While star-forming galaxies are likely the dominant force behind cosmic reionization, the relative contributions of different galaxy populations remains uncertain. Understanding how ionizing photons are produced and subsequently escape from these galaxies in the early Universe is crucial for reducing such uncertainties in existing models. In this dissertation, I conduct four observational studies of star-forming galaxies at z~3-5 to build a more complete picture of ionizing-photon production and escape. These works utilize observations of the rest-UV portion of the electromagnetic spectrum to examine the interplay between massive stars, neutral gas, and dust in star-forming galaxies. These studies are conducted at the highest redshifts feasible for this type of analysis by avoiding complete attenuation of relevant observational signals from the neutral intergalactic medium. I present the redshift evolution of rest-UV spectral features out to z~5 using a data set from the DEIMOS 10K Survey, alongside comparisons to similar analyses performed at z~2-4. I find a relationship between Lya emission line strength and low-ionization absorption line strength and conclude that the neutral-gas covering fraction of a galaxy simultaneously modulates the two. The redshift evolution of this relation indicates that the ionizing photon production efficiency xi_ion of this z~5 sample is elevated as compared to z~2-4 samples. I examine the effects of foreground contamination in direct measurements of the Lyman continuum (LyC) at z~3 using data from the Keck Lyman Continuum (KLCS) Survey and novel Hubble Space Telescope imaging. After the removal of contamination, I measure an average escape fraction of ionizing radiation of f_esc=0.06+-0.01 at z~3, and present 13 confirmed, individual LyC leakers. Using this cleaned KLCS sample, I measure trends between f_esc and different galaxy properties in order to elucidate which galaxy populations contribute most strongly to the ionizing emissivity during reionization, while simultaneously determining which galaxy properties are most appropriate to use as a proxy of f_esc at z>6. I attempt to measure the correlation between f_esc and star-formation rate surface density, an important diagnostic of f_esc found in the local Universe, and make recommendations for the total sample size required to determine whether this correlation exists out to z~3. I present a negative correlation between f_esc and stellar mass, indicating that lower-mass galaxies (<~10^9 M_sun) likely drive reionization. I also find a negative correlation between f_esc and E(B-V), underlining the importance of the configuration of neutral gas and dust in determining the escape fraction of an individual galaxy. Finally, I present a lack of significant trend between f_esc and stellar age or specific star-formation rate, indicating that recent star-formation does not strongly affect the conditions of the neutral-phase interstellar and circumgalactic media at the masses of the KLCS sample. Continued explorations of the physics behind f_esc and xi_ion at z~3-5 are crucial for building a comprehensive understanding of the reionization process.

