Even though years have been spent studying how star formation is quenched in galaxies, we are still unsure of the physical mechanisms that drive the halt in forming stars. Quenching is presumed to be caused by feedback, which is the expulsion of gas and dust, from various sources inside the galaxy. Unfortunately, feedback is the crux of the current theory of galaxy evolution and how elliptical galaxies are formed. With this in mind, we have gathered a small selection of unusual late-stage merging galaxies at intermediate redshift. Not only are they very massive, but also compact, bright, blue, still forming stars, and have extremely fast outflows. Using data acquired from Keck/NIRSPEC, the Sloan Digital Sky Survey, and the Multiple Mirror Telescope, we examined forbidden emission lines of these extraordinary galaxies. We found strong indicators of broad line emission tracing the outflow, which is seen in both emission and absorption, indicating the outflows have multiple phases.

In this dissertation we study the properties of active galactic nuclei (AGN), which are powered by the accretion activity of supermassive black holes residing at the centers of galaxies. While observations propose that growth of AGN and galaxies are globally tied, we investigate whether this connection exists in individual galaxies. We also investigate various AGN selection techniques and star formation rate (SFR) estimates using multi-wavelength data from Chandra, Spitzer and rest-frame optical spectra from the Keck telescope.

We find that combining multi-wavelength identification techniques provides a more complete AGN sample, as each selection method suffers from selection biases. In particular, all selection techniques are biased against identifying AGN in lower mass galaxies. Once stellar mass selection biases are taken into account, we find that AGN reside in galaxies with similar physical properties (i.e., SFR) as inactive galaxies.

We find that while AGN are prevalent in both star-forming and quiescent galaxies, AGN of a given accretion rate are more likely to reside in star-forming galaxies. The probability of fueling an AGN does not strongly depend on SFR for a star-forming galaxy, though it decreases when star formation is shut down in quiescent galaxies. We also find no evidence for a strong correlation between SFR or stellar mass of the host galaxy and AGN luminosity. These results indicate that while both AGN and galaxy growth are reliant on the same fuel, enhanced star formation activity does not necessarily go hand-in-hand with increased AGN activity. While the star formation activity of galaxies can be traced with various indicators, our investigations indicate that extrapolations from mid-infrared data using calibrations based on local galaxies overestimates SFRs at higher redshift. We show that a combina- tion of mid-infrared and far-infrared data provide a more reliable SFR estimation than the mid-infrared data alone. We also find that the robustness of UV-based SFRs depends on the extinction correction method used. We find a relatively small fraction of z ∼ 2 galaxies have SFRs from infrared observations that are elevated relative to other SFR tracers, and we do not find any contribution from AGN in this excess.

This dissertation studies the properties of outflows driven by active galactic nuclei (AGNs) and their impact on the evolution of galaxies. AGNs are predicted by theoretical models of galaxy formation to provide the necessary feedback to produce realistic galaxies. In theoretical models, AGNs provide feedback by driving outflows that remove gas from the host galaxy, thereby quenching star formation in massive galaxies and producing scaling relations between supermassive black holes and galaxies. Despite being indispensable in theory, critical open questions remain for AGN-driven outflows from an observational perspective.

This dissertation first presents two studies using data from the MOSDEF survey, a large spectroscopic survey of galaxies when the Universe was about 3.5 billions years old (z~2). The first study focuses on the incidence and physical properties of AGN-driven outflows at z~2. We show that AGN-drive outflows are at least as prevalent at z~2 as they are in the local Universe. They are fast and extend to distances comparable to the size of the host galaxy. Using emission line ratio diagnostics, we find our data consistent with the picture of negative AGN feedback, where AGN-driven outflows suppress star formation, and find no evidence of galaxy-wide positive AGN feedback.

The second study focuses on the demographics of galaxies hosting AGN-driven outflows and the relation between outflow properties and the galaxy and AGN population at z~2. We show that AGN-driven outflows are a ubiquitous phenomenon occurring across the galaxy population and in different phases of galaxy evolution, both before and after quenching. By measuring the energetics and correlations of AGN-driven outflows, we find that the outflows are more energetic at z~2 than in the local Universe, where AGNs are more powerful on average. We find that the outflows remove gas at a rate comparable to or faster than gas is being converted into stars. This shows that AGN-driven outflows at z~2 are capable of regulating star formation in the host galaxy.

The third study in this dissertation presents integral field spectroscopy of a nearby ultraluminous infrared galaxy (ULIRG) and AGN Mrk 273. The study focuses on the extended ionized gas on scales of ~20 kpc. We detect for the first time highly ionized gas in one of the extended nebula surrounding the galaxy. From this, we show that shocks contribute significantly to the ionization of the gas in the extended nebulae, mixed with AGN photoionization. Our data is compatible with theoretical models in which AGNs drive a multiphase outflow, and slower-moving extended cold gas filaments form out of a more spatially confined but faster warm outflow. Our data suggests that AGNs play an important role in ejecting gas in the ULIRG phase of galaxy evolution.

The galaxy distribution in the joint space of stellar mass and star formation rate (SFR) is observed to be bimodal, with distinct star-forming (high SFR) and quiescent (low SFR) populations across many orders of magnitude in stellar mass. The absence of a significant population of intermediate-stage galaxies implies that galaxies tend to cease star formation and become quiescent relatively rapidly, and the physical origins of this rapid quenching are an open question in the field of galaxy evolution.

Recent progress in cosmological simulations of dark matter structure evolution, as well as in statistical modeling of how galaxies inhabit dark matter halos, suggest a nuanced coevolutionary relationship between galaxies and the dark matter halos galaxies reside in (the "galaxy–halo connection"), in which the statistical galaxy content of a halo depends on more than halo mass. One class of these galaxy–halo connection models are those that incorporate galaxy assembly bias, a general term for the dependence of galaxy properties on halo properties other than mass. For example, distinct stellar-to-halo mass relations for star-forming and quiescent central galaxies is a form of galaxy assembly bias that could manifest as an anticorrelation of galaxy SFR with clustering amplitude, if the trend is independent of stellar (or halo) mass.

In this dissertation we present the first measurements of galactic conformity, or the tendency of neighboring galaxies to share the star formation properties of an adjacent central galaxy, on two-halo scales beyond the local universe (to z~1). We then measure the clustering of isolated galaxies as a proxy for central galaxies, separately for star-forming and quiescent galaxies, to test predictions of galaxy–halo models reflecting galaxy assembly bias. Finally, we measure the joint dependence of clustering on stellar mass and SFR. With mock galaxy catalogs derived from simulations and an empirical galaxy evolution model we quantify and compensate for the effects of systematic biases on these measurements. A parallel theme of this dissertation is demonstrating how existing galaxy surveys beyond the local universe are at the cusp of probing the volumes needed for statistically significant tests of theoretical predictions of various models of galaxy evolution.