ultraviolet- and optically-selected galaxies, 24 μm-selected galaxies, and powerful radio galaxies. In the IRAC bands, submillimeter-selected galaxies are brighter and redder than ultraviolet-selected galaxies, suggesting they have higher dust content, higher stellar mass, a higher contribution from an active nucleus, or some combination of these factors. The near-infrared colors of submillimeterselected galaxies are most similar to those of high-z radio galaxies, objects which are known to contain powerful, obscured active nuclei. However, submillimeter-selected galaxies are fainter in the MIPS 24 μm band than high-z radio galaxies, suggesting that the dust in the submillimeter-selected galaxies is not heated to such high temperatures as in the radio sources. We find that the typical stellar mass of submillimeter-selected galaxies is larger by a factor of 3–4 than that of high-redshift ultraviolet-selected galaxies, roughly similar to the typical stellar mass of optically-selected high-z galaxies, and lower than the typical stellar mass of powerful high-z radio galaxies. These comparisons suggest that submillimeter-selected galaxies are among the more massive galaxies of their epoch, but not necessarily the most massive, as has been suggested in the literature. However, systematic errors in the stellar masses of any of the high-redshift galaxy samples of a factor of a few, which are certainly possible, can alter this conclusion.

Thesis Files