Rest-frame ultraviolet (UV) and optical spectroscopy provides valuable information on the physical properties of the neutral and ionized interstellar medium (ISM) in star-forming galaxies. Such observations probe both the systemic interstellar component originating from H II regions, and the multi-phase outflowing component that is associated with star-formation feedback and can ultimately contribute to the circumgalactic and even intergalactic medium (CGM and IGM, respectively). In this dissertation, I investigate the physical properties of ionized gas in star-forming regions, and the kinematics and evolution of the multi-phase outflowing ISM/CGM in distant star-forming galaxies spanning the redshift range z ~ 1-4. This work consists of three studies that examine different aspects of the ISM and CGM, which collectively improve our understanding of the gas content in galaxies and the processes associated with the formation of massive stars near the peak of the star-formation-rate (SFR) density in the universe. I present a comparison of kinematics between the low- and high-ionization absorption features at z ~ 1, and demonstrate that the apparent larger blueshift of interstellar C IV relative to the low-ionization features is likely a result from the nature of resonant transitions instead of an evidence of the faster motion of the highly ionized gas. I further investigate the origin of the highly ionized gas by examining the correlations between the spectral properties of C IV and various galaxy properties. Both the blueshift and equivalent width (EW) of C IV are modulated by SFR and specific SFR, suggesting a direct connection between the highly ionized gas and the formation of massive stars. Nebular emission features provide valuable insights into the physical conditions of the ionized gas in H II regions, as well as the properties of young, massive stars. I show that the nebular C III] emission at z ~ 1 is much weaker compared to the detections from galaxies observed during the epoch of reionization (z > 6), and explore the factors that modulate the strength of this nebular feature. In combination with the results from photoionization models, I further infer the gas-phase metallicity and abundance pattern in the z ~ 1 star-forming galaxies based on the observed rest-frame C III] EW. Studying the lower-redshift analogs of the z > 6 C III] emitters is an alternative way to obtain more detailed information on the physical properties of these extreme-emission-line-galaxies (EELGs). By assembling a sample of EELGs at z ~ 1-2 and examining C III] and other nebular emission features, I aim to characterize the physical conditions of the z > 6 galaxies that are likely responsible for the cosmic reionization. Finally, with carefully constructed samples and uniform measurements, I investigate the evolution of the ISM/CGM at z ~ 2-4 as probed by rest-UV spectroscopy. I discover redshift-independent correlations among Lya emission, low-ionization interstellar absorption lines, and dust extinction. I further show that the covering fraction of neutral gas decreases with increasing redshift at multiple fixed galaxy properties. Gaining a full understanding of galaxy evolution requires further studies of the ISM/CGM in a systemic manner at higher redshifts. With exceptional capabilities in the near-IR and excellent spectroscopic sensitivity, the next generation of large telescopes will enable rest-UV and rest-optical spectroscopic studies of star-forming galaxies out to z > 10. Answering key questions regarding the interplay among massive stars, their contribution to the ionizing background, and feedback will deliver a clear picture of the formation and evolution of these distant star-forming galaxies.

Understanding the regulation and environment of star formation across cosmic time is critical to tracing the build-up of mass in the Universe and the interplay between the stars and gas that are the constituents of galaxies. Three studies are presented in this thesis, each examining a different aspect of star formation at a specific epoch. The first study presents the results of a photometric and spectroscopic survey of 321 Lyman break galaxies (LBGs) at z = 3 to investigate systematically the relationship between Lyα emission and stellar populations. Lyα equivalent widths were calculated from rest-frame UV spectroscopy and optical/near-infrared/Spitzer photometry was used in population synthesis modeling to derive the key properties of age, dust extinction, star formation rate (SFR), and stellar mass. We directly compare the stellar populations of LBGs with and without strong Lyα emission, where we designate the former group (Lyα equivalent widths greater than 20 Å) as Lyα-emitters (LAEs) and the latter group (Lyα equivalent widths fewer than 20 Å) as non-LAEs. This controlled method of comparing objects from the same UV luminosity distribution represents an improvement over previous studies in which the stellar populations of LBGs and narrowband-selected LAEs were contrasted, where the latter were often intrinsically fainter in broadband filters by an order of magnitude simply due to different selection criteria. Using a variety of statistical tests, we find that Lyα equivalent width and age, SFR, and dust extinction, respectively, are significantly correlated in the sense that objects with strong Lyα emission also tend to be older, lower in star formation rate, and less dusty than objects with weak Lyα emission, or the line in absorption. We accordingly conclude that, within the LBG sample, objects with strong Lyα emission represent a later stage of galaxy evolution in which supernovae-induced outflows have reduced the dust covering fraction. We also examined the hypothesis that the attenuation of Lyα photons is lower than that of the continuum, as proposed by some, but found no evidence to support this picture.

