The Keck Baryonic Structure Survey (KBSS)
The Keck Baryonic Structure Survey (KBSS) is a large, targeted spectroscopic survey designed to jointly probe galaxies and their gaseous environments at the peak of galaxy assembly (z~2-3). The survey comprises 15 independent fields centered on a bright background quasar; the total survey area is 0.24 square degrees, comparable to many of the legacy fields. The KBSS galaxy sample is selected from deep optical and near-infrared imaging and subsequently followed up with spectroscopic observations in the rest-UV (with Keck/LRIS) and rest-optical (with Keck/MOSFIRE) bandpasses. My role since MOSFIRE's commissioning in 2012 has been to lead the near-infrared component of the KBSS survey, which now encompasses observations of more than 1100 individual galaxies.
Above image courtesy Gwen Rudie (Carnegie)
Near-infrared spectroscopy of KBSS galaxies was conducted in the early 2000s with Keck/NIRSPEC, but follow-up with MOSFIRE over the last several years has been significantly more efficient, thanks to its more sensitive detector and multiplexing capability. The figure above compares the spectra of two different galaxies taken with NIRSPEC (on the left) and MOSFIRE (on the right). With MOSFIRE, we now have multiple strong-line measurements, including H-alpha and [N II] as shown, for hundreds of individual galaxies. These measurements are critical for constraining galaxies' gas-phase metallicities, electron densities, and ionization parameters.
Near-infrared spectoscropy of individual high-z galaxies
The KBSS contains more than 700 galaxies at z~2-3, with quality near-infrared spectroscopic observations of ~380 individual systems. The above figure comes from a paper submitted in August 2016, in which I detail the nebular properties of the z~2-3 KBSS galaxies and conclude that the primary difference with respect to local galaxies is an increase in the overall degree of excitation. At the same time, high-z KBSS galaxies appear to be more chemically evolved (with higher N/O and O/H) than local galaxies with similar excitation conditions, meaning that galaxies in the early Universe must have harder ionizing radiation fields than z~0 galaxies at fixed oxygen abundance. The most likely explanation for this trend is a systematic difference in the star-formation histories of galaxies at z~2-3 and z~0, even at fixed stellar mass.
Measuring O/H in high-redshift galaxies
The nebular spectra of galaxies originates in the ionized gas surrounding young, massive stars and thus reflects the combined properties of both the gas and stars. Both local galaxies and high-redshift galaxies occupy relatively tight loci in multi-dimensional line-ratio space, which implies strong correlations between the physical properties driving their spectra. However, it remains unclear if such correlations (for example, between ionization parameter and metallicity) are redshift-invariant, limiting the usefulness of empirical abundance calibrations based on z~0 samples.
There is also evidence to suggest that many of the emission line ratios observed for high-excitation nebulae respond more sensitively to changes in the shape and normalization of the ionizing radiation field than to changes in the gas-phase oxygen abundance. This effect is more pronounced at high-redshift because nearly all z~2-3 galaxies exhibit high levels of nebular excitation. Thus, efforts to measure their O/H must also consider their special ionization and excitation conditions.
Part of my ongoing work with the KBSS sample involves using a combination of BPASSv2 stellar population models and photoionization modeling with Cloudy to obtain self-consistent estimates of O/H, N/O, and ionization parameter U for individual galaxies. In a paper published earlier this year, we proved the utility of this approach using deep composite spectra constructed from the rest-UV and rest-optical spectra of a representative subset of KBSS galaxies.