The second study focuses specifically on galactic-scale outflowing winds in 72 star-forming galaxies at z = 1 in the Extended Groth Strip. Galaxies were selected from the DEEP2 survey and follow-up LRIS spectroscopy was obtained covering SiII, CIV, FeII, MgII, and MgI lines in the rest-frame ultraviolet. Using GALEX, HST, and Spitzer imaging available for the Extended Groth Strip, we examine galaxies on a per-object basis in order to better understand both the prevalence of galactic outflows at z = 1 and the star-forming and structural properties of objects experiencing outflows. Gas velocities, measured from the centroids of FeII interstellar absorption lines, are found to span the interval -217, +155 km s^-1. We find that approximately 40% (10%) of the sample exhibits blueshifted FeII lines at the 1σ (3σ) level. We also measure maximal outflow velocities using the profiles of the FeII and MgII lines; we find that MgII frequently traces higher velocity gas than FeII. Using quantitative morphological parameters derived from the HST imaging, we find that mergers are not a prerequisite for driving outflows. More face-on galaxies also show stronger winds than highly inclined systems, consistent with the canonical picture of winds emanating perpendicular to galactic disks. In light of clumpy galaxy morphologies, we develop a new physically-motivated technique for estimating areas corresponding to star formation. We use these area measurements in tandem with GALEX-derived star-formation rates to calculate star-formation rate surface densities. At least 70% of the sample exceeds a star-formation rate surface density of 0.1 solar masses yr^-1 kpc^-2, the threshold necessary for driving an outflow in local starbursts. At the same time, the outflow detection fraction of only 40% in FeII absorption provides further evidence for an outflow geometry that is not spherically symmetric. We see a 3σ trend between outflow velocity and star-formation rate surface density, but no significant trend between outflow velocity and star-formation rate. Higher resolution data are needed in order to test the scaling relations between outflow velocity and both star-formation rate and star-formation rate surface density predicted by theory.

Galactic winds are further explored in the third study of this thesis, where we present a study at z = 1 of the prevalence and kinematics of ultraviolet emission lines from fine-structure FeII* transitions and resonance MgII transitions. Utilizing a multiwavelength dataset of 212 star-forming galaxies, we investigate how the strength and kinematics of FeII* and MgII emission lines vary as a function of galaxy properties. We find that FeII* emission is prevalent in the sample; composite spectra assembled on the basis of a variety of galaxy properties all show FeII* emission, particularly in the stronger 2396 and 2626 Å lines. This prevalence of emission is in contrast to observations of local galaxies; the lack of FeII* emission in the small star-forming regions targeted by spectroscopic observations at z = 0 may imply that FeII* emission arises in more extended galaxy halos. The strength of FeII* emission is most strongly modulated by star-formation rate, dust attenuation, and [OII] equivalent width, such that systems with lower star-formation rates, lower dust levels, and larger [OII] equivalent widths show stronger FeII* emission. MgII emission, while not observed in a spectral stack of all the data in our sample, is seen in 30% of individual objects. We find that objects showing MgII emission have preferentially larger [OII] equivalent widths, bluer U-B colors, and lower stellar masses than the sample as a whole. Active galactic nuclei are not likely responsible for the MgII emission in our sample, since we have excluded active galaxies from our dataset. We also do not observe the NeV emission line at 3425 Å characteristic of active galaxies in our co-added spectra. We find that the kinematics of FeII* emission lines are consistent with the systemic velocity. This result does not necessarily imply that these lines arise from star-forming regions, however, as an optically thin galactic wind could show blueshifted and redshifted FeII* emission lines centered around 0 km s^-1. We note that FeII* emission arising from extended gas is consistent with the hypothesis that slit losses are responsible for the lack of FeII* emission in local samples. We propose that dust is primarily responsible for the correlations between FeII* strength and galaxy properties, as objects with lower star-formation rates and larger [OII] equivalent widths also exhibit lower dust attenuations, on average. The strong MgII emission seen in systems with larger [OII] equivalent widths, bluer U-B colors, and lower stellar masses may also be the result of low dust attenuation in these objects. Larger studies composed of high signal-to-noise observations will be critical for testing the hypothesis that dust is the primary modulator of fine-structure and resonance emission.

One of the foremost goals in the study of cosmological reionization is understanding the nature of the sources of the ionizing photons. The search for leaking ionizing radiation from high-redshift star-forming galaxies has resulted in dozens of promising candidates, yet few confirmed detections. In this thesis, I present results from a survey for z~2.85 ionizing Lyman-Continuum (LyC) emission in the HS1549+1933 field and place constraints on the amount of ionizing radiation escaping from star-forming galaxies. Using a custom narrowband filter (NB3420) tuned to wavelengths just below the Lyman limit at z>=2.82, I probe the LyC spectral region of 49 Lyman break galaxies (LBGs) and 91 Lyman-alpha emitters (LAEs) spectroscopically confirmed at z>=2.82. I also present high-resolution, UVJH follow-up HST observations of 16 z~3 candidate LyC emitters identified with the NB3420 filter. With these follow-up data, I obtain high spatial-resolution photometric redshifts of all subarcsecond components of the high-redshift galaxies in order to eliminate foreground contamination and identify robust candidates for leaking LyC emission. I find only one object with a robust LyC detection that is not due to foreground contamination. A comparison with representative samples of LBGs indicates that the most exceptional aspect of the stellar population fit to this object is its young age (<50 Myr). I obtain a contamination-free estimate for the comoving specific ionizing emissivity at z=2.85, indicating (with large uncertainties) that star-forming galaxies provide roughly the same contribution as QSOs to the ionizing background at this redshift. The results of my thesis work show that foreground contamination prevents ground-based LyC studies from obtaining a full understanding of LyC emission from z~3 star-forming galaxies. Future progress in direct LyC searches is contingent upon the elimination foreground contaminants through high spatial-resolution observations, and upon acquisition of sufficiently deep LyC imaging to probe ionizing radiation in galaxies at the faint end of the luminosity function.

There is compelling evidence that the growth of supermassive black holes and the stellar populations of their host galaxies are intricately linked. At the same time, the exact relationship between the buildup of stellar mass and the growth of supermassive black holes is still not well understood. At z ~ 2, both star-formation and Active Galactic Nuclei (AGN) activity in the universe were at peak levels, and this epoch is ideal for exploring the coevolution of stars and supermassive black holes. This dissertation examines high-redshift galaxies, looking at the particular properties of star-forming galaxies, the stellar populations of AGNs and the relationship between black hole and star-formation activity at z ~ 2 - 3. I have used the magnification afforded by gravitational lensing of z ~ 2 star-forming galaxies to measure important physical properties of the stars and gas in these galaxies. Using near-IR spectroscopy, I have calculated the metallicities, ionization parameters, star-formation rates, and dynamical masses for two galaxies which help to explain the differences between local- and high-redshift star-forming galaxies. The third and fourth chapters of this work cover the analysis of a sample of z ~ 2 - 3 AGNs drawn from the UV-selected Lyman Break Galaxy (LBG) survey. I present a rest-frame UV composite spectrum for this sample of AGNs. This spectrum shows many emission and absorption features, such as HI Lyman-alpha, NV1240, NIV]1483,1486, CIV1548,1550, HeII1640, and CIII]1907,1909. Redshifted SiIV1394 absorption provides evidence for outflowing high-ionization gas in these objects at speeds of 10^3 km/s. Finally, by using optical, near-IR, and mid-IR photometry, which covers the rest-frame UV to near-IR portions of the galaxies spectral energy distributions, I have modeled the SEDs of the AGNs that comprise this sample. I have developed a unique dual-component modeling approach that allows me to correct for the presence of an AGN. Based on these results, I have explored the relationship in the host galaxy between AGN activity, maturity of the stellar population, and regulation of star formation.

The gas-phase metallicity of the interstellar medium is a powerful probe of the cycle of baryons into and out of galaxies. Constraining the scaling of metallicity with global galaxy properties such as stellar mass (M_*) and star-formation rate (SFR) at multiple epochs provides insight into galaxy growth across cosmic history and the origin of the present-day galaxy population. In this dissertation, I investigate the evolution of the physical conditions of ionized gas in star-forming regions, including metallicity, over the past 12 billion years of cosmic history. This work is contained in five studies that collectively improve our knowledge of galaxy metallicities over the redshift range z=0-3. I present measurements of the mass-metallicity relation at z~2.3 using a novel high-redshift data set from the MOSFIRE Deep Evolution Field (MOSDEF) survey. I further show that there is a relation among M_*, SFR, and metallicity for z~2.3 star-forming galaxies, unambiguously demonstrating the existence of this relation at z>1 for the first time. Knowledge of the physical conditions of line-emitting gas, including the electron density and ionization state, is required for robust estimates of metallicity from strong optical emission lines. I show that the electron density of star-forming regions increases by an order of magnitude from z~0 to z~2.3, and place constraints on the evolution of ionization state. Obtaining unbiased galaxy metallicity estimates additionally requires proper treatment of the various line-emitting sources falling within spectroscopic apertures. I characterize systematic metallicity biases from z~0 global galaxy spectra using a model framework that treats galaxies as ensembles of HII and diffuse ionized gas regions of varying metallicities. The resulting corrections increase the accuracy of the z~0 baseline for evolutionary studies. Finally, I present the first temperature-based metallicity determination at z>2 from a detection of the auroral emission line [OIII]4363. Measurements of auroral lines provide an independent estimate of metallicity that can be used to construct metallicity calibrations appropriate at high redshifts. Observational facilities coming online in the near-future will enable temperature-based metallicity measurements for large samples of high-redshift galaxies, providing unprecedented accuracy in metallicity measurements and a more complete understanding of gas flows and galaxy growth.

Understanding the nature and evolution of the high-redshift universe is crucial in forming complete models of galaxy evolution and large-scale structure formation. In this dissertation I investigate several aspects of the high-redshift universe, including the structure of a massive galaxy protocluster, understanding the nature and interplay between massive stars and ionized gas in high-redshift galaxies. First, I present an analysis of densely sampled spectroscopic observations of galaxies within the SSA22 protocluster at z~3.09 which reveal two distinct structures separated in redshift space and segregated on the sky. An analysis of similar structures within cosmological N-body simulations reveals that such a distribution of galaxies can only be explained as two nearby overdensities which will remain distinct as they evolve to z=0. Based on the N-body simulations, I find that the opportunity to observe such a phenomenon is incredibly rare, with an occurrence rate of 7.4 h^3 Gpc^−3. In this dissertation I also investigate the differences between local and high-redshift galaxies suggested by the offset towards higher [OIII]5007/Hb and [NII]6584/Ha on the `BPT' diagram. I analyze combined rest-UV and rest-optical spectra of high-redshift galaxies. Crucially, rest-UV spectra provide a powerful constraint on the population of massive stars within high-redshift galaxies, which is an important driver powering the observed rest-optical emission lines. To investigate this origin of the offset on the BPT diagram, I construct two composite spectra composed of high-redshift galaxies at different locations on the BPT diagram. This analysis demonstrates that high-redshift galaxies that are more offset typically have younger stellar ages, lower stellar metallicities, higher ionization parameters, and are more alpha-enhanced compared to high-redshift galaxies that lie along the local sequence. In addition, this analysis reveals that even galaxies that are entirely consistent with the local nebular excitation sequence appear to be alpha-enhanced. This suggests that a similarity in the location of high-redshift and local galaxies in the BPT diagram may not be indicative of a similarity in their physical properties. I further expand upon this analysis by fitting the joint rest-UV and rest-optical properties of individual galaxies in the sample. An important aspect of analyzing individual galaxy spectra is a quantitative limit on how well we can fit the spectra. By introducing noise to model galaxy spectra which has known properties, and binding its best-fit properties, I determine that galaxy properties can be accurately reproduced if the spectrum has a SNR>4. The best-fit properties of individual galaxies in our sample reveals that they have comparable ionization parameters to those of local HII regions the share the same nebular metallicity. In addition, I find that all galaxies in the sample show evidence for being alpha-enhanced resulting in harder ionizing spectra compared to local galaxies. These results point toward the observed offset on the BPT diagram being primarily caused by a harder ionizing spectra at fixed nebular metallicity.

The study of galaxies across multiple epochs is essential for understanding the evolution of the universe. One key time period to study is ? ∼ 2, when star formation activity in the universe is at its peak. Comparing local galaxies to those in this more active time period is a critical way to learn about galaxy evolution by examining the differences and/or similarities in galaxy properties. In this thesis, I study the rest-frame optical emission-line and host galaxy properties of star-forming galaxies at ? ∼ 2 in the MOSFIRE Deep Evolution Field (MOSDEF) survey to better understand the evolution of galaxies over the past 10 Gyr of our universe’s history.

First, I investigate correlations between the emission-line properties and the physical and chemical properties of ? ∼ 2 star-forming galaxies in the MOSDEF survey. It is necessary to understand the known offset of ? > 1 galaxies on the [O III]?5008/H? vs. [N II]?6585/H? ([N II] “BPT”) diagram compared to their local counterparts because strong rest-optical emission-lines are commonly used to infer a variety of galaxy properties (e.g. gas-phase oxygen abundance). To investigate the physical driver of this shift, I defined two populations of ? ∼ 2 MOSDEF galaxies on the [N II] BPT diagram, one on and one off (i.e., offset from) the local sequence. I find that these two groups remain separated on the [O III]?5008/H? vs. [S II]??6718,6733/H? ([S II] BPT) diagram and the [O III]??4960,5008/[O II]??3727,3730 vs. ([O III]??4960,5008+[O II]??3727,3730)/H? (O 32 vs. R 23 ) diagram, which suggests that star-forming regions in the more offset galaxies are characterized by harder ionizing spectra at fixed nebular oxygen abundance. Such a phenomenon may be tied to ?-enhancement and massive stars that are chemically “young.”

Second, I compare the ? ∼ 2 MOSDEF survey with the ? ∼ 2 portion of the Keck Baryonic Structure Survey (KBSS) that has been observed with MOSFIRE. Like MOSDEF, KBSS studies a large sample of star-forming galaxies at ? ∼ 2 with the MOSFIRE instrument; however, there are notable differences in survey construction and key results (e.g., the magnitude of the offset from the local star-forming sequence on the [N II] BPT diagram). Using consistent spectral-energy-distribution (SED) modeling for both surveys reveals that the MOSDEF ? ∼ 2 targeted sample has a higher median stellar mass, lower star-formation rate (SFR) and specific SFR, and redder U−V color compared to KBSS. However, the subsets of the surveys that have been analyzed in previous work with high S/N spectra and multiple emission lines detected are strikingly similar. Aside from stellar population age, all sample properties investigated agree within the median uncertainties. Additionally, applying uniform stellar Balmer absorption correction and emission-line fitting techniques for both samples results in the same offset on the [N II] BPT diagram. I find that the previously believed differences in key results between the two surveys can be attributed toutilizing different SED and emission-line fitting techniques.

Third, I analyze the completeness of the MOSDEF ? ∼ 2 survey. Specifically, I use SED modeling and composite spectra created from spectral stacking to test if the subset of the MOSDEF ? ∼ 2 star-forming galaxies with high S/N spectra is representative of the complete sample of star-forming galaxies. I find that the host galaxy and emission-line properties (most notably offset from the local SDSS sequence on the [N II] BPT diagram) are very similar, indicating that the smaller spectroscopic samples are representative of the full catalog of star-forming galaxies. Additionally, comparing galaxy properties obtained through SED modeling reveals that the ? ∼ 2 sample observed by MOSDEF is representative of all ? ∼ 2 galaxies that met the selection criteria for the MOSDEF survey. Taken together, these results reveal that the emission-line trends established using high S/N ? ∼ 2 detection samples of star-forming galaxies in MOSDEF studies to date are representative of the rest-optical-magnitude-limited star-forming galaxy population at ? ∼ 2.

Fourth, I focus on two actively merging galaxies in the MOSDEF survey at ? = 1.89. We model the SEDs of the merging galaxies to find that they are both massive with low SFRs and similar stellar population ages. Additionally, the star formation in both galaxies began and peaked within a few hundred Myr of each other, suggesting that their bursts of star formation may be connected. For one of these galaxies, GOODS-S 43114, S�rsic profile fitting and a relatively low velocity dispersion estimate indicates that it is a face-on disk; therefore, it will likely undergo a large structural change as it evolves into a massive, slowly-rotating elliptical galaxy in the present day.

Finally, as an earlier part of research I used far-IR ??? ??ℎ??/PACS spectra to investigate the profiles of six OH doublets for a large sample of 178 local galaxies. I assembled ancillary data to probe AGN luminosity, radiation field hardness, dust temperature, and dust obscuration, and find correlations between the EW(OH) and these galaxy observables. Additionally, I comment on how the origin of emission for these OH doublets, whether from radiative pumping by infrared photons or from collisional excitation, may influence these relationships.